Vous êtes sur la page 1sur 10

SPECIAL ARTICLE

Where to Invest to Accelerate Agricultural Growth


and Poverty Reduction

Seema Bathla, Sukhadeo Thorat, P K Joshi, Bingxin Yu

T
This study aims to understand the drivers that helped he road to poverty alleviation is as tenuous as choosing
India achieve the challenging targets of the Millennium from diverse economic, social, and infrastructural
developmental programme for public investment. Public
Development Goal of reducing poverty before 2015.
investment, if rationally targeted, helps in reducing poverty
Have increased public investments or farm subsidies directly and/or indirectly through pro-poor growth (Fan 2008;
contributed to reducing rural poverty, directly through Sen 1997; Ahluwalia 1978). Empirical evidence on the effective-
various public spending schemes or indirectly through ness of public spending for poverty reduction is needed to make
judicious investment decisions that stimulate growth and reduce
increased agricultural land productivity? Utilising a
poverty. The allocation of appropriate public spending remains a
structural equation to answer this question for the major policy challenge for governments in developing countries.
period 198182 to 201314, it was found that education The seminal work by Fan et al (1999) empirically estimated the
and agricultural research and development produced impact of various public spending programmes on reducing pov-
erty in India during the period 19731993. The study revealed
the highest marginal returns for promoting agricultural
that improvements in agricultural productivity and increases in
income, while investments in rural infrastructure agricultural wages and non-agricultural employment opportu-
development and health provisions are the most nities have significantly contributed towards rural poverty re-
effective in reducing rural poverty. duction. The study also argued that investing in agricultural re-
search and extension services, rural infrastructure (especially
roads), education, and rural development, all targeted directly at
the poor, facilitated these three productivity-augmenting and
poverty-reducing factors. The same exercise was further extend-
ed by Fan et al (2008) up to 199899 with three improvements.
These included: (i) extending the study period; (ii) undertaking
decadal analysis for the 1960s, 1970s, 1980s, and 1990s; and
(iii) including input subsidies in the analysis. This exercise
further confirmed the continuing importance of agricultural
research, education, and rural infrastructure in promoting agri-
cultural growth and reducing poverty. Subsidies for credit, ferti-
lisers, electricity, and irrigation also contributed to the accelera-
tion of agricultural growth and reduction of poverty during the
1970s and 1980s, but their impact decelerated during the 1990s.
The present study is an attempt to further extend the analysis
up to 2014. The specific objectives are to: (i) study the composition
of and trends in public spending by social and economic cate-
gories, including subsidies on electricity, fertilisers, irrigation,
and credit from 198182 to 201314; (ii) identify the pathways
through which rural poverty can be reduced, and the role of
The authors are grateful to Shenggen Fan, Madhur Gautam, and
Joachim von Braun for their suggestions and revisions at various stages
public investment in the same; and (iii) estimate the marginal
of the study. impact of various types of public spending on agricultural
gross domestic product (GDP) and rural poverty reduction.
Seema Bathla (seema.bathla@gmail.com) teaches at the Centre for
the Study of Regional Development, Jawaharlal Nehru University,
New Delhi where Sukhadeo Thorat (thoratsukhadeo@yahoo.co.in) Conceptual Framework and Model Estimation
is professor emeritus. P K Joshi (p.joshi@cgiar.org) and Bingxin Yu Figure 1 (p 38) portrays the conceptual framework for the struc-
(byuifpri@gmail.com) are at the International Food Policy Research tural equation model (SEM). Public expenditure on investments
Institute, New Delhi and Washington, DC.
and subsidies is assumed to impact agricultural growth and
36 SEPTEMBER 30, 2017 vol liI no 39 EPW Economic & Political Weekly
SPECIAL ARTICLE

poverty through several channels: by improving technology RWage = f3(GDPGNA, AY, ELECT, ROAD, EDU, Health
and the availability of inputs and by increasing irrigation, rela- Status, MGNREGA) (3)
tive prices, wages, and non-farm employment. The use of NFEmpl= f4(GDPGNA, AY, RWage, ROAD, EDU, ELECT,
various farm inputs is influenced by the availability of re- RurDev Exp, Vill IndExp, MGNREGA) (4)
sources and the price at which these are available. Subsidised TT = f5(AY, World Price, GDPGNA, ELECT, Trend) (5)
prices incentivise farmers to adopt and use inputs. FERT = f6(Subsidiesfertiliser, credit, power, irrigation,
The impact of government investments and subsidies on TOT, IRRI, Agri R&D, ROAD) (6)
agricultural growth and rural poverty is examined by analys- IRRI = f7(Irrigation Exp, TT, Power Subsidy, Rain) (7)
ing the complex interlinkages between growth, productivity, ELECT = f8(Energy Exp) (8)
input use, the rural non-farm economy, and rural employ- ROAD = f9(Roadtransport Exp) (9)
ment and wages as discussed in the literature (Mogues et al EDU = f10(Education Exp) (10)
2012; Hazell et al 2000). The conceptual framework and Health Status (IMR) = f11(Health Exp) (11)
model is built sequentially, in three steps. First, it identifies Rural poverty is determined by agricultural income, that is,
the determinants of rural poverty through a poverty equa- GDP in agricultural and allied activities (GDPA) per unit of
tion. Second, it identifies public spending channels that influ- cropped area (AY ), rural wages (RWage), terms of trade; ((TT)
ence the identified determinants of poverty along with other three-year moving average); non-farm employment (NFEmpl),
factors. And, finally, it estimates the marginal impact of population density, and weather conditions (represented by the
selected public spending heads on agricultural productivity annual rainfall index). Wages from non-agricultural employ-
and rural poverty. ment is an equally important source of income in rural areas.
Figure 1: Analytical Framework Estimating the Impact of Public Urbanisation (captured through growth in the GDP of the non-
Expenditure and Subsidies on Productivity and Poverty agricultural sector or in per capita income in the non-agricul-
Public spending ture sector) drives jobs outside of agriculture, resulting in an in-
Investments
crease in the share of non-farm activities in the income portfo-
Subsidies lios of rural households. Tracking terms of trade helps meas-
Other exogenous variables:
Food
population growth, agro- ure the impact of changing agricultural prices relative to non-
ecological factors, urban
security and Agriculture agricultural prices to test the underlying hypothesis that the
employment inputs growth, macro and trade
programmes policy, world price
poor, who are the net buyers of food, may be negatively af-
Social services: Infrastructure: Technology:
fected by higher agricultural prices.
education, irrigation,
health, nutrition, power, road
Fertiliser, The second equation is a land productivity function, which
seeds, R&D
welfare
takes into account conventional inputs, namely land (LAND), la-
Agricultural Productivity
bour (LABOUR), irrigation (IRRI ), rainfall (RAIN ), and ferti-
liser use (FERT ), along with the public expenditure on agricul-
Non-agricultural tural R&D and other variables. The latter explains agricultural
employment:
(MGNREGA, village Non-farm Farm Food income through the education level of the rural population
industry, rural wages wages prices
development) (EDU ), road density (ROAD), and electricity use (ELECT). A
dummy variable is also used to capture the impact of the gov-
Rural Poverty
ernments flagship programmethe National Food Security
Dotted lines indicate indirect effects. Mission (NFSM )on productivity. The mission, initiated in
Figure 1 represents a system of equations that models the re- 2006, is expected to raise foodgrain productivity.
lationships between government spending and input subsidies Equation (3) captures the impact of various factors on rural
and agricultural growth through different pathways. Equa- non-farm wages and the resultant poverty alleviation. The
tion (1) explains rural poverty as determined by agricultural wage function is determined by land productivity (AY ), elec-
growth, changes in farm wages, non-farm employment, and trification (ELECT ), road density (ROADS), and education
terms of trade (agriculture prices relative to non-agriculture level (EDU). Some of these variables capture the impact of govern-
prices). This equation further endogenises agricultural growth, ment expenditure on poverty via improvements in farm and
terms of trade, rural wages, and non-farm employment, as non-farm activities. Non-agricultural GDP growth (GDPGNA) is
reflected in equations (2) to (5). Each of these equations are included to control for the effects of urban labour demand on
linked to input subsidies and public investment areas, such as rural wages. Health status is measured in terms of the infant
agricultural R&D, rural roads and transport, rural electricity, mortality rate (IMR) in rural areas to gauge the impact of
education, irrigation, and health as seen in equations (6) to expenditure on public health and nutrition (Health Status),
(11). The role of technology in areas using high yielding varie- which influences both farm and non-farm wages by improving
ties could not be used as they have been universally adopted the productivity of workers. The impact of another flagship
since 2000. programmethe Mahatma Gandhi National Rural Employment
Poverty = f1(AY, TT, RWage, NFEmpl, Pop Density, RAIN) (1) Guarantee Act (MGNREGA)on agricultural wages has been
AY = f2(Agri R&D, Land, Labour, IRRI, ELECT, EDU, captured through a dummy variable from 200506. The
FERT, Road, RAIN, NFSM) (2) programme guarantees each individual 180 days of employment
Economic & Political Weekly EPW SEPTEMBER 30, 2017 vol liI no 39 37
SPECIAL ARTICLE

in a year, and it is expected to have a positive influence on both Equation (8) models agricultural electricity consumption
farm and non-farm wages. This employment programme, in (ELECT) as a function of government expenditure on rural energy
conjunction with the NFSM, which again offers high subsidises, (Energy Exp). As is the case in equation (7), the agricultural
is expected to have a significant impact on poverty in rural areas. electricity subsidy is not included to avoid double counting.
Its impact is also taken into account in the non-farm employment Equations (9), (10), and (11) capture the relationships between
function in equation (4). past investments and improvements in roads, education, and
Non-farm employment is determined by the GDPGNA, land health. Public investments are expected to increase road density
productivity (AY ), wage rate (RWage), rural road density and years of schooling and decrease the infant mortality rate.
(ROADS), electrification (ELECT ), and education level (EDU ). Since the impact of investments on roads, education, and irriga-
The variables, road density and electrification are taken as tion usually lasts more than one year, we have considered these
infrastructure variables to capture their impact on investments investments as capital stock using a 10% depreciation rate. It is
in off-farm activities and the search for suitable jobs. Govern- also possible to determine the optimal lag length in variables in
ment expenditure on rural development and village industries place of stock using adjusted R2 or Akaike information criterion
is also taken into account. Most of the expenditure on rural (AIC) as done in Fan et al (2008). But, one has to deal with a
development is towards creating infrastructure and roads that loss in degrees of freedom and also address the problem of
may increase off-farm employment. The impact of MGNREGA high correlations among the lagged independent variables.
in generating off-farm employment is tracked through a dummy.
Education is highly correlated with other explanatory varia- Structural Equation Model
bles, and it is therefore excluded from the equation. The system is estimated using a Structural Equation Model
Equation (5) models the terms of trade. It is hypothesised (SEM). SEM is a methodology increasingly used to describe com-
that growth in farm output would increase the aggregate supply of plex systems in a multivariate setting (Kline 2011; Widaman
agricultural products and hence reduce commodity prices, and Thompson 2003). It provides a flexible framework to in-
which in turn would help the poor. A world price index of five vestigate more than one causal process among the variables.
commodities is included to gauge the impact of international By estimating multiple equations simultaneously, it has the
trade on domestic agricultural prices. ELECT is taken to gauge advantage of permitting the evaluation of networks of direct
the impact of infrastructure, and the demand-side effects on ag- and indirect effects along with different error structures. It
ricultural prices are captured through GDPGNA. Equation (6) is models the relationships among unobservable latent variables
modelled to determine fertiliser use per hectare in agriculture by allowing multiple measures to be associated with a single
by taking government subsidies for fertilisers (FERT ), credit latent variable. In this analysis, the specifications are based
(CREDIT ), irrigation (IRRIGATION), and power (POWER) into on relevant theory and literature with an aim to account for
account. Other variables include irrigation, road density, TT, socio-economic factors that are not captured in the model.
and agricultural research (Agri R&D). Most of these variables im- Double-log functional forms are used for all the equations in the
pact agricultural production through fertiliser use. The impact of system. State dummies are added to each equation to capture
research on productivity is captured indirectly through ferti- state-level unobservable, specific effects. We added a trend in
liser use. As elicited in Fan et al (2008), improved irrigation the trade equation to highlight the impact of supply-side vari-
and the development of new seeds through agricultural re- ables other than agricultural income and world prices. The
search would increase farmers demand for fertilisers. The ex- endogeneity problem, which generally occurs in time series
planatory variable, that is, TT is taken to be a three years mov- models, is controlled for by using variables in the lagged form or
ing average or lagged one year, as supply response is generally by respecifying it using the instrumental variable method.
based on last years price. The total poverty and productivity effects (both direct and
Equation (7) indicates the relationship between government indirect) of public expenditure and subsidies are obtained by
spending on major and medium irrigation projects (Irrigation taking into account two components: (i) the estimated elasticities
Exp) and the percentage of cropped area under irrigation of the variables in the poverty equation; and (ii) the elasticities
(IRRI). It would have been useful to consider cropped area as of other variables in the poverty equation that are affected by
per canal and tube well irrigation to capture the impact of the variables in other equations. For example, the effect of
government investment in canal irrigation on private investment agricultural R&D on poverty may work through various channels
in irrigation tube wells. Due to paucity of data across many through improved farm productivity, and increased productivity
states, the variable is considered in totality. Farmers investments can reduce poverty through terms of trade, input use, and
in irrigation are influenced by agricultural prices and power wages. The total effect of agricultural R&D on poverty is
subsidies. A favourable price structure, in turn influenced by summarised in equation (12).
power subsidies, is expected to instigate farmers to invest in Poverty Poverty AY AY FERT
= ( + )
tube wells and other machinery. Since expenditure on irrigation AgRD AY AgRD FERT AgRD
subsidies is subsumed under the total expenditure on Poverty TT AY AY FERT
+ ( + ) (12)
irrigation and flood control, it was excluded from the equation. TT AY AgRD FERT AgRD
Rainfall and TT are highly correlated, and hence the latter Poverty AWAGE AY AY FERT
+ ( + )
was dropped. AWAGE AY AgRD FERT AgRD

38 SEPTEMBER 30, 2017 vol liI no 39 EPW Economic & Political Weekly
SPECIAL ARTICLE

Based on the estimated elasticities, the marginal effect of Figure 2: Government Spending on Economic and Social Services
(` billion at 200405 prices)
poverty and different types of government subsidies and
10,000
expenditures are expressed as (i) increase in agricultural GDP
9,000
(rupees per unit of public spending averaged during 201112 to 8,000
Total expenditure
201314); and (ii) reduction in poverty headcount (number of 7,000
rural poor brought out of poverty per unit of spending). It enables 6,000
us to compare the relative benefits of additional units of 5,000
expenditure across different types of subsidies and investment 4,000
Social
Development
items. These can be taken as useful indicators while setting 3,000

priorities for government expenditure plans for accelerating 2,000


Economic
1,000
farm production and mitigating poverty.
0

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
2009
2010
2011
2012
2013
Trends in Public Expenditure and Subsidies in India Source: Authors calculations using data from Finance Accounts, Government of India.
This study is carried out at the state level and covers 17 major
states in India, spanning about 30 years from 198182 to expenditure has consistently outgrown population growth,
201314. The relevant data are drawn from different sources. and per capita development expenditure increased from `1,513
Time series data on state statistics is extracted from Finance in 1981 to `7,270 in 2013.
Accounts, Statistical Abstract of India, Agricultural Statistics at Nearly 57% of development expenditure went to social services
a Glance, Fertilizer Statistics of India, EPW Research Foundation in 201314, mostly education and health. Expenditure under
(EPWRF), and the annual reports of state electricity boards. economic services is further subdivided into various subheads.
Poverty estimates, based on National Sample Survey Office An analysis of the average share of various expenditure heads
(NSSO) consumption expenditure surveys, are taken from the for 19812014 reveals that nearly 27% was allocated to irrigation
Planning Commission and interpolated. We use data from 19 and flood control, followed by agriculture and allied activities
major states.1 The public expenditure series, state domestic (20%), rural development (14%), and rural transport (10%)
product, and other data sets given in nominal prices are con- (Figure 3, p 41).
verted into real prices at the 200405 base using income defla- Table 1: State Government Expenditure at 200405 Prices (` billion)
tors from the EPWRF. The wage rate has been converted into real Average 198189 199099 200013 19812013 Annual % Share
Growth Rate of
prices using the customer price index (CPI) published by the
(19812013) Capital
Ministry of Consumer Affairs. Splicing was done to convert Expenditure
the index to the common base of 200405. Annexure 1 (p 46) Total expenditure 1,399 2,391 5,692 3,521 6.73 12.3
furnishes the relevant variables, estimation method, and the Development 1,025 1,566 3,619 2,289 6.10 17.9
Social 512 815 1,968 1,221 6.54 5.56
sources for the data. The variables are specified on per hectare
Economic 513 751 1,651 1,068 5.61 31
or per capita basis. Compared to the previous study by Fan et
% share in
al (2008), some of the variables have been redefined or economic services 100 100 100 100
re-estimated due to better access to data, and a few new vari- Agriculture &
ables are included to improve the analysis.2 allied services 21.98 21.99 17.75 19.21 4.59 9.1
Public expenditure in India is broadly categorised into Rural development 15.12 14.82 13.02 13.68 4.96 6.7
Irrigation and
development and non-development expenditure; these are flood control 32.86 26.88 22.69 24.92 4.00 55.7
further bifurcated into revenue (current) and capital expen- Rural energy 1.06 4.22 4.87 4.23 12.37 28.9
ditures. Development expenditure refers to public spend for Rural road and
the promotion of economic development and social welfare, transport 9.74 8.70 11.75 10.84 6.59 9.3
and non-development expenditure refers to expenditure used Village and
small industry 2.86 2.33 1.35 1.75 2.19 45.6
to maintain governmental operations. All expenditure related % share in agriculture
to agriculture and rural development is generally development & allied services 100.00 100.00 100.00 100.00 - -
expenditure, directly charged from the revenue account. Capital Crop husbandry 23.17 27.98 30.25 28.64 5.67 2.50
expenditure refers to expenditure on capital formation such as Soil & water
conservation 5.98 6.93 5.24 5.76 4.04 20.90
assets creation, machinery, and construction. Animal husbandry 11.65 11.32 10.57 10.92 4.26 2.30
The total real public expenditure of all the states has Dairy development 13.32 8.45 3.67 6.29 -1.59 2.10
expanded by more than seven times from `1,088 billion in Fisheries 2.37 2.53 2.59 2.54 5.05 17.70
198182 to `9,288 billion in 201314, growing at a rate of Forestry & wildlife 22.09 19.64 17.58 18.76 3.60 10.60
6.73% per year (Table 1 and Figure 2). Development expenditure Food, storage &
warehousing 2.73 5.67 11.43 8.72 - -
increased at 6.1% annually, much slower than non-development
Agricultural research
expenditure, which grew at 8.3%. As a result, the share of & education 6.32 6.21 7.01 6.71 5.08 1.80
development expenditure declined continuously from 75% in Cooperation 11.07 10.20 11.06 10.85 4.41 21.70
1980 to below 60% in 2004; it then recovered to around 67% Others 1.29 1.07 0.60 0.82 - 17.1
in 201314. It is also promising to see that development Source: Authors calculations using data from Finance Accounts, Government of India.

Economic & Political Weekly EPW SEPTEMBER 30, 2017 vol liI no 39 39
SPECIAL ARTICLE
ELOOLRQ
Figure 3: Percentage Share of Various Services in Indias Economic Figure 4: Agricultural Expenditure and Input Subsidies (` billion)
Expenditure, 19812013
1,800
40 Irrigation 1,600
35 Fertiliser
Agriculture and allied Rural development
30 1,400 subsidy
25 1,200
Power
20 1000 subsidy
15 800 Credit
10 subsidy
Rural energy 600 Irrigation
5 subsidy
Rural road-transport 400
0
Agriculture
1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

2011

2013
200
expenditure
0

1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
Source: Authors calculations using data from Finance Accounts, Government of India.

Although the amount spent under every head within economic Source: Authors calculations using data from Finance Accounts, Government of India.
services has more than doubled, it lags behind the growth rate
Figure 5: Public Investment and Input Subsidies in Agriculture (` billion)
of total expenditure on social services. The highest growth
rate, over the last three decades, is found in rural energy, leading 300 1,400

to an increase in the shares of rural infrastructure like energy 250 1,200


and transport in economic service spending (Table 2, p 42). It 1,000
200
is a matter of concern that the shares of agriculture and irriga- 800
tion and flood control fell substantially. Expenditure on agri- 150
Public investment agriculture 600
culture and allied activities grew at a slow pace. The main sub- 100
400
heads related to agricultural and allied services are soil conser- 50
Input subsidy 200
vation, animal and crop husbandry, R&D, forestry, livestock devel-
0 0
opment, and cooperation. Within rural development, expendi-
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
ture on employment programmes constitutes a major share. Ex-
penditure on medium irrigation has a higher share within the Data Source: Input subsidies estimated using Finance Accounts, Annual reports SEB and
Fertilizer Statistics of India; Public investment in agriculture from NAS, CSO.
head of irrigation and flood control. Expenditure on dairy de-
velopment and rural employment has decelerated since the and is slightly lesser when irrigation expenditure is also con-
1990s, revealing that resources were diverted to activities like sidered. Input subsidies dropped sharply in the mid-2000s due
food storage and other rural development programmes. to a hike in the power tariff and changes in government priorities,
Within social services, education and sports received more but it surged swiftly afterwards, partly in response to the food
than half the total social service expenditure, followed by public crisis of 200708. Total input subsidies averaged `738 billion in
health (15.12%). Among all the social services, the percentage the 2000s, more than triple the amount allocated in the early
share of health and nutrition expenditure drastically 1980s. This growth rate surpassed population growth, and the
decreased from 18% to 12% from 1981 to 2013. Both revenue average per rural resident input subsidy nearly quadrupled
and capital expenditure increased at almost the same rate, from `150 in the early 1980s to `1,256 in 201314.
which was 6.7% during 19812013. However, revenue and capital Fertiliser and power subsidies, the major items among input
expenditure on social services increased at much higher annual subsidies, reached `390 billion and `345 billion, respectively,
rates of 6.4% and 8.6% compared to that of economic services, by 201314. Government support for power as well as fertilis-
which increase at rates of 5.2% and 6.4%. ers exhibited a brisk increase at almost 7.5% per annum over
Capital expenditure constituted a significant part of irrigation three decades. The power subsidy lost its position as the larg-
and rural road transport spend in the preceding decade. Both est among the input subsidies and its share dropped from 41%
irrigation and rural energy expenditure, largely capital in nature, in 198182 to 37% in 199091; it increased again to as high as
witnessed an increase in current expenditure since 1990, perhaps 62% during the early 2000s and then fell to 32% in 201314.
due to the rapid rise of subsidies to departmental commercial On the other hand, the share of fertiliser subsidies rose from
undertakings and state electricity boards or distribution losses. 32% in 198182 to 48% in 200910 and fell to 32% in 201314.
In the other economic expenditure heads, resources seem to Nearly 80% of the total allocated amount for input subsidies
have shifted from capital formation to government operational was used to provide power and fertiliser subsidies in 201314,
expenses and subsidies because almost all the expenditure whereas irrigation constituted only a 13% share in the total.
increase has been driven by soaring current expenditure, Figure 5 shows that input subsidies have increased far more
implying mounting bureaucracy and inefficiency in distribut- rapidly than public expenditure on capital formation in agri-
ing public expenditure. culture.3 They were about the same size (nearly `100 billion)
Agricultural subsidies cover many kinds of activities, the in 198182, but the gap quickly enlarged over time, partly due
major ones being irrigation, electricity, credit, and fertiliser to a stagnation in public investment in the 1990s and a decline
use. Expenditure on agriculture and allied services is usually in the marginal efficiency of irrigation investment.4 This reveals
overshadowed by subsidies as the amount spent on subsidies is that expenditure on farm subsidies limits the resources available
usually twice the size of agricultural expenditure (Figure 4) for financing public investment in irrigation and for that matter
40 SEPTEMBER 30, 2017 vol liI no 39 EPW Economic & Political Weekly
SPECIAL ARTICLE
Figure 6a: Regional Variations in Development Expenditure per Rural Figure 6b: Regional Variations in Input Subsidies (`/ha; averaged 198182
Resident (`, averaged 198182 to 201314) to 201314)
8,500 6,500
7,500 5,500
6,500
5,500 4,500
4,500 3,500
3,500
2,500
2,500
1,500 1,500
500 500

Maharashtra
BiharJharkhand

UPUttarakhand

West Bengal

Odisha

Assam

Rajasthan

MPChhattisgarh

Kerala

Punjab

Karnataka

Tamil Nadu

Haryana

Gujarat

Jammu and Kashmir


Andhra Pradesh

Himachal Pradesh

Maharashtra
BiharJharkhand

UPUttarakhand

West Bengal

Odisha

Assam

Rajasthan

MPChhattisgarh

Kerala

Punjab

Karnataka

Tamil Nadu

Haryana

Gujarat

Jammu and Kashmir


Andhra Pradesh

Himachal Pradesh
Source: Estimates based on data given in Finance Accounts, Annual reports SEB and Source: Estimates based on data given in Finance Accounts, Annual reports SEB and
Fertilizer Statistics of India. Fertilizer Statistics of India.

in agriculture, which does not appear to be the case during the Like public spending under various heads, the distribution
last decade. of subsidies across states has been exceedingly unequal as
Large spatial variations in the allocation of public expenditure well, since the average per hectare input subsidy in agriculture
and subsidies have been identified across states. With the ex- ranges from less than `1,000 in Assam and Odisha to almost
ception of Jammu and Kashmir, and Himachal Pradesh, per `5,000 in Punjab and Tamil Nadu from 1981 to 2013 (Figure
capita development expenditure is much higher in the eco- 6b). Generally speaking, a high per capita development
nomically advanced states, namely Punjab, Haryana, Tamil expenditure is associated with a high income level and a
Nadu, Karnataka, Gujarat and Maharashtra at nearly `4,000. higher rate technology adoption, and states with advanced
The per capita expenditure in poorer and less affluent states economies in the north-west tend to have a higher subsidy per
like Assam, Bihar, Odisha, Uttar Pradesh and West Bengal is hectare rate than their eastern counterparts. This underscores
half of what richer states spend. The difference is more than that current expenditure and subsidy distribution has exacer-
three times between the lowest and the highest spending states, bated inequality in the agricultural sector, contributing to
namely BiharJharkhand and Maharashtra. This explains the imbalances in input use, stagnation of productivity, and
higher incidence of poverty in the poorer and also the agricul- income growth in many states.
turally dominant states compared to the states with a higher
per capita income. However, the expenditure on development Results and Discussions
has grown at a faster rate in each state during the 2000s than The SEM estimation results are presented in Table 3 (p 43).
in the 1990s, which is a welcome sign. The annual rate of growth The coefficients of the poverty equation reveal that rural
in per capita development expenditure in BiharJharkhand poverty is significantly influenced by agricultural income,
(11.2%), Odisha (8.9%), and Assam (7.64%) has exceeded non-farm employment and wages, terms of trade, and annual
that of many advanced states (Table 2 and Figure 6a). rainfall. Relative prices are also significant. The coefficient
of terms of trade is significant at a 5% level and the elasticity
Table 2: Annual Rate of Growth in Development Expenditure
States Development Expenditure Development Expenditure
is 0.98 for the period of 198182 to 201314. This is closely
per Rural Resident followed by non-farm employment, which has an elasticity of
198189 199099 200013 198189 199099 200013
0.93, and non-farm wages at 0.55. High land productivity in
Andhra Pradesh 5.33 4.08 9.04 3.24 3.03 8.79
terms of income per hectare has helped reduce poverty
Assam 7.39 0.52 9.26 5.17 -1.04 7.64
significantly. Better annual rainfall also reduces poverty
Gujarat 7.57 7.61 7.54 6.05 5.93 6.51
Haryana 5.60 6.17 9.92 3.44 4.18 8.75
by ensuring income from agriculture. These results validate
Himachal Pradesh 8.54 7.55 6.47 6.63 5.95 5.14 the findings of the two studies conducted earlier on the
Jammu & Kashmir 8.15 5.01 5.81 5.51 3.02 3.72 impact of these variables on poverty reduction. Together,
Karnataka 6.63 5.36 9.34 5.05 4.01 8.45 they explain nearly 72% of all variations in the incidence
Kerala 3.65 5.39 7.92 3.29 4.42 6.43 of rural poverty as indicated in the R square. It reiterates
Maharashtra 6.86 5.54 8.65 5.05 4.05 7.51 that agricultural output per hectare, or land productivity
Odisha 6.13 2.47 10.22 4.40 1.14 8.91 facilitated by better rainfall conditions, and remunerative
Punjab 6.24 1.69 5.93 4.52 0.49 5.07 farm prices relative to the non-farm sector, and non-farm
Rajasthan 7.24 5.20 7.96 4.83 2.69 5.97
employment with better wages, have been the main
Tamil Nadu 5.41 1.94 9.87 4.10 2.38 9.16
sources of poverty reduction in rural areas during the
West Bengal 5.92 5.71 7.71 3.20 4.53 6.86
study period.
BiharJharkhand 5.36 1.47 14.16 3.97 1.26 11.26
MPChhattisgarh 5.61 4.38 9.81 3.51 5.64 6.99
The determinants of these factors that have reduced rural
UPUttarakhand 6.96 1.73 9.38 4.88 -0.01 7.38 poverty during 198182 to 201314 has been captured through
Total (of 17 states) 6.31 4.10 9.05 4.44 3.02 7.33 the SEM results. Rising agricultural income per hectare has
Source: Calculated by authors using data from Finance Accounts, Government of India. contributed to reducing rural poverty. The estimated results
Economic & Political Weekly EPW SEPTEMBER 30, 2017 vol liI no 39 41
SPECIAL ARTICLE

from the agricultural income (land productivity) equation The prices received by farmers for their produce relative to
suggest that it is positively influenced by spending on industrial products (terms of trade) are significant for poverty
agricultural R&D, land, labour, fertilisers, irrigation, electricity, reduction. The factors that influenced the relative prices are
and education. Among all the variables, elasticity is found to world prices for agriculture and rural infrastructure, repre-
be relatively high for education, represented by the number sented by power consumption. World prices had a positive
of years of schooling of rural population in each decade, and significant influence, which could be seen mainly during
followed by irrigation. The elasticity of the education variable the post-reform period when India embarked upon economic
is high at 0.42, suggesting that a 10% increase in literacy reforms. Comparative advantages in many agricultural
would raise agriculture income by 4.2%, followed by 0.18 for commodities coupled with a relatively greater openness to
irrigation. Public spending on agricultural R&D has been international trade and a favourable price regime have ena-
shown to have a positive impact on agricultural productivity, bled better price integration. While world prices and infra-
with an elasticity of 0.076. Rainfall also has a positive structure (electricity) have had a positive impact on the
impact on poverty. prices received by farmers, an increase in land productivity
Non-farm employment, which is an important factor for and the growth in the share of non-farm income has reduced
poverty reduction, is determined by agricultural land pro- the chances of farmers receiving relatively high prices for
ductivity, road density, growth in the size of gross domestic their products. Agricultural terms of trade are affected nega-
product from non-agriculture, rural development expendi- tively by land productivity as an increase in the commodity
ture, and employment under MGNREGA. Rural non-farm supply depresses farm prices. The effect of increasing non-
wages, which are equally important, are positively influ- farm incomes on the price received by farmers through
enced by agricultural land productivity, growth in non-farm increased demand is also not significant.
income, the education level of workers, and their health status. Fertiliser consumption and irrigation emerged as impor-
Employment under MGNREGA has a positive, but insignificant tant factors in agriculture productivity. Fertiliser use was
effect. Similarly, the road network, which facilitates the determined by irrigation and fertiliser subsidies, irrigation
mobility of labour, turns out to be more important for non- level, terms of trade, and agriculture R&D. In another equation,
farm employment than for increasing wages. The results in which fertiliser subsidy and terms of trade were excluded,
reveal that rural non-farm employment and non-farm wages power and irrigation subsidies, irrigation intensity, and agri-
together act as significant factors for poverty reduction, cultural R&D, including infrastructure in the form of roads,
and they are both enhanced by land productivity, size of turned out to be major factors that influenced fertiliser
non-farm income, and the education level and health status consumption. Public spending on major and medium irrigation
of workers. works, power subsidies, and terms of trade are positive
factors for irrigation use. Compared to power and credit sub-
Table 3: Determinants of Poverty and Agriculture Income, 198182
to 201314 sidies, fertiliser and irrigation subsidies appear to have a
Estimation Method: Maximum Likelihood; Log Likelihood =2056.73; N=495 greater positive impact on fertiliser use, while the power
(1) Poverty = 13.86* - 0.297 AY** - 0.98 TT* - 0.55 RWage* - 0.93 NFEmpl* subsidy is important for irrigation use. Lastly, education,
+ 0.19 RAIN** (R2 0.71) road density, health, and agricultural power consumption,
(2) AY = 7.24* + 0.076 Agri R&D* + 0.088 Labour* + 0.18 IRRI* + 0.071 ELECT* which are essential for agricultural productivity and poverty
+ 0.42 EDU* + 0.086 FERT* - 0.004 ROAD + 0.061 Land + 0.049 Rain* reduction, are significantly determined by the public spend
+ 0.082 NFSM* (R20.96) on these heads.
(3) RWAGE = 1.56*** + 0.097 AY*** + 0.26 GDPGNA* - 0.03 ELECT + 0.028 To sum up, agricultural productivity facilitated by better
ROAD + 0.47 EDU* -0.52 HS* + 0.88 MGNREGS (R2 0.88) rainfall conditions, remunerative farm prices, and non-farm
(4) NFEmpl = - 2.75* + 0.29 GDPGNA* + 0.16 AY* - 0.001 RWage + 0.065 employment at better wages have been the main sources of
ROAD** + 0.021 ELECT + 0.36 RurDev Exp* - 0.099 Vill Ind. Exp* + 0.061
poverty reduction in rural areas during the period from 1981
MGNREGS* (R2 0.89)
to 2014. Land productivity in turn has improved through
(5) TT = -10.65* - 0.27 AY* + 0.25 World Price* -0.24 GDPGNA** + 0.11 ELECT*
spending on agricultural R&D, fertilisers, irrigation, electricity,
+ 0.01* Trend (R2 0.65)
and education, beside land and labour. Increased fertiliser use
(6a) FERT = -1.13* + 0.54 Fertiliser Sub + 0.014 Power Sub + 0.035 Irrigation Sub
was facilitated by irrigation and fertiliser subsidies, increased
+ 0.16TT* + 0.045 IRRI** + 0.07 Agri R&D* + 0.15 ROAD* (R2 0.95)
irrigation level, terms of trade, and agriculture R&D, while
(6b) FERT = -0.85* + 0.01 Credit Sub + 0.056 Power sub* + 0.17 Irrigation Sub*
+ 0.18IRRI* + 0.53 Agri R&D* (R2 0.92)
irrigation use has been augmented by public spending on irri-
(7) IRRI = 1.54* + 0.093 Irrigation Exp*0.15 TT** + 0.092 Power Sub* (R2 0.88)
gation, power subsidies, and terms of trade. Rural non-farm
(8) ELECT = 5.54* + 0.24 Energy Exp* (R2 0.89) Road employment and non-farm wages, which also act as significant
(9) ROAD = -2.93* + 0.34 RoadTransport Exp* (R2 0.88) factors in poverty reduction, were enhanced by land produc-
(10) EDU = -2.93* + 0.63 Education Exp* (R2 0.91) tivity, growth in non-farm incomes, education, and the health
(11) HS = 7.36* - 0.49 HealthNutrition Exp* (R2 0.89) status of workers. Improvements in agricultural research,
*,**,*** indicate significance at 1%, 5%, and 10% levels. State effects are significant. education, health status, and power and road networks, which
Variables are specified on a per hectare basis; public expenditure is per capita; In equation
6 separate equations are estimated due to multicollinearity between fertiliser and credit
helped both in land productivity and poverty alleviation, were
subsidies. Test diagnostics of the model are not given for want of space. facilitated by increased public spending.
42 SEPTEMBER 30, 2017 vol liI no 39 EPW Economic & Political Weekly
SPECIAL ARTICLE

Returns from Agricultural Growth and Poverty Reduction is relatively lower than that of other public investments. The
Next, we discuss the marginal impact of the various categories marginal returns on the four subsidies are generally below
of public spending and subsidies on productivity growth and unity. Irrigation and fertiliser subsides hold fifth and sixth
poverty reduction; this will allow policymakers to compare rank, higher than the credit and power subsidies which come
the effects of one additional rupee spent on expenditure items at the ninth and tenth places, respectively. However, irrigation
or subsidies. It is helpful to prioritise government expenditure and fertiliser subsidies still yield relatively higher marginal
to further augment productivity and reduce poverty. returns compared to credit and power subsidies.
Table 4 presents the estimated total marginal effect on The ranking of the marginal impacts of various spend cate-
agricultural income and poverty reduction. The first column gories on rural poverty is quite different from the ranking for
provides the elasticity of each spending item, which gives the agricultural GDP. Expenditure on rural development ranks
percentage change in productivity or poverty corresponding first, which could be due to the big push on spending on
to one percent change in government spending. These elas- employment programmes during the 2000s. This is followed
ticities provide a measure of the relative growth and poverty- by health, energy, and education, in that order. However, there
reducing benefits that could accrue from additional expendi- are common themes in the rankings of spend categorises
ture on different items, where the increases are proportional according to their marginal impacts on agricultural growth
to existing levels of expenditure (Fan et al 2000). The esti- and poverty reduction; spending on health and power ranks
mates are based on a decomposition of various direct and high for both poverty and growth. One possible explanation is
indirect components, using each expenditure head and its that expenditure on power and health helps to facilitate
corresponding estimate of elasticity in the model as illustrated growth while simultaneously cutting down poverty through
in equation (12). To illustrate, productivity and wages each income generation, productivity, and other pathways. Input
account for almost 40% of the total impact on rural poverty. subsidies have a relatively low impact on poverty reduction.
As stated above, marginal returns are quantified by multiply- Similar to the returns to agricultural growth, the poverty
ing the relevant elasticities (direct and indirect) with the reduction from spending on irrigation was also lowest when
ratio of poverty or agricultural income to the relevant gov- compared with other spending heads.
ernment expenditure item in the recent year.
We now discuss the marginal returns on the 11 heads of public Policy Implications
spending and subsidies in terms of farm income. The selected First, this paper examines the pattern and composition of
11 heads of public spending had a positive impact on agriculture central government expenditure and subsidies across 17 states
income; however, not all had a similar marginal impact. The from 198182 to 201314. A structural equation model is then
marginal impact was found to differ quite significantly among applied to the data to relate government spending with devel-
the social and economic expenditure heads. The amount of opment goals of improving agricultural production and elimi-
public spending on education, agriculture R&D, health and nating poverty. Based on the estimated elasticities, and the
energy rank first four, in this order. The returns on the first direct and indirect pathways through which agricultural
three heads of spending, namely education, R&D, and health, income and poverty are influenced, the marginal effects of
stood as high at 2.64, 2.38 and 1.74, indicating 264%, 238% government investment and subsidies are ranked. The results
and 178% returns from additional public spending on these have policy implications for prioritising government spending.
services. Power comes next, implying that spending on rural Public expenditure on agriculture and irrigation grew at nearly
electricity yields relatively higher returns as compared to 4% per annum, while that on input subsidy (irrigation, ferti-
expenditure on roads or irrigation. Investment on rural devel- liser, power and credit) increased at 6.53% per annum. A high-
opment had relatively higher marginal impact than on roads er rate of growth close to 8% per annum on each head has
or irrigation. Irrigation investment has a positive impact, but it taken place between 2000 and 2013. Nevertheless, the share
Table 4: Agricultural Growth and Poverty ReductionReturns to of agriculture, irrigation and flood control fell substantially
Investments and Input Subsidies, 19812014 in total expenditure on economic services, which is a matter
Expenditure Elasticity Returns in Agricultural Returns in Rural
GDP Poverty Reduction
of concern.
Variable GSDPA Poverty (` per Rank (Decrease in Number of To accelerate the increase in agricultural income, more
` Spent) Poor per ` Million Spent)
expenditure on education, agricultural R&D, health, and energy
Education investment 0.557 -0.144 2.64 1 77 4
is essential as these bring high returns. Similarly, expenditure
Agriculture R&D 0.083 -0.006 2.38 2 20 6
Health investment 0.248 -0.134 1.74 3 105 2 on education and health lead to higher agricultural incomes
Energy investment 0.039 -0.028 1.15 4 95 3 and have an impact on poverty reduction through higher
Fertiliser subsidy 0.051 -0.007 1.01 5 14 8 wages and better non-farm employment. Undoubtedly, addi-
Irrigation subsidy 0.012 -0.001 0.92 6 4 10 tional spending on agricultural research, education, health, and
Rural development 0.062 -0.371 0.84 7 545 1
rural energy is the right strategy to ensure agricultural growth
Road investment 0.121 -0.074 0.82 8 56 5
and poverty alleviation.
Credit subsidy 0.018 -0.005 0.74 9 19 7
Power subsidy 0.040 -0.033 0.70 10 5 9 The low marginal impact of additional expenditure on irri-
Irrigation investment 0.014 -0.001 0.12 11 1.5 11 gation needs to be considered with some caution. An analysis
Based on averages from 201112 to 201314 using direct and indirect effects. of agroecological regions indicates that returns on irrigation
Economic & Political Weekly EPW SEPTEMBER 30, 2017 vol liI no 39 43
SPECIAL ARTICLE

investment tend to be higher in rain-fed areas where the irri- Credit and power subsidies, however, hold fewer prospects
gation level is low. The marginal impact in highly irrigated for agricultural growth. Input subsidies do entail a high
regions has possibly reached an upper ceiling, beyond which cost to the government, but at the same time, they incenti-
investment in irrigation will not yield high returns (Fan et al vise farmers to increase their use of inputs through lower
2000). Additionally, low returns may be related to the low prices and may, hence, increase agriculture productivity. In
marginal efficiency of investments in major and medium view of large inter-state disparities in the distribution of sub-
irrigation projects in some states. In this context, an increased sidies, proper targeting is also necessary so that subsidies
outlay of `500 billion under the Prime Minister Krishi reach needy farmers. Recent studies (Chirwa and Dorwad
Sinchayee Yojana is a welcome policy initiative. Rural devel- 2013) in African countries have indicated that better target-
opment is another expenditure head which has been shown ing and rationing of subsides induce rational use. Although
to be significant in terms of accelerating farm income and credit subsidies have a low return on agricultural producti-
lessening rural poverty.5 vity, it shows a high marginal return in poverty reduction.
Additional expenditure on fertiliser and irrigation subsidies This implies that it is useful to retain credit subsidies for
would also positively contribute to agricultural productivity. the poor.

Notes place of three-stage least squares (3SLS) to had a positive impact on employment and sev-
1 The newly created states, namely Jharkhand, capture the impact of latent variables. eral other aspects of the rural economy, such as
Chhattisgarh and Uttarakhand, are merged with 3 Data on gross capital formation in agriculture wages, agricultural productivity, and gender
their respective parent states, that is, Bihar, on public account, published in the NAS, CSO empowerment. In a span of 10 years, it is re-
Madhya Pradesh and Uttar Pradesh. refers to capital spending mainly on major and ported to have generated 19.86 billion person-
medium irrigation works (for details, see Gula- days of employment, benefiting 276 million
2 Literacy rate has been replaced with number of
ti and Bathla 2001). workers (Himanshu 2016; Desai et al 2015).
years of schooling; villages electrified has been
replaced with electricity consumption in agri- 4 The decline in public investment in irrigation
culture; public expenditure on rural develop- was attributed to a few extraneous forces such References
ment and village and small industries have as escalation of irrigation costs, the impact of Ahluwalia, M S (1978): Rural Poverty and Agricul-
been included. Expenditure on rural roads the environmentalist movement, and the fed- tural Performance in India, Journal of Devel-
has been broadened to include transport and eral character of the Indian state which led to opment Studies, Vol 14, No 3, pp 298323.
bridges. The infant mortality rate in rural areas inter-state river disputes (Shetty 1990; Mishra Bathla, Seema (2014): Public and Private
is taken to represent the effect of public invest- and Chand 1995; Bathla 2014). It could also be Capital Formation and Agricultural Growth:
ment on health and nutrition. Dummy varia- explained by a steep fall in overall capital ex- State-Wise Analysis of Inter-Linkages during
bles are specified to capture the impact of penditure, of which agriculture and irrigation Pre- and Post-Reform Periods, Agriculture Eco-
MGNREGA and the National Food Security seem to have borne the brunt. nomics Research Review, Vol 27, No 1, pp 1936.
Mission (NFSM). Public investment in irriga- 5 One of the important components of rural Chirwa, E and A Dorward (2013): Agricultural
tion excludes expenditure on flood control. development expenditure, namely MGNREGA. Input Subsidies: The Malawi Experience, New
Structural equation model (SEM) is used in Studies have shown that the programme has Delhi: Oxford University Press.

Money, Banking & Finance


March 25, 2017
Introduction Saibal Ghosh, Partha Ray
MONETARY POLICY
Reflections on Analytical Issues in Monetary Policy: The Indian Economic Realities A Vasudevan
Negative Interest Rates: Symptom of Crisis or Instrument for Recovery C P Chandrasekhar
Vulnerability of Emerging Market Economies to Exogenous Shocks Bhupal Singh, Rajeev Jain
BANKING CHALLENGES
Indian Banking: Perception and Reality Ashima Goyal
Non-performing Assets in Indian Banks: This Time It Is Different Rajeswari Sengupta, Harsh Vardhan
Were Public Sector Banks Victimised through AQR? A Strategic Orientation Perspective Shubhabrata Basu, Moovendhan V
Did MGNREGS Improve Financial Inclusion? Saibal Ghosh
How Efficient Are Indias Cooperative Banks? Evidence from DCCBs Sarthak Gaurav, Jisha Krishnan
Role of Fintech in Financial Inclusion and New Business Models Ajit Ranade
FINANCIAL MARKETS
Determinants of Bid-ask Spread in the Indian Government Securities Market Golaka C Nath, Sahana Rajaram,
Priyanka Shiraly, Manoj Dalvi
Long-run Performance of Seasoned Equity Offerings: New Evidence from India Soumya G Deb

For copies write to: Circulation Manager,


Economic and Political Weekly,
320322, A to Z Industrial Estate, Ganpatrao Kadam Marg, Lower Parel, Mumbai 400 013.
email: circulation@epw.in

44 SEPTEMBER 30, 2017 vol liI no 39 EPW Economic & Political Weekly
SPECIAL ARTICLE
Desai, S, P Vashishtha and O Joshi (2015): Mahatma Fertilizer Statistics of India, New Delhi: Fertilizer Others, Economic & Political Weekly, Vol 30,
Gandhi National Rural Employment Guarantee Association of India. No 25, pp A64A79.
Act: A Catalyst for Rural Transformation, Central Statistics Office (various years): Estimates of Mogues, T, Y Bingxin, S Fan and L McBride (2012):
New Delhi: National Council of Applied National Accounts Statistics, New Delhi: The Impact of Public Investment in and for
Economic Research. Government of India. Agriculture: Synthesis of the Existing
Fan, S (2008) (ed): Public Expenditures, Growth Ministry of Finance (various years): Finance Evidence, ESA Working Paper No 1206, Food
and Poverty: Lessons from Developing Countries, Accounts, New Delhi: Government of India. and Agriculture Organization of the United
New Delhi: Oxford University Press. Gulati, A and S Bathla (2001): Capital Formation Nations.
Fan, S, P B R Hazell and T Haque (2000): Targeting in Indian Agriculture: Trends, Composition Sen, A (1997): Agricultural Growth and Rural
Public Investments by Agro-Ecological Zone to and Implications for Growth, Economic & Poverty: In Growth, Employment and Poverty,
Achieve Growth and Poverty Alleviation Goals Political Weekly, Vol 36, No 20, pp 1697708. Change and Continuity in Rural India,
in Rural India, Food Policy, Vol 25, No 4, Hazell, P, S K Thorat and S Fan (2000): Impact of Growth, Employment and Poverty: Change and
pp 41128. Public Expenditure on Poverty in Rural India, Continuity in Rural India, G K Chadha and
Fan, S, P B R Hazell and S K Thorat (1999): Link- Economic & Political Weekly, Vol 35, No 40, A N Sharma (ed), New Delhi: Vikas Publishing
ages between Government Spending, Growth pp 358188. House.
and Poverty in Rural India, Research Report Himanshu (2016): A Decade of MGNREGA, Mint, Shetty, S L (1990): Investment in Agriculture:
10, Washington DC: International Food Policy 3 February. Brief Review of Recent Trends, Economic &
Research Institute. Kline, R B (2011): Principles and Practice of Struc- Political Weekly, Vol 25, Nos 78, pp 38998.
Fan, S, A Gulati and S K Thorat (2008): Invest- tural Equation Modelling, 3rd ed, New York: Widaman, K F and J S Thompson (2003):
ment, Subsidies and Pro-poor Growth in Rural Guilford Press. On Specifying the Null Model for Incremental
India, Research, Agricultural Economics, Mishra, S N and R Chand (1995): Public and Private Fit Indices in Structural Equation Modelling,
Vol 39, No 2, pp 16370. Capital Formation in Indian Agriculture: Com- Psychological Methods, Vol 8, No 1,
Fertilizer Association of India (various years): ments on the Complementarity Hypothesis and pp 1637.

Annexure 1: Exogenous and Endogenous Variables Used in the SEM and Data Sources
RAIN Annual rainfall index (Fertilizer Statistics of India)
GDPGNA Annual rate of growth in non-agriculture state domestic product (SDP) or non-agriculture income per capita (National Accounts
Statistics (NAS), Government of India)
World Price World Food Price Index, 2005 = 100, includes cereal, vegetable oils, meat, sea food, sugar, bananas, and oranges price indices
(International Financial Statistics (IFS))
Population Rural population (Census of India)
Health status (IMR) Proxied by infant mortality rate in rural areas (Sample Registration System (SRS), Registrar General, India)
TT Terms of trade based on three years moving average; estimated taking SDP agriculture current/constant price divided by non-
agriculture SDP at current/constant price at 200405 (EPW Research Foundation and NAS)
Poverty Rural population falling below poverty line based on the headcount ratio (Planning Commission, based on the Lakdawala approach)
AY Agricultural incomeGDP per NSA (NAS and Agricultural Statistics at a Glance)
NAWage Rural wage rate (NSSO)
NFEmpl Percentage of non-farm employment in total rural employment (NSSO)
Agri R&D Public expenditure on agriculture R&D, soil conservation, crop and animal husbandry (Finance Accounts)
Labour Agriculture labourall age group as per UPSS (NSSO)
IRRI Percentage of cropped area irrigated by public and private sources (Agricultural Statistics at a Glance)
ELECT Electricity consumption in agriculture per hectare (annual reports of state electricity boards and Agricultural Statistics at a Glance)
EDU Number of years of schooling of rural workers bifurcated by educational categories and then divided by total population (estimated
using NSS data)
FERT Fertiliser consumption per hectare (Fertilizer Statistics of India)
ROAD Road density measured in km per 000 sq km (Statistical Abstract of India and Rural Development Statistics)
Land Gross cropped area (Fertilizer Statistics of India)
NFSM Dummy for National Food Security Mission from 2006
RWage Non-agricultural wage rate (NSSO)
MGNREGA Dummy for MGNREGA from 2006
RurDev Exp Public expenditure on rural development (Finance Accounts)
Vill IndExp Public expenditure on village industry (Finance Accounts)
Fertiliser Sub Fertiliser subsidy estimated using unit rate based on all India estimates and then apportioned across states based on consumption
(Fertilizer Statistics of India)
Power Sub Power subsidy based on unit cost of supply minus agriculture tariff * consumption of power in agriculture (annual report of SEB)
Credit Sub Credit subsidy (only interest) based on loan outstanding of Primary Agricultural Credit Society, Rural Regional Banks, and commercial
banks. based on differences in the lending rate of agricultural and non-agricultural sectors (NAFSCOB)
Irrigation Sub Irrigation subsidy (1) based on operation and maintenance (O&M) expenditure: receipts + interest income/receipts; Second estimate
= irrigation subsidy (1) + 1% of cumulative capital expenditure (exclude expenditure on flood control. Capital stock base year: 1981
(Finance Accounts))
Irrigation Exp Public expenditure on minor, medium, major irrigation and command area development (Finance Accounts)
Energy Exp Public expenditure on rural power; bifurcated as per consumption in agriculture (Finance Accounts)
Roadtransport Exp Public expenditure on rural road transport, roads, and bridges. Division as per the share of rural population in total (Finance Accounts)
Education Exp Public expenditure on education, sports, art, and culture (Finance Accounts)
Health-nutrition Exp Public expenditure on medical and public health and social and welfare nutrition (Finance Accounts) Public spending on each head
comprises revenue and capital expenditure. The latter is measured as a stock variable based on accumulated expenditure at the end of
1981, that is, the base year in this study. The annual depreciation is taken to be 10%.

Economic & Political Weekly EPW SEPTEMBER 30, 2017 vol liI no 39 45

Vous aimerez peut-être aussi