EMPLOYMENT GROWTH NEXUS (JOBLESS GROWTH)

A Case Study of Pakistan

A thesis submitted for the partial fulfillment of the requirement for the degree of Doctor of Philosophy

BY AZAD HA IDER

SCHOOL OF ECONOMIC SCIENCES FEDERAL URDU UNIVERSITY OF ARTS, SCIENCE & TECHNOLOGY, ISLAMABAD 2010

CERTIFICATE

This is to certify that Mr. Azad Haider has completed his dissertation in partial fulfillment of the requirements for the degree of Doctor of Philosophy entitled, Employment Growth Nexus (Jobless Growth) A Case Study of Pakistan under our supervision. In our opinion, this dissertation fully meets the requirements, in scope and quality, for the degree of Doctor of Philosophy.

Senior Supervisor

Foreign Supervisor

_________________ Dr. Ather Hussain Akbari

___________________ Dr. Yigit Aydede

ii

ABSTRACT

The slow growth in employment during the present and past few economic recoveries around the world has generated interest among economic researchers in the phenomenon of jobless growth. In developing countries the potential of economic growth to generate employment has often been questioned. The present study is assesses this potential in Pakistan which has relied mainly on economic growth in the past to generate employment. Using a production function, an employment demand model is derived in which Gross Domestic Product (GDP) is the main explanatory variable. This model is estimated using annual time series data for the period 1974-2008, first at the national level and then by pooling data for the major sectors of the economy. Econometric estimates of the parameters of this model have been used to calculate a threshold level of GDP growth below which growth in employment is expected to be stagnant. The analysis of national data indicates that the economic growth in Pakistan has generally been slower than the level of growth required to generate employment growth. This result is mainly attributed to below threshold level of growth in the manufacturing sector of the economy which is regarded as one of the main engines of economic growth by economic planners in Pakistan. Among the potential factors behind the above results are labour productivity, sectoral shifts in the labour market, and the macroeconomic policies of the last three decades. Directions for future research and policy implications of results are also discussed in this study.

iii

Dedication
I dedicate this dissertation to my Parents and late brother, Farrukh Hayat, who provided me encouragement and moral support to pursue a doctoral degree.

iv

ACKNOWLDGMENTS
In the name of Allah, the most gracious, the most merciful, praises and thanks are first and foremost due to the Almighty Allah, the Lord of the Universe, who gave me the strength, patience, and the ability to complete this dissertation. It is difficult to overstate my gratitude to my senior supervisor, Dr. Syed Ather H. Akbari. With his enthusiasm and expertise in explaining the economics concepts clearly and simply, he helped to make the discipline of economics fun for me. Throughout my thesis-writing period, Dr. Akbari provided encouragement, sound advice, good teaching, good company, support and many good ideas. I would also like to thank Dr. Yigit Aydede, my foreign supervisor, for his valuable guidance especially on the econometrics analysis. My appreciations are also due to Dr. Syed Nawab Haider Naqvi, Dr. Rehana Siddiqui, Dr. Abdul Salam, Dr. Ather Maqsood, Dr. Waseem Shahid Malik and Saeed Ahmed Sheikh whose classes strengthened my interest in the field of economics and provided me the inspiration to conduct applied research. I also thank Dr. Mumtaz Hussain of the International Monetary Fund for his valuable guidance and encouragement. Words are inadequate to express my full and sincere gratitude to my parents and siblings. A special thought is devoted to my parents for a never-ending support especially to my mother for her understanding, endless patience and encouragement when it was most required. Finally, I also gratefully acknowledge The Higher Education Commission (HEC) of Pakistan for providing me financial assistance to pursue a doctoral degree in economics.

Azad Haider
v

Table of Contents
Certificate ............................................................................................................................... ii Abstract .. ............................................................................................................................iii Dedication ............................................................................................................................... iv Acknowledgements ................................................................................................................. v Table of Contents................................................................................................................... vi List of Tables ........................................................................................................................... x List of Figures ....................................................................................................................... xii

CHAPTER 1
Introduction ............................................................................................................................. 1 1.1 Importance for Pakistan: ................................................................................................ 3 1.2 A Brief Review of Pakistans Current Employment Policy: ....................................... 6 1.3 Research Questions: ..................................................................................................... 10 1.4 Method of Analysis and Data Used: ........................................................................... 12 1.5 Organization of the Study: ........................................................................................... 13

CHAPTER 2
Literature Review ................................................................................................................. 15 2.1 European and North American Studies: ..................................................................... 16 2.1.1 Regional Studies on Europe and North America:............................................... 16 2.1.2 Country Specific Studies in Europe and North America: .................................. 25 2.2 African and Asian Studies: .......................................................................................... 37 2.2.1 African Studies on Jobless Growth: .................................................................... 37 2.2.2 Asian Studies: ....................................................................................................... 42 2.3 Studies on the Relationship between Economic and Employment Growth in Pakistan: ...................................................................................................................... 48 2.4 Summary:...................................................................................................................... 53

vi

CHAPTER 3
Theoretical Background, Method of Analysis, and Economic Model.......................... 55 3.1 Theoretical Background of the Relationship between Employment and Output Growth: ........................................................................................................................ 55 3.2 Method of Analysis: ..................................................................................................... 60 3.3 Employment Demand Model: ..................................................................................... 61 3.4 Some Issues in the Estimation of Employment Demand Equation: ......................... 68

CHAPTER 4
Econometric Results of Employment Demand Model for Pakistan ............................. 71 4.1 The Employment Demand Equation:.......................................................................... 72 4.2 Sources of Data and Their Description:...................................................................... 75 4.3 Econometric Results of Employment Demand Model: ............................................. 77 4.3.1. Econometric Issues Relating to the Time Series Estimation of Employment Demand Model: .............................................................................................................. 78 Stationarity of Data .................................................................................................... 78 Autocorrelation .......................................................................................................... 80 Multicollinearity......................................................................................................... 80 4.4. Interpretation of Econometric Results: ...................................................................... 81 4.4.1 A Comparison of the Threshold Level of GDP Growth Rate with Actual Annual GDP Growth Rate: .................................................................................. 84 Calculation of Threshold Level of GDP Growth ..................................................... 84 4.5 Employment and Economic Growth Relationship during Periods of Slow and Fast Economic Growth: ...................................................................................................... 87 4.5.2 Threshold Level of GDP Growth in the Fast Economic Growth Period: ......... 90 4.6 Summary:...................................................................................................................... 91

vii

CHAPTER 5
Sectoral Analysis of Employment Demand in Pakistan ................................................. 92 5.1 Augmented Employment Demand Equation Showing Sectoral Effects: ................. 95 5.2 Econometric Issues Relating to Estimations Based on Pooled Data: ....................... 96 5.3 Results of the Unit Root Test for Stationarity of Pooled Data: ................................. 97 5.4 Econometric Results of the Augmented Employment Demand Equation: .............. 99 5.5 Summary:.................................................................................................................... 106

CHAPTER 6
Why Jobless Growth in Pakistan ..................................................................................... 108 6.1 Possible Factors Responsible for Jobless Growth: .................................................. 109 6.2 An Investigation of Factors Responsible for Jobless Growth in Pakistan: ............ 112 6.2.1 Labour Productivity Growth in Pakistan:.......................................................... 112 6.2.2 Structural Change in Pakistans Economy: ....................................................... 116 Groshen and Potters (GP) Measure ....................................................................... 117 Liliens Measure ...................................................................................................... 120 Rissman (1997) Measure ......................................................................................... 121 Aaronson, Rissman and Sullivan Measure ............................................................. 123 6.2.3 Monetary and Fiscal policies: ............................................................................ 126 6.2.4 International Trade and Globalization:.............................................................. 130 6.3 Summary:.................................................................................................................... 133

CHAPTER 7
Conclusions and Policy Implications ............................................................................... 135 7.1 Summary of Conclusions Regarding Presence of Jobless Growth in Pakistan: .... 135 7.2 Key Factors Responsible for Jobless Economic Growth in Pakistan: .................... 140 7.3 Some Policy Implications of this Study: .................................................................. 141 7.4 Limitations of the Study and Some Directions for Future Research: ..................... 144 viii

Appendix I ........................................................................................................................... 147 Appendix II .......................................................................................................................... 151 Appendix III ........................................................................................................................ 153 Appendix IV ........................................................................................................................ 157 Bibliography ........................................................................................................................ 158

ix

List of Tables
Table 1.1: Some Labour Market Indicators in Pakistan (1974-2008) ................................... 4 Table 2.1: Sector Wise Elasticities of Employment with Respect to GDP in Pakistan ..... 49 Table 2.2: National and Sectoral Elasticity of Employment with respect to GDP in Pakistan, 1970s, 1980s, and 1990s ...................................................................... 50 Table 3.1: Share of Capital Goods in Total Imports, Pakistan (1974-2008) ...................... 66 Table 4.1: Regression Results of Employment Demand Model ......................................... 77 Table 4.2: Unit Root Test Results (ADF Test) ..................................................................... 79 Table 4.3: Residuals Based Co-Integration Test Results (ADF) ......................................... 80 Table 4.4: Correlation Matrix (Employment is Dependent Variable) ................................ 81 Table 4.5: Periods of Economic Growth in Pakistan .......................................................... 88 Table 4.6: Econometric Results of Employment Demand Model Incorporating the Effects of Slow and Fast Economic Growth .................................................................... 89 Table 5.1: Sectoral Shares of GDP and Employment in Pakistan (1950-2008, %) ........... 93 Table 5.2: Sectoral Hours Worked Per Week in Pakistan (Averages, 1974 to -2008) ...... 94 Table 5.3: Unit Root Test for Employment, GDP and Total Number of Hours Worked Per Week ...................................................................................................................... 98 Table 5.4: Econometric Results of Employment Demand Equation: FGLS Estimates .. 100 Table 5.5: Residuals Based Co-Integration Test Results (ADF) ....................................... 101 Table 5.6: Sector Wise Elasticities of Employment with respect to GDP in Pakistan .... 102 Table 5.7: Threshold and Average Annual GDP Growth Rates in Non-Agricultural Sectors of Pakistan .............................................................................................. 104 Table 6.1: Employment to Population Ratio and Total Number of Hours Worked Per Week Per Person Employed in Pakistan ........................................................... 115 Table 6.2: Employment Growth Rates by Sectors in Peak and Trough of Economic Growth, and in the Years Following them. ....................................................... 118 Table 6.3: Comparison of Three Measures of Sectoral Allocation in Slow and Fast Economic Growth Periods ................................................................................. 125 Table 6.4: Econometric Results of Employment Demand Equation by Using OLS Method .............................................................................................................................. 128 Table 6.5: Residuals Based Co-Integration Test Results (ADF) ....................................... 129 Table A5.1: Sectoral Compositions of Pakistan Economy ................................................ 153

Table A5.2: Sectoral Distribution of Employment according to Rural and Urban Areas, Pakistan (1974 - 2008)..................................................................................... 154 Table A5.3: Sectoral Distribution of Informal Employment according to Rural and Urban Areas, Pakistan (2001 - 2008) ......................................................................... 155 Table A5.4 : Sectoral Employment and GDP Growth Rate of Pakistan (1974-2008) ..... 156 Table A6: Measures of Sectoral Change............................................................................. 157

xi

List of Figures
Figure 3.1: Phases of Business Cycle.................................................................................... 57 Figure 4.1: Growth Rates of Employment and Total Number of Hours Worked Per Week (1973-74 to 2007-08) ........................................................................................... 73 Figure 6.1: Pakistan Employment, GDP and Labour Productivity Growth Rates ........... 112 Figure 6.2: Lilien, Rissman and Aaronson et.al Measures of Sectoral Variations in Employment, Pakistan (1967-2008) ................................................................. 124 Figure A4: Estimates of Economic Growth Cycles in Pakistan Based on HP Method: .. 152

xii

CHAPTER 1 Introduction
The main purpose of this study is to assess the employment growth potential of economic growth in Pakistan. This purpose is achieved by estimating the elasticity of employment demand with respect to Gross Domestic Product (GDP), within the context of production theory, for the period 1974-2008. The analysis is conducted at national and sectoral levels of Pakistans economy. 1 Following an empirical verification of the strength of the relationship between employment and economic growth, the study also explores a set of factors that may determine the strength of this relationship in the context of Pakistans economy. Recent interest in the relationship between employment and economic growth began at the end of the global recession of 1991 when employment recoveries in many countries were found to be slower than the recoveries observed at the end of previous recessions. The phrase jobless growth was first introduced by the New York Times in the 1930s when at the end of the Great Depression, job growth was slower than the GDP growth.2 This term gained prominence after the 1991 recession in the global economy to describe the economic recovery that did not result in immediate job creation in most economies. Jobless growth is defined in two broad ways in this study. In the first definition, "it refers to a situation whereby the overall economy is growing, but the absolute

To be consistent with terminologies used in the Pakistani literature, sectoral level data include data on agriculture; industry defined as mining, manufacturing, construction and electricity, and services sectors. Appendix Table A5.1 provides a list of the components of each sector. 2 For reference, please see U.S. Heads for Third Straight Jobless Recovery. Morning Edition, National Public Radio. 16 Oct 2009 cited by http://en.wikipedia.org/wiki/Jobless recovery .

employment level is stagnant or falling, rendering near-zero or negative employment growth rates. Second, the term may be used to describe a situation whereby the overall economy is growing, while the rate of unemployment is rising. Each of these two definitions represents a very strict and broad interpretation of the term, with the latter not ruling out the possibility of an increase in the level of employment" (Altman 2003:12). The issue of jobless economic growth in developing countries is important because, economic growth in most of them is the main focus of economic policies to fight wide-spread poverty levels. However, efforts to reduce poverty are destined to fail unless jobs are created for the unemployed and poor. As Fields (2004) points out "the poor are poor because they earn little from the work they do. If growth does not produce enough jobs that are also high-productivity and high-paying, its purpose to foster development and alleviate poverty will eventually be defeated". Indeed, recent data on developing countries indicate a weakening of the relationship between overall employment and economic growth (Rada, 2008). Khan (2007) has identified several factors responsible for jobless growth in a developing country. These factors include the transition towards a market-based competitive environment that necessitates economic reforms, which in turn lead to worker layoffs and adoption of a cautious approach to new hiring; a structural shift in the economies towards less employment-intensive sectors, which may be the result of greater competitiveness of an economy; adoption of more capital-intensive production; and reduced availability of credits for small- and medium-sized enterprises, which tend to be more labour-intensive. In addition, labour productivity increases accompanied with economic growth, expansion of international trade, and wait unemployment which is also

referred as just-in-time employment practices by Schreft and Singh (2003), are cited as possible factors causing economic growth to be jobless.

1.1 Importance for Pakistan: An investigation of the relationship between employment and economic growth rates is important for Pakistan. As Table 1.1 shows, over the period 1974-2008, employment growth in Pakistan has been generally slower than the growth in its real GDP. On the whole, the unemployment rate matched economic growth rate over the period. Since the mid-1990s, the unemployment rate has been high. The rapid rate of economic growth in the current century has been accompanied by high unemployment rates. This indicates that economic growth in Pakistan has not been able to absorb new entrants into its labor force. 3 Even in the periods where employment growth rate came close to the GDP growth rate, the unemployment rate was 6 percent or higher. A crude measure of employment elasticity, response of employment growth to GDP growth, is also provided in the same Table by calculating the ratio of annual percentage changes in employment and GDP. An overall inelasticity of employment with respect to GDP growth rate is observed over the four decades.

Arif, Kiani and Sheikh (2002) also find that labour absorptive capacity of Pakistans economy has declined over time.

Table 1.1: Some Labour Market Indicators in Pakistan (1974-2008)

Employment Growth Rate Unemployment Rate GDP Growth Rate Productivity Employment Growth Elasticity Rate

Period
1974-80 1981-85 1986-90 1991-95 1996-00 2001-05 2006-08

3.30 2.23 2.73 0.94 3.25 2.28 4.17

2.65

2.70 3.80 3.20 5.00 6.00 8.00 6.40

5.20

5.56 6.78 5.80 4.65 3.27 5.23 5.84

5.29

2.25 4.55 3.07 3.71 0.02 2.95 1.67

2.64

0.59 0.33 0.47 0.20 0.99 0.44 0.71

0.50

1974-08

Source: Based on Economic and Labour Force Survey of Pakistan (various issues).

It is by and large notion that the key objective of any economic policy is, to enhance material wellbeing and the quality of peoples lives and this can be accomplished with an economic growth which will produce income generating employment opportunities (Loots, 1998). This does not seem to be the case in Pakistan. A systematic investigation of the employment and GDP growth relationship in Pakistan that will quantify this relationship can help shed light on the desirability of relying on economic growth as a tool for future employment growth in the country. This is essential because until recently, economic planning in Pakistan has been largely economic growth oriented (Pakistan 2007a). One desired outcome of this policy was employment expansion which in turn could help reduce income inequality and poverty. However, with the launching of Vision 2030 by the Government of Pakistan in August 2007, the policy focus has now changed towards an employment-intensive economic growth (Pakistan 2007c). A comprehensive strategy has been designed to create "A developed, industrialized, just and prosperous Pakistan through rapid and sustainable development in a resource constrained economy by developing knowledge inputs. A 4

major objective of this strategy is to place employment and employability of labour at the centre of economic and social policies" (Pakistan 2007a, p.6). By estimating the elasticity of employment with respect to economic growth within an economic framework, and by providing evidence on the presence or absence of jobless growth, the present study will shed light on whether a shift from economic growth led employment towards employment led economic growth can be supported by past data on Pakistans economy. Apart from employment expansion, employment policies in Pakistan also focus on enhancing worker productivity through education and training, and self-employment through microfinance programs (Pakistan, 2007a). Enhancement of worker productivity can weaken (alter) the relationship between employment and economic growth rates (Majid, 2000). In particular, if worker productivity is rising, one would expect employment levels to grow slower than national output thereby creating a jobless growth type of situation in the economy. If this is true, then an employment generation policy should emphasize expansion of industries where specific skills of newly trained unemployed workers are in demand. Table 1 data also showed that the productivity growth rate in Pakistan has generally varied directly with economic growth. However, no clear relationship appears to exist between productivity and unemployment rate. The present study will investigate productivity growth as a possible cause of slow response of employment to GDP growth. Whenever a country emerges from a recession, public expectations are raised that employment will rise. However, if the link between economic growth and employment is weak, economic planners in a democratic country must convince general public why employment does not grow as expected when the economy recovers. The present study

will provide useful information to policymakers whose job becomes more challenging during bad economic times. The following section provides an overview of the countrys current employment policy.

1.2 A Brief Review of Pakistans Current Employment Policy: Economic growth or development is a multifaceted phenomenon as it is an outcome of numerous social, economic and institutional factors such as capital, labour (quantity as well quality), technology, natural resources, entrepreneurship, and infrastructure. It is also influenced by the political environment and governance as well as the strength of public institutions and civil services, which in turn help determine the governments approach to fiscal, monetary, and trade issues. The focus of any research aimed at providing inputs to policymakers should be on decomposing the development experience of fast growing economies of the world into its critical sub-components and then using this experience to visualize the future track of economic development in relation to Pakistan. In 2005, the Government of Pakistan launched a Medium Term Development Framework, 2005-10 (MTDF) and Poverty Reduction Strategy (PRSP-I), which emphatically states that Pakistan must become a developed, industrialized, just and prosperous nation within one generation. 4 It must do so in a manner that sustains its development paradigm of a good quality of life and opportunity for all its citizens to reach their true potential. The MTDF also states that in spite of resource constraints,
4

"High pro-poor economic growth, social development, good governance and protection of the vulnerable are the four basic themes of the employment and poverty reduction strategy of the MTDF. It aims at achieving employment oriented pro-poor growth by accelerating productive economic activities. Expansion of employment opportunities in agriculture, SMEs, and housing and construction that provide employment to the poor segments of the society are the key instruments. Some targeted public support-programs, such as, Tameer-e-Pakistan and Khushal Pakistan aim at income and employment generation" (Pakistan 2007b, p.10). 6

Pakistan can reach the development levels, which it needs and deserves, by deploying knowledge inputs and human capital. Employment generation has now become a central focus of macro and sectoral policies and budgetary allocations in Pakistan (Box-1). The MTDF stresses on "creating a just and sustainable economic system for reducing poverty and achieving the Millennium Development Goals (MDGs) by the target year of 2015. It incorporates a paradigm shift towards enhancing competitiveness not only by means of higher investments but also through knowledge inputs to maximize total factor productivity (TFP)" (Pakistan 2007b, p.2).

Box-1 "Key Elements of the Global Employment Agenda (GEA) Reflected in Pakistans Employment Generation and Skills Development Strategy Making employment central in economic and social policy making, Emphasis on both the quantity and quality of jobs generated in the economy, Increasing employment intensity of growth and creation of decent employment through appropriate macro, trade, labour market and sectoral policies, Strengthen institutions to create an efficient and equitable labour market, Active involvement of employers and workers organizations and civil society in formulation of employment policy, Create favourable conditions for growth of the private sector especially SMEs and other labour absorbing sectors, Create a competitive and productive world class labour force, and Strengthen institutional machinery for employment policy making, HRD and monitoring labour market development".
Source: Sustainable Economic Growth, Employment and Human Resource Development, Ministry of Labour and Manpower (Policy Planning Cell) 2007, p.5

The MTDF aspires to attain an average growth rate of 8.0 per cent per annum in real GDP during the period leading up to 2030. By then, sector share of agriculture in GDP is expected to fall from 21.6 percent to 10 percent, that of manufacturing is expected to rise from 18.2 to 30 percent, while that of Services will remain stagnant at about 52 percent. In fact, the share of agriculture in GDP has fallen from 39 percent in 1969-70 to 20.9 percent in 2006-07. During the same period, the share of service sector increased from 38.4 percent to 53 percent. The manufacturing sectors share rose from 15.9 percent in 2001-02 to 19.1 per cent in 200607. To reach the quantitative targets of 30 percent GDP share by 2030, within an overall average GDP growth rate of 7 to 8 percent until 2030, the manufacturing sector needs to grow at an average rate of about 10 percent.5 Vision 2030 document does not emphasize economic growth as necessary for a higher level of economic development, but it recognizes that they both feed into and feed upon the social and infrastructure development. Thus, it is important that backward and forward linkages among the key variables of the macroeconomic framework are properly identified and adequately quantified so as to make the framework consistent, mutually enforcing, and a catalyst for overall economic activity. Finally, the Vision 2030 framework places special attention on development of human capital and its quality. As discussed above, employment generation has become a direct target of policymakers to achieve economic growth, while in the past employment generation took a back seat and was totally dependent on economic growth. The new approach towards employment generation under the employment agenda of the Vision 2030 is highlighted in Box 1. In light of this change in focus, the present study is

Pakistan Economic Survey, 2006-07.

important as it will provide some quantitative evidence on the desirability of such a change. Effectiveness of an employment policy is crucially linked with sufficiently available, reliable, and disaggregated data on the employed and unemployed. Realizing the importance of such information and regular monitoring of the developments in the labour market, the Federal Bureau of Statistics (FBS) has started conducting Labour Force Surveys (LFSs) on a quarterly basis. The quarterly reports mainly in the form of summary tables are now providing important insights on labour market indicators. These data can be used in ascertaining the outcomes of the targeted programs aimed at employment generation (Pakistan 2007b). However, it is too early to conduct such an analysis. To strengthen the system of data collection, the Government of Pakistan has launched a Labour Market Information System (LMIS) to collect labour market information in order to provide labour market diagnosis/monitoring and labour market intermediation in a participative and cost effective manner. "The Policy Planning Cell (PPC) of the Ministry of Labour, Manpower and Overseas Pakistanis (MOLMOP) is currently elaborating an institutional framework for the LMIS as a part of its employment policy currently under preparation; The MOLMOP has also initiated a project for better monitoring and analysis of the labour market" (Pakistan 2007b, p.9). The labour market of Pakistan is confronted with two main challenges. The first is the creation of employment opportunities to a level that is at least commensurate with absorbing fresh entrants into the labour market; their number increasing due to high rates of growth of the labour force. This has to be accompanied with the creation of conditions for decent work thus focusing on the quality of jobs and work opportunities that are 9

being generated in terms of income, productivity, better working conditions and respect for fundamental rights at work. The second relates to tackling the low absorptive capacity and declining employment elasticity of the economy posing indeed a serious challenge to policy makers (Pakistan 2007b). The focus of Vision 2030 on employment generation is an attempt to meet these challenges.

1.3 Research Questions: Our discussion in Sections 1.1 and 1.2 raises at least five important research questions that will be addressed in this thesis:

1. What has been the magnitude of the impact of GDP growth on employment growth in Pakistan? An answer to this question will shed light on the desirability of the recent change in policy focus from relying on economic growth to generate employment towards directly generating employment. 2. Is there a differential response of employment growth to GDP growth in Pakistan during periods of fast economic growth compared to slow growth periods. By answering this question, further insights can be gained on the sensitivity of employment growth to economic growth. 3. Was there a minimum threshold level of GDP growth that the country missed to achieve before the impact of GDP could be felt on employment? This question is related to the first question. It will help assess the extent to which past economic growth has really been jobless or job-generating. 4. Was there a sectoral imbalance in the impact of GDP growth on employment growth? Identification of sectors where employment response to GDP growth was

10

slower than the average can be important information for policy makers desiring to maintain a balance in employment generation across rural and urban regions based on economic growth. This information will also be useful as the manufacturing sector has been identified to have large multiplier effect on employment in the economy of a developing country (Mazumdar and Sarkar, 2007). Furthermore, over the past three decades, many developing countries including Pakistan have experienced a shift in employment towards the services sector. This shift has been observed in both rural and urban Pakistan (Appendix Table A5.2). Development economists are divided over the role of services sector as a sustainable source for future employment growth in developing countries due to their low per capita incomes and high income elasticity of demand for services. Hence, a current high potential of economic growth in services sector may not be viewed as sustainable. Thus, if employment is found responsive to economic growth nationally then this outcome will not be sustainable if it were derived solely by the service sector.6 A sectoral analysis is also important due to the varying component of informal economy in each sector.7 5. What were some major causes of jobless growth, if any, in Pakistan? If structural changes in the economy, or increases in labour productivity have caused economic growth to be jobless, then employment generation programmes aimed at facilitating worker re-training and establishment of industries which will hire these workers will help enhance employment opportunities.

6 7

Appendix Table A5.4 provides sectoral growth rates of GDP and employment. Sectoral components of the informal economy are provided in Appendix Table A5.3.

11

1.4 Method of Analysis and Data Used: Using an aggregate production function, an employment demand model that expresses employment as a function of Gross Domestic Product (GDP) will be derived. A Cobb-Douglas production function will be assumed. The model will be expressed in logarithmic form. Econometric estimates of the parameters of this model will be used to calculate a threshold level of GDP growth below which growth in employment is zero. A comparison of the actual GDP growth with the threshold level will be drawn to understand if past economic growths in Pakistan can be viewed as jobless or jobgenerating. The analysis will be performed first at the national level; and then separately for seven sectors of the economy. The period of analysis of this study is from 1974 to 2008.8 During this period, Pakistan has experienced: five short periods of democratic governments followed by longer periods of two military regimes, international oil crises, economic reforms, Afghanistan crisis, economic sanctions imposed on it due to its production of a nuclear bomb, and natural disasters, all of which may have affected the data used in this study. The data have been obtained through various published and unpublished sources of the Government of Pakistan. For some periods, these data were missing and had to be interpolated, using the linear interpolation method which is inbuilt in EViews 7.1 software.

A choice of this period avoids any data inconsistency resulting from the 1971 division of Pakistan.

12

1.5 Organization of the Study: The presentation of the study is organized as follows. In Chapter 2 a review of previous international studies that have examined the relationship between employment and economic growth is provided according to geographic regions. This review mainly focuses on those studies which investigate presence or absence of jobless growth. In Chapter 3, an empirical model of employment demand in Pakistan is derived using a production function approach. This model relates national employment with GDP. Its econometric estimation provides the elasticity of employment with respect to GDP which is a measure of the degree of employment response to a change in national output. Some econometric issues related to the estimation of the employment demand model are also identified in Chapter 3. The main focus of Chapter 4 is to analyze the relationship between employment and GDP at the national level. Emerging trends in the employment growth, GDP growth and hours worked per week in Pakistan are discussed in the same chapter. Estimates of employment elasticity are obtained based on econometric estimates of an expanded employment demand model. The model estimates are also used to estimate a threshold level of GDP growth below which employment growth is expected to be zero. The actual GDP growth rate is compared with this threshold level. This comparison is also performed in periods identified as periods of fast economic growth. Chapter 5 breaks down the national level analysis of Chapter 4 by dividing the economy into seven broad sectors. These sectors include: Agriculture, Manufacturing, Construction, Electricity, Transport, Trade & Finance and All Other Services sectors.9 A

Mining sector has a small (less than 1 percent) contribution towards GDP and employment and hence, the combined sector of mining and manufacturing is dominated by manufacturing industries.

13

sectoral analysis is necessary due to the presence of a large informal sector which employs about 70 percent of the working age population, because of which employment elasticity estimates based on national data may not be useful from an employment policy perspective. The chapter also estimates separate thresholds for seven different sectors of the economy. The analysis of Chapter 5 is based on pooled data comprising cross sectional and time series components. The use of pooled data for estimation gives rise to certain econometric issues which are also discussed in the same chapter. Chapter 6 discusses a set of various possible causes of this result in light of the relevant literature. Some of these causes are broadly investigated for Pakistan depending on the availability of data. Finally, Chapter 7 concludes the study and discusses some policy implications of econometric results of the study. Some shortcomings of the present study and directions for future research are also discussed in this Chapter.

14

CHAPTER 2 Literature Review

In developing countries, the necessity of economic growth to reduce poverty levels has been emphasized in literature and employment generation as a consequence of economic growth is a mean to achieve this goal. However, very few studies in development economics literature have evaluated the impact of economic growth on employment generation in developing countries. In developed countries, recent interest in the issue of jobless growth emerged among economists after the recession of early 1990s. Studies conducting empirical investigation of this issue started to emerge only towards the end of 1990s. Most of these studies focus on estimation of the relationship between employment growth and growth in Gross Domestic Product (GDP). Jobless growth is inferred in these studies on the basis of the magnitude of the GDP, or output, elasticity of employment demand. 10 Some studies have gone a bit further and have used the output elasticity of employment to compute the threshold level of economic growth, defined as the minimum level of economic growth required to have any positive growth in employment. Finally, some analysis has also been based on the growth accounting framework borrowed from the macroeconomics literature.

10

GDP elasticity of employment, employment elasticity, intensity of employment elasticity and employment- to-output elasticity have the same meaning.

15

2.1 European and North American Studies: The phenomenon of jobless growth has been extensively investigated for European countries and the United States. Some studies have been conducted at the regional level where data on a group of countries are analyzed while making references to some particular countries. Other studies are purely country specific.

2.1.1 Regional Studies on Europe and North America:

A positive relationship between employment and economic growth can be suggested on the basis of economic theory. Padalino and Vivarelli (1997) provide one such evidence for G-7 countries. 11 Their study approaches the issue from both theoretical and empirical perspectives. Econometric analysis of the relationships between real GDP, employment and working time is conducted using aggregate as well as manufacturing sector data for the period 1960 to 1994. The theoretical framework of their article is based on the French regulation theory which seeks to single out the characteristics of what the authors call the post Fordist era, in influencing the link between economic and employment growth. 12 The jobless growth hypothesis is contrasted with the more optimistic view which continues to maintain that economic growth entails employment growth, in light of different ways of looking at technological change. They show
11

G-7 countries include "Canada, France, Germany, Italy, Japan, the United Kingdom and the United States" http://www.actuaries.org.uk.
12

Post-Fordism is the name given to "the dominant system of economic production, consumption and associated Socio-economic phenomena, in most industrialized countries since the late 20th century. It is contrasted with Fordism, the system formulated in Henry Ford's automotive factories, in which workers work on a production line, performing specialized tasks repetitively. Definitions of the nature and scope of Post-Fordism vary considerably and are a matter of debate among scholars". "Post-Fordism is very much driven by information technology. Advancement in computer technologies allows for just-in-time manufacturing. There is no longer a need to stock-up on a given product. Products are made and then they are out the door. The key to production flexibility lies in the use of informational technologies in machines and operations". Retrieved September 2009, http://en.wikipedia.org/wiki/Post-Fordism and www.absoluteastronomy.com/topics/Fordism.

16

empirically that aggregate economic growth contributes to employment. To analyze the jobless growth phenomenon in G-7 countries, these authors estimated the output elasticities of employment for the entire economy and also for manufacturing sectors using data for the period 1960-1994. Separate estimations were made for the Fordist period before the two oil shocks (1960-73) and for the post-Fordist period (1980-94), in order to investigate any structural break during the 1960-94 periods. The findings suggested that as a whole, the North American economy experienced employment intensive growth. The economy of United Kingdom experienced jobless growth. France and Germany also faced jobless growth when total working time was taken into account. In the manufacturing sector, European countries faced a problem of jobless growth in terms of the number of employees as well as in terms of working time, while North America still experienced positive employment-intensive growth. A similar study was conducted by Pini (1996) for the G-7 countries (excluding Canada) and Sweden for the period 1960-95. This study focused on changes in the output elasticity of employment over time. France and Sweden experienced a decrease in output elasticity of employment during the post Fordist period (1979-95) while the reverse occurred in Germany and Japan. Italy, United Kingdom and the United States experienced similar aggregate elasticities of employment to output in the two periods. Negative elasticities were found for Italy and Sweden during 1990-95, meaning that economic growth led to job destruction instead of job creation. All countries experienced a decrease in elasticities, except for Japan, as compared with the 1980s. When data were disaggregated by sectors, all countries were found to have negative employment elasticities in manufacturing sector and positive in the services sector. 13

13

These sectoral results are consistent with the previous studies by the same author (Pini, 1995 and 1996).

17

In investigating the impact of output on employment, the above studies did not isolate the impact of changing labour productivity on employment. To disentangle this impact from the impact of output growth, Pianta et al. (1996) included the labour productivity variable in the employment demand model. This variable was calculated as the difference between GDP growth and employment growth. Their data included 36 manufacturing industries in the G-7 countries, excluding Canada, over the period 198092. A Value-added measure of GDP was used and it was found that growth in this variable had a significant effect on employment growth only in the United States and in Germany. The other four countries namely France, Italy, the United Kingdom and Japan, experienced a positive relationship between value added growth and employment but this relationship was found to be statistically insignificant. In an earlier study, Pianta (1995) had also found a positive relationship between employment and value added in Germany and Italy, but this relationship was also statistically insignificant in Italy. The study used data on 20 manufacturing industries based on the European Community Innovation Survey. 14 Two proxy variables for aggregate input productivity, including research and development intensity and the innovation expenditure, were also used in this model and were found to be negatively correlated with employment both in Germany and Italy. Hence, the aggregate input productivity improvement can be viewed as a source of jobless growth. Formulation of any economic growth policy aimed at promoting employment opportunities for the population must distinguish between the short- and long-run impacts of economic growth on employment. Dopke (2001) used the co-integration and error correction techniques to estimate the short- and long-run relationship between

14

A standardised cross sectional analysis carried out in 1992.

18

employment and economic growth. He used data on 18 selected OECD countries for the period 1970-1999. To test for a possible co-integration relation, he used two residual based statistics. The first one was obtained using an Augmented Dickey-Fuller (1979) test, abbreviated as ADF test, to test the null hypothesis of "no co-integration." The second one was based on the Kwiatkowski, Phillips, Schmidt, and Shin (1992) test, abbreviated as KPSS test. In this case, the null hypothesis was co-integration between employment and GDP growth. The ADF test could not confirm the presence of cointegration between these variables, while the KPSS test indicated the presence cointegration between these variables and confirm that a long run relationship exist between employment and economic growth. However, in some cases the influence of real GDP on employment was found to be unexpectedly low.15 One standard approach to measure the impact of various factors on economic growth is the growth accounting approach attributed to Solow (1957). Growth accounting decomposes the growth rate of GDP into two main components, one that is due to increases in the amounts of observable factors mainly capital and labour - and the other which cannot be accounted for by observable changes in factor utilization, also called total factor productivity (TFP) growth considered to represent technological progress. In a report published by the Centre for Economic Performance (2002), its head, C. Pissarides, used the growth accounting approach to investigate jobless growth, in Europe, United States and Japan over the period 1980-2000. He found that since 1960,

15

At the end, Dopke also estimated Okuns (1962) coefficient for each country in his sample and found it to be statistically significantly different across countries which led him to conclude that economic growth affects employment growth differentially across countries in the OECD. In economics, Okun's law, named after economist Arthur Okun who proposed the relationship in 1962 (Prachowny 1993), describes a relationship between the change in the rate of unemployment and the difference between actual and potential real GDP. In the literature, some economists make use of Okuns coefficient to discuss jobless growth. One implication of Okun's law is that "an increase in labour productivity together with an increase in the size of the labour force can mean that real net output grows without net unemployment rates falling" (the phenomenon of jobless growth).

19

capital accumulation accounted for about one third of GDP growth in the European countries and the United States. 16 However, in the United States, the employment growth accounted for 40 percent of GDP growth and TFP growth for the remainder while in the European Union, the employment growth contributed only seven percent of GDP growth leaving about 60 percent of GDP growth explainable by TFP growth. However, if employment growth is measured by hours of work, its contribution to growth becomes slightly negative, leaving about 70 percent of economic growth attributed to TFP growth. In Japan, capital accumulation contributed about 50 percent of the economic growth, while employment contributed 14 percent if measured by persons and 4 percent if measured by hours worked, leaving about 36 to 46 percent contribution of TFP growth. In summary, while the rate of GDP growth was about the same in Europe and in the US, growth in the US was much more employment-intensive than in Europe and in Japan, which is attributed by the author to a higher population growth rate. The dominance of TFP growth rate in European growth rate is viewed as technological catching up. The higher rate of TFP growth, which is labour-saving, and lower rate of capital growth worked against employment generation in Europe. Although TFP growth also played an important role in Japanese growth, capital accumulation which is an indicator of larger investment partially offset the joblessness of economic growth in Japan. Kwiatkowski, Roszkowska and Tokarski (2004) analyzed the sources and consequences of jobless growth in 48 European countries including some members of the

16

The European Union is composed of 27 sovereign member states: "Austria, Belgium, Bulgaria, Cyprus, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, and the United Kingdom". European Countries, Europa web portal. Retrieved 18 September 2009, http://europa.eu/abc/european_countries/index_en.htm.

20

Commonwealth Independent States (CIS) over the period 1990-2000.17 These authors found a strong link (relationship) between employment and economic growth in Slovenia, the Central European and then by post Soviet Baltic region countries, but a weak relationship in countries of former Yugoslavia and CIS. In their view, the weak link between GDP and employment was due to limited socio-economic reforms in Bulgaria and Romania and the regional instability in former Yugoslavia and Albania. They also concluded that the limits to jobless growth differ between countries they are the lowest in the EU countries, a bit higher in Eastern Europe and the highest in the Baltic Republics. Finally, no conclusive evidence on the link between GDP and employment growth rates was found for the CIS countries of Asia. On the other hand, Verme (2006) observed that the phenomenon of jobless growth was present in seven low-income CIS countries which experienced sustained output growth since 1997, but did not experience growth in employment. The joblessness of economic growth in these countries could be the result of a capital-intensive growth, large potential for productivity gains among existing workers and the dual labour market framework. In these low income countries, in the periods of recession and recovery affect the agriculture and manufacturing sectors in different ways. The agriculture sector infrastructure is largely depleted, underinvested and disconnected from manufacturing sector, because of which this sector is unable to grow beyond subsistence. On the other hand, manufacturing sector only focus on capital intensive techniques and highly productive means of production and provides good wages for the few highly skilled workers. With governments and international organizations currently lacking focus on

17

CIS countries include: "Armenia, Azerbaijan, Belarus, Georgia, Kazakhstan, Kyrgyzstan, Moldova, Russia (former Soviet Union), Turkmenistan, Tajikistan, Ukraine, and Uzbekistan. Some of these countries are in Asia: Kazakhstan, Kyrgyzstan, Soviet Union, Tajikistan, Turkmenistan, and Uzbekistan". CIS Countries, Retrieved 18 September 2009. http://en.wikipedia.org/wiki/CIS_countries.

21

agriculture and manufacturing sector investments, jobless growth is likely to persist in the low-income CIS countries. Sengenberger (2006) investigated the phenomenon of jobless growth in SouthEast European (SEE) countries. 18 Employment was fairly sensitive, or elastic, to GDP growth in some countries of the SEE region. Elasticity of employment to output was in the normal range for Albania, Bulgaria, Croatia, Greece, and Moldova, at least during part of the transition period from the command to market economic system. During the 1994 economic crisis, GNP in Turkey fell by 6.1 percent while employment continued to grow (World Bank 2006b, p.25), thereby giving a negative employment elasticity. In 2001, Turkish GNP fell by 9.5 percent, with employment shrinking by 3.7 percent until 2003. In the years following the 2001 economic crisis, GDP expanded by 25 percent in real terms, but unemployment could not be reduced below 10 percent. Apparently,

agriculture acted as an employment safety net. Jobless growth continued in 2003 and 2004, both at national and sectoral levels. Employment decline was substantial in Macedonia, the only country in the region identified as having low GDP growth after 1999, as a result of which negative employment elasticity was found in that country too. No employment elasticity values are available for Bosnia Herzegovina, Serbia and Montenegrin. However, available data in these countries also show employment decline despite economic growth. Studies on Central and Eastern Europe (CEE) found the lowest and declining employment intensity of GDP growth (GDP elasticity of employment) in the world after
18

For the purposes of this paper "SEE comprises the twelve nation states, they include Albania, Bosnia and Herzegovina, Bulgaria, Croatia, Greece, Romania, Serbia, Moldova, Montenegro, the Former Yugoslavia, Republic of Macedonia, Slovenia, and Turkey. Montenegro declared its independence from Serbia on 2 June, 2006. As most of the statistical data presented in this paper relates to the period prior to the separation, Serbia and Montenegro are treated as one state in part of the analysis. Where this is the case, reference will be made to the SEE-11". Retrieved 18 September 2009, http://www.ilo.int/public/english/region/eurpro/ankara/events/istanbul/ist-background.

22

1991 (Rutkoswki et.al 2005; Cazes and Nesporova 2006).19 During 1991-95, elasticity of employment demand with respect to GDP was 0.24, which fell to 0.01 during 1995-99 period, and then to -0.19 during 1999-2003. A United Nations study concluded no evidence of a relationship between output growth and changes in unemployment in 20032004 in the CEE countries (United Nations Economic Commission for Europe, 2005b, p. 33). World Bank economists attributed the joblessness of economic growth in CEE and SEE to overprotected, rigid labour markets and excessive social benefits (Rutkowski and Scarpetta, 2005). However, this explanation of jobless growth is refuted by the lack of employment improvement in these regions experienced after protection standards were scaled down and new flexible forms of labour laws were introduced. A cross-country study conducted by Kapsos (2005) also arrived at a similar conclusion. The above studies were based on aggregate data on each economy. Onaran (2007) estimated a labour demand equation based on the pooled data of manufacturing sector in selected countries of Central and Eastern European (CEE) region.20 He investigated the effects of domestic factors, such as wages and output, and international factors, such as exports, imports, and FDI, on employment during posttransition recovery. Jobless growth was prevalent in manufacturing sector, irrespective of wage developments, in most countries. Although the output elasticity of labour demand was mostly positive, its low value indicated a weak relationship between employment and

19

CEE includes the following former socialist countries, which extend east from the border of Germany and south from the Baltic Sea to the border with Greece: "Baltic states, Estonia, Latvia, Lithuania, Poland, CzechRepublic, Slovakia, Hungary, Romania, Bulgaria, Albania ,states of former Yugoslavia: Slovenia, Croatia,BosniaHerzegovina, Serbia, Kosovo(disputedindependence), Montenegro, Macedonia". Retrieved 18 September 2009, http://danpritchard.com/wiki/Central_and_Eastern_Europe.
20

These CEE countries included the Czech Republic, Hungary, Poland, Slovakia, Slovenia, Lithuania, Bulgaria, and Romania.

23

GDP, especially in the medium and low skilled sectors of the Czech Republic, Bulgaria, and Romania. The process of transition resulted in downsizing of firms and productivity enhancement which in turn resulted in significant employment losses in the new millennium, despite the achievement of economic growth in these countries. The transition economies of Europe also present an interesting case for investigating the phenomenon of jobless growth. Rutkowski et al. (2005) analyzed the links between employment and economic growth in nine transition economies of Europe during the period 1992 to 2002. They concluded that these countries were developing symptoms of a jobless growth. They also investigated whether jobless growth in those Central and Eastern Europe countries could be accounted for by a particular set of macroeconomic conditions. To study the response of employment to economic growth, they identified two main economic phases. The first phase was termed the transitional recession experienced during the period 1990-97. The second phase started with the 199899 financial crisis in Russia. A rise in productivity was observed during both phases, although output had begun to rise in the second phase. The productivity increase was accompanied by a drop in employment, even in the second phase, in most countries. The authors of this paper held changes in monetary and fiscal policies to be responsible for jobless growth. Their econometric analysis indicated that the poor employment

performance since the 1998 recession was perhaps due to the worsening of the fiscal stance, as indicated by the significant effect of the primary budget surplus variable. The employment elasticity of output did not change significantly in the post 1997 period, but the partial effect of output rise on employment was offset by the worsening of the fiscal stance.21

21

Their regression included a slope dummy for the post-1997 period interacted with GDP. That term should capture any change in the employment-output relationship that occurred during the second phase.

24

2.1.2 Country Specific Studies in Europe and North America:

During the early 1990s, a number of countries belonging to the Organization for Economic Cooperation and Development (OECD) experienced severe recession. 22 In most of them, economic recovery was associated with a sluggish job growth (see, for example, OECD, 1994, 53-55).23 This caused many economists to investigate joblessness in recoveries in individual countries. One member country of the OECD for which the phenomenon of jobless growth has been studied extensively is Finland whose economy was hit the most in 100 years by the recession of 1990s. Real GDP in Finland fell by almost 15 per cent during 1990s recession. Unemployment rose to record levels, and the banking system was hit by a serious crisis. During 1993-98, with growing GDP as compare to European Union (EU) the Finnish economy showed signs of recovery. However, the rate of unemployment remained above the EU average during this period. Productivity growth in Finland is described as "excessive" both during the depression years from 1992 to 1993 when GDP decreased, and in 1994 which was viewed as the first year of recovery. However, the period of "excessive" productivity growth lasted only three years. Hence, a statistical test to confirm any jobless growth during this period was necessary. This test was provided by Sauramo (1999a) who used quarterly data starting from the first quarter of 1975 and ending in the second quarter of 1998. He estimated a simple structural Vector

22

There are 34 members of OECD most of which are high-income economies with a high Human Development Index and are regarded as developed countries such as, "Australia, Austria, Belgium, Canada ,The Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, Korea, Luxembourg, Mexico, Netherlands, New Zealand, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Turkey , United Kingdom and United States". Retrieved 18 September 2009, http://www.oecd.org/countrieslist.
23

In OECDs Jobs study (OECD, 1994, p.55) the experiences from different countries are summarized as follows: "On balance, while the recovery in employment has been slower in some countries than in the past, this would appear to reflect an initially weaker rebound in output rather than jobless growth as such".

25

Autoregressive (VAR) model to investigate whether economic growth in that country can be characterized as jobless and found no statistical evidence in this regard for the entire period. However, there was some evidence of jobless growth during 1992-94 which could be attributed to rapid growth in labour productivity during these years. After this period, labour productivity in Finland slowed down. While the above study was based on aggregate data, Sauramo (1999b) investigated the presence of jobless growth using disaggregated quarterly data on the manufacturing sector in Finland over the period 1976-1998. This analysis was also based on the estimation of simple structural VAR models, but this time the author also assessed the role of technological and structural changes in affecting the demand for labour in Finland. The VAR model is appropriate for this analysis. Improvements in labour productivity that occurred in the manufacturing sector, especially in the metal industry, and improvements in service sector technology - largely innovation- appeared to have caused growth in these two sectors to be jobless. In the manufacturing sector, company restructuring and labour re-allocations contributed to productivity growth which may have resulted in jobless growth in that sector. Regional disparity between employment and output relationship in Finland has also been of interest in the literature. Kangasharju & Pehkonen (2001) linked this disparity to variations in industrial structures. They used a Generalized Method of Moments (GMM) technique24. The highest employment decline occurred in regions where agriculture and public sector industries dominated, while the least decline occurred where the private sector was dominant. Private services played a major role in

24

Generalized Method of Moments (GMM) estimator by Arellano and Bond (1991), where the lagged values of the variables (in levels) are used as instruments to remove the correlation between the regressors and the error term.

26

determining the relationship between changes in employment and output in Capital and Eastern-Northern region and in the Middle Finland region, contributing about 50 percent and 66 percent of the total growth, respectively. The authors also found that growth centres (such as Helsinki, Turku, Tampere, Oulu, and Jyvaskyla) have an impact on employment in their neighbouring regions. The second main contributor, in the case of Middle Finland was industry which contributed 21 percent of the total. 25 For the EasternNorthern and the Capital region, public services were the second main contributor about 34 percent of the total. Bockerman (1998) used data on labour districts for the period 1988-1995 and found a strong employment and output relationship in southern Finland, where economic activity is largely concentrated, while the link disappeared in some districts of the eastern, northern, and western parts of the country. He also found that various exogenous factors, such as the migration process and the quality of jobs, played a more important role in employment fluctuations in the northern part of the country than elsewhere. Another European country for which the employment and economic growth relationship has also been studied extensively is Poland where unemployment was the most important issue faced by economists after economic transition of the early 1990s. Kwiatkowski, Kucharski and Tokarski (2002) aimed at establishing the possible links between GDP and unemployment between 1993 and 2001. They used three main indicators: the number of employed; the number of unemployed; and the flow between employment, unemployment and productivity. Their analysis suggests that in Poland, the unemployment rates dropped but the employment growth rate was slower than the GDP growth rate during 1995-98, which was a period of fast economic growth. During periods

25

In European countries the terms industry and manufacturing are used interchangeably.

27

when employment grew, the GDP growth rate was at least 4.8 percent. These authors also predicted that by 2010, a GDP growth rate of at least 6 percent would be required to create jobs in the Polish economy. They also emphasized the importance of macroeconomic policies to stimulate short-term growth in employment. Kwiatkowski and Kwiatkowska (2006) concluded that a faster productivity growth in Poland, that increased its global competitiveness, also increased its potential for experiencing a jobless growth over the period 19912004. During the period, the countrys GDP growth was 4.3 percent, labour productivity growth was also 4.3 percent and employment growth was zero percent. The growth of labour productivity was much higher in Poland than in the remaining EU countries. These authors estimated the threshold GDP growth rate in Poland to be around 4.3 percent which was much higher than the 1 to 2 percent rate estimated for other EU countries. The above study also provided some evidence of jobless growth on fifteen EU countries. In some of them, labour productivity growth rates were higher than GDP growth rates which led to employment declines (Latvia, Estonia, Lithuania, Slovakia, Slovenia, the Czech Republic, Sweden, and Finland). Wolnicki, Kwiatkowski and Piasecki (2006) also attribute the jobless economic growth in Poland mainly to labour productivity growth during the late 1990s and early 2000s. Using the interpretation of Harrod-Domar (1946) as proposed by Barro and Sala-iMartin (1992, p.22) these authors provide an explanation of jobless growth. "The longrun employment growth rate is determined by the difference between the GDP growth rate and productivity growth rate. A positive GDP growth rate is possible when there is no growth in employment, or even where there is a negative rate of employment growth.

28

Hence, joblessness is entirely possible when productivity grows at faster rate than that determined by GDP (guaranteed) rate of growth". The econometric analysis in the above study was based on GDP and job growth data obtained for the period 1990-2002 for Poland and fifteen European Union (EU-15) nations.26 The period of early to mid-1990s in Poland can be viewed as a period of transition from a command to a market economy. Based on an earlier study, Kwiatkowski, Roszkowska and Tokarski (2004), they calculate a threshold level of GDP growth defined as the rate below which employment growth rate is zero. At first an employment growth rate equation is derived from an employment demand equation in which log of employment is dependent on log of GDP and time trend. 27 A threshold level of GDP growth rate is then calculated based on the econometric estimates of employment growth equation in which employment growth rate is set equal to zero. Poland was found to have the highest threshold rate of GDP growth at about 4.38 percent, while in the EU-15, used for comparison, it ranged between 1-2 percent.28 During 2002-03, the unemployment rate in Poland was nearly three times higher than the average unemployment rate in the EU-15 countries which could be directly attributed to the drop in the overall demand for labour resulting from a significant rise of labour productivity which in turn led to higher rates of GDP growth. Labour productivity during this period rose by 60 percent in Poland, while it rose by only 20 percent in EU-15. In sum, a rise in labour productivity in Poland was largely responsible for its jobless growth.

26

Poland is included in the European Union which is composed of 27 sovereign member states. The authors compared Poland with other EU-15 countries namely Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom
27 28

The coefficient of time trend variable is interpreted as the drop in labour demand when GDP is zero. The threshold rate of GDP growth is the one below which employment growth is zero.

29

The phenomenon of jobless economic growth has also been studied in Sweden whose economy has a large public sector. Economic growth in that country during the early 2000s did not provide any boost to employment growth thus causing its politicians to term their countrys economic growth experience to be jobless. Lund (2006) confirmed jobless growth in both private and public sectors of Sweden during 1976-2003. Public sector had been experiencing jobless growth since early 1990s. The private sector was divided into two sectors; industry and service. As the level of real GDP grew, Sweden experienced a decreasing proportion of service sector unemployment in overall national unemployment. This means that jobless growth did not occur in services sector. On the other hand, jobless growth was experienced by the manufacturing sector since late 1970s, which indicates that jobless growth is not a new phenomenon. She attributed jobless growth during the period 1987-1993 to technological advances. These advances led to enhanced effectiveness and efficiency of capital resulting in a substitution of workers with machinery. Another study by Swane and Vistrand (2006) examined the jobless growth phenomenon for Sweden and some selected OECD countries over the period 1980-2004, which they divided into two sub-periods: 1980-1993 and 1994-2004.29 This study based its analysis on the Okuns law. GDP elasticity of employment was about 0.7 percent for Sweden which remained stable over time. For other OECD countries, the employment elasticity ranged from 0.2 to 0.8. The employment and GDP move together in the long run and their stable relationship over time suggested that the joblessness of economic growth in Sweden identified in the previous study was a temporary situation. The
29

There are 34 members of OECD and the author compares Sweden with 21 selected OECD countries namely Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Iceland, Ireland, Japan, Italy, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Spain , Switzerland, United Kingdom and United States. The author provides no reason for separating the 1980-1993 periods from the rest.

30

following country specific studies are all from the United States. Although the term jobless growth was used in the public circles of the United States since the early 1990s, most studies on the United States are found in literature since 2001. Schreft and Singh (2003) identified two jobless recoveries in the United States economy by examining employment growth in the first year after the recession of 1991 and after the recession of 2001. Employment growth was only 0.14 percent after the first recession, but it was 2.7 percent after the second recession. One possible explanation for the failure of employment growth to resume in the first year of the recovery period was attributed to unusually weak economic growth especially immediately after the 1991 recession. The analysis was based on the Okuns law (1962). Groshen and Potter (2003) found that the economic recovery after 2001 recession was reflected in stable growth in output but no corresponding rise in employment. A review of sectoral reallocation in manufacturing sector during 2001-03 suggested that structural change in the economy caused a permanent reallocation of workers from some industries to others which led to slow response of aggregate employment to growth. The task of finding new jobs in other firms or industries is generally difficult and timeconsuming as workers need time to learn new skills. This situation was further complicated by the weakening of financial markets. In such a situation, firms may postpone new job creation due to the risks involved in growing their businesses or undertaking new ventures. The authors found some support for this interpretation in the shortfall to low job creation than to widespread job elimination. Thus, improved financing options for riskier ventures and a resolution of uncertainties about economic conditions are the two main ingredients which require for a return to job growth.

31

Schweitzer (2003) pointed out that the U.S economic expansion of 1990s began with such surprisingly slow employment growth that economist called it the jobless recovery. A review of the trends in employment, unemployment, and labour force participation rates all indicated jobless recovery. The unusually low level of labour force participation led the author to term the recovery as less jobseeker. He suggested that not all recoveries were the same, so economic policies needed to adapt to the situation at hand rather than following a predetermined recovery path. Bailey and Lawrence (2004) noted that the economic recovery in the United States which began in the third quarter of 2001 was more jobless than 1991 recovery. On the basis of their econometric model estimates, they predicted that real GDP, real compensation of employees and real profits in the US will be higher in 2015 largely due to lower prices of imports associated with off-shoring. They held that the decline in US manufacturing employment had occurred mainly because the manufacturing sectors productivity grew faster than its GDP growth. In addition, off-shore activities and the rise of trade were also responsible for jobless growth in the US manufacturing sector. Schreft and Singh (2003), reviewed earlier, found that job creation in the US took sometime after the 1991 recovery began. Bailey and Lawrence suggested that the waiting period in case 2001 recovery may be even longer. 30 They viewed international competition to be the main cause. Andolfatto and MacDonald (2004) noted that after the last two recessions of 1991 and 2001 in the United States, employment growth lagged the GDP recovery by several quarters. According to them, a recession followed by a jobless recovery is what neoclassical theory also predicts. This is because new technology impacts different

30

This prediction is confirmed by latter studies some of which are reviewed below.

32

sectors of the economy differentially, and the resulting sectoral adjustments in the labour market take time to have an effect. Their explanation for the jobless recovery include the varying size and scope of technological innovation, delayed effect of adopting new technology and a slow reallocation of labour across different firms. 31 The authors explored the properties of a Real Business Cycles (RBC) model in which growth is driven by technological advances that improve factor productivity. They do not claim, however, that all technological advances lead to jobless recoveries. Glosser and Golden (2004) consider the substitution between employment and weekly hours of work in the US manufacturing sector as a possible source of jobless recovery. The multiequation vector auto regressions methodology was used to get the impulse response functions of both employment and weekly hours. The results revealed a marked change in labour input adjustments since 1979, Weekly hours worked by workers rose somewhat, while employment dropped considerably which suggests employers potential reactions to a skill-upgrading of jobs under greater market pressures in order to control cost. For policy purposes, the results of the above study imply that active labour market policies would likely facilitate a more robust job creation instead of longer work hours following growth in output. Aaronson, Rissman, and Sullivan (2004a) noted that job recoveries after recession slowed down since 1990s. During each of the five recessions of the 1960s, 1970s, and 1980s, non from employment took less than four month to recover and exceed its previous level of the recession and in the recoveries it took on average 26 month to recover. However, after the recession of early 1990s, employment growth was much

31

In a previous study, Greenwood, MacDonald, and Zhang (1996) had developed a real business cycle model to include the effect of varying size of technological innovation and labour re-allocation and found that a jobless recovery can be the result of a technology-driven recession. Their explanation hinged critically on the above three assumptions.

33

weaker. It took 14 months for the number of jobs to return to the level reached just before the trough and an additional nine months before it exceeded the previous expansions peak. Job growth was even more disappointing after the recession of 2001. After 26 months into that recovery, nonfarm payrolls were still 0.5 percent lower than those of November 2001, the day the National Bureau of Economic Research (NBER) declared that the recession had ended. Many analysts (Lilien, 1981; Abraham and Kurtz, 1986; Rissman, 1997, etc) attributed this surprisingly weak employment performance to structural changes in the US economy. According to them, "an accelerating pace of structural change greatly increased the number of workers forced by job loss to make major career transitions. Because securing a new job in a different economic sector is a time consuming process, temporarily higher unemployment and a restrained employment growth was observed. Of course, in the long run, reallocating workers to their most productive use greatly benefits the economy. However, in the short run, reallocation is costly". Aaronson, Rissman, and Sullivan (2004b) do not support the view that the need to reallocate labour across industrial sectors has been particularly great during the last two recessions (1991, 2001) or the jobless recoveries that followed. These authors emphasize the need to focus future analysis on changes in occupational mix of employment rather than the industrial mix because, according to their findings, industrial restructuring of employment declined during the two recoveries. Gomme (2005) also explained the two jobless recoveries of early 1990s and 2000s with the probabilities of job finding and job separation. "Flows into employment are called job finding; flows out of employment are referred to as job separations. The probabilities of job finding and job separation can be calculated from monthly data on the

34

levels of employment and unemployment". The author analyzed the cyclical behavior of these probabilities in the postwar U.S. economy. During the recession of early 1990s, both probabilities showed the same behaviour, while both probabilities showed the different behaviour in the 2001 recession. The probability of finding a job was similar to that experienced in the early 1990s recession. However, the probability of job -separation behaved quite differently. Thus, focusing on changes in the job-separation probability either the firing decisions of firms or the decisions of workers to quit their jobs - is unlikely to provide a satisfactory explanation of the jobless recovery of 2001. Although the job separation probability followed the pattern of previous business cycles during 199091, the job-separation probability was unusually low during 2001, a fact that by itself would tend to maintain the level of employment. Thus, explanations of the jobless recoveries based on the job-finding probability are more likely to be on target. Hemraj, Madrick and Semmler (2006) interpret the recent phenomenon of jobless growth in the US in terms of Okuns theory. They measured output by GD P volume at constant prices and used quarterly data from 1961-2001. They estimated the Okuns coefficient for U.S (0.37), Germany (0.22), France (0.17) and for U.K (0.31). They demonstrated that a declining response of job growth stems from a decline in the Okun coefficient. They also show that, in Germany and France, this coefficient had not been falling but rising. The previous higher response of job growth to economic growth in the US has thus been reversed. All studies that have been reviewed above ignored the fact that employment growth may also be affected by the output gap, i.e., the difference between actual and potential GDP which is an indicator of macroeconomic disequilibrium. Seyfried (2004) considered this possibility for the US by examining the relationship between employment

35

growth and output gap. In addition, he also analyzed the relationship between employment and economic growth. State level data were used for the period 1990-2003. Seemingly Unrelated Regression (SUR) technique was used to estimate state-specific employment demand models while fixed effects technique was used for the pooled regression. His results indicated that although economic growth had some immediate impact on employment, its effects continued for several quarters in most of the states considered. He found that nine of the ten states showed a positive and significant relationship between employment growth and GDP gap, with the exception of California which was attributed in part due to its fiscal crisis. State level employment elasticities with respect to the GDP gap ranged between 0.20 and 0.56 and with respect to the GDP growth ranged between 0.31 and 0.61. The national level elasticities, based on the pooled regression, were found to be 0.39 and 0.47, respectively. In short, this study showed low response of employment to GDP growth in ten of the United States. It is concluded from most of the US studies that the last two recoveries, of 1991 and 2001, were different from the ones observed in the past due to a weak post-recession (in the period of recovery) employment performance that led economic commentators to coin the term jobless recovery (see Schreft and Singh, 2003; Groshen and Potter, 2003; Gomme, 2005). While there is a continued debate in the United States as to whether the two jobless recoveries of the early 1990s and the early 2000s should be attributed to accelerated structural changes in the labour market, enhanced productivity growth, economic and political uncertainty, or institutional factors, it is increasingly recognized that the strong job growth that was once a typical feature of recoveries may be a thing of the past.

36

2.2 African and Asian Studies: One common limitation of studies which focus on the less advanced countries of Africa and Asia is data reliability. This problem is further complicated by the presence of a large informal sector whose employment changes are not captured in official statistics. For example, Sinha and Adam (2004) recognized that there could be a rise of employment in India that is not captured by official figures due to the presence of a large informal sector. 32 With increased competition in the formal sector that has led to severe cost-cutting measures including a reduction in employment, the informal sector has been providing livelihoods for the new unemployed. Hence, any study in the developing country that investigates jobless growth based on formal sector data is likely to result in misleading conclusions. Hence, all developing country studies tend to make clear distinction between formal and informal sectors of their economies.

2.2.1 African Studies on Jobless Growth:

Most African studies on the issue of jobless growth are on South Africa where unemployment rate has been in double digits since early 1980s. The most cited empirical study on South Africa is the one conducted by Loots (1998). Using data for the period 1970 1996, he estimated the GDP elasticity of employment in the formal nonagricultural sector to be less than unity - in the range of 0.49 during 1970-80, 0.76 during 1980-90 and -0.55 during 1990-96, while on national level the GDP elasticity of employment was 0.16 over the period 1990-1995 - thus leading him to conclude that economic growth did not stimulate employment in South Africa.

32

The informal sector in India contributes more than 60 percent of GDP (NAS) and covers about 87 percent of the workforce (Sinha, Sangeeta, Siddiqui, 2003).

37

In the same study, the author also estimated the GDP elasticity of total formal and informal employment in South Africa to be in the range of 0.77 during 1970-80, 1.01 during 1980-90 and 0.16 during 1990-95. Although the GDP and unemployment relationship was negative - a one per cent increase in the real GDP led to a 1.26 percent decrease in the unemployment rate - the R-square value is 0.16 which is very low and indicated that only 16 percent of the change in the unemployment rate could be explained by economic growth. The low elasticity figures and the weak relationship between GDP and unemployment show inability of the South African economy - especially the formal economy - to create sufficient employment opportunities. Toit and Koekemoer (2003) confirmed the presence of jobless growth in South Africa by using a neoclassical model of wages and employment for skilled and unskilled labour. They investigated the structural nature of the problem of rising unemployment since 1970s with a view to investigate if economic growth was jobless in South Africa. Their model of wages and employment was based on the systems approach followed by Layard and Nickell (1985, 1986) and Nickell (1988). A Cobb-Douglas cost function was estimated as representative of the South African production structure. The period of analysis was 1970 to 2000 and all data series were employed in the first difference form in error correction models, which captured the short- as well as long-run dynamics of the labour market. A single equation residual based Co- integration (ADF) estimation procedure was chosen. The estimated elasticity of labour demand with respect to the GDP
growth was found to be 0.48 in case of skilled labour, and 0.39 in case of unskilled. These low responses of employment to GDP growth are indicators of jobless growth in South Africa.

Altman (2003) defined jobless growth in two broad ways: Firstly, "jobless growth may refer to a situation whereby the overall economy is growing, but the absolute 38

employment level is stagnant or falling, rendering near-zero or negative employment growth rates. Alternatively, the term may be used to describe a situation whereby the overall economy is growing, while the rate of unemployment is rising". In his empirical analysis, he found no strong correlation between GDP and employment growth in South Africa over the period 1994-2002. He also found that higher GDP growth generally accompanied higher unemployment rate. A shift away from formal to informal sector was observed where there are lower returns to education, lower wages, and fewer contractual benefits for the employees. His calculations suggested that as of 2001, the South African economy would have needed about 450,000 net new jobs each year while on average only 240,000 net new jobs had been created annually since 1994, almost all of which were in the informal sector. This higher rate of desired job creation meant annual employment growth of 3.9 to 4.4 per cent. Due to dramatic structural changes taking place in South Africa, it was difficult to calculate precisely the rate of GDP growth that would have been needed to achieve this minimum target of employment growth. Hence, to give an indication of the needed GDP growth rate, the author cited literature on other countries that had transformed less dramatically. "In the South-East Asian economies, for example, employment growth rates of 2.5 to 4 per cent would normally be associated with GNP growth rates around 5 to 8 per cent or more (Mazumdar & Basu, 1997). In these countries, employment and growth depended on the rapid expansion of low-cost manufactured exports, made possible by a low cost structure. On the other hand, developed economies that have overcome high unemployment have relied substantially on the expansion of high-valued market services (such as business and professional services) and low- value social and personal services, such as health, education and child care. Ireland is one such example, where employment growth rates generated in this way

39

have averaged 2.6 per cent per annum during 1991 to 1997, while GDP grew by more than 8 per cent per annum" (OConnell 1999). "South Africa has a capital-intensive structure of production with many supply constraints limiting employment and investment multiplier effects. This is reinforced by globalization, where it is mainly the high-skill, capital-intensive export industries which are benefiting" (Altman 2001). At the same time, "service sector industries are expanding as in Ireland, and entry is easier for formal and informal Small Medium and Micro Enterprises (SMMEs)" (Lewis 2001). Hence, Irelands example was used by the author to conclude that the South African economy would generate less than 1 per cent annual growth in formal market -based employment in the short -to-medium term. Bhorat and Oosthuizen (2007) analyzed South African data from 1995-2005. Their study can be divided into two parts. In the first part, they provide a descriptive analysis of GDP and employment growth. GDP and employment growth had positive relationship over the period 1967-1993 but from 1994-2002 these variables exhibited a negative relationship. This means employment fell during economic growth in most of the post 1990 period. The authors also used two simple performance indicators Target Growth Rate and the Employment Absorption Rate or EAR.33 They concluded that employment grew just 29.3 percent during 1967-2002, but it would have needed to grow by 66.7 percent of the net entrants to the labour market. The economy was able to provide just 44 jobs for every 100 economically active individuals that entered the labour market during the period.

33

"The target growth rate measures how fast employment would have had to expand in order to provide work for all net entrants to the labour market over a given period. The employment absorption rate is the ratio between actual employment growths and the desired or target employment growth rate"(p.10-11).

40

In the second part of their analysis, the authors obtained their GDP elasticity of employment by running an OLS regression of the log of employment demand on the log of wage rate, user cost of capital and GDP. They estimated this elasticity for the nonagriculture sector using formal employment data for the period 1980-98, as obtained from the official South African Reserve Bank (SARB). Their autocorrelation adjusted results of the regression model yielded the GDP elasticity of employment to be 0.29 for 1980-89 and -0.20 for 1990-1998 in the formal sector of South Africa. Extending their analysis to individual sectors of the economy including the agriculture sector, they also tested the jobless growth phenomenon on disaggregated data for the period 1995-2005. All sectors had a positive employment growth except for the mining and agriculture sectors which were hence identified as experiencing jobless growth. Another African study (Fofana, 2001) investigated jobless growth in the private sector of Cote d Ivoire. A co -integration analysis was conducted to estimate employment elasticities with respect to GDP over the period 1975-1995. He also used foreign aid, investment and public expenditure as explanatory variables in the model. The study concluded that employment and economic growth do not move together in the long run. In sum, his econometric findings showed a presence of jobless growth in the Ivorian modern private sector. 34

34

Their study also showed that the employment elasticities with respect to foreign aid and investment were statistically significant, but only the latter had a positive effect on employment. Public expenditure also had statistically insignificant impact on employment.

41

2.2.2 Asian Studies:

Khan (2007) has identified several factors responsible for jobless growth in developing countries. These factors include: the transition towards a market-based competitive environment that necessitated economic reforms which in turn led to worker layoffs and adoption of a cautious approach to new hiring; a structural shift in the economies towards less employment-intensive sectors which may be the result of greater international competitiveness of an economy; adoption of more capital-intensive production; and reduced availability of credits for small and medium size enterprises who tend to be more labour intensive. Some of these factors have been the focus of Asian studies on jobless growth which are reviewed in this Section. Gupta et.al (2005) presents detailed evidence on India. Economic growth in countries with per capita incomes similar to Indias has normally been led by the manufacturing sector. However, in recent years, the services sector has been the engine of economic growth rather than manufacturing.35 These authors used both national and sector wise data from 1960 to 2000 to estimate employment elasticities in the pre-Reform period (1983-84 to 1987-88) and the post-Reform period (1993-94 to 1999-2000).36 Their calculations indicated that between these two periods a sharp decline in the overall employment elasticity from 0.6 to 0.16. They also noted sizeable reductions in employment elasticity in the agriculture sector from 0.87 to 0.01, in manufacturing sector

35

Rodrik and Subramaniam (2004) provided a very optimistic view about the Indian economy, that Indian economy would grow faster than China, at the rate of 7 to 8 percent, for two reasons. The first was the presence of a higher level of well developed institutions in India because of it democratic political system. Second reason was the faster growth in the labour force of India, at 2 percent annually, than in that of China and other competitor countries.
36

Market-based economic reforms were introduced in India during 1988-89. The purpose of these reforms was to integrate the Indian economy with global economy by adopting liberal policies such as liberalization of trade and foreign exchange rate, and increased privatisation. It was argued that these reforms would benefit wealthy people and bypass the poor. The Ninth Plan stated one of its objectives as growth with social justice and equity.

42

from 0.59 to 0.33 and in construction sector from 2.81 to 0.82. These reductions occurred because the growth rates of GDP in agriculture and manufacturing sectors were slower after 1997 than they were in 1980s. In addition, employment elasticity increased in the post reforms period in a number of service sector industries including finance, insurance, real estate and business services. In another study on India, Muzumdar and Sarkar (2007) decomposed the determinants of employment elasticity into three key factors (i) the trend in the domestic real exchange rate, abbreviated as DRER. 37 (ii) The wage-employment trade-off; and (iii) the trend in the share of wages. They analyzed the formal manufacturing sector data for the period 1974-75 to 2001-02 and found that employment elasticity showed a cyclical pattern in four sub-periods, such as reform period of 1986-96, and the immediate postreform period of 1996-2002. During the first period, 1974-80, the employment elasticity was 0.99 but it declined sharply to -0.16 during the 1980-86. During 1986-96, employment elasticity rose to 0.33 but it experienced a significant drop again during 1996-2002 to -1.39. This last drop was attributed by the authors to both factors isolated in their decomposition exercise: (i) DRER further turning against manufacturing and (ii) the wage growth at the expense of employment growth. Both of these factors were important as indeed they were in the period of jobless growth. But when considering the magnitudes involved, these authors concluded that the quantitative impact of the DRER (price) factor was stronger than the wage-employment trade-off. 38

37

DRER also define as, the relative movement over time of the producer price index relative to the consumer price index.
38

The relative movements of producer and consumer price indices; sometimes called the domestic real exchange rate, or DRER, "translates the wage bill growth into real terms in terms of consumer prices. The share of wages and the wage-employment trade-off are both labour market variables. They are tied together in neo-classical economics by the supply functions of labour and of capital working through the production function. Together, they determine the share of wages, the level of employment, and the wage per worker. Economists recognize the importance of expectations both in the determination of the wage per worker and the share of wages. Thus, the difference between the neo-classical tradition, stressing the dominance of

43

Paul and Pellissery (2007) attributed jobless growth in India to recent policy changes on the demand side of the labour market. They suggested that the policy shift away from capital intensive methods may have generated greater mismatches in the labour market, thereby creating a significant impact on the economic growth itself. The decline in economic growth can have cyclical effects on employment generation, especially since the Indian education system was unable to create the employable labour force. The paper emphasized the need for appropriate employment strategies at regional levels by focusing on the appropriate human capital demands of their labour markets. Karmakar (2008) reviewed the effects of market liberalization and trade openness on the employment generating potential of service sector. Indian services sector employs about 26 percent of labour force, while its share in the national GDP is over 55 per cent. This was a clear indication of jobless growth in this high growth sector. A disaggregation of data showed that, finance, insurance, real estate and business, contributed 23.4 per cent to the total incremental employment generated in services sector during the five-year period ending in 2004-05. 39 About 16.8 million workers found jobs in services sector during the five-year period. However, despite the low overall elasticity of employment in the country at just 0.48, the disaggregated data showed that the employment elasticity was reasonably high, and was increasing in certain service sectors. An elasticity of employment of 0.94 in finance, insurance, real estate and business services followed by

factor supply functions, and the post-Keynesian tradition, emphasizing the importance of decisions originating on the employers side, have been reduced. The relative importance of the wage-employment trade-off is that a negative sign of the value signifies that there is a decline towards wage growth, while a positive value indicates that employment growth is preferred. Thus, other things being equal, a positive value of the first term would favour an increase in employment elasticity, while a negative value would signify that the bias towards wage growth reduces employment elasticity. The DRER effect is the difference between producer price growth and the consumer price growth. Given real output growth, a negative value of the DRER is similar to a leakage from the growing pie, which has to be shared between wage growth and employment growth: ceteris paribus, it depresses employment elasticity".
39

The gross incremental employment was around 60.82 million during this period.

44

construction sector at 0.88, trade 0.59 and transport 0.27. The employment elasticity was 1.52 in agriculture and 0.34 in manufacturing sector over the period 1999 to 2005. The author of the above study also argued that, contrary to developed countries, Indian service sectors employment elasticity was lower than that in its m anufacturing sector. Services sector benefited from outsourcing since 2000. Her statistical analyses indicated that India continues to exhibit a strong revealed comparative advantage, RCA, in services relative to goods sector. 40 Finally, the author of the above study also argued that developments in services contribute to economic growth in numerous ways, by offering cheaper final services to both individual and business consumers and by providing inputs into productive activities. Hence, a liberal and pro-competitive service regime can be a new source of growth and employment generation. Another country for which jobless growth has been investigated is Turkey. Pamukcu and Yeldan (2005) allude to the possibility of jobless growth in Turkey when evaluating the outcomes of that countrys macroeconomic policy. Turkey had a high economic growth rate but also a high unemployment rate during the 2001 recovery period leading the authors to conclude that problem of jobless growth exist in the Turkishs economy. Between 2002 and 2004, unemployment rate continuously increased in Turkey from 6.5 to 10.3 percent, and employment averaged a negative 0.1 percent growth, even

40

"Indias RCA in services has been rising sharply since mid -1990s, increasing the gap with the goods sector drastically in the past decade. A time-series analysis of the service sectors RCA indicated that Indias current strength in commercial service exports comes from business services including finance, management and other professional services among others. Between 1996 and 2000, Indias RCA in business services grew most dramatically, rising by 327 per cent. The recent trend of increased specialization of Indias exports of services in a selected set of sub -sectors within the services sector reflects the change in composition of exports". 45

though GDP grew at 7.5 percent rate. 41 Employment decline was observed in seven of the eleven quarters although GDP grew throughout the period. Tezcek (2007) followed up on the above study and considered three possibilities for jobless growth in Turkey including: 1. Agriculture Displacements in the pre and post 1980s period: Until 1950s, about 80 percent of total labour force in Turkey was employed in agriculture. Since then, agricultural employment continued to decline. Hence, the decline in agricultural employment cannot be a reason for jobless growth that was observed in 2001. 2. Structural adjustment policies adopted during 1980-2000: These policies

resulted in a deregulation of labour markets and greater openness of Turkish economy to rest of the world which may have made employment practices more competitive. 3. Production re-structuring that led to productivity increases in the post-2000 period: Turkey initiated a restructuring of its production technology in 1997. Production activities aimed at increasing productivity of labour rather than increasing employment. One measure adopted in this regard was the increased mechanization of workplace and less reliance on labour. As a result, between 1997 and 2005, the rate of labour productivity increased by 52.9 percent but the rate of employment declined by 17 percent in Turkish manufacturing sector. The change in production technology is provided as a third possible reason by authors for a jobless growth in Turkey. Another Asian country for which jobless growth has been investigated is Indonesia. Islam and Nazara (2000) estimated GDP elasticity of employment growth in Indonesia and the threshold level of GDP growth required to absorb the new entrants into
41

Unemployment was a severe problem in Turkey, especially among the young urban labour force, reaching 26 percent in 2003-04.

46

its labour force. They used province wise as well as national data from 1979 to 1996 and estimated the employment elasticity using different statistical techniques such as, simple descriptive analysis, the Ordinary Least Squares (OLS) method, Generalized Least Squares (GLS) method and the random effect model. Their estimates varied from a low of 0.49, based on descriptive statistical analysis, to a high of 0.66, based on econometric analysis. Indonesia required between 3.47 per cent to 4.68 per cent threshold levels of GDP growth to absorb its 2 million new entrants to the labour force each year. 42 Islam and Nazaras study suffered a caveat in that it did not account for any possible nonstationarity of data. Hence their results could be misleading. Ghannam (2008) used the data of Saudi private firms and investigated the relationship between employment and economic growth. He used co-integration/error correction models and Granger causality techniques to determine the direction between the two variables in both short and long runs. The study covered the period 1973-2002. His co-integration test confirmed a long-run equilibrium relationship between economic growth and employment in Saudi Arabian private sector. Error correction model and Granger causality test results indicated a unidirectional causality from economic growth to employment, and not vice versa, in the short and long runs. The authors suggested that the Saudi private firms should focus on the quality of hired labour to increase their productivity via technological improvements in order for employment to play important role in economic growth.

42

A widely cited statistic is that every one per cent growth in GDP leads to the creation of 400,000 jobs. This led a recent (1999) ILO Mission to Indonesia to conclude that the economy would have to grow at 5 per cent to absorb new entrants to the labour force, but even such a growth rate would not be able to cope with the backlog of the unemployed and underemployed.

47

2.3 Studies on the Relationship between Economic and Employment Growth in Pakistan: No study has investigated the issue of jobless growth in Pakistan directly. Some studies have estimated the employment and economic growth relationships, and a few of these have estimated the output elasticity of employment. Some of these studies are descriptive while others are empirical. Results of these studies can be used to infer the nature of economic growth, jobless or job generating, as was done for other countries in this literature review. Baqai (1979) estimated the employment elasticity of output in various sectors of Pakistans economy. The elasticity was found to be 0.33 in agriculture, 0.35 in Mining and manufacturing and 0.33 for the overall economy. Many other studies, for example Kemal & Irfan (1983), Kemal (1990), Chaudhary & Hamid (1994, 1998), Aslam and Zulfiqar (2008), also provided estimates of employment elasticities which are reported in Table 2.1. Nationally, the calculated values are in the range of 0.33 and 0.41. Only Aslam and Zulfiqar (2008) conducted their study for the post-2000 period. Except for Aslam and Zulfiqar (2008) all studies have the same caveat, as some studies reviewed earlier in this Chapter, that they do not account for the possibility of data non-stationarity which makes the reliability of their results questionable.

48

Table 2.1: Sector Wise Elasticities of Employment with Respect to GDP in Pakistan
Elasticity

Sectors
Agriculture Mining &Manufacturing Construction Electricity Transport Trade &finance Overall(National)

Chaudhary & Hamid

Hyder

Baqai

Kemal

Planning Commission*

Aslam& Zulfiqar

0.44 0.50 0.82 0.32 0.34 0.44 0.37

0.46 0.22 0.89 0.28 0.48 0.40 0.40

0.33 0.35 0.33

0.12 . 0.18 0.31 .

0.37 0.02 0.87 0.54 0.45 0.57 0.41

0.13 0.20 0.60 0.24 0.22 0.32 0.11

Source: * Based on Anwar (2004), State Bank of Pakistan, Table.9. All other studies are reported in Aslam and Zulfiqar (2008) who do not provide the time period of coverage. These studies were not available to me at the time of writing.

Majids (2000) report was based on a study he conducted in 1997 -98 for the International Labour Organization (ILO), United Nations Development Program (UNDP) and the School of Policy, Planning and Development (SPPD). The basic objective of his report was t o provide input to the development of an employment policy for Pakistans Ninth Five Year Plan period, 1997-98 to 2001-02. He emphasized the importance of manufacturing sector and of macroeconomic policy environment in creation of an economic growth that is job-intensive. Following important observations were made in the study: 1. 1970s and late 1990s were periods of low economic growth. 2. Pakistans employment growth has been lower than its GDP growth in all decades except in the early 1970s. Since the mid 1980s, volatility of employment growth has increased. During the high economic growth of the 1980s, employment growth was slow. 49

3. GDP growth has been a stable component of the employment elasticity and the elasticity of employment with respect to GDP growth has generally followed the employment growth. The first two patterns noted above are indicative for jobless growth. The volatility in employment growth appeared to be the cause of structural changes in the economy of Pakistan. Hence, it will be instructive to undertake an examination of sectoral changes on employment growth in order to more fully understand the sources of jobless growth. Majids (2000) report also emphasized the importance of estimating employment elasticities in assessing the employment situation in Pakistan. The study provided national and six sectoral estimates of the employment elasticities over the three different decades (1970s, 1980s, 1990s) using a descriptive analysis of data. These estimates are reported in Table 2.2

Table 2.2: National and Sectoral Elasticity of Employment with respect to GDP in Pakistan, 1970s, 1980s, and 1990s Sectors
Agriculture Mining &Manufacturing Construction Electricity Transport Trade &finance Overall(National) 1970s 1980s 1990s

1.05 0.70 0.89 1.51 0.55 0.90 0.68

0.49 0.17 1.05 -0.39 0.48 0.37 0.38

0.48 -0.10 1.81 0.32 0.14 1.22 1.22

Source: Majid (2000)

50

The elasticity estimates of Table 2.2 may be misleading as the issue of possible data stationarity is ignored. Furthermore, no controls are introduced for the impact of other variables on employment. However, these are the only available consistent estimates of the elasticity of employment demand with respect to GDP growth in Pakistan. These estimates indicate that a weak relationship existed between employment and GDP growth at the national level until the 1980s, when the national elasticity values were lower than unity, but the relationship became stronger in the 1990s. The strengthening of this relationship is largely attributed to Construction and Trade & Finance sectors. Structural changes in the transport sector and in the combined mining and manufacturing sectors appear to be responsible for a weakening of employment and GDP growth relationships in these sectors while it appears to indicate its declining capacity to absorb labour in the agriculture sector. 43 Majids report also discussed two important issues with the growth and development experience in Pakistan. The first one was the insensitivity of economic growth to income distribution. A structural change in the economy may have spurred economic growth but it has not improved the lives of a majority of the population by anywhere near what that growth process could have achieved. The second issue was the governments inability to establish a sustainable pattern of public finance. The country has historically relied on borrowing from short term private sources, loans by domestic non bank sectors and heavily depends on foreign grants.

43

Agricultural output growth rates were reasonable in the 1980s and 1990s but slightly declined in 1990s. Employment growth and productivity growth have declined during this time, leaving a more or less constant elasticity of employment. However, the low employmentlow productivity growth situation is sensitive to growth conditions in the rest of the economy and the sector needs to be protected on this count. An overview of the employment situation in agriculture suggests that over the 1970s, 1980s, and 1990s, a profitability-driven cropping pattern changes based on the new technological package that may have increased labour requirements in the sector. However, the changing distribution of operated holdings, mechanization and tenure shifts may have partially adapted to, as well as reduced, these increases.

51

As a result, Pakistan now faces a profile of increasing future debt service payments. These two issues are also mutually reinforcing. A growth process insensitive to the distribution is complemented by the unwillingness or the inability of the State to tax the better-off sections of society. This in part makes the public finance structure narrow and precarious. For the development of an employment growth strategy, the above study recommended that the country should focus on achieving the growth of GDP in the manufacturing sector which boost the employment and promote the dynamic sub-sectors within the small-scale enterprise sector. The limitation of increased labour absorption in agriculture should also be recognized by the policy makers. Other labour market studies in Pakistan which have discussed employment generation include Kazi (1987), Nadvi (1990), and Kemal and Mehmood (1993). They all suggest that informal sector has been an important source of employment generation in Pakistan. This conclusion highlights the importance of a sectoral analysis of employment growth relationship in a developing country, as was stated in the beginning of this Section, to isolate the impact of informal sector on the official employment data. Some studies (Haq, 1993 and Chaudhary and Hamid, 1998) have discussed that fiscal policy and economic planning were not focused towards employment generation in Pakistan. Haque and Waqar (2006) suggest greater reliance on domestic commerce sector to generate employment opportunities, as it can be the most-pro-poor sector, given greater concentration of small businesses in this sector. In addition, expansion of this sector can also help generate employment opportunities in the other sectors of the economy, such as transport, construction, retail shops and other service industries.

52

Most of the above studies did not perform any econometric estimation and none has discussed or analyzed employment and economic growth in the context of jobless economic growth. Hence, the present study is an attempt to fill this gap in Pakistans labour market literature.

2.4 Summary: Most investigations of jobless growth in empirical literature are based on estimates of the output elasticity of employment. Some studies have used these estimates to calculate the threshold level of GDP growth, i.e., the rate of GDP growth below which employment growth will not occur. The actual growth rate in GDP is then compared with this rate. The theoretical background to most studies is providing by the theory of business cycle. When the economy starts to recover from a recession, employment grows at the same time or soon thereafter. If employment takes longer to recover, the recovery is viewed as jobless. However, some studies in literature have also based their analysis on the seminal work of Okuns to analyze the phenomenon of jobless growth. A dec lining value of Okuns coefficient is viewed as an indicator of jobless growth in such studies. Other studies have simply interpreted a weak employment elasticity value as an indicator of jobless growth. In summary, a review of literature demonstrates the presence of jobless growth in developing countries (for example, India, South Africa, Cote d Ivoire) as well as in developed countries (for example, the United States, Poland, Finland, and most of the European countries). This phenomenon has been attributed to technological changes, structural changes and changes in macroeconomic variables.

53

Empirical studies on jobless growth in developing countries are somewhat limited in their scope due to data limitations. Employment data used in these studies may not be meaningful since they do not take into account the general trend towards a decrease in the working time per employee (Maddison, 1991). Lack of consideration of this tendency may lead to over-evaluation of employment elasticities (Padalino and Vivarelli, 1997). In addition, as clearly stated by Saget (2000), attention should be paid to the impact of the unofficial economy on GDP and employment data, because it may alter their relationship. This issue is particularly important in case of developing countries.

54

CHAPTER 3 Theoretical Background, Method of Analysis, and Economic Model

The previous chapter provided a review of the empirical literature on the issue of jobless growth. Theoretical background in most of those studies is provided by the theory of economic growth and business cycle which is reviewed in the next section (Section 3.1). The method of analysis employed in estimating the nature of economic growth, jobless or job generating, in Pakistan is discussed in Section 3.2. Following this method of analysis, an employment demand model is derived in Section 3.3 using a Cobb-Douglas production function. Section 3.4 describes some issues in the estimation of employment demand equation derived in Section 3.3.

3.1 Theoretical Background of the Relationship between Employment and Output Growth: In economics literature, the relationship between employment and economic growth can be best described under the background of economic growth theories, business cycle theories and Okuns (1962) law. Economic growth refers to a change in economic output, measured by the real Gross Domestic Product (GDP) of a country. According to the neoclassical theory, the long term economic growth in an economy depends on the growth in labour force, growth in human capital, growth in physical capital and on technological improvement. The short-run economic growth depends on the conditions in goods and services market of the economy. Factors that may change aggregate supply are changes in factor prices and factor productivity. Factors that may change aggregate demand include changes in private investments, government spending, foreigners demand for goods and services and changes in household consumption 55

resulting from changes in personal income. Unexpected changes in either aggregate demand or in aggregate supply are responsible for causing short-term fluctuations in economic growth around its long-term trend and such fluctuations are called business cycles, which in turn are the main source of fluctuations in employment. Business cycle research has a long history, much of which was initiated by US researchers working at the National Bureau of Economic Research (NBER) in the early 1920s. These earlier researchers included Mitchell (1927), Mitchell and Burns (1938) and Burns and Mitchell (1946) who were particularly interested in the widespread and persistent fluctuations in the general level of economic activity.44 Classical economists hold that due to perfectly competitive markets, periods of unemployment cannot persist for long because wages will fall in order to get rid of the excess supply. Keynesians on the other hand believe that wages are not changed very easily, especially not downwards. Thus, excess demand and supply may persist for a while. An outstanding characteristic of the economic system in which we live is that it is subject to severe fluctuations in respect of output and employment. However, it is not violently unstable. Indeed, it seems capable of remaining in a chronic condition of subnormal activity for a considerable period without any marked tendency either towards recovery or towards complete collapse. Moreover, the evidence indicates that full, or even approximately full, employment is of rare and short-lived occurrence. Fluctuations may
44

The best definition to date of the business cycle is that proposed by Burns and Mitchell: "Business cycles are a type of fluctuations found in aggregate economic activity of nations that organize their work mainly in business enterprise: a cycle consists of expansions occurring at about the same time in many economic activities, followed by similarly, general recessions, contractions, and revivals which merge into the expansion phase of the next cycle; this sequence of changes is recurrent but not periodic; in duration business cycles vary from more than one year to 10 or 12 years; they are not divisible into shorter cycles of similar character with amplitudes approximating their own"(Burns and Mitchell, 1946, p.3)

56

start rapidly but wear out before proceeding to great extremes, and an intermediate situation is viewed as normal. It is upon the fact that fluctuations tend to wear themselves out before proceeding to extremes and eventually to reverse themselves, that the theory of business cycles having a regular phase was founded.45 National output (GDP) may fluctuate more rapidly than unemployment in the economy for several reasons. As unemployment increases, some unemployed workers may get discouraged and drop out of the labour force after which they are no longer counted in the unemployment statistics. Also, the employed workers may work shorter hours and labour productivity may decrease, perhaps because employers retain more workers than they need. The different phases of business cycles are presented in Figure 3.1. Figure 3.1: Phases of Business Cycle

45

John Maynard Keynes (1936)

57

In the above Figure, a business cycle has been divided into two broad periods: contraction, also called recession, and expansion. During the recession period, the economic output as a whole is in decline. More specifically, recession occurs after the business cycle has peaked, but before it becomes a trough. The economy enters this phase because the bottlenecks that occur at the peak, due to limited resources, result in higher prices of factors of production and of final goods and services that eventually dampen the investment demand. In most economies, a recession is said to occur when the real GDP has declined for two or more consecutive quarters. For most people, a recession in the economy can be the source of harsh economic conditions because as the economy pushes into a recession people start losing their jobs. While no recession lasts without end, it is very difficult to judge just how long a down trend will persist. During the period of expansion, the economy moves from a trough to a peak. This period is divided into two sub-periods, i.e., recovery and prosperity. In this period, business activity surges and gross domestic product expands until it reaches a peak. The economy enters this phase after reaching trough as it starts to replace its existing plant and equipment that begin to wear off after sometime. This replacement investment generates an economic activity, more workers are hired, additional income is generated thereby creating additional demand for goods and services as well as that of the factors of production, and "economic recovery" takes place. Recovery turns into prosperity when the economic output exceeds the output at the previous peak of the business cycle. The whole period of expansion lasts on average about three to four years but has also been known to last anywhere from 12 months to more than 10 years. Andolfatto and MacDonald (2004) noted that after the last two recessions of 1991 and 2001 in the United States, employment growth lagged the GDP recovery by several

58

quarters. According to them, a recession followed by a jobless recovery is what neoclassical theory also predicts. This is because new technology impacts different sectors of the economy differentially, and the resulting sectoral adjustments in the labour market take time to have an effect. Their explanation for the jobless recovery include the varying size and scope of technological innovation , delayed effect of adopting new technology and a slow reallocation of labour across different firms. 46 When economies have been in the recovery phase of a business cycle with growing GDP, employment has grown at the same time, or soon thereafter. However, lately this fact concerning the relationship has been questioned and the concern of jobless growth has been raised. The long-run relationship between employment and GDP growth may be obtained from the Okuns (1962) law which is primarily based on empirical observations that can vary depending on the country and time period under consideration. For the United States, Okun found that a one point increase in the unemployment rate is associated with two percentage points of decline in real GDP. More recently, Abel and Bernanke (2005) also verified this relationship. Reasons for a weak relationship between GDP and unemployment rate could include a reduction in labour productivity if employers retain more workers than needed during an economic downturn, reduction in work hours, discourage worker effect which tends to increase the official unemployment rate, and a possible reduction of multiplier effect of money circulation from workers. One implication of Okun's law is that an increase in labor productivity or an increase in the labour force can result in growth in real output without a fall in the unemployment rate which is the same as the jobless growth phenomenon.

46

In a previous study, Greenwood, MacDonald, and Zhang (1996) had developed a real business cycle model to include the effect of varying size of technological innovation and labour re-allocation and found that a jobless recovery can be the result of a technology-driven recession. Their explanation hinged critically on the above three assumptions.

59

3.2 Method of Analysis: The previous chapter provided a review of the empirical literature that has investigated the issue of jobless growth in different countries. Most studies have used the concept employment elasticity of demand with respect to GDP to infer the presence or absence of any jobless growth in an economy. The employment elasticity of demand with respect to output is obtained as the ratio of percentage changes in employment to GDP. This ratio indicates the extent of the response of employment in the economy to a change in national output. Some studies have calculated the arc elasticity of employment directly by using annual data on employment and GDP. Others have estimated employment elasticity by estimating an employment demand equation. The value of employment elasticity is then used to directly infer any jobless growth in the economy. It is also used to calculate the minimum GDP growth that must take place in an economy to generate any new employment. The actual rate of growth in GDP is then compared with this threshold GDP growth rate. The present study will use the latter approach to investigate if the average GDP growth in Pakistan over the period 1974 to 2008 was jobless. 47 At first, an employment demand model will be derived mathematically based on the production function approach which is standard in the microeconomics literature. The employment demand equation considers employment demand as derived from the demand for final output, or GDP, in the economy. The value of employment elasticity of demand with respect to GDP will be used to calculate the threshold level of the GDP growth rate. This is the level of growth rate which must be achieved by the economy to generate any growth in employment. Finally, the actual GDP growth rate in the economy

47

A direct calculation of the arc elasticity of demand from raw data suffers from two problems. First it considers only the extreme values of employment and GDP, thereby ignoring any interim effects on data. Second, its value is highly unstable. Hence, an econometric estimation is preferable (Gupta and Singh, 2005).

60

over the period 1973-74 to-2007-08 will be compared with the threshold level of GDP growth rate. The data used in this study span the period 1974 to 2008. Separate estimates will be obtained using time series data for the entire economy and using data for seven broad sectors of the economy in pooled form. The next Section derives the empirical model of employment demand.

3.3 Employment Demand Model: Most empirical studies on employment demand are based on production functions. Some of these studies include Fair (1969), Hamermesh (1986) and Hazeldine (1981). Employment demand equation is derived from a specified production function to express labour employment as a function of output. An economy wide employment demand equation can be derived from an aggregate production function in which the national output (usually the Gross Domestic Product, GDP) is expressed as a function of two inputs, i.e., labour and capital. An employment demand equation will be derived from the production function with some economic assumptions so that the employment demand is dependent on the final output and not on input prices. 48 Although the demand equation is derived for an individual firm, the analysis is conducted for the economy which is viewed as a collective producing unit. A two input-single output production function for a single firm can be written in a general form as follows:

X= f (Z1, Z2)
48

(1)

An alternative derivation of employment demand equation in which factor price ratio appears as an explanatory variable is provided in Appendix I.

61

Where X denotes the output level while Z1 and Z2 are labour and capital inputs, respectively. In the short run, a firms capital is assumed to be fixed so that the production function may be written as follow, short run

X = f (Z1,

2)

= f (Z1)

(2)

Assuming, invertability, equation (2) may be solved for Z1 to give the following expression,

Z1 = f (X)

(3)

Which shows the firms desired level of labour input ( Z1) as a function of its output level (X). Here we see that the above equation (3) is the employment demand function derived from the technical relationship between the firms inputs and output as represented by its short run production function given by equation (2). This simple model is based on the usual assumption of technical efficiency of the production function, in the sense that equations (1) and (2) show the maximum output level for each possible combination of inputs (Henderson and Quandt, 1980, p.66). Here we see that in the employment demand function there is no role of the wage rate in affecting the firms desired level of labour input. The analysis in this study is based on such a labour demand equation, which has also been estimated by Greenaway et al. (1999) for UK, Hine and Wright (1998) for UK, and Milner and Wright (1998) for Mauritius. In the context of jobless growth, this function has been estimated by Kwiatkowski et.al (2004) and Wolnicki et al (2006), among others. This equation was also used by Stehrer (2004) for analyzing trade effects on employment in the Organization for Economic Co-Operation and Development (OECD) countries. 62

To derive the employment demand equation for Pakistan, a Cobb-Douglas (1928) production function is assumed.49 This production function at a given time period, t, can be written as follows:

Where, Xt = Firms actual output at time period t. A(t) = A measure of input productivity assumed to grow over time in Hicks neutral

fashion50 = Labour input at time t = Capital input at time t and are exponent s of output respectively. If A (t) = there is an absence of technical change. If technical change is present, it .51 and which measure the labour and capital elasticity of

is assumed that it occurs exponentially at a constant rate

Where,
49

In a Cobb-Douglas production function, the elasticity of substitution between inputs is constant at unity. More flexible specifications of production functions include Constant Elasticity of Substitution and Translog production functions. Zahid,Akbar and Jaffry (1992) find that the elasticity of substitution between capital and labour in Pakistani manufacturing sector is generally low and consistent with Cobb Douglas production function. Battesse, Malik and Sultana (1993) find that in food manufacturing, elasticity of substitution is not statistically significantly different from unity. Kalim (2009) finds the overall elasticity of substitution in manufacturing to be 0.96. All of these studies are based on Constant Elasticity of Substitution and translog specification of production functions.
50

Hicks-neutral technical change (Hicks, 1932) refers to a technical change that does not affect the factor ratio in the sector which is progressing. Hence, such a change does not affect the factor price ratio in that sector, if the production function is homogeneous. According to Krugman (1999), Hicks-neutral technical change comes up in practice only for aggregate production functions.
51

Charles, Knox and Marie (1976), See Katzner (I 970) for further details.

63

 = Positive constant t = Time trend So that the production function can be re-written as under

(4)

The equation (4) is non-linear in terms of time trend which captures shifts in the production function brought about by the technical progress leading to change in the efficiency of production factors. The above equation (4) can be linearized by taking natural logarithms on both sides which yields the following equation: (5) Since lne =1, the equation (5) becomes (6) Rearranging equation (6) and solving for we get,

Therefore,

(7) 64

where,

The above equation (7) shows that for any given level of output, the firms desired level of labour input in a given time period is negatively related to its use of capital input, in other words the production isoquant drawn between capital and labour is negatively sloped. The positive relationship of employment demand with the level of output indicates the scale effect on labour demand: for any given quantity of capital used in a particular time period employment demand will rise with an increase in output. Ball and St Cyr (1966) suggested growth in capital stock can be assumed to be similar to technical progress which can be captured by the time trend variable. Kwiatkowski et al (2004) and Wolnicki et al (2006) assumed the same in their empirical study on jobless growth in 48 Commonwealth Independent States (CIS) and for European countries, respectively. An important question to ask is whether capital stock growth can be captured by the time trend variable in a developing country like Pakistan where markets for capital goods may not function as efficiently as in a developed country. In the absence of reliable data, it is assumed in this study that most capital goods that are used in production of goods and services in developing countries are imported from industrialized developed countries. A recent study, Caselli and Feyrer (2007), has found that the marginal product of capital across developed and developing countries is remarkably similar. Using this result, the authors draw the implication that the flow of capital stock from rich to poor countries is not affected by international aid and credit

65

restrictions. 52 In Pakistan, the share of capital goods in annual imports has varied only slightly, mostly within one standard deviation of the mean, since the 1970s (Table 3.1).53

Table 3.1: Share of Capital Goods in Total Imports, Pakistan (1974-2008)

Capital Goods ( Millions) 63000 106908 205304 471355 736292 1312345 2033200 4928404 Total Value ( Millions) 189013 347662 580622 1238788 2298670 4096906 6075036 14826697 Share (Percentage) 33 31 35 38 32 32 33 33

Years 1974-80 1981-85 1986-90 1991-95 1996-00 2001-05 2006-08 1974-08

Source: Pakistan Economic Survey, 2008-09.

Based on the findings of the study noted above, and because of the stability of the share of imported capital goods in annual imports in Pakistan, it appears safe to assume that the availability of capital stock in Pakistani industries has not been a significant hurdle in industrial production and that the capital stock grew smoothly over the period 54.

Hence, assuming capital stock is growing at a constant rate, simplifies to

equation (7)

where,

52

They attribute lower capital ratios in developing countries to lower availability of complementary factors and lower efficiency, as well as to lower prices of output goods relative to capital.
53

Only during 2000-02 this share fell below one standard deviation probably due to the international sanctions that were imposed on Pakistan soon after the 1997 nuclear explosion. These sanctions were removed in 2002.
54

Zahid, Akbar and Jaffry (1992) find there is idle capital stock in large scale manufacturing sector of Pakistan.

66

= Collecting the time trend variable, the above equation becomes: 55 (8) where,

The coefficient

in Equation (8) captures the total effect of the growth of

capital stock and technical progress. Equation (8) is a particular form of Equation (3) which results from the assumption of a Cobb-Douglas production function with a constant rate of growth of capital input. The stochastic form of the basic model (8) is obtained by adding an error term, to capture the influence of such random factors on employment as the influence of weather on agriculture, international demand for goods and services, unexpected changes in raw material prices, changes in political events, and so on. The equation for econometric estimation may be written as follows: (9) where, is employment demand. Measurement of this variable for econometric

estimation of the model will be discussed in the next Chapter. The equation will be

55

We substitute the in equation (7) for capital stock b ecause we assume that capital changes over time with a proportional rate , so = = = t therefore,

67

estimated using aggregate data assuming the economy to be a collective producing unit comprising several individual units or economic sectors. It will be estimated first using a time series national data for the economy of Pakistan, and then using sectoral data for the economy that will be organized in a pooled form.56 The model will also be modified to control for certain data characteristics that may have additional influence on the employment demand in an econometric estimation based on pooled data. The estimates of the model will be used to obtain a threshold level of GDP growth defined as the minimum level of GDP growth needed to generate employment in the economy. Calculations of the threshold level will be discussed in Chapter 4 where the augmented model is provided.

3.4 Some Issues in the Estimation of Employment Demand Equation: The issues relating to the estimation of the employment demand equation obtained above can be divided into two sets. One set of these issues relates to the way the employment demand equation is derived and its interpretation. The second set of issues relates to the data set used for its estimation. While we discuss the first set of issues in this section, the second set of issues will be discussed in the econometric estimation chapters. First, there is the issue of the two-way relationship between employment and output. From the perspective of an economy-wide production function, the use of labour and complementary factors of production generates national output or GDP. Hence, the faster the growth of labour, ceteris paribus, the faster the growth rate of output. In an econometric estimation, this two-way causality between independent and dependent
56

Pooled data comprise several independent cross section data gathered at different points in time.

68

variables creates the problem of endogeneity. It is assumed in this study that employment determination is a two-step procedure. In the first step, a firm determines its output level based on market demand. In the second step, it chooses its combination of inputs that will minimize the cost of producing that level of output. In this sense, the employment demand equation is a conditional demand for labour employment conditional on the demand for final product. 57 Second, some economists have argued (Islam, 2004; Islam & Nazara, 2000) that the notion of employment elasticity is endogenous to the policy regime. A given policy regime could be more or less conducive to the growth of employment. For example, existing policy initiatives could encourage labour-using technology or they could impart a capital bias in production processes. The clear implication is that the elasticity of employment is not really an exogenous variable. It is affected by the market forces as well as by the economic policies. Disentangling the two effects can be a complex task. In the case of Pakistan, economic policy has focused mostly on capital biased technology to promote economic growth which in turn was expected to create employment. While data limitations prevent an explicit incorporation of the changes in capital stock, the time trend variable has been included to capture the effects of changes in capital stock and technology over time. A third complexity in the measurement of employment elasticity is specific to Pakistan.58 It can be argued, following Islam and Nazara (2000) that the impact of GDP on employment is not symmetrical, i.e., while economic growth may promote employment, economic contraction may not result in unemployment. The authors
57

Lunenberger (1995) uses the term conditional factor demand to describe an input demand function in which the final output appears as one argument.
58

Islam and Nazara (2000) point out that this issue is specific to the Indonesian economy. We can also apply this argument to Pakistan, because its economys features are similar to Indonesia.

69

suggested that it appears to be true in Pakistan because of the so-called unemployment as luxury hypothesis. In the absence of any comprehensive unemployment benefits, unemployment during an economic contraction in Pakistan becomes a luxury that only those with adequate non-labour income can afford. During the 1997-98 Asian financial crisis, workers in Indonesia responded to a major recession by re-allocating their labour services to the agricultural sector and the informal sector rather than remaining openly unemployed (Islam and Nazara, 2000). This means that in an economic downturn the use of any employment elasticity coefficient can overstate the impact on unemployment (Groshen and Potter, 2003). To account for possible difference in the response of employment in different phases of business cycle, the present study will identify the slow and fat growth periods of Pakistans economy over a business cycle and capture their effects by including a dummy variable in the employment demand model. This dummy variable will be interacted with the GDP variable.

70

CHAPTER 4 Econometric Results of Employment Demand Model for Pakistan

The previous chapter derived an employment demand equation based on a Cobb-Douglas production function. An econometric estimate of this model based on national data will provide the elasticity of employment with respect to GDP. The magnitude of this elasticity will indicate the strength of the relationship between employment and GDP, which can be used to infer the extent to which GDP growth in Pakistan has induced employment growth during the period between 1974 and 2008. This elasticity value can also be used to obtain a threshold level of the GDP growth rate, i.e., the minimum growth rate in GDP required to generate any employment growth in the economy. The present chapter first specifies the employment demand equation in a form that will be estimated econometrically. This is done in Section 4.1. Section 4.2 defines the variables included in the model and also describes the sources of data used for each variable. Econometric issues related to the employment demand estimation are discussed in Section 4.3. Section 4.4 discusses the econometric estimates of the employment demand equation and also uses the estimated elasticity of employment demand with respect to GDP to calculate the threshold level of GDP growth rate with which the annual real GDP growth rates in Pakistan that were experienced over the period 1974 - 2008 are compared. Section 4.5 extends this analysis by estimating the elasticity of employment demand with respect to output observed during fast growth periods and using the same to calculate the threshold levels of GDP growth rates for the two fast growth periods observed during the period of analysis. A summary of findings is presented in Section 4.6.

71

4.1 The Employment Demand Equation: The basic employment demand equation was derived in the previous chapter, assuming a Cobb-Douglas production function and a constant rate of growth of capital. This equation is re-written below: (9)

For an empirical estimation of the above equation, it is important to first define the labour input (Z1t). In this study, Z1t is defined to be a product of the number of employed workers at period t and the hours worked by these workers in that period. This definition of labour input is a more complete definition of labour service in the sense that it incorporates the effect of varying hours demanded per worker by the firm. Mathematically, Z1t = H t.Et (10)

Where Ht is the hours worked per worker in period t and E t is the number of workers employed in that period. To achieve any particular level of Z1t, the firm can adjust Ht and Et. With the above definition of Z1t, the employment demand equation can now be rewritten as follows:

which gives the following:

Therefore, (11) 72

Our derivation of the above equation suggests that, in its econometric estimation, the coefficient of variable should be restricted to unity. However, in this study the

equation will be estimated without any restriction on the coefficient of this variable to control for differential preferences of employers for full-time and part-time employees who work different hours during a week. Hence, the coefficient of the variable

measures the elasticity of employment demand with respect to the weekly hours worked in the economy. A statistical test of significance will reveal if the coefficient of the variable differs from unity. A value higher than unity would imply that to achieve a certain growth in total hours worked by employees, firms rely heavily on employing more workers. A lower than unity value would imply the opposite. The elasticity value can also be negative, in which case workers and hours will be viewed as substitutes indicating that there is a tradeoff between employment and hours of work. When one reviews the trend in the relationship between employment and hours worked in Pakistan over time, an overall negative relationship is observed indicating a possibility of substitution between them (Figure 4.1 below). Figure 4.1: Growth Rates of Employment and Total Number of Hours Worked Per Week (1973-74 to 2007-08)
10 8 6 4 2 0 -2 -4 -6 -8

Growth rate (%)

1982

1984

1998

2000

1974

1976

1978

1980

1986

1988

1990

1992

1994

1996

2002

2004

2006

Emp growth

Hours growth

Source: Labour Force Survey (various issues). 73

2008

Hours worked are often manipulated in developed countries to affect the level of employment in the labour force. Caballero and Hammour (1998) found that between 1960 and late 1970s the average weekly hours per worker in France fell gradually from around 45 to 40 and further down to 39 hours per worker since 1981. The official purpose of this reduction was to increase employment by decreasing the unemployment rate in the country. Moreover, a fourth week of paid vacation was introduced in 1969 and a fifth week in 1981. The government elected in 1981 was committed to implementing the 35hour a week initiative by 1985. A further reduction in the work week to 32 hours was pursued in 1986 but this reduction was not then successful in expanding employment. Since the early 1990s, the French government encouraged reduced hours with reduced pay through agreements between employers and unions. Similar government policies have been implemented in Belgium. In Germany, the legal workweek has been shortened through union agreements. Such proposals were also under discussion in Italy. Erbas and Sayers (2001) also show that, during the period 1970-2000, it may have been possible to increase employment in the G-7 countries with a policy that combined a reduction in the workweek with an employment subsidy. Werner (1998) has discussed that in the US, employment and working hours grew in the same direction but not so in Europe. To date, evidence on the substitutability of hours worked and employment in developing countries is lacking. Therefore, inclusion of the hours worked variable in the estimation of employment demand model for Pakistan will provide useful information that is missing for a developing country. Another important variable that can affect employment, especially in a developing country, is its population growth. This is especially true in case of developing countries where the phenomenon of high population growth rates is well known. A rapid

74

growth in population results in a rapidly growing young population entering the labour force. As a result, levels of both employment and unemployment can change in the same or in opposite directions. In addition, as indicated by Walterskirchen (1999) and Beaudry and Collard (2002), a more rapidly expanding supply of labour can lead to more employment-intensive and lower productivity growth. Hence, in an econometric study on a developing country, it is important to disentangle the effect of GDP from the effect of population growth on employment. Others, such as Pasinetti (1981), have also used population variable to control for the effect of labour supply on employment. Based on the above discussion, the final employment demand equation which will be estimated econometrically in this Chapter is specified as follows: (12)

4.2 Sources of Data and Their Description: The period of analysis in this study spans over the years 1974 to 2008. Data on real GDP at constant factor costs (base year-2000) are obtained from the International Monetary Fund (IMF, 2009). Data on employed labour force are obtained from various issues of the Pakistan Economic Survey and are based on the Labour Force Survey (LFS) of Pakistan which is conducted annually by the Federal Bureau of Statistics. 59 All paid and self-employed individuals aged 10 years and above who worked at least one hour during the survey week are considered. 60

59 60

Pakistan Economic Survey is published annually by the Government of Pakistan, Ministry of Finance.

Paid employment includes persons with a job, at work and not at work. According to Pakistan Labour Force Survey at work are those "who during the reference week performed some work for wage or salary, in cash or in kind and persons with a job. Not at work are those who having already worked in their present

75

Population data are also obtained from the same sources as above and are estimates of population based on population censuses that were conducted in 1972, 1981, and 1998.61 The hours worked per week data are obtained from Federal Bureau of Statistics (1998) and are also based on the LFS. These data are frequency distributions reported for the reference week of LFS in intervals of weeks worked variable. Averages are calculated using these frequency distributions. As the LFS was not conducted in years 1976-78, 1980-82, 1984, 1989-90, 1996, 1999, 2001, 2003, 2005, there are some missing values. These values have been interpolated using a linear interpolation method which is inbuilt in EViews 7.0 Beta software. 62 Majid (2000) notes that an important weakness of any analysis on labour markets in Pakistan is the internal limitation of the data produced on population, labour force and employment. The 1998 population census was conducted after seventeen years. Although LFS is conducted annually, population coverage of the LFS is known to be low, with the result that both participation and employment rates based on LFS are underestimated. 63

job and temporarily not at work during the reference week due to some reasons, such as absence, leave, illness and strike etc, but had a formal attachment to their jobs".
60

Self-employment includes four types of persons: "(1) persons at work, (2) persons not at work but with an enterprise at work, (3) persons who during the reference week performed some work for profit or family gain, in cash or in kind, and (4) persons with a enterprise, a farm or a service undertaking who were temporarily not at work during the reference week due to some specific reasons, such as absence, leave, illness, and strike etc". According to the Right to Education Project, the minimum working age in Pakistan is 14 years (http://www.right-to-education.org ). However, the Labour Survey in Pakistan collects data on individuals aged 10 years and above.
61 62

The 1998 census was the latest available population census conducted in Pakistan.

The Linear interpolation method simply computes a linear approximation based on the previous nonmissing value and the next non-missing value. The interpolated value is then calculated as:
63

Since 2007, LFS is being conducted on a quarterly basis.

76

Data quality and coverage in developing countries are known to be low and Pakistan is no exception. It is important that in the absence of regular and better quality data, results of the present study be interpreted with care and be considered as indicative of a general trend. The best alternative to the published data used in this study are surveybased data for whose collection more time and resources are required.

4.3 Econometric Results of Employment Demand Model: Table 4.1 provides econometric estimates of employment demand model using the ordinary least squares (OLS) technique. Time series data for the period 1974 - 2008 are used to estimate the model.

Table 4.1: Regression Results of Employment Demand Model Variable Constant GDP (lnXt) Hours (lnHt) Population (lnPopt ) Time (t) ) R2 (adjusted) Durbin Watson (DW) Stat. No of Observations Coefficient t-statistic
7.455 0.257 -1.404 - 0.617 0.0255 0.9923 1.85 35 5.17 2.70 -5.38 -3.68 7.93

P-value
0.0000 0.0113 0.0000 0.0009 0.0000

Before interpreting the results of Table 4.1, it is important to discuss some of the econometric issues relating to the above estimation, their implications, and remedy. 64

64

A study by Ramsey (1969) and Ramsey & Alexander (1984) showed that the Regression Specification Error Test (RESET) could detect specification error in an equation which was considered a priori to be misspecified. According to this test, if the F test statistics is greater than the F critical value, we reject the null hypothesis that there is no specification error in the model. A Ramsey RESET was applied on the estimated model. F-stat was found to be 3.10 which is lower than h critical F-value of 4.18. Hence the hypothesis of no specification error cannot be rejected and the conclusion is reached that the model is correctly specified.

77

4.3.1. Econometric Issues Relating to the Time Series Estimation of Employment Demand Model:

There are three major issues relating to estimations based on time series data which can affect the reliability of econometric estimates. These issues include: nonstationarity of series, multicollinearity, and autocorrelation. Each of the above issues, their implications and related tests are discussed below. Stationarity of Data "A stochastic process is said to be stationary if its mean and variance are constant over time and the value of covariance between two time periods depends only on the distance or lag between the two time periods and not on the actual time at which the covariance is computed" (Gujarati, 2006, p.496). If data are non-stationary, regression estimates are likely to be spurious. Previous time series studies that have calculated employment elasticity in Pakistan using descriptive analysis did not test for data stationarity (for example, Baqai, 1979; Kemal, 1990 and Chaudhary & Hamid, 1998). In contrast, the present study has used the Augmented Dickey Fuller (1979) test (ADF test) to test for stationarity of data series used. The results are presented in Table 4.2. The null hypothesis being tested is that the data have a unit root.

78

Table 4.2: Unit Root Test Results (ADF Test) Level Variables Non
LnE LnX LnPOP LnH 6.80 3.37 -0.47 -0.91 Intercept 0.019 -1.19 -3.14** -2.46 Intercept &Trend -2.31 -1.65 -4.51* -3.74** Non -1.81*** -0.71 -7.66*

1st Difference
Intercept -5.24* -3.80* -7.69* Intercept &Trend -5.16* -3.91** -7.56*

Order of Stationary
I(1) I(1) I(0) or Trend stationary I(1)

Notes: The null hypothesis is that the data have a unit root. *, **, *** means level of significance at (1, 5, 10 %). E = employment, X = GDP, POP = population, H = hours worked per week and Ln = natural Logarithm.

As seen in the above results, the variables are stationary at different orders. First differences of the employment, GDP and hours worked variables appear stationary indicating that these variables are integrated of order one, i.e., I(1). On the other hand, the test results show mixed results on the population variable depending on whether the data are assumed to have intercept and/or trend. Moreover, first differencing does not make the data stationary. It is likely that it has a trend stationary process. To address the concern that the non-stationarity of the data may result in spurious or meaningless regression outcomes, a residual based co-integration test was also applied. Its results using the residual based ADF test are provided in Tables 4.3. If residuals of the employment demand model are found to be stationary, the series are co-integrated and results are not spurious.

79

Table 4.3: Residuals Based Co-Integration Test Results (ADF)

t-Statistic Augmented Dickey-Fuller test statistic Test critical values:* 1% level 5% level 10% level -5.783743 -5.6532 -4.8601 -4.4749 Prob. 0.0000

* The Null hypothesis indicates non-co-integration. The critical values are calculated by using the software provided by James G. MacKinnon http://www.econ.queensu.ca/faculty/mackinnon/jbes/. See MacKinnon (1994).

The above co-integration test result shows that residuals of the estimated regression, i.e., the employment demand model, are stationary in their levels. Hence, the series are co-integrated so that a linear combination of the series is converging in the long-run, indicating that econometric results of the employment demand model are not spurious. Autocorrelation Autocorrelation occurs if the error terms of the OLS regression based on the time series data are successively correlated. Presence of autocorrelation causes the OLS estimates to be inefficient. The detection of autocorrelation is based on Durbin and Watson (DW, 1950) test statistic. As was shown in Table 4.1, the calculated DW statistic of 1.85 is close to the critical value of 2, implying no auto-correlation in the error terms. Hence OLS estimates are efficient. Multicollinearity The issue of multicollinearity arises when the independent variables are correlated with each other. In the presence of high multicollinearity, the variances of the OLS

80

coefficients are high which can result in lower t-values, thereby leading one to wrongly conclude insignificance of a variable in the regression. The following correlation matrix (Table 4.4) shows some sign of multicollinearity in the model. 65 However, all variables in the regression model are found to be statistically significant at 0.05 level of significance as revealed by the t values that exceed the critical value of 2 (Table 4.1). Hence, following Kennedy (1998), multicollinearity is not viewed as affecting the interpretation of regression results. 66 Kennedy (1998) has also noted that multicollinearity is a common problem in estimations based on Cobb-Douglas production using time series data.

Table 4.4: Correlation Matrix (Employment is Dependent Variable)

Variables Employment (LnE) GDP (LnX) Hours (LnH) Population (LnPop) Time Trend (t)

Employment (LnE) 1.0000 0.9850 -0.7931 0.9823 0.9912

GDP (LnX)

Hours (LnH)

Population (LnPop)

Time trend (t)

1.0000 -0.7398 0.9974 0.9927 1.0000 -0.7320 -0.7361 1.0000 0.9937 1.0000

Source: E-views output on correlation matrix.

4.4. Interpretation of Econometric Results: We now turn to the interpretation of the econometric results of employment demand model reported in Table 4.1. Since the model is run in log-linear form, each coefficient is the elasticity of employment with respect to the corresponding variable
65

Multicollinearity was also confirmed when the R-square of a regression of one independent variable on others was found to be higher than the R-square of the original regression of employment demand.
66

Multicollinearity does not depend on any theoretical relationship among any of the regressors; rather it depends on the approximate linear relationship in the data set at hand.

81

except t. All t-values indicate that the elasticity of employment with respect to each independent variable included in the model is statistically significant at 0.05 percent level of significance. The main variable of interest is the GDP variable. It has the expected positive sign. Hence, GDP growth causes employment growth. However, the magnitude of the coefficient of log of GDP variable is only 0.257, indicating an inelasticity of employment with respect to GDP. Over the period 1974 - 2008, a one percentage change in annual GDP growth caused employment to increase by only about 0.26 percentage on average, keeping the effect of other variables in the model constant. Hence, economic growth in Pakistan appears to be largely focused on projects that did not generate employment over the period. The negative value of the elasticity of employment with respect to hours worked per week reflects a trade-off between employment of workers and hours of work in the labour markets of Pakistan. The absolute value of this elasticity is 1.404 and the t- test indicates that this value exceeds unity in the true model, thereby indicating a significant trade-off between employment and hours worked, keeping the effects of other variables in the model constant. 67 The population variable also has a negative and statistically significant coefficient. A Wald test of statistical significance of the difference between two coefficients revealed that the difference between the coefficients of population and GDP variables is

67

The calculated t-value for this test obtained as ( -1) / s.e ( ) was equal to -9.212 which in absolute terms exceeds the critical t-value of 2 at 0.05 level of significance.

82

statistically significantly different from zero. 68 Hence, the two coefficients differ from each other in the true model. The magnitude of the population variable is less than unity, at -0.617. This result implies that the population variable has its own effect on employment rather than through its effect on GDP per capita in Pakistan. The negative impact of population in the model is contrary to the suggestions of Walterskirchen (1999) and Beaudry and Collard (2002) that a more rapidly expanding labour supply can lead to more employment-intensive and lower productivity growth in a developing country. The lower population growth and expansion of post-secondary education in recent years, and continued availability of capital are all sources of productivity growth that may have resulted in a negative effect of population on employment demand. An investigation of these causes is beyond the scope of this study and can be pursued in future research. Based on the theoretical discussion of Chapter 3, the time trend variable captures the effect of changes in the capital stock and of technical progress. Its estimated positive sign and low magnitude probably suggest a low complementarity between capital and labour and a lack of employment intensity of technical progress in Pakistan, keeping the effects of all other variables constant in the model. A high value of the constant term indicates a large effect of all excluded

variables on the employment demand over the period. These variables could be the domestic political and economic conditions, changing economic policies, worker training, etc. When more data are available, a future study can investigate the effects of these variables.

68

The value of F-statistic in the Wald test was found to be 12.38 with degrees of freedom 1 and 30 in the numerator and denominator, respectively. The critical value of F-statistics at 0.05 percent level of significance is 4.17.

83

4.4.1 A Comparison of the Threshold Level of GDP Growth Rate with Actual Annual GDP Growth Rate:

Econometric estimation of the employment demand model yielded an elasticity of employment demand with respect to GDP as 0.26. Thus, over the period 1974 - 2008 a GDP growth rate of one percent was responsible for only a 0.26 percent growth rate in employment. One can use this elasticity estimate to calculate a threshold rate of GDP growth, i.e., the minimum GDP growth rate required in a given year to achieve a positive growth rate in employment during that year. The method of this calculation is discussed as follows.

Calculation of Threshold Level of GDP Growth The augmented employment demand equation, Equation 12, whose estimation was provided above, is re-written below: (12)

Taking the first-order derivative of the above equation with respect to time, we obtain the following equation: = + (13)

where

= (lnEt ) / t is the annual growth rate of employment in an

economy during year t. Similar to employment growth rate, annual growth rates in all other variables are obtained by taking their respective time derivatives. Thus, real GDP growth rate, is the

is the growth of hours worked per week in an economy, while

is the growth rate of population. From equation (13), we can estimate the GDP

84

growth rate below which employment will be stagnant, ceterus paribus. This growth rate will be called the threshold level of GDP growth rate, Kwiatkowski et al (2004) also

based his calculation on the same method to calculate threshold level of GDP growth for 48 countries of Western, Central, and Eastern Europe and some selected countries which were parts of the former Soviet Union. 69 Letting =0, one can derive as follows:

(14)

The expected signs of

are negative and those of

are positive.

The GXt* is calculated at average values of GH t, and GPOPt. Another simple calculation of threshold GDP has been suggested by Islam (2004). The threshold GDP growth is obtained as a ratio of labour force growth and employment elasticity. This ratio provides the GDP growth desired in order to absorb its surplus labour.70 Now we consider the results of estimated employment demand equation LnEt = 7.455 +0.257 LnXt - 1.405 LnHt - 0.617 Ln Popt + 0.026 t

Based on the method discussed above, the threshold level of GDP growth rate for Pakistan, calculated at the average values of annual growth rates in hours worked and

69

Their period of analysis was 1990-2000.

70

For example, if labour force growth is 2 per cent per annum and the employment elasticity is 0.5, a GDP growth rate of 4 per cent would be required merely to absorb annual increments in the labour force. Thus, in order to have an employment growth that is enough to absorb the economys backlog of the unemployed and surplus labour, the required GDP growth would have to be higher than four percent.

85

population over the estimation period using the formula provided above, is found to be 6.03 percent. In other words, Pakistans economy must grow at least by 6.03 percent in a given year to generate any employment in that year, while keeping the effects of growth in other variables in the model constant. A 99 percent confidence interval constructed around this value indicates that the range of this threshold is 5.75 to 6.32. The average annual growth rate of GDP in Pakistan during the entire period of analysis, i.e., 1974 2008 was around 5.28 percent which lies outside the 99 percent confidence interval constructed for the threshold level. Hence, it may be concluded that the average annual growth rate in Pakistans economy has fallen below the level necessary to generate jobs in the economy. Another more direct way of calculating the threshold level of GDP growth rate, as noted above is to consider the rate required to absorb the surplus labour in the economy which its growing labour force generates. 71 With the current annual labour force growth rate of 2.8 percent in Pakistan and an elasticity of employment demand value of 0.26 with respect to GDP, as computed in this study, the GDP must grow at 11 percent to absorb all new labour. This method of calculating threshold GDP also leads to the same conclusion: the annual growth rate of Pakistans eco nomy has not been sufficient to generate employment. The value of threshold growth rate in Pakistans GDP is close to that reported for Pakistan by Amjad (2007) who found the same to be in the range of 7 to 8 percent. In addition, the 2008 issue of Pakistan Economic Survey, published by the Government of Pakistan, also reported the threshold level of GDP growth rate to be in the range of 6 to 8 percent. During the period 1974 - 2008, the annual employment growth rate in the

71

Islam(2004)

86

country has been 2.65 percent while the labour force growth has been 2.8 percent. With a 13 percent unemployment growth rate, the employment in the economy must grow at a rate higher than 2.8 percent if all new entrants into the labour force are to be absorbed. A review of the literature on other countries, as provided in Chapter 2, showed the threshold levels of GDP growth rates for Poland to be 4.38 percent, Sweden 2.58 percent, and Finland 2.61 percent. Most of the other European countries had threshold levels between 1 and 2 percent (Wolnicki et al, 2006). Hence, the rate calculated for Pakistan falls within the range reported for other developing countries.

4.5 Employment and Economic Growth Relationship during Periods of Slow and Fast Economic Growth: The analysis conducted so far in the present study has shown a weak relationship between GDP growth and employment growth in Pakistan. It has also shown that on average, the annual GDP growth was slower than that required for employment generation. Based on literature discussed in Chapter 2, the response of employment to economic growth can vary during periods of fast and slow economic growth. One would expect economic growth to be more employment intensive during periods of fast economic growth, when more demand is created in the economy and businesses are expanding, than when economic growth is slow. The empirical analysis carried out in this Chapter can be extended to study the employment generation during periods of fast economic growth in Pakistan experienced over the 1974 2008 period.

87

To proceed with the analysis, the trend growth in GDP is first filtered out to isolate cyclical movements. Irregular shocks are then removed from these movements and periods of fast and slow economic growth identified as periods of rising and falling growth in real GDP, respectively, which reach the peak and trough in the interim. Many methods of detrending an economic time series and identifying the turning points of its cyclical component have been suggested in literature. In the present study, the most common method, the Hodrick- Prescott (1997) method is used. 72 The method is described in Appendix II. Results are reported in Table 4.5. 73

Table 4.5: Periods of Economic Growth in Pakistan74

Cycle
1949-69 1969-1991 1991-2008

Slow Growth
1949-50 to 1959-60 11 years 1969-70 to1978-79 10 years 1991-92 to 2001-02 11 years

Fast Growth
1960-61 to 1968-69 9 years 1979-80 to 1990-91 12 years 2002-03 to 2007-08 6 years

Source: Based on authors own calculations using real GDP data at 1999-2000 prices.

To analyze the behaviour of employment during the two periods of economic growth, the employment demand model is re-estimated by adding a slope dummy

72 73

For details of HP methodology, please see Appendix II.

Arby (2001) also used the same method to identify fast and slow economic growth periods during 19492001. He used real GDP data at prices of 1980-81.The present study has used these data until 2008 at prices of 1999-2000.
74

For the purpose of this study, a period of slow economic growth is considered to be the one of declining de-trended annual value of GDP, free of irregular shocks while a period of fast growth is the opposite, Some economists, for example Arby (2001), have called these two periods as periods of recession and recovery which is contrary to their conventional definitions which are based on actual real GDP.

88

variable that identifies the periods of fast economic growth in Pakistan during the period of analysis (1974 to 2008). The dummy variable has been interacted with the GDP variable. The modified form of employment demand equation is written as under: (15)

Where, D = 1 in fast growth period = 0 in slow growth period All other variables enter the regression model as before. The joint slope coefficient is interpreted as the elasticity of

employment with respect to output during the period of fast economic growth. OLS results of the modified model are provided in Table 4.6.

Table 4.6: Econometric Results of Employment Demand Model Incorporating the Effects of Slow and Fast Economic Growth Coefficient Variable Constant 7.1373 GDP (lnXt) 0.1148 Hours (lnHt) -1.0724 Population (lnPopt ) -0.3804 Time (t) 0.0263 ) Interaction variable (Base: Slow economic growth) * D GDP (lnXt) R2(adjusted) DW Statistic No of Observations Note: * Based on Table 4.5.
0.0017

t-statistic
5.6241 1.2077 -4.2503 -2.2992 9.2698

P-value
0.0000 0.2369 0.0002 0.0289 0.0000

3.1489

0.0038

0.9941 1.63 35

89

Table 4.6 results show that all variables, except for the GDP coefficient, are statistically significant and have the same signs as in the previous estimation. The GDP coefficient now provides the elasticity of employment with respect to output during the period of slow economic growth. The statistical insignificance of this elasticity value indicates no effect of GDP growth on employment growth during period of slow economic growth. . However, the effect is statistical significant but only slightly higher during the period of fast growth. The employment elasticity rises to only about 0.117 when the economy is expanding rapidly, which is even lower than the average value found for the entire period of analysis.

4.5.2 Threshold Level of GDP Growth in the Fast Economic Growth Period:

The modified employment demand model results during the period of fast economic growth, based on Table 4.6, is written as under: LnEt = 7.137 +0.117 LnXt - 1.072 LnHt - 0.38 Ln Popt + 0.026 t From the above equation, the threshold level of annual GDP growth, i.e., the one below which employment growth is zero, is estimated to be 7.91 percent. In other words, employment will be generated during the period of fast economic growth if the economy grows at a rate faster than 7.91 percent which is higher than the value found for the overall period. The 99 percent confidence interval for this threshold is found to be 7.66 to 8.23 percent. The actual rates of GDP growth during the two fast growth periods (197990, and 2002-2007) were 6.5 percent and 6.4 percent, respectively, which were below the threshold values found for fast growth period 75. Hence, it may be concluded that over the

75

The average growth rates for the two periods are based on various issues of Pakistan Economic Survey.

90

period 1974 to 2008, the economic growth in Pakistan was jobless even during the period of fast economic growth. 4.6 Summary: The analysis conducted in this chapter reveals that economic growth in Pakistan cannot be viewed as contributing towards more employment in the economy . Overall, a 5.75 to 6.32 percent growth rate in GDP is required to achieve any growth in employment, but the GDP growth rate has generally been below this level. During fast economic growth rate period, the threshold level of GDP rises. Results of this Chapter may be questioned on the basis that, similar to any developing country, a large part of non-agricultural employment in Pakistan (about 67 percent) is found in the informal sector on which no data are available to perform the employment demand analysis. However, the sectoral component of the informal economy also varies (Appendix III, Table A5.3). Hence, if the analysis is conducted for different industrial sectors of the economy, the effect of informal sector on the results of present analysis can be narrowed. This is done in the next Chapter.

91

CHAPTER 5 Sectoral Analysis of Employment Demand in Pakistan

In Chapter 4, an econometric investigation of jobless growth in Pakistan was conducted using national data. It was concluded that economic growth in Pakistan has been jobless over the past four decades. The present Chapter extends the same analysis using separate data on seven sectors of Pakistans economy. As has been the case of many developing and developed countries, Pakistans economy has also experienced a structural shift in employment as employment shifted from agriculture towards manufacturing and then towards the service sector (Table A5.3). During 2007-08, the service sector accounted for most of the national output, with its GDP comprising about 53 percent of national GDP (Table 5.1). It has been argued by some development economists (Gupta and Singh, 2005; Karmakar, 2008; ) that due to low per capita incomes in developing countries, growth in the service sector is likely to be mainly jobless. This is because of the high income elasticity of demand for services. Hence, reliance of a growth policy on service sector to generate employment so that the benefits of growth are shared by a large section of the society will yield results that are unsustainable. The present study will also investigate the sensitivity of employment to economic growth in services sector of Pakistan. An estimation of employment elasticity for each sector of the economy can also suggest which sector of the economy needs more focus if the overall (national) response of employment to economic growth is to be enhanced.76 Finally, an analysis of sectoral data helps in reducing the sensitivity of national results to the presence of informal economy.

76

A simple illustration may be useful in indicating the magnitude of employment elasticity (and output growth) that may be desirable for an economy in order to quickly absorb its surplus labour. "With a labour force growth of 2.5 per cent per annum and an overall employment elasticity of 0.4, a GDP growth of 6 per cent would be required merely to absorb the annual additions (not including the already unemployed pool of

92

Table 5.1: Sectoral Shares of GDP and Employment in Pakistan (1950-2008, %) Sector Agriculture Industry Manufacturing Construction Electricity Services Transport Trade &finance All other services Share GDP Employment GDP Employment GDP Employment GDP Employment GDP Employment GDP Employment GDP Employment GDP Employment GDP Employment 1950
53.2 9.6 8 1.4 0.2 37.2 5 12.3 19.8 -

1960
45.8 60.47 15.5 15.39 12.4 13.6 2.5 1.44 0.5 0.35 38.7 24.14 5.7 2.04 13.5 7.6 19.6 14.5

1970
38.9 57.03 22.7 19.91 16.5 15.57 4.2 3.93 2 0.41 38.4 23.06 6.3 4.73 15.6 9.89 16.5 8.44

1980
30.6 52.65 25.6 20.32 17.5 14.66 5.1 4.92 3 0.74 43.8 27.04 6.8 4.73 16.8 11.94 20.3 10.37

1990
25.8 51.15 25.6 19.81 18.1 12.84 4.1 6.38 3.3 0.59 48.6 29.04 9.5 4.89 18.8 11.93 20.4 12.22

2000
25.9 48.42 23.3 18.03 17 11.55 2.5 5.78 3.9 0.7 50.7 33.55 11.7 5.03 21.2 13.5 18.3 15.02

2008
21.3 44.65 24.3 20.1 20.7 13.11 2.1 6.29 1.4 0.7 53.8 35.25 10.3 5.46 23.7 14.62 19.9 15.17

Source: Various issues of Pakistan Economic Survey. As shown in Table 5.1, while manufacturing sector has experienced a rise in its contribution towards GDP since 1960, its share in total employment dropped after rising in 1970. The share of employment in electricity sector remained almost stagnant during this period, although it too showed a rise in its GDP share. The rising GDP share of manufacturing accompanied with a decline in its employment share is the result of a bias towards capital intensity of production in this sector as has been noted in a recent study by Kalim (2009). State support for the use of

workers) to the labour force. On the other hand, if this hypothetical economy could achieve a high growth of its more labour intensive sectors (e.g., labour intensive manufacturers, construction, and services), the overall employment elasticity could perhaps be raised (say, to 0.6) and a lower GDP growth (say, of six per cent) could enable it to achieve the same objective" (viz., the absorption of surplus labour in modern sectors) (Islam, 2004, p.5)

93

capital, as identified in Mahmood, Ghani and Muslehuddin (2007) appears to be the main cause for this outcome. In Pakistan, hours worked per week by employees vary significantly by sectors (Table 5.2). In some sectors (transport and trade &finance) total number of hours worked per week is above 50 hours. On the other end, a considerable proportion of workers, about 28 percent worked 56 hours or more a week. About 14 percent of the employed persons worked less, while 86 percent more, than 35 hours a week - the duration representing full time employment. Some broad averages by sectors are provided in Table 5.2.

Table 5.2: Sectoral Hours Worked Per Week in Pakistan (Averages, 1974 to -2008)

Sector Agriculture Manufacturing Construction Electricity Transport Trade &finance All other services Total Pakistan Economy

1974 46.75 45.24 45.67 46.68 47.63 48.41 46.17 46.87

1980 46.19 46.56 46.98 47.33 49.10 48.41 46.51 46.64

1990 43.36 47.32 45.70 45.88 49.41 47.57 45.72 45.14

2000 45.14 46.81 44.28 43.82 51.37 48.71 44.30 46.28

2008 40.69 47.95 46.24 45.03 52.95 51.12 47.66 45.12

Source: 50 years of Pakistan in Statistics Volume-II (1947-1997) and Labour Force Survey (various issues).

The next Section 5.1 presents an augmented employment demand equation that will be estimated using sectoral data for the economy of Pakistan. The sectoral data are 94

pooled for the period 1974 to 2008 to estimate the employment demand equation. Section 5.2 discusses the econometric issues involved in the estimation of pooled cross sectiontime series data. Stationarity results of Pakistani data are discussed in Section 5.3 while econometric results are interpreted in Section 5.4 which also discusses the employment elasticity in each sector based on which threshold growth of GDP is obtained for each sector. Section 5.5 summarizes the results.

5.1 Augmented Employment Demand Equation Showing Sectoral Effects: The final employment demand equation that was estimated in the previous Chapter to obtain national elasticity of employment demand with respect to GDP is rewritten below:

In the present Chapter, sectoral time series data are obtained to estimate the above equation. Seven sectors of the economy are considered, including: Agriculture, Manufacturing, Construction, Electricity, Transport, Trade & Finance and All other services.77 Data are organized in a pooled form giving rise to 245 observations. The employment demand equation with sector dummy variables, Di, is written below:

77

The grouping of sectors was based on data availability. A detailed list of the composition of each sector is provided in Appendix III, Table A5.1.

95

where i represents 6 of the sectors listed above, excluding agriculture which is considered as the base. In the above equation, the dummy variable (Di) takes on a value of 1 for sector i and zero otherwise. The dummy variable is interacted with the GDP and hours worked variables. This interaction allows a direct test of significance of the differential impact of each variable on employment demand in relation to the base, or the seventh, sector. Agriculture is used as the base sector. Although over time, employment has shifted in Pakistan away from the agriculture sector, it still remains the major employer, with about half of the employed labour force working there.

5.2 Econometric Issues Relating to Estimations Based on Pooled Data: Pooled data consist of a time series for each cross-sectional unit in the data. In the econometric analysis of present Chapter, data are pooled for seven sectors of the economy for the period 1974-2008.
78

In the analysis of pooled data issues of data stationarity and

autocorrelation arise due to the time series component of data and the issue of heteroscedasticity arises due to the cross section component. To test for the presence of stationarity, the econometrics literature provides different tests such as, graphical, correllogram and unit root tests of stationarity. Different tests of unit roots have been proposed in recent econometric literature, for example, Levin, Lin and Chu (2002); Im, Pesaran and Shin (IPS, 1997); Breitung (2000); Fisher type tests using ADF and PP tests (Maddala and Wu, 1999, Choi, 2001, and Hadri,

78

Gujrati (2006) provides good discussion of econometric issues arising in estimations based on pooled data.

96

2000).79 In these six tests, some provide a common root, such as the Levin, Lin and Chu test; and Breitung and Hadri test. 80 Others provide an individual root , such as, Im, Pesaran and Shin (1997) test also called the IPS test; and Fisher type tests using ADF and PP tests (Maddala, 1999 and Choi, 2001).81 For the purpose of this study, the IPS test is more valid, because it tests for the overall stationarity of data as well as for the stationarity of each cross section series. A detailed discussion on IPS can be found in M.H. Pesaran (1997). Presence of autocorrelation and heteroscedasticity also render the OLS estimations inefficient. The Newey-West Heteroscedasticity and Autocorrelation Consistent (HAC) standard errors method corrects for the wrong standard errors and t values obtained in the presence of heteroscedasticity and autocorrelation.
82

Given the

large sample size in this study, exceeding 50 observations, the HAC estimation by Newey-West is appropriate (Gujrati, 2002). 5.3 Results of the Unit Root Test for Stationarity of Pooled Data: Results of the IPS test for the stationarity of variables used in the employment demand model of this Chapter are provided in Table 5.3.

79 80

For a detail discussion on Unit root tests see Wooldridge (2002) and Baltagi (2008).

Common root indicates that tests are estimated assuming a common autoregressive (AR) structure for all of the series in a pooled.
81

Individual root is used for tests which allow for different autoregressive (AR) coefficients in each series of the pooled.
82

Newey and West (1987)

97

Table 5.3: Unit Root Test for Employment, GDP and Total Number of Hours Worked Per Week
Im, Pesaran and Shin (IPS) Cross sections Agriculture Manufacturing Construction Electricity Transport Trade &finance All other services Overall Unit root Im, Pesaran and Shin (IPS) Cross sections Agriculture Manufacturing Trade &finance Construction Electricity Transport Trade &finance All other services Overall Unit root Im, Pesaran and Shin (IPS) Cross sections Agriculture Manufacturing Construction Electricity Transport Trade &finance All other services Overall Unit root Intercept Statistics Prob
-1.0849 -3.7833* -1.1431 -1.1563 -1.5707 0.2340 -1.4664 0.02448 0.7095 0.0070 0.6862 0.6773 0.4863 0.9707 0.5355 0.5098

Employment Level Intercept Statistics Prob

0.3833 -2.1173 -1.1507 -6.9106* 0.4925 0.4944 0.4628 0.72681 0.9789 0.2394 0.6835 0.0000 0.9836 0.9838 0.9820 0.7663

1st difference Intercept Statistics Prob

-4.5872* -5.0762* -6.8064* -5.6091* -4.5943* -5.7551* -6.5476* -11.5930* 0.0009 0.0002 0.0000 0.0001 0.0010 0.0000 0.0000 0.0000

Intercept &Trend Statistics Prob

-3.1968*** -3.1903*** -2.1451 -4.4814* -5.7098* -3.4772** -2.0066 -4.07946* 0.1030 0.1033 0.5034 0.0073 0.0002 0.0591 0.5717 0.0000

Intercept &Trend Statistics Prob

-4.5637* -5.0205* -6.7426* -5.6879* -4.5971* -5.7196* -6.0412* -10.1431* 0.0051 0.0016 0.0000 0.0003 0.0049 0.0003 0.0002 0.0000

GDP@ Intercept Statistics Prob

-0.5455 -0.3907 -1.9597 -1.5420 -1.8597 0.0088 -1.9993 0.98077 0.8697 0.8992 0.3023 0.5006 0.3465 0.9530 0.2858 0.8366

Level Intercept &Trend Statistics Prob

-2.5308 -2.3809 -2.5476 -0.1667 -1.3559 -1.6634 -2.4748 0.82279 0.3123 0.3815 0.3049 0.9913 0.8558 0.7454 0.3377 0.7947

1st difference Intercept Intercept &Trend Statistics Prob Statistics Prob

-7.1211* -2.859*** -4.5025* -5.2016* -5.2027* -4.8005* -6.3837* -10.7682* 0.0000 0.0615 0.0011 0.0002 0.0002 0.0005 0.0000 0.0000 -7.0919* -2.7757 -2.8312* -5.7358* -5.5456* -4.7162* -6.3551* -8.7656* 0.0000 0.2158 0.1995 0.0002 0.0004 0.0034 0.0000 0.0000

HOURS Level 1st difference Intercept &Trend Intercept Intercept &Trend Statistics Prob Statistics Prob Statistics Prob
-1.9714 -4.6181* -4.3002* -4.5721* -4.5944* -0.8601 -3.5782** -4.2296* 0.5927 0.0040 0.0089 0.0045 0.0043 0.9489 0.0496 0.0000 -3.2601** -4.8932* -6.1293* -3.5457** -10.198* -8.6358* -3.2890** -11.8649* 0.0258 0.0004 0.0000 0.0146 0.0000 0.0000 0.0252 0.0000 -3.198*** -4.7936* -6.0135* -3.364*** -10.113* -9.1655* -3.354*** -10.6643* 0.1032 0.0031 0.0001 0.0784 0.0000 0.0000 0.0783 0.0000

(*, **, ***) means statistically significant at 1, 5, and 10 percent levels of significance, respectively. Hence, the series are considered stationary. Employment and hours worked per week data are stationary at I (0), while the GDP data are stationary at I (1). @GDP data are stationary at I(0) according to Levin ,Lin and Chu test (Test statistics: -2.02349 & Probability value is 0.0215).

98

The IPS unit root test shows the employment data in all sectors have different levels of stationarity. The data on employment are trend stationary in levels but in each sector it is stationary in first difference and also stationary in first differences with an intercept. GDP data are stationary in first differences in each sector and also in the overall pooled data. The data on total number of hours worked per week are stationary in levels, except in agriculture and trade &finance sectors. In the overall pooled data, this variable is stationary in levels. In summary, the IPS test results show that in pooled data, employment and number of hours per week data are trend stationary in levels but also stationary in first differences with an intercept and GDP data are stationary in first differences. To account for the non-stationarity of GDP data, a residual based co-integration ADF test will be performed after presenting econometric estimation results in the next Section. This test will indicate whether econometric results are spurious.

5.4 Econometric Results of the Augmented Employment Demand Equation: The augmented employment demand equation (equation 16) was estimated with pooled data. The Feasible Generalized Least Square (FGLS) method was used as it corrects for heteroscedasticity, present due to cross section component of data, and any contemporaneous correlation that may be present due to the time series component of the data. For robust variance, the methodology of Beck and Katz (1995), which is also called a Panel Corrected Standard Error (PCSE), is used. 83

83

According to Gujarati (2006), panel data are a special type of pooled data, which are also called longitudinal data or micro panel data, in which the same cross sectional unit, for example a family or a firm, is surveyed over time. All tests applicable to panel data are also applicable to the general form of pooled data.

99

Table 5.4 provides econometric results of employment demand equation using the FGLS approach.

Table 5.4: Econometric Results of Employment Demand Equation: FGLS Estimates Variable
Constant GDP (Xt) Hours (Ht ) Population (Popt ) Time (t) ) Coefficient 3.371023 -0.07635 0.635056 -0.58049 0.037442 t-statistic 3.066692 -1.77954 3.161675 -4.45186 11.04457 P-value 0.0024 0.0765 0.0018 0.0000 0.0000

Interaction variables (Base sector is agriculture)*

GDP(Mining & Manufacturing) GDP(Construction) GDP(Electricity) GDP(Transport) GDP(Trade &Finance) GDP(All other Services) Hours (Mining & Manufacturing) Hours (Construction) Hours (Electricity) Hours (Transport) Hours (Trade &Finance) Hours (All other Services) R
2

0.105492 0.565023 0.287625 0.244561 0.285162 0.372586 -0.72916 -2.29621 -2.00685 -1.39097 -1.32689 -1.62049 0.983237 245

1.970137 16.32795 3.07024 5.650384 7.381531 7.561574 -3.9752 -20.6644 -7.12157 -9.55886 -9.97152 -9.57022

0.0500 0.0000 0.0024 0.0000 0.0000 0.0000 0.0001 0.0000 0.0000 0.0000 0.0000 0.0000

No of Observations

* The interaction variable in a sector is the product of its dummy variable with the corresponding variable (GDP and hours worked per week). All variables, except for the time trend variable, are entered in log form.

Before interpreting the econometric results, we address the concern that the nonstationarity of the GDP data, as found above, may result in spurious or meaningless regression outcomes. A residual based co- integration test was applied. Its results using ADF are provided in Table 5.5. If residuals of the employment demand model are found to be stationary, the series are co-integrated and results are not spurious. 100

Table 5.5: Residuals Based Co-Integration Test Results (ADF)

t-Statistic Augmented Dickey-Fuller test statistic Test critical values:* 1% level 5% level 10% level -11.14020 -5.6532 -4.8601 -4.4749 Prob. 0.0000

* The Null hypothesis indicates non-co-integration. The critical values are calculated by using the software provided by James G. MacKinnon http://www.econ.queensu.ca/faculty/mackinnon/jbes/. See MacKinnon (1994)

The above co-integration test result shows that residuals of the estimated regression, i.e., the employment demand model for sectoral analysis, are stationary in their levels, i.e, I (0). Hence, a linear combination of the series is co-integrated indicating that the econometric results of the employment demand model are meaningful and variables of the model are co-integrated in the long run. We now turn to the interpretation of the econometric results which were reported in Table 5.4. The high value of R-square suggests that the regression model is a good fit. Values of t-statistics suggest that all coefficients are statistically significantly different from zero. This is true for most variables at 0.05 percent level of significance, but for the GDP variable it is true at 0.10 percent level of significance. Since all variables are entered in log form, except for the time trend variable, the coefficient of each variable is the elasticity of employment with respect to the corresponding variable. Coefficients of interaction variables measure by how much each sectors elasticity of employment with respect to the corresponding variable differs from that in the base sector which, as before, is the agriculture sector. All such coefficients are statistically significantly different from

101

zero indicating a statistically significant difference between the elasticity of employment in each sector from that in agriculture sector. In each of the included sector, the elasticity of employment demand with respect to GDP is higher than it is in agriculture. However, the opposite is true when employment elasticity with respect to hours worked per week in each sector is compared with that in agriculture. Since the absolute values of the negative coefficients of each dummy interaction variable with hours are higher than the positive elasticity in agriculture, they all have negative employment demand elasticities with respect to hours. This result perhaps indicates that employers in those sectors substitute between the number of workers and their working hours, but not so in agriculture. The main elasticity of interest in this study is the elasticity of employment demand with respect to GDP. Based on the results of Table 5.4, this elasticity value is computed for each sector and presented in Table 5.6.

Table 5.6: Sector Wise Elasticities of Employment with respect to GDP in Pakistan Sectors
Agriculture Mining &Manufacturing Construction Electricity Transport Trade &finance All other services Elasticity

-0.076 0.029 0.488 0.211 0.168 0.208 0.296

Source: Based on Table 5.4. Each sectors elasticity is obtained by adding the coefficient of its dummy interaction variable with GDP to the coefficient of GDP variable.

102

As Table 5.6 results indicate, the agriculture sector has a negative elasticity of employment demand with respect to GDP.84 A one percent output growth in agriculture causes a negative employment growth in agriculture of 7.6 percent. In other words, employment demand in agriculture sector of Pakistan drops when its output grows. A possible reason for this result could be the presence of surplus labour in agriculture sector which has resulted in a negative marginal product of labour in this sector. 85 Future research can investigate this issue in more detail. 86 The elasticity of employment demand with respect to GDP is positive in all nonagricultural sectors. Hence, output growth in non-agricultural sectors appears to have an employment-enhancing effect. The strongest response of employment to output growth is found in the construction sector, while it is the least in the combined sector of mining and manufacturing. Given the above estimates of employment demand elasticities, it is clear that for employment generation, public policy in Pakistan can rely upon economic growth in only non-agriculture sectors of the economy. However, how much output growth is needed in each sector before it begins to employ additional workers? To answer this question, one must calculate the all important threshold levels of GDP growth for each sector whose calculation was discussed in Chapter 4. Results of this calculation for each non-

84

That means higher GDP will lead to lower agricultural employment. This, in essence, is in line with the structural change theory proposed by Chenery and Syrquin (1975) which suggests that "increases in agricultural GDP will have two counteracting influences on agricultural employment. On the one hand, the expansion of the agricultural sector will boost employment in that sector, but on the other hand, the expansion of the economy as a whole could also decrease agricultural employment as workers reallocate their services to non-agricultural activities".
85

The surplus labour appears to have pushed agriculture production to the inefficient third stage of production.
86

Kapsos (2005) has estimated elasticity of employment in agriculture for Nepal, Thailand and Indonesia for the period 1991-2003 and has found that it has turned negative in recent years. Perugini and Signorelli (2007) also reported similar finding for Italy, noting that manufacturing and agriculture sectors have low employment elasticity (even negative in agriculture) as compared to services sector.

103

agricultural sector are provided in Table 5.7. The same Table also provides the actual output (GDP) growth in each sector for comparison.87

Table 5.7: Threshold and Average Annual GDP Growth Rates in Non-Agricultural Sectors of Pakistan Sectors
Mining &Manufacturing Construction Electricity Transport Trade &finance All other services Threshold growth rate in GDP* Average Annual Growth Rate of GDP (1974 - 2008)

19.231 1.287 1.968 4.620 3.142 2.224

5.759 3.819 3.857 6.375 6.236 5.349

*Calculated as the GDP growth rate below which employment growth will be zero.
Source: Calculations by the author based on results of Table 5.4. The method of calculation was discussed in Chapter 4.

Results of Table 5.7 show that the combined sector of mining and manufacturing in Pakistan had the highest level of threshold GDP growth rate during 1974 - 2008. GDP must rise by about 19 percent in this sector to generate employment. The actual average rate of growth of GDP in manufacturing has been about 6 percent during 1974 - 2008. Hence, on average, GDP growth in the mining and manufacturing sector was not enough to help generate employment growth.88 However, GDP growth rates in all other sectors were above their threshold levels.

87

Complete sectoral average growth rates of employment and GDP can be found in Appendix Table A5.4.
88

Mazumdar and Sarkar (2007) also found GDP growth rate to be jobless in Indian manufacturing.

104

Less than half a percentage of national employment in Pakistan is in the mining sector. The large threshold growth rate in the combined sector of manufacturing and mining is an indication of the strong employment barrier faced in the manufacturing sector which is viewed as a high capital intensity sector. One of the major factors considered responsible for high capital intensity of manufacturing production is the factor market distortion in favor of capital caused by deliberate policy measures aimed at attracting domestic and foreign investment.89 This is considered as a root cause of low employment in the large-scale manufacturing sector of Pakistan by Hussain (1974); Kemal (1981); Zahid, Akbar and Jaffry (1992) and Kalim (2009). The last two studies also report excess capacity in the manufacturing sector which implies a potential for job creation in this sector.

The mining and manufacturing sectors together employ about 13 percent of employed labour force in Pakistan and their share in national GDP is about 22 percent.90 Hence, these two sectors together occupy an important position in the economy of Pakistan. The combined growth rate in these two sectors was rapid, 5.7 percent over the period of analysis. In some years the growth in this sector was even in double digit. Any growth in this part of the economy can have a direct as well as an indirect effect on

89

Some of these measures include low (0-5 percent) import duty on machinery, no withholding tax on imported machinery, no sales tax on imported machinery, a debt to equity ratio of 70:30 on foreign and 80:20 on local machinery. These and further details have been provided by Shah and Ahmed (2003) and also on the following web site: www.fauji.org.pk/webforms/InvestmentClimatePak.Mahmood, Ghani and Muslehuddin (2007) maintain that while industrial and trade policy reforms in recent years have exposed domestic enterprises to greater internal and external competition, most of these enterprises continue to seek government support.
90

Anwar (2004) pointed out that "growth in large scale manufacturing sector is mainly due to the utilization of excess capacity ranging from 30-40 percent created by large investments in the mid 1990s in thermal power generation through independent power projects (IPPs), cement, sugar, automobile and consumer electronics. Furthermore, employment elasticity of large scale manufacturing sector is very low (0.02) relative to other sectors. Thus, this pattern of growth, together with stagnant investment, does not seem to be pro-poor since it is not likely to generate sufficient employment to offset the large increases in labour force over the years".

105

employment in other sectors. The direct effect can come from the additional demand for supporting goods and services produced in other sectors. The indirect effect can come from the additional demand for goods and services which results from additional income that accompanies any combined employment generation in the mining and manufacturing sectors. Manufacturing sector also has spillover effects on other sectors of the economy through technical progress (Gupta and Singh, 2005).91 The economy wide result (reported in Chapter 4) that the average annual GDP growth rate in Pakistan has been jobless can be explained by the jobless growth in the mining and manufacturing sectors combined. Future research should explore any multiplier effect of growth in mining and manufacturing sectors on employment growth in other sectors of the economy.92

5.5 Summary: Econometric analysis of sectoral data conducted in this Chapter reveals that output growth in agriculture sector of Pakistan, the most dominant employer, cannot be viewed as contributing towards more employment in the economy. This sector appears to have absorbed so much labour that labours marginal product in it has become negative. Any future employment generation in Pakistan will have to rely on non-agricultural growth. In the non-agricultural sectors, although the mining and manufacturing sectors (combined) grew at a rapid rate, this rate was not enough to generate employment largely due to high

91

Another important way in which manufacturing benefits the whole economy is through its role in international trade and in balance of payments. This is because of the fact that a large part of international trade takes place in manufacturing products.
92

Kaldor (1967) point out that manufacturing sector is an engine of economic growth because of its multiplier effects on other sectors.

106

capital intensity of production. This appears to be the main cause of the overall jobless growth in Pakistan that was concluded in the previous Chapter. Output growth in all other sectors generated a positive employment growth.93

93

Mazumdar (2003) acknowledges the importance of employment generation in the manufacturing sector in a developing country which also highlights the importance of a sectoral analysis undertaken in the present study. He first stressed the centrality of manufacturing output growth for total economic growth in developing countries, based on the Kaldors law of economic development. However, if the elasticity of employment with respect to output is low, the economy can end up with an enclave type of development in which the impact of even a respectable rate of manufacturing growth has only a limited effect on the rest of the economy (p.564).

107

CHAPTER 6 Why Jobless Growth in Pakistan

Detailed econometric investigations conducted in the previous two chapters revealed an overall weak sensitivity of employment growth to economic growth in Pakistan. Over the period 1974 to 2008, a GDP growth rate of one percent contributed to only a 0.26 percent growth rate in employment when national data were used. The national annual threshold level of GDP growth rate was estimated to be 6.03 percent, i.e., for employment generation in any given year, Pakistans economy should have grown by at least 6.03 percent in that year. The average annual GDP growth rate in Pakistan during the entire period of analysis was around 5.28 percent lower than the estimated threshold GDP growth rate of 6.03 percent. Hence, it may be concluded that the average annual growth rate of Pakistans economy has fallen below the level necessary for job creation. . This result was found to be driven mainly by the mining and manufacturing sectors which had the highest level of threshold GDP growth rate at 19 percent while experiencing only a 6 percent average annual growth rate during the 1974-2008 period. GDP growth rates in all other sectors were above their threshold levels. In addition to the employment barrier present in the manufacturing sector resulting as a byproduct of policy measures to encourage the use of capital, several other important changes occurring in the economy may also be responsible for the slow response of employment to GDP growth. It is important to investigate those changes for an appropriate employment policy response. An in-depth investigation of these factors warrants a separate study. In the present Chapter, only a broad investigation is pursued.

108

Economics literature has identified several possible factors responsible for jobless growth. These are discussed in Section 6.1. The impacts of these factors in case of Pakistan are discussed broadly in Section 6.2 by using mostly descriptive tools of data analysis. The impact of macroeconomic policies is analyzed in the same Section using an econometric model. Section 6.3 presents a summary of findings.

6.1 Possible Factors Responsible for Jobless Growth:

Five major factors responsible for jobless growth can be identified from economics literature.94 These reasons are listed below, followed by their discussion: 1. Increase in labour productivity 2. Structural change in labour market (Sectoral Change ) 3. Monetary and fiscal policies 4. International trade and globalization (Outsourcing and Off-shoring) 5. Just-in-time employment practices The first four causes are broadly explored in this Chapter using available data. . Labour productivity, measured as the average product of labour or output per worker, rises when there are factors that suppress employment growth, for example technological changes, while output is expanding (Francis and Ramey 2003; Gong and Semmler 2006; Hemraj et .al 2006). Indeed, empirical research shows that strong labour productivity growth is not sufficient to solve problems of acute poverty or
94

Tezcek (2007) analysis of the Turkish economy concludes that the Turkish economy faces the problem of jobless growth due to three main reasons. One reason is Agriculture displacement (pre and post 1980s), the second is Structural adjustment policies (1980-2000) and the third is productivity increase in the new millenium. Hemraj et.al (2006) identified possible factors responsible for jobless growth in the US economy. Rada (2008) and Rutkowski et.al, (2005) identified such factors for Central and Eastern European countries.

109

underdevelopment (UN, 2008). If productivity rises faster than economic growth, the problem of jobless growth can arise (Indiresan, 2002; Tezcek, 2007; Rada, 2008). Structural change in the labour market, or the sectoral change, is said to occur in a labour market when there are changes in the composition of aggregate demand for goods and services, or when there are changes in the productivity of labour, that result in an industrial shift in labour demand. When the labour market is undergoing structural change, workers may lose jobs because their current skills are no longer in demand. Hence, if an economic growth is accompanied by structural change, there is a potential for this growth to be jobless. This possible explanation of jobless growth in the United States was first suggested by Aghion and Howitt (1994) and then later empirically studied by Rissman (1997), Groshen and Potter (2003) and Aaronson et.al (2004c).95 Monetary and fiscal policies can cause economic growth to be jobless if they increase the interest rate. Interest rate is defined as the opportunity cost of capital which can rise due to a monetary contraction, or due to a fiscal expansion that increases demand for loan-able funds in financial markets. A rise in interest rate reduces private investments. Hence, if an increase in aggregate demand is caused by an expansionary fiscal policy, private firms would attempt to meet this increase by restructuring to increase the productivity of existing labour and capital rather than hire new ones. These firms could also withhold their new investment plans in the anticipation of a rise in future tax burden that the government may impose in order to finance future interest payments. 96

95

Some studies have also examined the relationship between sectoral changes business cycles. These studies including Lilien (1982), Abraham and Katz (1986), Campbell and Kuttne (1996), Davis (1987), Loungani, Rush, and Tave (1990) decompose changes in labour productivity into a component due solely to changes within plants and a component due to sectoral reallocation of labour across plants and analyze the cyclical behavior of both components.
96

This outcome is predicted by the Ricardian equivalence rule as discussed by Barro (1989).

110

Expansion of international trade and greater economic integration of the world can also cause economic growth in a country to be jobless. Since the 1990s, the world has become more economically integrated because many countries have moved towards a uniform economic climate in which there is an increased role of markets in allocations of resources and of goods and services, there is an increased role of private sector and there are lower barriers to trade in resources and in goods and services. As a result, there is now a greater movement of goods, and factors of production between countries. Lowering of trade barriers and flexible economic policies in many developing countries has also enabled many manufacturers in developed countries to take greater advantage of lower labour costs in developing countries by outsourcing of production. This phenomenon has caught the attention of many economists who argue outsourcing to be a major source of jobless economic growth experienced by many developed countries since the 1990s. 97 Empirical research also suggests that the spread of new information technologies promotes increased international services trade which could be another source of outsourcing of economic activities to developing countries. Finally, jobless economic growth may also be attributed to just-in-time employment practices. Firms can decide when to hire, taking into consideration the cost associated with hiring too early or too late. More flexible employment practices would thus delay hiring, allowing firms to see when everyone else is going to hire before beginning their own hiring if they are uncertain about the strength of a recovery. This

97

Using the data on Ireland, Gorg and Hanley (2005) argue that countries adopt outsourcing when imports fall short of exports. Nevertheless, outsourcing remains a major and significant driver of the wage differential. Following Feenstra and Hansons (1996,) work, similar evidence has been produced for a number of European countries, e.g., Geishecker and Gfrg (2004) for Germany and Egger and Egger (2003) for Austria. These economists discuss the issue of outsourcing in their study and point out that the outsourcing is a cause of net employment decrease. (See,Egger, Pfaffermayr, & Weber (2003), Head and Ries (2002) , Kletzer (2000), Deardoff (2001) , Jones & Kierkowski (2001), Kohler (2001), Arndt (1997), and Krugman ( 1995)).

111

type of approach by firms can result in extended periods of jobless growth accompanying economic growth (Schreft and Singh, 2003). 6.2 An Investigation of Factors Responsible for Jobless Growth in Pakistan: An in-depth analysis of the factors identified in the above section warrants a separate study. In the present study, only a broad investigation of some of those factors will be conducted to provide some policy directions and directions for future research.

6.2.1 Labour Productivity Growth in Pakistan:

The time series data on growth rates of labour productivity, employment and GDP are presented in Figure 6.1 for the period 1965 to 2009. Labour productivity in this Figure is measured as the ratio of GDP and employment where employment is measured as the number of persons working.

Figure 6.1: Pakistan Employment, GDP and Labour Productivity Growth Rates (1974-2008)
14 12 10 Growth Rate 8 6 4

2
0 -2 -4 -6 Employment Growth Year GDP Growth Productivity Growth
1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Source: Pakistan Economic Survey and Labour Force Survey of Pakistan (various issues). 112

As observed in Figure 6.1, during most of the periods of high GDP growth rates in Pakistan, employment growth rate has been low. From 1970s until early 1990s, GDP growth rate was remarkably high in some years, reaching above 6 percent, while employment growth rate generally hovered around 2 percent in those years. However, in 1971 when the GDP growth rate fell close to zero, employment growth rate rose to approximately 4 percent. In early 1990s, the employment growth rate became negative while the GDP growth rate even exceeded 6 percent in one year. After the recession of early 1990s, the employment rate did shoot up for three years, but has been below GDP growth rate since then. As was seen in Table 1.1, in periods of high GDP growth rates, the unemployment rates have also been high indicating the inability of GDP growth to absorb new labor force entrants. During 1994-1996, the GDP growth rate increased continuously while the employment growth rate continued to decrease, reaching almost zero in 1996. Since 2000, employment and GDP have shown upward trends, but the direction of relationship between them has not been consistent, with employment growth rate exceeding GDP growth rate in 2007 but then dropping to under 1 percent. A calculation of the correlation coefficient between the two data series resulted in a value of -0.026 for the period. In sum, one can conclude a low and negative correlation between GDP and employment growth rates over the entire 35 year period. GDP growth rate equals the sum of changes in employment and labour productivity growth rates (Kapsos, 2005). Hence, if GDP growth rate rises while employment growth rate is falling, labour productivity growth rate should also rise and vice versa. From Figure 6.1, it is also observed that the growth rates of GDP and labour productivity followed identical paths in most of the period, while employment and GDP growth rates went in opposite directions. During 1970s, the average annual labour productivity growth rate was 1.73 percent while the average GDP growth rate was 4.84 113

percent - much lower than the average 6.87 percent growth rate of GDP experienced in the preceding decade. During the 1980s, an average 4.47 percent growth rate of labour productivity was accompanied by a 6.87 percent growth rate of GDP on average, which was the highest ever recorded during a decade in Pakistan. However, in the 1990s, the annual average annual productivity growth rate fell to 1.8 percent and exhibited large fluctuations. Some of the years also saw negative growth rates in labour productivity. The negative growth in labour productivity may be attributed to the employment generation schemes introduced by the civilian governments of Peoples Party and Muslim League, which caused employment to grow more rapidly. Lower productivity growth of 1990s was also accompanied by a fall in the average annual GDP growth rate to 3.98 percent. Finally, during the 2000-09 period, a 2.41 percent average annual growth rate of productivity, which was a significant improvement over the previous decade, was accompanied by a 5.03 percent average growth rate of GDP. During 1990-1993 and in the year 2001, the average annual productivity growth rate was higher than the corresponding GDP growth rate but employment growth rate was negative, implying that productivity growth rate was achieved at the expense of employment reduction. The correlation between GDP growth rate and productivity growth rate is found to be high, at around 0.78 during the entire period. When employment and productivity growth rates are compared in Figure 6.1, a negative relationship is found with a correlation coefficient of -0.67, demonstrating a trade-off between employment and productivity growth rates. These findings are consistent with the result that economic growth in Pakistan has been largely jobless. Further insight into the trade-off between productivity and employment growth rates can be obtained by investigating the relationship between working hours per person and the employment to population ratio. This is done in Table 6.1. 114

Table 6.1: Employment to Population Ratio and Total Number of Hours Worked Per Week Per Person Employed in Pakistan
Years 1973-74 1980-81 1985-86 1990-91 1995-96 2000-01 2005-06 2007-08 Employment /Population Ratio 29.75 30.11 27.4 26.39 24.69 25.33 26.85 28.49 Hours per Person Employed per Week 47.77 46.84 46.96 46.37 46.66 46.11 44.34 45.12

Source: Various issues of Pakistan Labour Force Survey.

Table 6.1 data show an overall negative relationship between the employment to population ratio and the hours per employed person per week. Hence, labour employment and hours worked may be viewed as substitutes. This finding of our descriptive analysis is consistent with the econometric result of Chapter 4 showing a negative employment elasticity of hours valued at -1.404 and is also similar to Jorgenson and Vu (2005), who reported that Pakistans productivity and GDP growth rates declined during 1995 -2003 as compared to during 1989-1995. An earlier study by Mahmood and Siddiqui (2000) had also arrived at the same conclusion for manufacturing industries. In summary, the descriptive analysis presented above shows that during the period 1974 to 2008, GDP and employment growth rates had a negative correlation, as did employment and labour productivity growth rates, while GDP and labour productivity growth rates had a positive correlation. Hence, increases in labour productivity may be viewed as one source of jobless economic growth observed in the econometric analysis of

115

data for the period 1974 to 2008. At least in the short- and medium-term, a tradeoff is observed between productivity and employment growth rates. The next section investigates structural change in the economy of Pakistan which can be another reason for jobless growth in its economy.

6.2.2 Structural Change in Pakistans Economy:

If economic growth is accompanied by structural changes in the economy, workers may have to retrain and update their skills. Hence, some unemployment may occur in the economy, when workers are switching their skills, thereby weakening the relationship between employment and GDP growth rates. Structural change of an economy is often proxied by employment shifts between industrial sectors. An in-depth analysis of structural change in the economy requires disaggregated employment data within each of the seven sectors listed in Chapter 5. Unfortunately, such disaggregated data are not available. Hence, only some broad patterns of sectoral change are analyzed for the present study. Four measures of structural change in the economy are commonly used in the literature. These measures have been proposed by Lilien (1982), Groshen and Potter (2003), Rissman (1997), and Aaronson, Rissman and Sullivan (2004c) and are discussed below. The measure proposed by Groshen and Potter (GP) is presented and discussed first. Because of the similarities between the other three measures, they are discussed after the GP measure.

116

Groshen and Potters (GP) Me asure Following Groshen and Potter (2003), industrial sectors that continue to experience slower than average employment growth during and after periods of slow growth as well as those sectors that continue to experience faster than average employment growth during and after slow economic growth can be considered as undergoing structural change. The statistic suggested by these authors, called the GP statistic, measures the percentage of such sectors in all sectors of the economy. In present study, Pakistans economy was divided into seven sectors (Chapter 5). Annual data are used for each sectors employment, although it will be preferable to use monthly data bu t they are not available. First, employment growth rate in each sector is compared with the average employment growth rate during different periods of economic growth as shown in Table 6.2.98 The signs of sectoral changes concluded in columns (11) and (12) are for the start of slow and fast economic growth periods as they are for one year after growth rate peaks and troughs. To account for any randomness in employment fluctuations that could affect employment growth in these years, employment changes at the peak and trough of economic growth are also compared and signs of sectoral change concluded in columns (9) and (10). To help the intuition of the reader, an example has been provided in the notes below Table 6.2.

98

When monthly data are used, GPs measure is based on a recession period that starts one month after the business cycle peak and an 11-month post-recession period that begins the month after the business cycle trough (Groshen and Potter, 2003).

117

Table 6.2: Employment Growth Rates by Sectors in Peak and Trough of Economic Growth, and in the Years Following them. Employment Growth Rates in Peak and Trough Peaks 196869 (1) 3.86 -0.21 2.42 2.54 -0.88 -2.73 -2.56 1.59 199091 (2) -2.33 -1.23 -0.67 9.49 3.69 2.75 2.83 -0.42 Troughs 197879 (3) 1.92 5.42 8.61 16.42 2.30 3.83 4.18 3.31 200102 (4) -2.99 8.12 3.20 6.70 7.18 4.92 4.64 1.64 Employment Growth Rates one year after Peak and Trough Peaks Troughs Effects of Sectoral Change In Peak and Trough 196869 (9) No No +ve +ve -ve No No 199091 (10) -ve No No +ve +ve +ve +ve 1year after Peak and Trough 1969- 199170 92 (11) (12) +ve -ve No No -ve -ve No -ve No +ve No +ve +ve +ve

Sectors Agriculture, forestry, hunting and fishing Manufacturing, Mining and quarrying Construction Electricity, gas, water and sanitary Transport and communication Trade and Finance All Others services Total Employment Growth Rate
Notes: 1. 2. 3.

196970 (5) 3.93 1.12 3.12 -3.39 -0.46 1.90 -4.00 2.29

199192 (6) -2.70 -5.13 2.57 11.02 3.75 3.36 6.28 -0.26

197980 (7) 2.84 -9.13 -2.29 5.57 -1.48 -2.83 30.02 2.79

200203 (8) -1.42 8.61 3.02 1.41 7.01 5.83 5.10 2.74

The year after each peak and trough represents the start of decreasing and increasing economic growth periods, respectively. Employment growth rates in peak and trough are also considered to cover for any possibility of randomness in the results based on slow and fast growth periods. For details, see Aaronson, Rissman and Sullivan (2004c). For agriculture, the positive effect in Column (11) is concluded as follows: Employment growth one year after peak (3.93) exceeded the average for all sectors (2.29) and so did the employment growth one year after the trough (2.84 and 2.79). Hence, a positive effect of sectoral change on employment in agriculture is concluded. No effect in Column (9) is concluded as follows: Employment growth in peak (3.86) exceeded the average growth for all sectors (1.59), but employment growth in trough (1.92) was slower than the average (3.31). Hence, no sectoral change is concluded. The same method is followed for other sectors.

118

As can be observed in Table 6.2, when one considers employment growth rates from one year into slow economic growth to one year into fast economic growth (one year after economic growth peaks and one year after it troughs as reported in columns 11 and 12), the effect of structural change in the economy appears somewhat stronger in the 1991-92 downturn of growth than it was in the 1969-70 downturn. During the 1969-70 downturn, a sectoral employment change was observed in four sectors while this was the case in five sectors during the next downturn. Electricity, water and sanitary sectors did not experience any structural shift, collectively. If one considers employment growth rates over the half cycle (in peak and trough as reported in Columns 9 and 10), then the structural shifts are found to be even less pronounced during 1969-70 as this occurred only in three sectors. However, the electricity, water and sanitary sector does show structural shift under this method. Only the transport and communication sector is found to have experienced structural shifts under both methods during both downturns. All other sectors show mixed results. 99 In conclusion, the analysis of structural change based on aggregated employment data for seven sectors of the economy yields some mixed results. However, it does appear that structural changes were more pronounced at the time of 1991 downturn of economic growth than in 1969-70 downturn. This result is largely due to significant shifts in employment from agriculture towards services sector as recorded in the GP method. The GP method also recorded significant changes in the transport sector. The Yellow Cab scheme introduced in the mid-1990s by the regime of former Prime Minister Nawaz Sharif, which made it easier for investors in the transport sector to import vehicles from

99

Groshen Potter (2003) also provided a descriptive statistic based on the correlations between the difference of employment growth rate in each sector from the national average before and after recession. This statistic will not be meaningful for present study due to small number of observations (only seven).

119

abroad, may have caused an expansion of this sector. When more disaggregated data are available, one can perform an in-depth analysis of employment shifts within each of the seven sectors to investigate if these shifts caused production in each sector to become more or less capital intensive. While one weakness of the above analysis is its use of aggregated data, another weakness is that it is based on only four data points which may not capture full fluctuations in employment during the period. Other measures such as those provided by Lilien (1982), Rissman (1997) and Aaronson et.al (2004c) are improvements over the GP measure. All three measures consider deviations of annual employment from a standard level, but differ in the measurement of this deviation. These methods are discussed next. Their results are presented together after their discussions. Liliens Measure Lilien (1982) holds that in the absence of structural change, employment in all sectors will grow at the same rate. By contrast, when labour is being reallocated across industries, expanding industries will grow faster than average and contracting industries will grow slower. Lilien proposed a measure of structural change based on the standard deviation of employment growth rates across industrial sectors calculated as follows:

Where GEit is employment growth in sector i at time t, GEt is the combined employment growth rate for all sectors, it may be viewed as the national average growth

120

rate in employment, and Sit is the i-th sectors share in total employment at time t. 100 If all sectors grow at the same rate, Liliens measure would be zero. The measure is always positive and larger, the more an individual sectors employment growth rate exceeds the average. The variable is called the Lilien measure of structural change.

Some economists, such as Abraham and Katz (1986) have criticized the Lilien measure. They note that employment growth in some sectors, such as the commodityproducing sectors, typically declines faster during economic downturns than employment growth in service-producing sectors. This pattern implies an increased dispersion in sectoral employment growth during contractions and a reduced dispersion during expansions, even if there is no actual impact of this change on aggregate employment. Consequently, sectoral change as measured by Lilien captures both the process of sectoral change and the normal employment flows of the business cycle. The measure does not tell us which sector is positively or negatively affected by recession or recovery. Hence, we cannot be certain that a high measured value of dispersion in employment growth is a signal of anything other than low economic activity. Rissman (1997) Measure Rissman (1997) tried to incorporate Abraham and Katzs criticism of the Liliens measure. The Rissman measure is based on a decomposition of the time series of sectoral employment share growth rates into three components. The first component reflects the long-term growth trend of employment in each sector. The second component, as noted by Abraham and Katz (1986), is the predictable movement of employment into and out of

100

A sectors employment growth is related to its share of employment by the following mathematical relationship: Ln (Sit) = Ln (E it / Et) = GE it GEt.

121

certain industries over the business cycle. The third component is the unexpected movement of workers across sectors or industries, i.e., changes across sectors that occur for reasons distinct to business cycles or long-term secular reasons. These unexpected movements, which Rissman calls idiosyncratic shocks, could be thought of as transformational changes in a firm or industry due to reforms, reorganizations, or other factors that might shift inputs to more valuable uses. Employment changes like these are likely to be the most upsetting to the labour market because they are unpredictable (unlike long term trends) but are everlasting characteristics of the conditions (unlike cyclical trends). Similar to Lilien (1982), Rissman proposed a measure of sectoral change based on the estimates of idiosyncratic shocks, . Specifically,

The term

is sector is acyclic employment share at time t1. This

employment share is hypothetically what the sectors employment share would have been if the national employment cycle were held constant at a value of zero, i.e., national employment was stagnant. The acyclic employment share would depend only on the sectors long-term trend and i.e., random, idiosyncratic shocks. The s are estimates of

the idiosyncratic shocks for each sector obtained from the H.P filter estimation exercise, as was described to decompose the GDP series in Appendix B relating to Chapter 4. In short, this measure relies upon unanticipated variations in the composition of sectoral employment growth. Long-term structural change affects the measure only through its effect on the acyclical employment shares. Because of its construction, the measure directly addresses the Abraham and Katzs critique of Liliens measure. 122

Aaronson, Rissman and Sullivan Measure

Aaronson, Rissman and Sullivan (2004c) provide a measure of sectoral change that is somewhat in between those proposed by Lilien and Rissman. Their measure is given by:

The above measure calculates variations in the composition of sectoral employment growth that are unrelated to the normal shifts that occur as the result of business cycle. It is a broader measure of sectoral change in that it considers long-term change in a sector employment share separately as a sectoral shift.

Figure 6.2 plots the above three measures of structural change obtained for Pakistans economy. The data are provided in Appendix IV. Annual dispersions in sectoral employment, using the above three measures, are provided in Figure 6.2. Liliens method shows that sectoral employment growth deviations around the national growth have been positive in all years indicating that Pakistans economy has been experiencing structural changes in all years. However, the measure did not show a systematic pattern until 2003. It dropped during the 1969-70 recession while it rose during the 1991 recession indicating more pronounced structural changes during the later recession.

123

Figure 6.2: Lilien, Rissman and Aaronson et.al Measures of Sectoral Variations in Employment, Pakistan (1967-2008)
1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00

Employment Deviations

1969

1977

1985

1993

2001

1967

1971

1973

1975

1979

1981

1983

1987

1989

1991

1995

1997

1999

2003

2005

Year

Aaronson et.al

Rissman

Lilien

Source: Based on own calculations presented in Appendix IV, Table A6. Liliens measure is based on sectoral employment growth deviation from the national employment growth. Rissmans measure is based on shifts in the employment composition that are unrelated to fluctuations in economic growth. Finally, the Aaronson et.al measure is similar to that of Rissman, but it also includes long-term change in sector employment as a sectoral shift.

The other two measures of employment dispersion are lower because of the way they are measured. These two measures also indicate an overall structural change in the economy, although the evidence is weak in recent years. Hence, it may be concluded that structural changes took place during the two periods of upturns in Pakistan economy. Finally, the data plotted in Figure 6.2 show that all three measures of structural shift are sensitive to the pace of economic growth.

2007

124

A summary of the three measures of sectoral change is provided in Table 6.3 during periods of slow and fast economic growth. On average, more employment shifts took place between sectors during the first ten years of slow growth as was also true for the first ten years of fast growth. Similar pattern is observed during periods of fast growth.

Table 6.3: Comparison of Three Measures of Sectoral Allocation in Slow and Fast Economic Growth Periods Average Dispersion of Employment Growth Sectoral Change Measure Aerosen et.al Rissman Lilien Slow Economic Growth 1969-70 to 1991-92 to 1978-79 2001-02 0.123 0.117 0.438 0.047 0.035 0.302 Fast Economic Growth 1979-80 to 2002-03 to 1990-91 2007-08 0.134 0.117 0.444 0.017 0.011 0.175

Source: Based on own calculation presented in Appendix IV, Table A6.

In summary, the different measures show that in general, the economy of Pakistan underwent structural changes during periods of slow and fast economic growth. However, there is an indication of stronger structural changes in the 1970s than in earlier periods. When more disaggregated employment data are available, a future study may investigate the quantitative impact of a structural change on the relationship between employment and GDP growth rates of Pakistan.

125

6.2.3 Monetary and Fiscal policies:

Monetary and fiscal policies can cause economic growth in a country to be jobless if they result in an increase in the real interest rate or in labour costs. As discussed earlier in this Chapter, a contractionary monetary policy, or an expansionary fiscal policy, can result in an increase in interest rates. An expansionary fiscal policy can also cause labour costs to rise. Akbari and Rankaduwa (2006) have provided a review of monetary and fiscal policies in Pakistan over the period 1973 - 2005. Both policies have gone through several periods of contraction and expansion. While international and domestic economic events have influenced them, they have also largely been dependent on the type of political regimes in place. More consistency was observed in fiscal and monetary policy variables during the presence of military regimes, especially during the last military regime which remained in power from 1999 to 2008, than during civilian regimes each of which lasted for short periods, mostly 2 to 4 years. Over the period 1999-2005, the first six years of the last military regime, fiscal policy became consistently contractionary when the governments budget deficit reduced from about 7 percent of GDP to about 3 percent. Monetary policy became expansionary during this period with a rise in money supply growth rate from under 5 percent to 15 percent. This consistent pattern of monetary and fiscal policies was also accompanied by several market reforms in the country which had actually started in the early 1990s. These market reforms included enhanced role of private sector and liberalization of interest rates, exchange rates and international trade. The central bank also became more autonomous during this period. A detailed analysis of the impacts of macroeconomic policies and of market reforms on the relationship between employment and GDP growth rates warrants a 126

separate study. For the purpose of the present study, a broad analysis will be conducted by augmenting the employment demand equation of Chapter 4 to incorporate the effects of macroeconomic policies and trade liberalization. A similar model was estimated for transition economies of Europe by Rutkowski et.al (2005). Three independent variables are added to the employment demand model of previous Chapters. The first two are money supply and fiscal budget balance to control for the effects of monetary and fiscal policies on the GDP elasticity of employment demand. A third variable, proportion of the sum of imports and exports in total GDP is included to control for the foreign trade effect. Rationale and results of this variable will be discussed in the next Section. Nominal values of all variables are converted into their real values using data on Consumer Price Index. 101 The augmented form of the employment demand equation is re-written below:

(16)

Each observation is for the year t. As before, of persons, week, and ln is log of GDP,

is log of employment in number

is the log of total numbers of hours worked per

is the log of total population of the country. The new variables are the ; budget deficit, . A dummy variable, ; the sum of imports and exports as a , is also added to the model to identify

logs of money supply, proportion of GDP,

any shift in employment after the introduction of market and macroeconomic policy

101

Data sources: Money Supply, Imports and Exports prices and Consumer Price Index data are taken from International Financial Statistics 2009 and Budget deficit data are from Pakistan Economic Survey (Various Issue).

127

reforms that took place in the 1990s. This variable is also interacted with the money supply, fiscal budget balance, and international trade variables to test for statistical significance of the change in their impacts on employment after 1990s. D t is given a value of one for the post 1989 period and zero otherwise. 102 Results of econometric estimation are provided in Table 6.4.

Table 6.4: Econometric Results of Employment Demand Equation by Using OLS Method Variable Constant GDP (Xt) Hours (Ht) Population (Popt ) Money Supply (MSt) Budget Deficit (BDt ) International Trade (OP t ) Time (t) Interaction variables (Base year is1990)* Dummy Variable (Dt) ) t) *Money Supply (MSt) (D (Dt)*Budget Deficit (BDt ) (Dt)*International Trade (OP t ) R2 Dwd Stat. No of Observations
Coefficient
10.52464 0.325217 -0.679397 -2.261729 -0.046624 0.003525 -0.070216 0.088808

t-statistic
5.794722 2.841063 -2.405950 -6.570361 -1.914136 0.331006 -2.717765 6.638536

P-value
0.0000 0.0093 0.0246 0.0000 0.0681 0.7436 0.0123 0.0000

3.442822 -0.306624 -0.021331 0.088657

2.856755 -4.009963 -1.206269 0.920945

0.0089 0.0005 0.2400 0.3666

0.9967 2.23 35

Before interpreting the econometric results, we again address the concern that the results may be spurious due to non-stationarity of the data. The employment and GDP series were found to be non-stationary in Chapter 4. The money supply and budget deficit

102

The model also follows the work of Boeri and Garibaldi (2004) on European New Members States and European Monetary Union (EMU).

128

series, which are introduced in the present Chapter, are found to be stationary at I(1) but the International Trade (Openness) series is stationary at level, i.e., it is I(0). As was done in the previous chapters, a residual based co- integration ADF test was applied. 103 The results are provided in Tables 6.5. If residuals of the employment demand model are found to be stationary, the series are co-integrated and results are not spurious.

Table 6.5: Residuals Based Co-Integration Test Results (ADF)

t-Statistic Augmented Dickey-Fuller test statistic Test critical values:* 1% level 5% level 10% level -8.081097 -5.6532 -4.8601 -4.4749 Prob. 0.0000

* The Null hypothesis is non-co-integration. The critical values are calculated by using

the software provided by J. G. MacKinnon http://www.econ.queensu.ca/faculty/mackinnon/jbes/. See MacKinnon (1994)

The above co-integration test result shows that residuals of the estimated regression are stationary in their levels, i.e, I (0). Hence, a linear combination of the series is co-integrated indicating that econometric results of the employment demand model are meaningful and variables of the model are co-integrated in the long run. The focus of the econometric analysis of this Section is on the effects of monetary and fiscal budget balance variables. Results of Table 6.4 show that money supply and international trade variables have statistically significant effects in the model while fiscal budget deficit does not. The negative effect of money supply on employment is strengthened since the 1990s, as shown by the t-value of the variable interacting the

103

As shown earlier, the employment and GDP series were stationary at I (1).

129

dummy variable and money supply, while the macroeconomic and market reforms of this period did not affect the influence of fiscal budget balance or the international trade, as the low t-value of the variable interacting the dummy variable and fiscal budget balance indicates. The GDP elasticity of employment is found to be 0.325 in the augmented model which is higher than what was observed in Chapter 4 (value = 0.26) when no controls for the effects of macroeconomic policy variables and for international trade were introduced. This means that the macroeconomic policies and international trade have weakened the effect of GDP on employment during the overall period of analysis. In summary, the negative effects of money supply on employment and on GDP elasticity of employment demand imply that monetary policy may have weakened the effect of GDP on employment thereby causing economic growth to be jobless over the three decades under study. A future study can investigate the impacts of other monetary variables such as, interest rate and inflation rate, on employment demand and on GDP elasticity of employment demand.

6.2.4 International Trade and Globalization:

When countries move towards economic integration, international trade expands as trade barriers are removed. Countries expand the production of goods in which they have comparative advantage while reducing the production of goods in which their trading partners have comparative advantage. Specialization of production on the basis of comparative advantage is a source of economic growth, but it is also a source of labour productivity growth and of structural shift in the economy both of which in turn can cause

130

the resulting economic growth to be jobless in a country.104 With a share of exports and imports at about 36 percent of its GDP, Pakistans international trade sector is a significant component of Pakistans economy. Hence it is important to investigate the role of this sector in employment generation. The econometric results reported in Table 6.4 showed a statistically significant and negative effect of Pakistans international trade on employment. As discussed above, the fact that introduction of this variable, along with the macroeconomic policy variables, increased the GDP elasticity of demand probably also indicates that macroeconomic policies and the international trade sector of the economy may have contributed towards joblessness in economic growth over the three decades under study. To what extent the expansion of international trade in Pakistan has led to economic and labour productivity growth and in structural shift, can be topics of future studies. Such studies could also estimate the international trade multiplier of employment. A byproduct of international trade and enhancement of private sector is the phenomenon of outsourcing. Sometimes a company in a developed country may lack comparative advantage in production of a good but may find it worthwhile to shift its production to a developing country where labour cost is low. 105 This phenomenon, known as outsourcing of production, has become more prominent since 1990s with the adoption of flexible economic policies in developing countries that allow for market driven means of production. Outsourcing of production from developed to developing countries has often been blamed for employment decreases in developed countries. For example, Gorg and Hanley (2005) found that Ireland lost much of its employment

104 105

When countries exploit their comparative advantage, some industries expand while others shrink.

This is especially true if investment in such production requires large amount of capital not available in the developing country.

131

opportunities due to outsourcing, particularly with respect to financial services being relocated to India. Kletzer (2000) found outsourcing led to job losses in the United States. This caused President Obama to announce that his government may levy a tax on those US companies whose majority of employees are located outside the country. "To encourage ... businesses to stay within our borders, it is time to finally slash the tax breaks for companies that ship our jobs overseas, and give those tax breaks to companies that create jobs right here in the United States of America. Now, the House has passed a jobs bill that includes some of these steps. As the first order of business this year, I urge the Senate to do the same, and I know they will. They will. People are out of work. They're hurting. They need our help. And I want a jobs bill on my desk without delay" (Obama, 28 January 2010).106 In the United Kingdom, about 100,000 jobs would have relocated abroad by 2005 as a direct consequence of outsourcing (Financial Times, 2005). 107 India has been one of the biggest beneficiaries of outsourcing and could be the worst hit by any move in developed countries to restrict outsourcing of jobs. American companies alone employ about 1.7 million people in India (IANS, 25 February 2009; 4 May 2009).

106

Obama raises red flag on outsourcing By Agencies Posted On Thursday, January 28, 2010 at 04:21:24 PM. In his first state of Union Address, US President Barack Obama said that the worst of the economy slump for the US is over. http://www.mumbaimirror.com/article/4/2010012820100128162124545309dffe2/Obama-raises-red-flagon-outsourcing.html.
107

A recent Financial Times article report's findings from an independent consultant that a two-speed Europe is emerging, with the UK and Ireland reporting the highest levels of worker replacement. Financial Times, August 16, 2005, Europe set to move 1 million jobs abroad in a decade (World News). Other studies that found evidence of job losses in Europe include: Egger & Egger (2003), Feenstra & Hanson (1996), these economist discuss the issue of off-shoring in their studies and find out that off-shoring is a cause of net decrease in employment.

132

Since the phenomenon of outsourcing mostly causes a shift of production from developed to developing countries, it is expected to generate a job -intensive, rather than a jobless, economic growth in a developing country which is at the receiving e nd. It could also increase the GDP elasticity of employment. Unfortunately, very little evidence exists to date as to how much employment generation and economic growth has taken place in developing countries due to production outsourcing from developed countries. In 2002, the Planning Commission of India reported that globalization offered little or no scope for employment growth in the organized sector of India (Indiresen, 2002). Data on outsourcing of production from other countries into Pakistan and the resulting impact on employment and economic growth are lacking and hence it is not possible to determine if this phenomenon has been the source of jobless growth or jobintensive growth in the country. A future study may collect and systematically analyze these data.

6.3 Summary: In this Chapter, some possible causes of jobless economic growth in Pakistan besides the employment barrier created in manufacturing sector due to deliberate policy creation of favorable prices of capital, were identified and interpreted. Based on a broad analysis of the impact of each cause, it may be concluded that labour productivity growth, structural shifts in the labour market, monetary policy and international trade may have caused economic growth in Pakistan to be jobless over the period 1974 to 2008. There is more evidence of an increase in structural change during 1968-1985. In the early 1990s, labour productivity increases were noticeable. As Krugman (1992, p.9) wrote productivity is not everything but in the long run it is almost everything so in the short run productivity may be harmful for employment but in the long run it is very important 133

for economic development as there is a two-way causation between productivity and growth.108 Monetary policy and international trade may have weakened the relationship between employment and GDP by causing labour productivity to increase during this period. Expansion of international trade could also have caused greater structural shifts. The analysis presented in this Chapter is broad. An in-depth analysis can be pursued in future using more detailed econometric modeling and disaggregated data for each sector.

108

Srinivasan (2005)

134

CHAPTER 7 Conclusions and Policy Implications

The main purpose of this study was to investigate whether economic growth in Pakistan is jobless or job intensive. An econometric analysis was conducted using national and sectoral data for the period 1974 to 2008 and possible reasons for joblessness of economic growth were explored. Section 7.1 provides a summary of the main conclusions in relation to the research questions which were raised in the first Chapter. Section 7.2 summarizes the investigation of the causes of jobless growth in Pakistan. Some policy implications of main findings are discussed in Section 7.3 while Section 7.4 discusses some limitations of this study and provides some directions of future research.

7.1 Summary of Conclusions Regarding Presence of Jobless Growth in Pakistan: An analysis of general trends in labour market indicators and economic growth rate in Pakistan over the past four decades showed that overall; the employment growth has been slower than growth in GDP. There appears to be a negative relationship

between labour productivity and employment elasticity according to these general trends. In the last decade, both of these variables have been volatile. 109 Jorgenson and Vu (2005) reported that Pakistans productivity and GDP growth rates both declined during 1995 2003 as compared to during 1989-1995.

109

This correlation does not isolate the impact of employment growth on GDP growth from other factors which was done in the employment demand model.

135

The rate of growth of employment declined in the 1990s, not due to a decline in economic growth from 1980s, rather due to a decline in the rate of growth of the labour force. To investigate the nature of economic growth, an employment demand model was derived from a production function. In this model, employment was expressed as a function of GDP. This model was estimated in log-linear form. The estimated employment elasticity of demand with respect to GDP was less than unity, 0.26, indicating a slow response of employment to economic growth over the period 19742008. Other studies have found employment elasticity with respect to GDP to be in the range of 0.11 to 0.41.110 The overall slow response of employment growth to GDP growth indicates that economic growth in Pakistan may have been largely dependent on capital intensive projects, which is largely the case in manufacturing sector. Arif, Kiani and Sheikhs (2002) longitudinal study found that the labour absorptive capacity of Pakistans economy has declined over time. 111 A major contribution of this study is its use of employment elasticity to calculate the threshold level of GDP growth for Pakistan below which employment is stagnant. This was found to be above 6 percent, while the annual average growth rate has been around 5 percent during the period of analysis indicating the jobless nature of economic growth over the period 1973-2008. The threshold level of GDP growth is higher than the

110

Baqai (1979), Kemal (1990), Chaudhary & Hamid (1994 and 1998), Majid (2000), Hyder (1994), Planning Commission of Pakistan (2004) and Aslam & Zulifqar (2008).
111

These authors also found that transition from unemployment to employment was slow and more so in case of females.

136

3.5 to 4.7 rate found for Indonesia by Islam and Nazara (2000), and the 4.4 percent rate found for Poland by Wolnicki, Kwiatkowski and Piasecki (2006).112 When the effect of fast economic growth was introduced into the employment demand model, it was found that employment elasticity dropped to only 0.11, giving an even higher threshold level of GDP growth rate of 7.9 percent. The slower response of employment during initial periods of fast growth indicates the possibility of a wait and see approach of employers before they start re-hiring workers which has been found to be true in other country studies (for example, Schreft et al ,2005). It could also be the result of structural changes in the economy towards labour-saving capital intensive means of production and a possible productivity growth when firms are restructuring. The above results were based on the aggregate national data. Based on sectoral analysis, a negative relationship was found in the agricultural sector between employment and GDP growth rates. Because of a low income elasticity of demand in the agriculture sector, future economic growth is not likely to cause significant growth in agriculture and more labour is expected to move to the non-agriculture sectors. Furthermore, any technical progress in this sector is also likely to be both land and labour saving. The low employment growth linkage in agriculture is contrary to the suggestions of those found in the literature that development of agriculture sector can be a recipe for poverty reduction (Majid, 2000). Kaldor (1968) has suggested that agriculture productivity can be increased by switching labour from agriculture to other sectors. Arif, Nazli and Haq (2000) suggest that a dynamic labour-intensive agriculture combined with a non-agriculture sector can lead to a broad spread of employment.

112

In the South-East Asian economies, employment growth rates of 2.5 to 4 per cent would normally require a GNP growth rate of 5 to 8 per cent or more (Mazumdar & Basu 1997).

137

Evidence for jobless growth is found in the manufacturing sector, which is often regarded as one of the main engines of economic growth.113 The threshold level of GDP growth in this sector is the highest among all sectors. High capital intensity of production appears to be the main reason for the lack of employment growth in this sector. The overall employment trend in manufacturing has been negative. This can be explained by the rise in trade and increase in productivity which has been the case in the manufacturing sector of developing countries as discussed by Schulze (2004) and Bailey& Lawrence (2004). Mahmood and Siddiqui (2000) also found an increase in productivity in the manufacturing sector of Pakistan. The share of the construction sector in employment is 6.2 per cent but it contributes only 2 per cent towards the GDP. Growth in this sector was found to be above threshold level of GDP growth. Two other studies, Pakistan (2007) and Ahmad (2006), report a shortage of 6.0 million houses in Pakistan and that the construction sector has the largest employment linkages to other sectors. Nearly 40 industries in the manufacturing sector are linked with construction related activities, in addition to the housing sub-sector. According to the above two studies, even if only 0.5 million housing units are constructed annually, at least 200,000 to 300,000 additional jobs will be created in the housing sector. The electricity and gas distribution sector (combined) does not appear to have much capacity to absorb labour due to its relatively small share in the overall economy, although its growth is also employment intensive (above threshold level) according to the findings of the present study.

113

Mining and manufacturing were considered as combined sector in this study. The share of mining in employment is very small (less than one percent).

138

The share of the services sector in employment is 36 per cent. Econometric results showed that the service sectors average growth rate was above the threshold level of GDP growth; hence, growth in this sector is also employment intensive rather than jobless. Trade and finance sectors, combined, contribute the most towards GDP among other services sectors, and they are also major employers of labour. The finance sector has significantly increased its labour absorption (Pakistan, 2007) and according to Ahmad (2006), this sector has little capacity to increase employment. Most activities in the trade sector are undocumented and as such this sector is mostly out of the tax net. A proper regulatory framework is needed for the wholesale and retail sectors which can absorb 0.25 to 0.45 million workers during the tenure of PRSP-II 2007-2010.114 Growth in the transportation sector, another component of services sector, is also employment intensive. This sector can also be a major employer in services sector, where about three million jobs can be generated in urban as well as in rural areas of the country (Ahmad, 2006). The share of all other industries in the services sector including

Ownership of Dwellings, Public Administration & Defense and Community, and S&P Services, is 15 per cent and econometric results of the present study show that these sectors average growth rate was above the threshold level of GDP growth; hence, growth in these sectors was employment intensive rather than jobless. Despite the impressive economic growth performance of services sector and its absorption of labour, its continued potential as employment source in future is questionable. Development economists are divided over the sustainability of services as potential source of employment in developing countries (Gupta and Singh, 2005;

114

Poverty Reduction Strategy of Pakistan

139

Karmakar, 2008). Income elasticity of demand for services, such as the trade finance services which are currently major contributors to GDP growth, is high because of which lower capita incomes cannot sustain the growth of this sector. The overall conclusion of employment elasticity analysis is that economic growth in Pakistan was jobless only in the manufacturing sector which has also experienced a decline in employment over the past decade. One may attribute this finding to the faster than average productivity growth in the manufacturing sector. Some authors, noted in the present study have found that factor price distortions in favor of capital, caused by deliberate policy measures aimed at increasing investment in this sector, have resulted in high capital intensity of production which has become a barrier to employment creation in this sector. However, at the same time, there may have been some spillover employment growth effects of manufacturing growth in the services sector, especially in the transportation sector thereby causing employment generation in the services sector. Future research can explore this possibility.

7.2 Key Factors Responsible for Jobless Economic Growth in Pakistan:

Besides identifying the high capital-intensity of production in manufacturing as being the likely reason for jobless growth in that sector, this study also investigated broadly other potential factors behind jobless growth in Pakistan and found labour productivity, sectoral reallocation or structural shifts in the labour market, and macroeconomic policies (fiscal and monetary) to be among the major causes. Sectoral reallocation in Pakistan during 1968-1985 was a main cause of the jobless growth during that period. In the early 1990s, increases in labour productivity in

140

Pakistan and most of the developing countries also appear to have led to jobless growth. Other studies that have also attributed jobless growth to productivity increases include Kwiatkowski and Kwiatkowska (2006) and Wolnicki et.al (2006) for Poland and Tezcek (2007) for Turkey. Finally, monetary and expansionary fiscal policies contributed

towards jobless growth in late 1990s and early 2000s. The negative effect of money supply on employment became stronger since the 1990s while the macroeconomic and market reforms of this period did not affect the influence of fiscal budget balance or the international trade on employment growth relationship. A similar conclusion was reached by Rutkowski et.al (2005) for transition economies of Europe and by UNDP (2008) for Asian countries. 7.3 Some Policy Implications of this Study:

The current labour policy in Pakistan aims at targeting employment growth directly instead of relying solely on economic growth to generate employment. However, one cannot deny the importance of aggregate demand in goods and services markets in promoting employment demand for which economic growth is necessary. Employment growth can be enhanced through increasing the GDP elasticity of employment which in turn would reduce the threshold level of GDP growth. In this regard, as per the evidence discussed in this study, priority should be given to the manufacturing sector which is found to have significantly higher level of threshold GDP growth than other sectors. This sector is likely to continue as a sustained source of future economic growth. It also has spillover effects on other sectors, such as the services sector. In particular, growth in transportation and storage sectors often relies on the growth in manufacturing sectors. Labour-intensive methods of production should be encouraged in this sector. The increasing demand for skilled workers in this sector is expected to slow down the

141

expansion of employment of low-skilled workers. Hence, the labour employment policy should also emphasize on worker training. Since the manufacturing sector is widespread in the country, provision of information regarding job availability and for greater worker mobility should also be the components of future employment policy. Consideration must also be given to the factor price distortions that currently favor the use of capital and have become employment barrier in this sector by raising the relative price of labour. Establishment of more small- and medium-sized labour-intensive manufacturing industries should be encouraged in order to address the problem of rural unemployment. In particular, agro-based industries such as food processing, meat processing, poultry processing, milk production, furniture industry, fibre industry, etc. should be encouraged in rural areas. Special tax incentives in rural areas that help reduce the cost of labour will also encourage expansion of manufacturing sector. This sector also has a large informal (unorganized) component where about 57 percent of its employment takes place (Gennari, 2004). Hence, any future sustainability of this sector as a potential employment source will also depend on how the informal sector grows. Economic growth in the services sector has been high and its labour absorption has been strong. However, this sector has uncertain potential of continued employment source since its growth may not be sustainable due to lower per capita income and high income elasticity of demand for services. Despite its overall uncertain employment potential, some of the service sub-sectors, such as the transport and storage, also have forward linkages to the manufacturing sector. Hence, employment and growth policies should also focus on improving employment conditions in these sectors. One important sector within services is the Information Technology (IT) which can absorb skilled labour. The use of information technology should be encouraged in all sectors in Pakistan

142

which will improve worker productivity and will also have a job-enhancing effect in the services sector. One other way of generating employment can be through the creation of special incentives for employers to substitute working hours with employees. Over the four decades for which data are analyzed in this study, working hours fell only in the agriculture sector while they rose in all other sectors and the econometric analysis of this study showed that labour and hours are substitutes in production. Employees in all nonagriculture sectors work more hours than the national average, working more than 50 hours per week in transport and trade and finance sectors. To create incentives for employers to use more workers, rather than require each worker to work more hours, possibilities for any reduction in the regular number of working hours and for increasing the overtime premium should be considered. Employment generation policies are prone to macroeconomic policies. The recent drift towards expansionary fiscal and restrictive monetary policies in Pakistan can have negative employment effects through their effects on interest rates. Muslehuddin (2009) suggests fiscal policy should focus on investment in labour-intensive infrastructure that can offset the negative employment effects. He also suggests a strong need for coordination of fiscal and monetary policies so that they do not offset each others effect. Finally, as Arif, Kiani and Sheikh (2002) longitudinal study has shown, women face lower probability of finding employment once they become unemployed. By investigating and removing employment barriers faced by women, as much as possible, and introducing special employment schemes for women in rural and urban areas, the unemployment issue for this large section of the society can be address.

143

7.4 Limitations of the Study and Some Directions for Future Research: The present study is the first to investigate the job intensity of economic growth in Pakistan. Most empirical studies have some limitations which should be taken into consideration when interpreting the results and this study is no exception. Some of the limitations of the present study are discussed below. Based on these limitations, some directions for future research are also provided.

First of all, in this study the focus has been on employment generation alone and not on the quality of jobs that may be generated by economic growth. As noted by Garibaldi and Mauro (2002), more jobs should not be viewed as necessarily good if they are not accompanied by increased productivity which causes higher wages thereby improving the standard of living, and/or lowering involuntary unemployment. A study on the quality of jobs can be based on survey data that are not available at this time.

The labour market data used in this study are based on the Labour Force Survey (LFS). The population coverage of the LFS is known to be low (varying from only 20,000 households in 1980s to about 32,000 households since the year 20,000 and beyond).

Many authors have raised the issue of the quality of data available in developing countries (Kenya, India, South Africa, etc) which can affect the reliability of the results of an empirical study (Bowen et al, 1994; Deininger and Squire, 1996 and 1998; Baird et al, 2008). To the best of my knowledge, the data I have used in this study are the best available data from government sources in Pakistan. Other empirical studies that have pursued different topics have also made use of these data. The results of present study are not expected to have been affected by the

144

quality of data in any different way from the results of other published studies that have used time series data. Again, the best alternative to these data will be survey data for which more time and monetary resources are required. The analysis conducted in this study was only limited to the formal sector. A large employment in developing countries is created in informal sectors on which no data are available. In Pakistan, estimates indicate that about 67 percent of employment in non-agriculture sector is in the informal sector (Pakistan, 2007). The sectoral composition of the informal economy has also changed in recent years (Table A5.3). As noted by Mahbub-ul-Haq Human Development Centre (2004), the largely jobless growth has lowered labour standards and increased the vulnerability of about 90 percent of the workforce engaged in the informal sector in South Asian countries. The annual Labour Force Survey of Pakistan began collecting data on the informal sector since 1996. The latest data are for 2008-09, but due to several interruptions in the survey, these data are available for only nine years which are insufficient for conducting a meaningful analysis of the impact of economic growth on employment that covers the informal sector as well. One important finding of the present study is that growth in the mining and manufacturing sectors (combined) was jobless over the last four decades. The mining sector is a small sector of the economy and hence the combined manufacturing and mining sector is dominated by the manufacturing industries. Growth in the manufacturing sector can have spill over effect on other sectors of the economy, especially the services sector and this effect has not been analyzed in the present study. Given that the growth in service sector has been employment intensive, it will be interesting to analyze how much of the service sector growth in employment was attributable to the growth in the manufacturing. It will also be

145

important to analyze separate data on manufacturing sector, and within that sector, disaggregated data on individual industries should be analyzed according to their geographic locations. An analysis by geographic locations can shed some light on the regional disparities in the effects of economic growth. A high value of the constant term in employment demand model of Chapter 4 indicates a large effect of all excluded variables on the employment demand over the period of analysis. These variables could be the domestic political and economic conditions, changing economic policies, worker training, etc. The exclusion of these variables may have biased the employment elasticity estimated in this study. This is the first study in Pakistan which has investigated the possible causes of jobless growth, but in a broad manner. An ad-hoc model was estimated to gauge the effect of macroeconomic policies. An in-depth analysis can be pursued in future using more detailed econometric modeling and disaggregated sectoral data. Such an analysis is necessary for an appropriate employment policy response. Lastly, an analysis based on survey data for specific industries and regions will shed some light on how individual industries have adapted to changing economic growth over time. Special attention may also be paid to any differential effects of the jobless nature of economic growth on employment of men and women.

Despite some limitations, the quantitative analysis conducted in this study has provided an important input for policy debate on the role of economic growth in improving the standard of living of the Pakistani population through employment creation. The study has also opened new pathways for future research on the job creating potential of economic growth in Pakistan.

146

Appendix I Derivation of Employment Demand Equation

The employment demand equation derived in the text can also be obtained by considering the cost minimization problem faced by a firm. Here we consider two factors of production

. So the general production function is given by

X=f( ,
where X is output, is a labour,

)
is a capital. To estimates the employment

demand equation, we have to minimize the cost function subject to output function. Assuming a Cobb-Douglas production function, which was used in most of previous studies, we state the problem as: function. Minimize C = +

Subject to where,

C = total expenditure on inputs, respectively. X is output of the firm. is labour, is capital.

are factors prices of labour and capital

incorporates technology changes as described in the text.

is share of labour, share of capital. The Lagrange function for the constrained minimization problem is written as follows: + (X 147

L =

Differentiate the above Lagrange function with respect to Z1, Z2 and . We get, First Order Conditions (FOCs)

(1)

(2)

=0

(3)

From equation (1) and (2) we get,

,
Therefore, (4)

=
Put the values of and

in equation (4) we get,

After some manipulation we find the value of Z2, we get,

From equation (3) we get,

X = Put the values of Z2 in the above equation, we get, 148

From the above equation we find the value of

Taking natural log on both sides we get,

+ t lne +

Where, lne = 1 and,

2 = If we impose an assumption on the above equation that the ratio of wages is constant, i.e., as follows: = + + 2t (6) do not vary, then the final demand for labour equation may be written

The last equation (6) is a labour demand equation which is similar to equation (8) derived in the text where represents the labour input in a given country and is a slope is time trend

represents the total gross domestic product (GDP) in a given year. coefficient and represents employment elasticity with respect to output. capturing the effect of technological changes.

is the intercept term which includes the

149

factor wage ratios in the above derivation. The equation is applicable if the factor wage ratios can be assumed fixed, which is possible in the context of Cobb-Douglas production function if capital to labour ratio remains constant under the assumption of constant returns to scale. However, the estimated equation in the text is based on a direct derivation of employment demand from the production function where the constant term and the time trend variables have different interpretations. No assumption regarding returns to scale is made.

150

Appendix II Methodology for Identifying Fast and Slow Economic Growth Periods in Pakistan
Hodrick Prescott (1997) developed a methodology to estimate business cycles. In the present study, their methodology is used to estimate the economic growth cycles. In summary, this method assumes that annual series of real GDP is an aggregate of three components including Trend, Cyclical movements, and Irregular movements. In symbolic form; Y t = T t + Ct + I t Where Tt = Long-run trend Ct = Cyclical movements It = Irregular movements (shocks) All variables are considered in natural log form. It is intended to isolate C t in the Y t series. This is done in two steps. In the first step, Tt is removed from Yt. For this purpose, an HP filter applied using E-Views software. This gives the following series: Zt = Yt Tt = Ct + It In the second step, HP filter is used again to remove irregular shocks (It) from Zt, to give Ct. The above method provides a de-trended log GDP series for Pakistan that is also isolated from irregular shocks. Slow and fast GDP growth rate periods are identified as

151

the periods in which the detrended annual GDP values, free of irregular shocks, are falling and rising, respectively, and are displayed in Figure A4. Figure A4: Periods of Slow and Fast Economic Growth in Pakistan Based on HP Method:

0.015

Peak
0.01

0.005

-0.005

-0.01

Trough
-0.015

1950 1952 1954 1956 1958 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

152

Appendix III

Table A5.1: Sectoral Compositions of Pakistan Economy

1. Agriculture
i. Major Crops ii. Minor Crops iii. Livestock iv. Fishing v. Forestry

2. Industry
2.1. Manufacturing i. Mining & Quarrying ii. Manufacturing a) Large-Scale b) Small-Scale 2.2. Construction 2.3. Electricity and Gas Distribution

3. Services
3.1. Transport, Storage and Com. 3.2. Trade and Finance i. Wholesale and Retail Trade ii. Finance and Insurance 3.3. All others Services i. Ownership of Dwellings ii. Public Administration & Defense iii. Community, S & P Services

153

Table A5.2: Sectoral Distribution of Employment according to Rural and Urban Areas, Pakistan (1974 - 2008)

Rural Sectors
1974 Agriculture Mining and Quarrying Manufacturing Construction Electricity Trade Transport Services Activity not described
72.08 0.13 9.32 3.41 0.23 5.9 2.94 5.7 0.29 100

1980
68.76 0.44 5.93 3.4 0.39 5.31 2.54 9.2 4.03 100

1985
70.94 0.29 8.52 4.36 0.23 6.22 3.07 6.3 0.06 100

1990
63.79 0.14 8.08 6.63 0.54 8.11 3.68 8.97 0.06 100

1995
61.94 0.11 6.78 7.47 0.56 9.56 3.81 9.73 0.05 100

2001
59.01 0.07 8.68 6.23 0.56 9.5 4.82 11.1 0.03 100

2005
53.45 0.15 8.69 7.99 0.5 12.09 5.98 11.1 0.04 100

2007
60.94 0.14 8.37 6.09 0.42 9.63 4.42 9.96 0.03 100

2008
61.72 0.25 8.35 6.51 0.43 10.55 4.3 7.86 0.03 100

Total

Urban
Agriculture Mining and Quarrying Manufacturing Construction Electricity Trade Transport Services Activity not described Total
6.2 0.19 25.74 6.41 1.23 28.24 10.3 21.26 0.44 100 7.39 0.28 18.29 6.43 1.18 23.94 8.65 26.41 7.43 100 6.83 0.17 26 7.68 1.32 29.37 8.18 20.35 0.09 100 7.63 0.17 22.35 6.59 1.55 28.83 9.07 23.75 0.07 100 5.8 0.14 20.1 6.49 1.53 30.76 8.47 26.58 0.13 100 5.19 0.07 25.08 5.67 1.33 29.37 8.26 25.03 0.01 100 6.32 0.02 24.71 5.91 1.25 29.5 8.22 24 0.06 100 6.21 0.07 23.89 6.75 1.36 31.15 7.92 22.39 0.26 100 5.77 0.37 23.9 6.87 1.3 32.99 7.42 21.3 0.08 100

Source: Labour Force Survey of Pakistan (Various Issues). Note: Totals may not add up to 100 due to rounding.

154

Table A5.3: Sectoral Distribution of Informal Employment according to Rural and Urban Areas, Pakistan (2001 - 2008)
Mining and Quarrying All other Services

Years

Class

Manufacturing

Electricity

Construction

Trade

Transport

Finance

Total

Rural 200102 Urban Total Rural 200304 Urban Total Rural 200506 Urban Total Rural 200607 Urban Total Rural 200708 Urban Total Rural 200809 Urban Total

0.02 0.01 0.03 0.03 0.01 0.04 0.09 0.01 0.1 0.1 0.01 0.11 0.09 0 0.09 0.09 0.03 0.12

9.53 11.33 20.86 9.96 10.65 20.61 10 11.34 21.34 9.89 10.46 20.35 10.19 10.68 20.87 10.2 11.3 21.5

0.02 0.01 0.03 0 0.07 0.07 0.02 0.01 0.03 0.02 0.01 0.03 0.02 0.02 0.04 0.02 0.01 0.03

9.89 3.97 13.86 9.66 3.72 13.38 9.73 4.06 13.79 10.15 4.41 14.56 9.7 4.51 14.21 10.77 4.73 15.5

15.24 18.72 33.96 15.38 19.24 34.62 15.24 19.25 34.49 14.83 19.75 34.58 15.7 19.64 35.34 17.08 22.53 39.61

7.11 4.62 11.73 6.22 4.93 11.15 6.65 4.43 11.08 6.25 4.59 10.84 6.79 4.24 11.03 6.83 4.06 10.89

0.23 0.43 0.66 0.37 1.03 1.4 0.4 1.04 1.44 0.39 1.2 1.59 0.51 1.27 1.78 0.57 1.27 1.84

9.25 9.62 18.87 8.58 10.14 18.72 8.22 9.51 17.73 8.59 9.35 17.94 8.15 8.48 16.63 4.98 5.54 10.52

51.29 48.71 100 50.2 49.79 100 50.35 49.65 100 50.22 49.78 100 51.15 48.84 100 50.54 49.47 100

Source: Labour Force Survey of Pakistan (Various Issues). Note: Totals may not add up to 100 due to rounding.

155

Table A5.4 : Sectoral Employment and GDP Growth Rate of Pakistan (1974-2008)

Sector

Growth Rate 1974

-80*

1981 -85

1986 -90

1991 -95

1996 -00

2001 -05

2006 -08

1974 -08

GDP Agriculture Employment GDP Manufacturing Employment GDP Construction Employment GDP Electricity Employment GDP Transport Employment GDP Trade &finance All other services Employment GDP Employment

3.31 1.72 5.55 9.84 9.75 4.46 8.16 30.5 5.98 5.93 6.16 5.79 7.44 -0.65

4.16 1.39 5.67 1.05 2.00 4.79 1.06 0.80 14.5 4.10 8.62 1.52 6.37 6.07

4.43 2.91 7.00 1.18 5.63 5.28 12.9 -0.46 4.21 1.45 5.63 3.34 5.73 2.10

4.01 -0.89 4.38 -3.13 4.22 3.34 8.56 7.48 6.14 1.62 4.72 4.80 4.72 5.15

3.80 3.86 2.23 5.08 -6.58 -1.24 2.51 0.01 5.06 3.02 5.42 1.75 0.83 3.04

1.80 -0.12 8.77 5.79 1.41 2.40

4.52 6.23 7.49 3.30 10.7 7.54

3.65 1.90 5.76 3.68 3.82 3.62 3.86 7.96 6.38 3.62 6.24 3.76 5.35 2.98

0.75 -16.9 1.36 3.05 4.84 6.07 4.07 4.71 3.65 6.47 5.55 3.40 7.63 4.60 7.81 2.98

Note: * indicates that annual averages of 1974-80 Source: Based on various issues of Labour Force Survey and Economic Survey of Pakistan.

156

Appendix IV

Table A6: Measures of Sectoral Change Year

1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Aaronson et.al
0.1727 0.0461 0.0271 0.0309 0.0275 0.3075 0.3519 0.0301 0.2280 0.2042 0.0175 0.0168 0.0170 0.4189 0.3947 0.0237 0.2818 0.2692 0.0266 0.0567 0.0132 0.0334 0.0174 0.0417 0.0290 0.0670 0.0283 0.0401 0.0092 0.0855 0.0172 0.0176 0.0558 0.0205 0.0862 0.0895 0.0645 0.0055 0.0077 0.0085 0.0084 0.0098

Rissman
0.0337 0.0745 0.0860 0.0449 0.0183 0.2412 0.2686 0.0380 0.2809 0.2486 0.0106 0.0104 0.0110 0.3503 0.3389 0.0172 0.2928 0.2520 0.0325 0.0343 0.0147 0.0117 0.0282 0.0204 0.0148 0.0259 0.0352 0.0224 0.0131 0.0546 0.0125 0.0143 0.0795 0.0277 0.0453 0.0557 0.0390 0.0123 0.0072 0.0052 0.0013 0.0031

Lilien
0.6017 0.3019 0.2695 0.2408 0.2841 0.8429 0.8806 0.2316 0.6595 0.6444 0.2050 0.1996 0.1951 0.9245 0.8907 0.1947 0.8011 0.7726 0.2543 0.3311 0.1645 0.2542 0.1860 0.3082 0.2428 0.3905 0.2690 0.3184 0.1465 0.3968 0.1856 0.1872 0.3346 0.2098 0.4504 0.4342 0.3787 0.0868 0.1205 0.1300 0.1501 0.1863

Source: Based on authors own calculation.

157

Bibliography

Aaronson, D; E. R. Rissman, and D. G. Sullivan (2004a) Assessing the Jobless Recovery Economic Perspectives, Federal Reserve Bank of Chicago, Vol: 28 (2), pp. 2-20. __________________________________________ (2004b) Further Evidence on Sectoral Reallocation and the Jobless Recovery Federal Reserve Bank of Chicago , Unpublished Manuscript. ___________________________________________ (2004c) Can Sectoral Reallocation Explain the Jobless Recovery? Economic Perspectives, Federal Reserve Bank of Chicago, Vol: 28 (2), pp. 36-49. Abel, A.B and B. Bernanke (2005) Macroeconomics- 5th Edition, Addison- Wesley Abraham, K. and L. F. Katz (1986) Cyclical Unemployment: Sectoral Shifts or Aggregate Disturbances? Journal of Political Economy, Vol: 94(3), pp. 507-522. Ahmad, N (2006) Employment Generation Strategy for PRSP II ILO, April 28, 2006. Aghion, P. and P. Howitt (1994) Growth and Unemployment Review of Economics Studies, Vol: 61 (3), pp. 477494. Akbari, A. H and W. Rankaduwa (2006) Inflation Targeting in a Small Emerging Market Economy: The Case of Pakistan SBP-Research Bulletin, Vol: 2 (1), pp.169190. Altman, M (2001) Employment Promotion in a Minerals Economy Journal of International Development, Vol:13, pp. 1-19, 2001. ___________(2003) Jobless or Job Creating Growth, Some Preliminary Thoughts Paper presented at the TIPS/DPRU Annual Forum, 8-10 September, Johannesburg. Amjad, R (2007) Decent Work: An Employment and Poverty Reduction Strategy for Pakistan Pakistan Development Forum, 25-27 April, 2007. Andolfatto, D. and G. M. MacDonald (2004) Jobless Recoveries Working Paper, 2004. Anwar, T (2004) Recent Macroeconomic Developments and Implications for Poverty and Employment in Pakistan: The Cost of Foreign Exchange Reserve Holdings in South Asia 10 Years of ASARC an International Conference , ASARC Working Paper 2004-14.

158

Arby, F. M (2001) Long-Run Trend, Business Cycles and Short-Run Shocks in Real GDP SBP Working Paper Series No.1, September 2001. Arellano, M. and S. Bond (1991) Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations Review of Economic Studies, Vol: 58, pp.277-297. Arif, G. M; H. Nazli and R. Haq (2000) Rural Non-agriculture Employment and Poverty in Pakistan The Pakistan Development Review, Vol: 39 : 4 Part II (Winter 2000) pp. 10891110. Arif, G. M; M. F. Kiani and K. H. Sheikh (2002) Labour Market Dynamics in Pakistan: Evidence from the Longitudinal Data The Pakistan Development Review, Vol: 41:4 Part II (Winter 2002) pp. 701 720. Arndt, S (1997) Globalisation and the Open Economy North American Journal of Economics and Finance , Vol: 8, pp.7179. Aslam, M. and K. Zulifqar (2008) Management of Human Resources: Employment Patterns and Dimensions of Unemployment in Pakistan National Conference on Socio- Economic Challenges Faced by Pakistan, International Islamic University Islamabad, June 2-3, 2008 (Unpublished Paper). Bailey, M. N. and R. Z. Lawrence (2004) What Happened to the Great US Job Machine? The Role of Trade and Electronic Offshoring Brookings Papers on Economic Activity, No. 2 (Fall), pp.201-260. Baird, S; J. Hamory, and E. Miguel (2008) Tracking, Attrition and Data Quality in the Kenyan Life Panel Survey Round 1 (KLPS-1) Center for International and Development Economics Research, Institute of Business and Economic Research, UC Berkeley. Ball, R. J. and E. B. A. St. Cyr (1966) Short Term Employment Functions in British Manufacturing Industry Review of Economic Studies, Vol: 33, No. 2, pp. 179-207. Baltagi, B (2008) Econometric Analysis of Panel Data 4th ed. Chicheter John Wiley. England. Baqai, M (1979) Pakistan's Pattern of Development and Prospects Pakistan Economics and Social Review, Vol: 17 (3-4), pp.1-62. Barro, R (1989) The Ricardian Approach to Budget Deficits Journal of Economic Perspectives, Vol: 3, pp.3754. _________ and X. Sala-I-Martin (1992) Convergence, Journal of Political Economy , Vol: 100, pp. 223-251. 159

Battese, G. E; S. J. Malik and N. Sultana (1993) Capital Labour Substitution in Large Scale Food Processing Industry in Pakistan: Some Recent Evidence The Pakistan Development Review, Vol: 32 (4) Part II (Winter 1993) pp.847-858. Beaudry, P. and F. Collard (2002) Why Has the Employment -Productivity Trade-off among Industrialized Countries Been So Strong? NBER Working Paper, No. 8754. Beck, N. and J. N. Katz (1995) What to do (and not to do) With Time -Series CrossSection Data American Political Science Review, Vol: 89, pp. 634-647. Bhorat, H. and M. Oosthuizen (2007) Employment Shifts and Jobless Growth Debate Human Resources and Development Review, 2008, pp.50-68. Bockerman, P (1998) Regional Evolutions in Finland Labour Institute for Economic Research, Working Papers, 142. Breitung, J (2000) The Local Power of Some Unit Root Tests for Panel Data in B. Baltagi (ed.), Nonstationary Panels, Panel Co-integration, and Dynamic Panels, Advances in Econometrics, Vol: 15, pp.161-178, Elsevier Science. Bowen, C. F; R. Radhakrishna, and R. Keyser (1994) Job Satisfaction and Commitment of 4-H Agents Journal of Extension , Vol: 32 (1). Burns, A. and W. C Mitchell (1946) Measuring Business Cycles National Bureau of Economic Research , New York, 1946. Caballero R. J. and M. L Hammour (1998) Jobless Growth: Appropriability, Factor Substitution, and Unemployment Carnegie Rochester Conference Series on Public Policy, Vol: 48 (0), pp. 5194. Campbell, J.R. and K.N. Kuttne (1996) An Anatomy of Labor Reallocation CarnegieRochester Conference Series on Public Policy , Vol: 44, pp.87-116. Caselli, F and J. Feyrer (2007) The Marginal Product of Capital The Quarterly Journal of Economics, Vol: May, 2007. Cazes, S. and A. Nesporova (2006) Introductory Paper for the Informal Meeting on Flexicurity in the West Balkan ILO, Geneva, 6 June 2006. Centre for Economic Performance (2002 ) Annual Review, 2000-01 Joint Publication by CEP/NBER, March 2002. Charles, B; L. Knox, and C. Marie (1976) Extended Hicks Neutral Technical Change The Economic Journal (Blackwell Publishing) Vol: 86 (344), pp. 845852. Chaudhary, M. A. and A. Hamid (1994) Ma gnitude of Unemployment and Underemployment in Pakistan presented in PakAIMS, Conference, Lahore. 160

___________________________ (1998) Unemployment Economic and Social Review, Vol: 36 (2), pp. 147-170.

in

Pakistan

Pakistan

Chenery, H. and M. Syrquin (1975) Patterns of Development, 1950-70 Oxford University Press, New York. Choi, I (2001) Unit Root Tests for Panel Data Journal of International Money and Finance, Vol: 20, pp. 249272. Cobb, C. W. and P. H. Douglas (1928) A Theory of Production American Economic Review, Vol: 18 (Supplement), pp. 139165. Durbin, J. and G. Watson (1950) Testing for Serial Correlation in Least Squares Regression Biometrica, Vol: 37, pp.409-428. Davis S.J (1987) Fluctuations in the Pace of Labor Reallocation Carnegie-Rochester Conference Series on Public Policy , Vol: 27, pp.335-402. Deardoff, A (2001) Fragmentation in Simple Trade Models North American Journal of Economics and Finance , Vol: 12, pp.121137. Deininger, K. and L. Squire (1996) A New Data Set Measuring Income Inequality. World Bank Economic Review, Vol: 10(3), pp. 565-91. Dickey, D.A. and W.A. Fuller (1979) Distribution of the Estimators for Autoregressive Time Series with a Unit Root Journal of the American Statistical Association , Vol: 74, pp.427-431. Dopke, J (2001) The Employment Intensity of Growth in Europe Kiel Institute of World Economics, Kiel Working Paper No. 1021. Egger, H. and P. Egger (2003) Outsourcing and Skill -Specific Employment in a Small Economy: Austria after the Fall o f the Iron Curtain Oxford Economic Papers, Vol: 55, pp.625-643. Egger, P; M. Pfaffermayr, and A. Weber (2003) Sectoral Adjustment of Employment: The Impact of Outsourcing and Trade at the Micro Level IZA Discussion Paper, No. 921, IZA Bonn. Elliott, G; T. J. Rothemberg, and J. H. Stock (1996) Efficient Tests for an Autoregressive Unit Root Econometrica, Vol: 64, pp. 813-836. Erbas, S. N. and C. L. Sayers (2001) Can a Shorter Workweek Induce Higher Employment? Mandatory Reductions in the Workweek and Employment Subsidies International Tax and Public Finance , Vol: 8, pp.485509.

161

Fair, R (1969) The Short -Run Demand for Workers and Hours Amsterdam: NorthHolland, 1969. Farsio, F. and S. Quade (2003) An Empirical Analysis of the Relationship between GDP and Unemployment Vol: 19, No. 3/4 2003 1. Feenstra, R. and G. Hanson (1996) Globalization, Outsourcing, and Wage Inequality American Economic Review, Vol; 86, pp. 240-245. Fields, G (2004) A Guide to Multisector Labour Market Models Paper Prepared for the World Bank Labour Market Conference , 1819 November 2004. Washington, D.C. Fofana, N. Z. F (2001) Employment and Economic Growth in the Cote d, Ivoire: An Analysis of Structural Determinants African Development Bank, 2001, Published by Blackwell Publishers, pp. 98-113. Francis, N. and V. A. Ramey (2003) The Sources of Historical Economic Fluctuations: An Analysis using the Long-Run Restrictions, manuscript, UC San Diego. Garibaldi, P. and P. Mauro (2002) Anatomy of Employment Growth in Economic Policy A European Forum, No. 34. April 2002. Geishecker, I. and H. Gfrg (2004) International Outsourcing and Wages: Winners and Losers Mimeo, University of Nottingham. Gennari, P (2004) The Estimation of Employment and Value Added of Informal Sector in Pakistan Paper presented at the 7th Meeting of the Expert Group on Informal Sector Statistics (Delhi Group) New Delhi, 2-4 February 2004. Ghannam, A. H (2008) The Relationship between Economic Growth and Employment in Saudi Private Firms Department of Economics, College of Administrative Sciences, King Saud University, Riyadh, Saudi Arabia. Glosser, S. and L. Golden (2004) The Changing Nature of Hours and Employment Adjustment in US Manufacturing, A Contributing Cause of the Jobless Recovery International Journal of Manpower , Vol: 25 No. 7, pp. 618-642. Gomme, P (2005) Accounting for the Jobless Recoveries Federal Reserve Bank of Cleveland, August 1, 2005. Gong, G. and W. Semmler (2006) Stochastic Dynamic Macroeconomics: Theory and Empirical Evidence Oxford University Press. Grg, H. and A. Hanely (2005) Labour Demand Effects of International Outsourcing: Evidence from Plant-level Data International Review of Economics and Finance , Vol: 14 (2005), pp.365376.

162

Greenaway, D; R. C. Hine, and P. Wright (1999) An Empirical Assessment of the Impact of Trade on Employment in the United Kingdom European Journal of Political Economy, Vol: 15, pp. 485500. Greenwood, J; G. MacDonald, and G. Zhang (1996) The Cyclical Behavior of Job Creation and Job Destruction: A Sectoral Model Economic Theory, Vol: 7, pp. 95112. Groshen, E. L. and S. Potter (2003) Has Structural Change Contributed to a Jobless Recovery? in FRBNY Current Issues in Economics and Finance , No.8 (www.newyorkfed.org/rmaghome/ curr_iss). Gujarati, D. N (2002) Basic Econometrics New York: McGraw-Hill, 4th edition, (January, 2003). ____________ (2006) Essentials of Econometrics New York: McGraw-Hill, 3rd edition, (2006). Gupta, S. D and A. Singh (2005) Will Services Be the New Engine of Indian Economic Growth? in Development and Change , Vol: 36(6), pp.1035-58. ____________________ and ILO (2005) Manufacturing, Services, Jobless Growth and the Informal Economy: Will Services be the New Engine of Indian Economic growth? University of Cambridge. Hadri, K (2000) Testing for Unit Roots in Heterogeneous Panel Data, Econometrics Journal, Vol: 3, pp. 148-161. Hamermesh, D. S (1986) The Demand for Labor in the Long Run in Ashenfelte, O. and R. Layard (ed.), Handbook of Labor Economics (New York: North-Holland Press, forthcoming 1986). Haq, M (1993) Social and cultural dimensions of change In Uner Kirdar and Leonard Silk (eds.), A World Fit for People, New York University Press. Haque, N. U. and I. Waqar (2006) Domestic CommerceThe Missing Link Islamabad: Ministry of Commerce, Government of Pakistan. Harrod, R. F (1939) An Essay in Dynamic Theory Economic Journal Issue , Vol: 39, pp. 14-33. __________ (1942) Toward a Dynamic Economics Macmillan, London. Hazeldine, T (1981) Employment Function and the Demand for Labour in the Shortrun In Zmira Hornstein, Joseph Grice and Alfr ed Webb, editors, The Economics of the Labour Market, HMSO, London.

163

Head, K. and J. Ries (2002) Offshore Production and Skill Upgrading by Japanese Manufacturing Firms Journal of International Economics , Vol: 58, pp.81105. Hemraj, T; J. Madrick, and W. Semmler (2006) Okuns Law and Jobless Growth The New School, SCEPA, Policy Note, no. 3. Henderson, J.M. and R. E. Quandt (1980) Microeconomics Theory: A Mathematical Approach 3rd edition New York: McGraw Hill, 1980. P.66. Hicks, J. R (1 932) The Theory of Wages London: Macmillan. Hine, R. and P. Wright (1998) Trade with Low Wage Economies, Employment and Productivity in UK Manufacturing Economic Journal, Vol: 108, pp. 15001510. Hodrick, R. and E. Prescott (1997) Postwar U.S. Business Cycles: An Empirical Investigation Journal of Money, Credit and Banking , Vol: 29, 1997, pp. 1-16. Human Development Report (1996) Economic Growth and Human Development UNDP, http://hdr.undp.org/reports/global/1996/en/. Hussain, I (1974) Employment Aspect of Industrial Growth in Pakistan Pakistan Development Review, Vol. 13, No. 5, pp. 211-221. Hyder, N.S (1994) Employment, Unemployment and Underemployment in Pakistan: An Overview of the Situation in Proceedings of the Pak-American Institute of sciences, Lahore, Pakistan, Im, K. S; M. H. Pesaran, and Y. Shin (1997) Testing for Unit Roots in Heterogeneous Panels Mimeo, Department of Applied Economics, University of Cambridge. Indiresen, P.V (2002) Vision 2020 The Jobless Growth Conundrum Business Line, Financial Daily from THE HINDU Group of Publications, Monday, Jul 01, 2002 International Monetary Fund (2009) International Financial Statistic s IMF, Washington D.C. International Monetary Fund (2009) World Economic Outlook October, 2009, IMF, Washington D.C. Islam, I. and S. Nazara (2000) Estimating Employment Elasticity for the Indonesian Economy ILO Technical Note, Jakarta: Indonesia. Islam, R (2004) The Nexus of Economic Growth, Employment and Poverty Reduction: An Empirical Analysis Recovery and Reconstruction Department, Geneva, ILO.

164

Jones, R. and H. Kierkowski (2001) Globalization and the Consequences of International Fragmentation In R. Dornbusch, G. Calvo, & M. Obstfield (Eds.), Festschrift in Honour of Robert Mundell, Cambridge, MA7 MIT Press. Jorgenson, D. and K. Vu (2005) Information T echnology and the World Economy Scandinavian Journal of Economics , Vol: 107 (4), pp. 631-650. Kaldor, N (1967) Strategic Factors in Economic Development, New York State School of Industrial and Labour Relations Cornell University , Ithaca NY. Kaldor, N (1968) Productivity and Growth in Manufacturing Industry: A Reply Economica, Vol 35 (November), 385-91. Kalim, R (2009) A Measure of the Elasticity of Substitution in the Manufacturing Sector of Pakistan The Lahore Journal of Economics , Vol: 16 (2), pp.43-56. Kangasharju, A. and J. Pehkonen (2001) Employment -Output Link in Finland: Evidence from Regional Data Finnish Economic Papers, Vol: 14 (1), pp.41-50. Kapsos, S (2005) The Employment Intensity of Growth: Trends and Macroeconomic Determinants Employment Strategy Dept, ILO, Geneva, Employment Strategy Paper No. 12/2005. Karmakar, S (2008) An Open Services Regime, Recipe for Jobless Growth Indian Council for Research on International Economic Relations, Working Paper No. 210. Katzner, D (1970) Static Demand Theory New York: Macmillan. Kazi, S (1987) Skill Information Employment and Earning in the Urban Informal Sector Pakistan Development Review, Vol: 26 (4), winter 1987, pp. 711-720. Kemal, A. R (1981) Substitution Elasticities in the Large-Scale Manufacturing Industries of Pakistan, Pakistan Development Review, Vol. XX, No. 1, Spring, pp. 1-35 __________ (1990) Protection, Industrial Development and Employment Generat ion in Pakistan HRD Working Paper, UNDP/ILO ARTEP, New Delhi, 1990. ________________Balance of Supply and Demand for Manpower GOP/UNDP/ILO ARTEP, Islamabad, 1990. _____________ and M. Irfan (1983) Employment and Manpower Project ions for the Sixth Plan Period Islamabad: Pakistan. _____________ and Z. Mahmood (1993) Labour Absorption in the Informal Sector and Economic Growth in Pakistan Friedrich-Ebert-Stiftung Islamabad, (Informal Sector Study No. 2).

165

Kennedy, P. E (1998) A Guide to Econometrics 4th ed. Cambridge, MA: MIT Press. Keynes, J. M (1936) " The General Theory of Employment, Interest and Money ", New York, Harcourt Brace Jovanovich, Inc. Khan, A (2007) Asian Experience on Growth, Employment and Poverty Report for the United Nations Development Program and International Labour Office , February 2007, 52pp. Kletzer, L (2000) Trade and Job Loss in U.S. Manufacturing, 19791994 In R. Feestra (Ed.), The Impact of International Trade on Wages, Chicago University Press, Chicago. Kohler, W (2001) A Specific-Factors View on Outsourcing North American Journal of Economics and Finance , Vol: 12, pp.3153. Krugman, P (1992) The Age of Diminished Expectations: US Economic Policy in the 1980s MIT Press, 1992. ___________ (1995) Growing World Trade: Causes and Consequences Brookings Papers on Economic Activity , I, pp.327377. ___________ (1999) The Accidental Theorist: And Other Dispatches from the Dismal Science W.W. Norton and Company, New York, London. Kwiatkowski, D; P. C. B. Phillips; P. Schmidt, and Y. Shin (1992) Testing the Null Hypothesis of Stationarity against the Alternative of Unit Root, Journal of Econometrics, Vol: 54, pp.159-178. Kwiatkowski, E., L. Kucharski, and T. Tokarski (2002) Bezrobocie i zartudnianie a PKB w Polsce w latach 1993-2001 [Unemployment and Labour use in Poland in View of GDP Changes (1993-2001)], Ekonomista 3. _____________ and W. Kwiatkowska (2006) The Lisbon Strategy, Competitiveness and the Labour Market, The Case of Poland International Trade and Finance Association, Working Papers 2006. _____________ S. Roszkowska, and T. Tokarski (2004) Granice wzrostu bezzatrudnieniowego w Europie i krajach WNP, Polish journal Ekonomista (The Economist). [Limits to Employment Non-Augmenting Growth in Countries of Europe and CIS], Ekonomista 1. Layard, R. and S. Nickell (1985) Unemployment, Real Wages and Aggregate Demand in Europe, Japan and the United States Carnegie Rochester Conference Series on Public Policy, Vol: 23, pp. 143- 202. ______________________ (1986) Unemployment in Britain Economica, 53: S121-69. 166

Levin, A; C.F. Lin, and C.S.J. Chu (2002) Unit Root Tests in Panel Data: Asymptotic and Finite-Sample Properties Journal of Econometrics, Vol: 108, pp.1-24. Lewis, J. D (2001) Policies to Promote Growth and Employment in South Africa World Bank, Informal Discussion Papers on Aspects of the Economy of South Africa, No. 16. Lilien, D. M (1982) Sectoral Shifts and Cyclical Unemployment Journal of Political Economy, Vol: 90(4), pp. 777793. Loots, E (1998) Job Creation and Economic Growth South African Journal of Economics, Vol: 66 (3), pp. 319- 336. Loungani, P; M. Rush, and W. Tave (1990) Stock Mar ket Dispersion and Unemployment Journal of Monetary Economics , Vol: 25, No. 3, pp. 367388. Lunenberger, D.G. (1995) Microeconomic Theory Mc-Graw Hill Inc. Lund, J (2006) Does Sweden Experience Jobless Growth? An Empirical Study of the Relationship between Unemployment and Growth in Sweden Jonkoping international Business School, Jonkoping University, Sweden (Unpublished Bachelor Thesis in Economics). Mackinnon, J. G (1994) Approximate Asymptotic Distribution Functions for Unit-Root and Co-integration Tests Journal of Business and Economic Statistics , Vol: 12, 1994, 167-176. Maddala, G. S. and Wu, S (1999) A Comparative Study of Unit Root Tests With Panel Data and New Simple Test Oxford Bulletin of Economics and Statistics, Vol: 61, 631-652. Maddison, A (1991) Dynamic Forces in Capitalist Development Oxford University Press, New York. Mahbub-ul-Haq, Human development Center (2004) Human development in South Asia 2003: The Employment Challenge Karachi: Oxford University Press. Mahmood, T; E. Ghani and Muslehuddin (2007) Efficiency of Large Scale Manufacturing in Pakistan: A Production Frontier Approach Pakistan Institute of Development Economics, Working Paper, 2007 (27). Mahmood, Z. and R. Siddiqui (2000) State of Technology and Productivity in Pakistans Manufacturing Industries: Some Strategic Directions to Build Technological Competence Pakistan Development Review, Vol: 39, 1-22. Majid, N (2000) Pakistan: Employment, Output and Productivity Issues in Development, International Labour Organization , Discussion Paper No. 33. 167

Mazumdar, D (2003) Trends in Employment and the Employment Elasticity in Manufacturing, 1971-92: an International Comparison Cambridge Journal of Economics, 27:563-582. ____________ and P. Basu (1997) Macroeconomic Policies, Growth and Employment: the East and South-East Asian Experience in AR Kahn & M Muqtada (eds.) Employment Expansion and Macroeconomic Stability under Increasing Globalisation London: ILO Studies Series, MacMillan. ____________ and S. Sarkar, (2007) Employment Elasticity in Organize Manufacturing in India Munk Centre for International Studies , University of Toronto, April, 2007. Milner, C. and P. Wright (1998) Modelling Labour Market Adjustment to Trade Liberalization in an Industrializing Economy Economic Journal, 108, 447, 50928. Mitchell, W. C (1927) Business Cycles: The Problem and Its Setting National Bureau of Economic Research , Inc, New York. _________and A. Burns (1938) Statistical Indicators of Cyclical Revivals, Chicago: National Bureau of Economic Research , Bulletin, 69, 1-12. Muslehuddin (2009) Global Financial Crisis: Implications For Macroeconomic and Development Policies in Pakista n Study Conducted at the Pakistan Institute of Development Economics (PIDE) under the Asian Development Bank (ADB-RETA) Project. Nadvi, K. M (1990) Employment Creation in Urban Informal Microenterprises in the Manufacturing Sector in Pakistan ILO/ARTEP Study, Karachi, July, mimeo. Newey, W.K. and K. West (1987) A Simple Positive Semi -definite Heteroscedasitcity and Autocorrelation Consistent Covariance Matrix Econometrica, Vol: 55, 1987, pp 703-708. Nickell, S (1986) Dynamic Models of Labour Demand In O. Ashenfelter, & R. Layard (Eds.), Handbook of Labour Economics, vol. 1. Amsterdam, 7 North Holland. __________(1988) The Supply Side and Macroeconomic Modelling In Bryant, R.C. et. al. (Eds.), Empirical Macroeconomics for Interdependent Economies, Washington: Brookings Institution: 202- 221. OConnell, P.J (1999) Astonishing Success: Economic Growth and the Labour Market in Ireland Employment and Training Department, International Labour Office , Employment and Training Papers, No. 44. Geneva. OECD (1994) The OECD Jobs Study: Evidence and Explanations, Part I - Labour Market Trends and Underlying Forces of Change Paris, OECD . 168

Okun, A.M (1962) Potential GNP: its Measurement and Significance American Statistical Association, Proceedings of the Business and Economic Statistics Section, 98-104. Onaran, O (2007) Jobless Growth in the Central and Eastern European Countries: A Country Specific Panel Data Analysis for the Manufacturing Industry Vienna University of Economics & B.A, Working Paper No. 103, March 2007. Padalino, S. and M. Vivarelli (1997) The Employment Intensity of Economic Growth in the G-7 Countries International Labour Review, Summer 1997, Vol: 136 (2) pp.199213. Pakistan, Government of (2007a) Pakistan Employment Trends, Skills, Labour Market Information and Analysis Unit Ministry of Labour, Manpower and Overseas Pakistanis, December, 2007.No.2. Pakistan, Government of (2007b) Sustainable Economic Growth, Emplo yment and Human Resource Development Ministry of Labour, Manpower and Overseas Pakistanis (Policy Planning Cell), Islamabad). Pakistan, Government of (2007c) Pakistan in the 21st Century, Vision 2030 Planning Commission of Pakistan Islamabad , August 2007. Pakistan, Government of (Various Issues) Labour Force Survey Islamabad: Federal Bureau of Statistics. Pakistan, Government of (Various Issues) Pakistan Economic Survey Islamabad: Finance Division. Pakistan, Government of (Various Issues) Statistical Year Book Islamabad: Federal Bureau of Statistics. Pakistan, Government of (2000) 50 Years of Pakistan in Statistics, 1947 -1997 Islamabad: Federal Bureau of Statistics, Volume-III. Pamukcu, T. and E. Yeldan (2005) Country Profile: Turkey Macroeconomic Policy and Recent Economic Performance Version -1, March 15. Pasinetti, L. L. (1981) Structural Change and Economic Growth. A Theoretical Essay on the Dynamics of the Wealth of Nations Cambridge, Cambridge University Press. Paul, G.D.P and S. Pellissery (2007) Jobless Growth to Inclusive Growth: Employability as an Alternative Planning Strategy ATLMRI Discussion Paper, 3/2007. Perugini C. and M. Signorelli (2007) Labour Market Performance Differentials and Dynamics in EU-15 Countries and Regions The European Journal of Comparative Economics, Special Issue on European Regions, Vol: 4, no. 2. 169

Pianta, M (1995) Technology and Growth in OECD Countries, 1979 -1990 Cambridge Journal of Economics, Vol: 19 (1), pp. 175-187. ________ R. Evangelista, and G. Perani (1996) The Dynamics of Innovation and Employment: an International Comparison Science, Technology Industry Review, 18: 67-93. Pini, P (1995) Economic Growth, Technological Change and Employment: Empirical Evidence for a Cumulative Growth Model with External Causation for Nine OECD Countries, 1960-1990 Structural Change and Economic Dynamics , 6: 185-213. _______(1996) An Integrated Cumulative Growth Model: Empirical Evidence for Nine OECD countries, 1960-1990 Labour, Vol: 10, 93-150. Prachowny, M. F. J (1993) Okun's law: Theoretical Foundations and Revised Estimates Review of Economics and Statistics, 75: 331-336. Rada, C (2008) Jobless Growth, A New Impediment to Development May 27, 2008, Available at http://rethinkingdevelopment.blogspot.com/2008/05/jobless-growth-new impediment-to.html. Ramsey, J. B. (1969). Tests for Specification Errors in Classical Linear Least Squares Regression Analysis, Journal of the Royal Statistical Society , Series B, 31, 350 371. Ramsey, J. B. and A. Alexander (1984). The Econometric Approach to Business -Cycle Analysis Reconsidered, Journal of Macroeconomics, 6, 347356. Rissman, E. R (1997) Measuring Labor Market Turbulence, Economic Perspectives, Federal Reserve Bank of Chicago, Vol: 21, No. 3, May/June, pp. 2 14. Rodrik, D. and A. Subramanian (2004) Why India can Grow at 7 Per Cent a Year or More Projections and Reflections Economic and Political Weekly, Special Article, April 17, 2004. Rutkowski, J. J. and S. Scarpetta (2005) Enhancing Job Opportunities: Eastern Europe and the Former Soviet Union World Bank, Europe and Central Asia Region, Washington D.C. Saget, C (2000) Can the Level of Employment be Explained by GDP Growth in Transition Countries (Theory Versus the Quality of Data) ILO Development Policy Group, Geneva, ILO. Sauramo, P (1999a) Jobless Growth in Finland? Evidence from the 1990 Labour Institute for Economic Research, Discussion Papers 159.

170

__________(1999b) Jobs and Growth in Finland: Industry -level Evidence from the 1990s Labour Institute for Economic Research , Discussion Papers 160. Schultz, T. P (2004) Evidence of Returns to Schooling in Africa from Household Surveys: Monitoring and Restructuring the Market for Education Journal of African Economies, Vol: 13, (Suppl.2). Schweitzer, M (2003) Another Jobless Recovery Federal Reserve Bank of Cle veland, Economic Commentary, March 1. Schreft, S. L. and A. Singh (2003) A Closer Look at Jobless Recoveries. Federal Reserve Bank of Kansas City, Economic Review, Second Quarter 2003, 88(2), pp. 45-73. Sengenberger, W (2006) Employment and Development in South-East Europe in the Context of Economic Globalization Sub-Regional European Conference on Globalization and Employment, Istanbul, 11 and 12 September 2006. Seyfried, W (2004) Examining the Relationship between Employment and Economic Growth in the Ten Largest States Southwestern Economic Review, pp.13-24. Shah, Z and Q. M. Ahmed (2003) Measurement of Cost of Capital for Foreign Direct Investment in Pakistan: A Neo-classical Approach A research paper submitted to Pakistan Institute of Development Economics (PIDE) to be presented in the Annual Conference of (PSDE) 2003. Sinha, A. and C. Adam (2004) Reforms and Informalisation: What Lies Behind Jobless Growth in India? EGDI and UNU-WIDER Conference, Unlocking Human Potential: Linking the Informal and Formal Sectors, 17-18 September 2004, Helsinki, Finland. ________ N. Sangeeta, and K. A. Siddiqui (2003) The Impact of Alternative Economic Policies on the Informal Sector: A Multisectoral Study NCAER, New Delhi. Solow, R. M (1956) A Contribution to the Theory of Economic Growth Quarterly Journal of Economics, February, 70: pp. 65-94. ___________ (1957) Technical Change & the Aggregate Production Function Review of Economics and Statistics, Vol. 39: 312-20. Srinivasan, T. N (2005) Productivity and Economic Growth The 21st Annual General Meeting and Conference, Pakistan Society of Development Economists , Islamabad. Stehrer, R (2004) Can Trade Explain the Sector Bias of Skill -Biased Technical Change? Vienna Institute for International Economic Studies, WIIW Working Paper no. 30. Vienna. 171

Swane, A. and H. Vistrand (2006) Jobless Growth in Sweden? - A Descriptive Study Masters Thesis in International Economics and Business, Stockholm School of Economics, Sweden, Unpublished. Tezcek, O (2007) An Alternative Approach to Overcome Jobless Gr owth in Turkey: A New Strategy of Industrialization Download free from, http://www.networkideas.org/feathm/mar2007/fa10_Ozlem_Tezcek.htm. Toit, C. and R. Koekemoer (2003) A Labour Model for South Africa The South African Journal of Economics, Vol: 71.1 March, pp.49-76. UNECE (2005b) Economic Survey of Europe 2005/2, Geneva . United Nation Development Programe (2008) Asian Development Bank (2008), Poverty Assessment Update Pakistan, 2008. Verme, P (2006) Constraints to Growth and Job Creation in Low -Income Commonwealth of Independent States Countries World Bank Policy Research , Working Paper 3893, April 2006. Walterskirchen, E (1999) The Relationship between Growth, Emplo yment and Unemployment in the EU Working Paper in European Economists for an Alternative Economic Policy, Workshop in Barcelona, 16 to 18 September 1999. Werner, H (1998) Labour Market Trends in the United States--Lessons We Can Learn . IAB Labour Market Research Topics No. 25, Available from
http://www.iab.de/topics25.pdf

Wolnicki, M; E. Kwiatkowski, and R. Piasecki (2006) Jobless Growth: A New Challenge for the Transition Economy of Poland International Journal of Social Economics, Vol: 33, No. 3, 2006, pp. 192-206. World Bank (1989) India: Poverty, Employment and Social Services A World Bank Country Study, World Bank, Washington D.C. _________ (2006b) Equity and Development World Development Report 2006, Washington D.C. __________ (2007) Pakistan: Promoting Rural Growth and Poverty Reduction Report: 39303-PAK, Sustainable and Development Unit, South Asia Region, Washington D.C. Wooldridge, J. M (2002) Econometric Analysis of Cross Section and Panel Data Cambridge, MA: MIT Press.

172

Zahid, N. S; M. Akbar and S. A. Jaffery (1992) Technical Change, Efficiency and Capital- Labour Substitution in Pakistans large- Scale Manufacturing Sector The Pakistan Development Review Vol: 31 (2), pp.165-188.

173