Policies of Ethnic Preference in Pakistan

Author(s): Charles H. Kennedy

Source: Asian Survey, Vol. 24, No. 6 (Jun., 1984), pp. 688-703
Published by: University of California Press
Stable URL: http://www.jstor.org/stable/2644399 .
Accessed: 23/08/2011 04:56
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

University of California Press is collaborating with JSTOR to digitize, preserve and extend access to Asian
Survey.

http://www.jstor.org

Policies of Ethnic Preference in Pakistan

-__

Charles H. Kennedy

Perhaps to an unusual degree Pakistan has suffered

since its birth from disabilities associated with regional differences and
resultant competing national demands. From Partition in 1947 until the
dismemberment of the state in 1971, Pakistan served as the ideological
battleground for antithetical visions of Punjabi and Bengali nationalisms.
Indeed, the consequences of such conflicting visions-the Civil War of
1971 and the emergence of Bangladesh-have marked the only instance
in the 20th century of a successful secessionist movement. Ominously
during the past decade, the specter of regionalism and possible future
secessionist sentiment has been voiced by disaffected Pathan, Baluch, and
Sindhi leaders.
Ironically, however, throughout its history Pakistan has adopted policies of ethnic preference formulated to address the demands of disaffected groups. Further, such policies have been rigorously implemented
and have lessened ethnic inequality in governmental employment. This
article analyzes Pakistan's ethnic quota. It is divided into four sections:
The first examines the rationale for adopting the quota; the second details
the history of ethnic preference policies in Pakistan; the third examines
the operation of the system; and the final section offers some cautious
lessons derived from the Pakistani experience.

Rationales
Pakistan is a multiethnic state. Since the secession of Bangladesh in
1971,' Pakistan has contained four major ethnic groups. In numbers, the
largest of these are the Punjabis, followed by the Sindhis, the Pathans,
and the Baluch.2 There is also a significant "tribal population." Each of
Charles H. Kennedy is Assistant Professor in the Department of
Government and Legal Studies, Bowdoin College, Brunswick, Maine. The research for this
article was generously supported by a research grant from the American Institute of Pakistan
Studies and by Bowdoin College. Earlier versions of this paper were presented to the 11th
Wisconsin Conference on South Asia and to the 19th Northeast Political Science Association Meeting (1982).
a 1984 by The Regents of the University of California
1. Before 1971 Bengalis constituted roughly 55% of Pakistan's total population.
2. Approximate populations of provinces in 1982 are: Punjab 54 million, Sind 19
688

CHARLES H. KENNEDY

689

these groups is defined by an admixture of linguistic and political

attributes-Punjabis in Punjab speak Punjabi as their mother language,
Sindhis in Sind speak Sindhi, Pathans in the North-West Frontier Province (NWFP) speak Pushtu, and the Baluch in Baluchistan speak Baluchi
or Brauhi. Tribal groupings live in all four of the provinces, but the
greatest concentrations are in the Federally Administered Tribal Areas
(FATA) and Azad Kashmir. However, there is significant slippage in such
definitions of ethnic identity. Considerable numbers of Punjabis,
Pathans, and Sindhis live outside their respective provinces. This is
particularlytrue in regard to the ethnically diverse major urban areas of
Pakistan-Karachi, Lahore, and Rawalpindi. Further, more Pathans than
indigenous Baluch live in Quetta, the capitalof Baluchistan. There is also
a slippage in regard to linguistic determinants of ethnicity. Pakistan has
one national language, Urdu, and one official language, English. The
great majority of Pakistanis, approximately 90%, speak or at least understand Urdu.i Regional languages are spoken by comparatively fewer
people, and the proportion of those domiciled in a given region who
speak the relevant regional language varies widely from region to region.
Indeed, it can be demonstrated that the self-declaration of mother
language is more indicative of political sentiment than of linguistic competence.5
However loose the definition of ethnic groupings in Pakistan may be,
indisputable gaps exist in the respective levels of development of the four
provinces. Generally speaking, the provinces of Punjab and Sind are the
most highly developed, while the other provinces are relatively less
developed. Indicators of such gaps include differentials in per capita
income,6 life expectancy,7 and levels of industrialization.8 Such gaps are
million, NWFP 14.6 million (including tribal areas, but excluding Afghan refugees), and
Baluchistan 3.25 million. Comparable figures for ethnic communities do not exist. Source:
Calculated by author using 1972 census data and assuming 3% annual growth rate per
province.
3. Selig Harrison, In Afghanistan's Shadow. BaluichNationalism and Soviet Temiptation
(Washington: Carnegie Endowment, 1980).
4. See Agha Iftikhar Husain, "Introduction of Urdu in Administration" in Agha Iftikhar
Husain (ed.), Studies in Public Administration of Pakistan (Islamabad:Pakistan Administrative Research Centre, 1979).
5. For instance, roughly one-third of the successful candidates for the CSS examination
who claimed Pakhtun as their mother language during 1974-77 did not speak or read the
language. Source: Compiled by author from relevant biodata on probationers.
6. ShahidJaved Burki calculates that in 1977 the mean per capita income in the two richer
provinces (Punjab and Sind) was 28% higher than that of the poorer provinces. "A Note on
Perspectives on Economic Development and Regional Inequalities in Pakistan" (paper
presented to the 32nd Annual AAS meeting, 1980), p. 4.
7. In 1977 the mean life expectancy in rural Baluchistan was 42 years, and in rural
NWFP 44 years. Mean life expectancy in the Punjab and the Sind was 60 years. ibid., p. 7.
8. In 1977 there were 3,000 "registered factories" with over 20 employees in the

690

ASIAN SURVEY, VOL. XXIV, NO. 6, JUNE

1984

widely perceived and politicized in Pakistan, and have resulted in the

perception of the actual or potential specter of Punjabi, or to a lesser
extent Sindhi, domination. Indeed, the most traumatic event in the
history of Pakistan, its dismemberment in 1971, was occasioned by the
perception of widening inequality between East and West Pakistan. Currently, Pakistanfaces several prospective Bangladeshes. The most serious
is the demand for greater provincial autonomy in Baluchistan, but there
are also significant separatist sentiments in the Sind, and intermittentthough recently dormant-demands for an independent Pathan statePakhtunistan.9
A third ingredient that has influenced the growth of the quota system
has been the inability of Pakistan to achieve balanced levels of institutional development. Pakistaninherited a political system from the British
that was crafted to suit the needs of a colonial power-i.e., its overriding
concern was to rule a subject people. As a consequence, a primary tool to
effect this control, the administrative system inherited by the new state,
demonstrated highly complex patterns of organization, well-established
forms of socialization for its members, and a remarkabledegree of institutional autonomy. Countervailing institutions, particularlythose responsible for ensuring governmental responsiveness to the demands of the
public (legislatures, electoral bodies, local governments, etc.) existed in
only an attenuated form. This developmental gap has persisted during the
37 years since independence. For example, Pakistan has been unable to
establish the precedent of orderly succession of political leaders. The
outcomes of the three national elections held in Pakistan- 1965, 1970,
and 1977-were each challenged by the losers as "rigged," and only in
the first did the "winner" remain in office.10 Similarly, Pakistan, ostensi-

Punjab and 2,889 in the Sind, while the NWFP had only 262 and Baluchistan 9. Pakistan,
Statistics Division, Enquiry on LabourWelfare, 1977 (Karachi, 1981), pp. 1-15. Similarly in
1977, of all "establishments with 20 or more employees excluding defence establishments,"
7,859 were found in the Punjab, 2,673 in the Sind, 1,046 in NWFP, and 290 in Baluchistan.
Pakistan, Reporton Annual EstablishmentInquiry, 1976-1977 (Karachi, 1981), p. 1.
9. Useful overviews of the nationalities question are found in Lawrence Ziring, Pakistan.'
Enigma of Political Development(Boulder: Westview, 1980), Khalid bin-Sayeed, Politics in
Pakistan (New York: Praeger, 1981), and in Selig Harrison, In Afghanistan's Shadow.
10. Actually the "winner," General Ayub Khan, was already in power in 1965. The 1970
election resulted in an overall plurality for Sheik Mujibur Rahman, and the eventual Prime
Minister Zulfikar Ali Bhutto received only a plurality in West Pakistan.The aftermath of the
1977 election resulted in charges by the Pakistan National Alliance of widespread rigging
and fraud on the part of the PPP. Partly as a consequence of these charges, Bhutto was
overthrown by a military coup that installed Pakistan'scurrent leader, General Zia-ul Haq, in
July 1977.

CHARLES H. KENNEDY

691

bly a republic, has been subject to extensive periods of martial law.

During such periods, elected nationalassemblies,"1albeit only marginally
consequential during periods of civilian rule, have been disbanded, and
political activity has been prohibited or placed under severe restraints.
Also, despite numerous attempts at reform, local governments are still
dominated by civilian bureaucrats deputed from the federal goverment
for service in the provinces. Functionally, such "district administration"
has performed the roles typically entrusted in other systems to elected
officials-revenue collection, revenue distribution, rule adjudication, and
the execution and formulation of local policy.12
Given this context, meaningful input into the policy-making process
can only be ensured by securing representation in the civilian bureaucracy. Therefore, to represent the interests of ethnic groupings within
Pakistan or to redress the consequences of inequalities between such
groupings requires perforce direct incumbency of "nationals" in the
civilian bureaucracy.It is this that makes the quota system so important to
federal policy making in Pakistan. It can be argued that denial of civilian
bureaucratic office in Pakistan is functionally equivalent to the denial of
political representation.
Combined, these three ingredients-ethnic diversity, unequal regional
or provincial development, and unbalanced institutional growth-explain
the nature of Pakistan's policies of preference. Indeed, Pakistan's quota
policy is governed by two partiallycontradictory principles: (1) the remedial or compensatory principle, and (2) the proportional representational
principle. The first provided legitimacy for the implementation of the
quota. Before 1971, the quota was designed to ameliorate inequalities
between East and West Pakistan,and after 1971, inequalities between the
Punjab and "less-developed" regions. The latter principle, that of proportional representation, has been invoked both as a goal and as a strategy for
effecting the compensatory principle-that is, since there is a loose fit
between ethnic groupings and provincial domicile, proportional representation favors less-developed regions or groups.

History of the Quota Policy

A number of factors, the most salient being the disparity of service
representation between East and West Pakistan immediately subsequent
11. The Majlis-i-Shura (Federal Council), established in January 1982, is an advisory
body wholly appointed by the Chief Martial law Administrator.
12. Major local government programs discontinued because of lack of effectiveness
include (in order of original implementation): Panchayats,Village Panchayats,Basic Democracies, Integrated Rural Development, and the People's Works Programme. Since 1979 the
Local Government Programme has been in effect.

692

ASIAN SURVEY, VOL. XXIV, NO. 6, JUNE

1984

to independence in 1947, led to the introduction in 1949 of a federal

quota system that applied to candidates seeking competitive entry to
officer level ranks in the federal bureaucracy.13This policy provided that
20% of the vacancies in the Central Superior Services were to be filled on
the basis of "merit" as a consequence of the Central Superior Services
Examination (CSS examination). The remaining 80% of the vacancies
were to be filled according to the following formula: East Pakistan 40%;
Punjab and Bahawalpur 23%; Karachi 2%; and Sind, Khairpur, NorthWest Frontier Province and Frontier States and Tribal Areas, Baluchistan,
Azad Kashmir, and Kashmir refugees, 15%.'4 As originally designed, the
quota had application to approximately one hundred vacancies per year.
Its thrust was to "prudently" increase the representation of Bengalis, who
were woefully underrepresented at the time of Partition, by in effect
creating separate competitive pools in the two wings. Advocates of the
quota perceived it as a temporary, but necessary, remedial expedient that
would be phased out five to ten years after its introduction.15 However,
the quota grew and prospered in the fertile soil of regional animosities
between East and West Pakistan. By 1956, the quota policy, which had
started as an administrative directive within the Establishment Division,
had grown to the status of a statutory exception to the "nondiscrimination
clause" in the first constitutional6 This status was reiterated in the 1962
Constitution.17 The quota also became a topic of frequent debate in the
National Assembly. Perhaps more important, however, the range of the
quota also steadily expanded. By the early 1950s the quota was used for
vacancies filled by the Federal Public Service Commission (FPSC) through
interview, and it also slowly became applicable to departmental and attached departmental recruitment for posts in the central government.
Indeed, by 1971, the quota was in use for approximately 2,000 entry-

13. The first application of the quota system for which written documentation exists
occurred in 1950 with application to that year's entering batch of recruits. Pakistan,Second
Reportof the Pakistan Public ServiceCommission
for the Period lstJanuiary to31st December1949
(Karachi, 1950), p. 7. However, according to members of the 1948 batch of recruits (those
who entered service in 1949), the quota was in effect a year earlier, in 1949. The terms of
the 1949 quota, slightly different from the 1950 quota, were: 15% merit, 43% East
Pakistan, 42% West Pakistan. West Pakistan was further subdivided into Punjab 23%;
NWFP, Sind, Baluchistan, and Northern Areas and Tribal Territories (NATT) 17%o;and
Karachi 2%. Source: Interviews.
14. Ralph Braibanti, "The Higher Bureaucracy of Pakistan," in Braibanti, et al., Asian
BureaucraticSystemsEmergentfrom the British ImperialTradition (Durham: Duke University
Press, 1966), p. 265.
15. Interviews with principals.
16. Article 17.
17. Article 240.

CHARLES H. KENNEDY

693

level positions in the federal government each year, an increase of over

2000% in 22 years.'8
Despite the phenomenal growth of the quota, which was designed to
keep the two wings together, centripetal tendencies became too great,
and the resultant civil war left Pakistandismembered. With the secession
of Bangladesh went the original rationale for the quota system. However,
instead of disappearing, the system has become increasingly vigorous during the past decade. In 1972-73, immediately following the secession of
Bangladesh, the quota was temporarily transformed into a confusing array
of six zones and four "provinces.'9 But by August 1973, it emerged in
its present form: 10% merit; 50% Punjab (including Islamabad);7.6%
urban Sind (Karachi, Sukkur, and Hyderabad); 11.4% rural Sind (areas in
Sind other than those above); 11.5% NWFP; 3.5% Baluchistan; 4%
Northern Areas and FATA; and 2% Azad Kashmir.20Why this particular
formula was chosen is open to conjecture. It is true that the percentages
roughly correspond to the respective populations of the regions2' and
that the modified system follows fairly closely the relative weightings of
the 1949 formulation, but no public sources indicate that such standards
were used in formulating the regional quotas. More important, the
broader questions of whether the quota should have been continued in
the aftermath of the 1971 war, and if continued, what form it should have
taken, seem to have had no formal public consideration, nor were they
addressed directly by any of the numerous administrative commissions of
the late 1960s and early 1970s.
In any event, following the war Prime Minister Bhutto's economic
policies served as a catalyst for the continued expansion of the quota
system. Between 1972 and 1975, Bhutto nationalized numerous industries (banking, insurance, heavy machinery, natural resource extraction,
18. 367 candidates entered through direct competition (CSS Examination), 912 entered
through Central Public Service Commission direct recruitment (by interview), and approximately 800 were admitted through departmental and attached departmental recruitment.
Source: Calculated by author.
19. As of January 1972, the quota was constitued as follows: Zone 1 = Quetta, Kalat,
and Tasbela;Zone 2 = Hyderabad and Khairpur Divisions; Zone 3 = Iahore Division and
districts of Rawalpindi, Gujrat, Sargodha,Lyallpur,Multan, and Sahiwal;Zone 4 = Bahawalpur Division and districts of Muzaffargarh, Dera Ghazi Kahn, Campbellpur, Jhelum,
Mianwali, and Jhang; Zone 5 = Agencies, states and tribal areas, including Added and
Special Areas Adjoining Settled Areas; Zone 6 = Peshawar and Dera Ishmail Khan Divisions (excluding Zone 5). Also recruitment was continued separately for the NWFP (Zones
5 and 6), Sind (Zone 2), Punjab and Bahawalpur (Zones 3 and 4), and for Karachi.
Percentile bases for such recruitment during 1972-73 were never finalized.
20. Pakistan, Establishment Division, memo no. F 8/9/72 TRV, August 31, 1973.
21. In 1973 approximate populations were: Punjab 55.5%, urban Sind 6.8%, rural Sind
13.8%, Baluchistan 3%, NWFP 12.3%, NATT 3.7%, and Azad Kashmir 4.4%.

694

ASIAN SURVEY, VOL. XXIV, NO.

6,

JUNE

1984

rice, cotton, textiles, cement, automobiles, etc.). As a consequence, such

industries, formerly in the private sector, became subject to the terms
and conditions of federal employment so that recruitment to autonomous
and semiautonomous corporations (188 such institutions existed in
1981)22 came under the quota. But the expansion was not restricted
solely to federal initiative. During the early 1970s the provinces established their own quotas for provincial recruitment. In Punjab, recruitment to provincial posts became subject to three reservations-Zone 1,
developed districts; Zone 2, undeveloped districts; and merit on an
all-Punjab basis.23 In Sind, provincial recruitment followed the federal
designation of "rural"and "urban."24And in the NWFP, provincial recruitment was divided into five zones, two of which are relatively less
developed and three relatively more developed.25 The operation of the
provincial quotas is analogous to the operation of the federal quota-that
is, recruitment to competitive posts in the provincial civil services,
direct recruitment to posts advertised by the provincial public service
commissions, departmental recruitment to provincial departments, and
recruitment to provincially run public enterprises have been made subject to the relevant quotas.
In the 1970s quotas also came to be applied to admissions to various
educational institutions. However, the patterns of relevant quotas differ
from province to province and from institution to institution. The Qaidi-Azam University-the only federal university-admits students on the
basis of the federal quota. In the provinces, the dominant pattern is
admission of provincially domiciled candidates on the basis of merit with
reservations for the children of military personnel and for those
domiciled in variously defined "undeveloped" regions. But sometimes
admission is based on intraprovincial quotas26 or is determined in an ad
22. Pakistan, International Symposiumon EconomicPerformanceof Public Enterprises(Lahore: Pakistan Administrative Staff College, 1981).
23. Zone 1 (developed) = Lahore Division and the districts of Rawalpindi, Gujrat,
Lyallpur, Multan, and Sahiwal. Zone 2 (less developed) = Bahawalpur Division and the
districts of Muzaffargarh, DG Khan, Cambellpur, Jhelum, Mianwali, and Jhang.
24. Urban = Karachi, Hyderabad, and Sukkur. Rural = everywhere else.
25. The "less developed" regions allow age relaxations for relevant candidates. Also
attached to the NWFP quota is a 10% provincial reservation for military personnel
domiciled in the province.
26. For instance, admission to Sind medical colleges (the Medicine Bachelor, Bachelor
Surgery program) is determined by the place of domicile within Sind. Admission to Dow
Medical College is restricted to those candidates living in Karachi;LiaqatMedical College is
restricted to those candidates domiciled in Hyderabad, Thatta, Badin, Tharparkar, Nawabshah, and Sanghar; Chandka Medical College is restricted to candidates from Larkana,
Sukkur, Shikapur, Khairpur, Jacobabad, and Dadu. Admission to the Medicine Bachelor,
Dental Surgery program at LiaqatMedical College (the only program of its type in the Sind)
is on an all-Sind basis. Similarly, admission to the Diploma of Associate Engineer degree
programs is based on a district quota.

CHARLES H. KENNEDY

695

hoc fashion by the institution itself.27

Since 1977 President Zia's regime has also introduced certain innovations into the quota system. The most far reaching has been the establishment of a 10% quota for former military personnel seeking competitive entry to governmental services through the CSS or provincial examinations.28 This, of course, has confused the regional rationale of the
quota. Further confusion has been introduced with episodic age relaxations by the FPSC for recruitment from minority communities.29 And
very recently a "sons of the soil" argument has been advanced by the
government of Baluchistan in its demands to the federal government to
establish two quotas within Baluchistan, one for native Baluch and one for
Pathan and Punjabi migrants.30
If all this sounds hopelessly complicated, that is correct. Currently,
recruitment to virtually every position in the federal and provincial
goverments is subject to a quota. What started out as a relatively limited
program designed to ameliorate the unhappy effects of Bengali underrepresentation in the secretariat, has in three decades been transformed
into a program of entitlement for virtually every governmental vacancy.

The Operation of the System

We turn our attention now to three sets of questions associated with the
operation of the quota system: (1) How is the quota system applied to the
selection process of recruits at the various levels of operation and how
27. For instance, Gomal University (NWFP) admits candidates for graduate degrees on
the basis of ten different categories-six intraprovincial and four based on special attributes
of candidates. Similarly, the Lady Health Visitor Course at the Public Health Nursing
School, Lahore, gives preference to "widows" and "destitutes."
The admission of Baluch candidates to educational institutions has involved direct
interprovincial agreements establishing reservations for Baluch nations in non-Baluch
institutions. For instance, in 1982 116 seats in various medical colleges, 88 seats in various
engineering colleges and universities, 15 seats in the College of Home Economics in
Karachi, 7 seats in the National College of Arts, Lahore, 50 seats in cadet colleges, 7 seats
in the College of Dentistry, Karachi, aqd 8 seats in various MBA programs were reserved
for Baluch students in institutions located outside their home province. Source: Government of Baluchistan, Services and General Administration Department.
28. Regularized in the Federal Public ServiceCommissionAct, February 1980.
29. The standard employed by the FPSC since 1981 is: "3 year relaxation to candidates
belonging to scheduled castes, Buddhist community, recognized tribes of the tribal area,
Azad Kashmir, Northern Areas/Districts of Gilgit, Ghizar, Skardu, and Diamir."
30. In the summer of 1982, the government of Baluchistan called for the modification
of the federal quota system as it applies to Baluchistan. The demands include dichotomizing
recruitment into two zones, Zone A and Zone B. Zone A would encompass the Tribal
Areas of Baluchistan (over 98% of the land area) and Zone B would encompass the Settled
Areas, the cities-a pattern similar to that in effect in the Sind. Also, demands have been
voiced to establish separate quotas for Baluchi and non-Baluchi speakers in each zone.

696

ASIAN SURVEY, VOL. XXIV, NO.

6, JUNE 1984

exactingly and enthusiastically is it implemented? (2) Is the bureaucracy

of Pakistanregionally representative and do such levels of representation
differ in regard to rank or type of governmental service? (3) How does
the quota system interact with the bureaucratic system of Pakistan and
what are the system's fixed costs?
HOW DOES THE SELECTION PROCESS WORK?
At the federal level the quota is applied to the selection process in three
distinct patterns. Pattern 1 applies to competitive entry through the
Central Superior Services Examinationadministered by the FPSC. In this
pattern the quota is applied both to entry-level competition and to the
competition for entry into occupational cadres. After the number of
vacancies is determined for a given year by the Establishment Division, an
open competition is held in which candidates from all domiciles participate. Ten percent of such vacancies are filled by candidates who score
highest on the exam regardless of domicile (the merit quota). The remainder are filled by the respective merit-rank within domiciles. For
example, if in a particularyear there are one hundred vacancies, 10 will be
filled on the basis of overall merit, 50 will be filled from the most
meritorious Punjabis, 19 from the meritorious Sindhis, etc. (as per the
federal quota cited earlier). Vacancies in occupational cadres are similarly
allocated on the basis of the quota, the preferences of individual candidates, and exam scores. This pattern applies to the recruitment for approximately 250 vacancies per year.3'
Pattern 2 is applicable to the selection of candidates who enter on the
basis of an interview, either administered by the FPSC or directly by the
ministry or department concerned. Here posts are advertised or listed
against regional vacancies. For example, if the Ministry of Defense needs
to recruit communication engineers, the ministry sets domicile requirements (which should be based on existing regional balances within the
ministry) for the relevant vacancies. In an illustrative case in 1976, the
ministry had to fill fourteen vacancies, and designated that seven should
come from the Punjab, four from the NWFP, and three from rural Sind.
Accordingly, the FPSC interviewed candidates only from such domiciles
and made appointments on the basis of these interviews.32 This pattern
applies to the recruitment of approximately 3,000 candidates per year.
Pattern 3 applies primarily to the recruitment of individuals to autonomous corporations and public enterprises. This pattern requires the
candidates to submit domicile certificates to the hiring authorities (rele31. Figures from 1978-81.
32. Example drawn from Pakistan, FPSC, Annual Reportfor the Year 1976 (Karachi,
1978).

CHARLES H. KENNEDY

697

vant personnel boards), but the domicile limits are typically not binding
on the actual hiring practices of such boards. However, considerations of
regional balance often play a significant, if informal, role in hiring decisions. In this pattern domicile serves more as a target than as a quota.
The operation of provincial quotas follows analogous patterns: Selection to the provincial civil services conforms to the first pattern; recruitment to other provincial posts the second pattern; and recruitment to
provincially administered public enterprises the third pattern. Selection
for educational institutions (predominantly under provincial control) is
determined by the independent decisions of college and university personnel boards in terms of their relevant institutional quotas.
Regardless of the relevant pattern the essential document that must be
presented by all candidates to establish regional status is the so-called
domicile certificate. An individual's domicile is determined by several
factors, including father's domicile, place of principal residence, place of
schooling, and place of prior recruitment. In most cases the determination of domicile is unambiguous; where it is in doubt it is determined by
relevant bureaucratic authorities, buttressed by duly authorized witnesses, and/or in rare cases by court proceedings. The stakes of domicile
declaration are high, and the motivation for tampering with or misrepresenting one's domicile is accordingly very much present. However, despite allegations in the press (primarily by disaffected Punjabis), instances
of blatant cheating on domicile certificates are infrequent.33 Even in the
absence of overt cheating, there still remains the perception in some
circles that candidates from less-favored regions are not "sons of the
relevant soil." That is, they are immigrants, however defined, to their
respective domiciles. Though I believe that this argument has little merit,
unfortunately there is no available information to test this hypothesis.
DOES THE QUOTA WORK?
To answer this question in an ideal manner requires information concerning the composition of the bureaucracy, given the operation of the
quota system, and comparable information concerning the hypothetical
composition of the bureaucracyin its absence. Complete data concerning
the former does not exist, and information on the latter is beset with
numerous imponderables. However, one can informally sidestep these
difficulties by comparing representation in different sectors of the federal
bureaucracy over time. Table 1 presents such an approach, and reveals
two significantpatterns. First, urban Sind and the Punjab, when compared
with quota percentages, are overrepresented in the bureaucracy, while
33. One inequitablepractice,discontinuedin 1980, wasto automaticallyassignwomen
candidatesto the domicilesof their fathers.

o &

Z,t
S

r-

?O

W> C)

VI\

C) a^

0 r-- 0

C-

?>

~~~~~00

0 0 y4
U U

>~~~~~~~~~~~~~r
<

:;

i *t.

m s-N

2 c

G.naSx>

- s &r?'jr-

rn

jIng g iE-C < E Xn

CHARLES H. KENNEDY

699

the other provinces are underrepresented. Second, and perhaps more

interesting, is the finding that the secretariat, which has been subject to
the quota for some thirty years, is more representative than are attached
departments, subject to the quota for approximately ten years, and both
are more representative than autonomous corporations, which at the time
of my data (1974) were not subject to the quota. That is, the quota has
made the bureaucracy demonstrably more representative, or alternatively, the quota has lessened the overrepresentation of urban Sind and
perhaps the Punjab, while increasing the representation of other provinces.

A more detailed analysis of posts subject to the competitive examination is possible. As I have shown elsewhere,34 candidates from the Punjab
consistently score higher on the CSS examinination than candidates from
other provinces. Therefore, in the absence of the quota system, it can be
demonstrated that Punjabis would dominate the bureaucraticsystem to an
extent fargreater than is currently the case. Specifically, it can be demonstrated that such domination would be particularly severe in the case of
selection to the prestigious All-Pakistan Unified Grades.35
Therefore, all available evidence points to the fact that the quota
system has made the bureaucracy more regionally representative. Indeed, though there is some slippage, the ostensible goal of the quota-to
increase regional equality in the bureaucracy-seems to have been addressed by the operation of the system. And at least at the higher levels
of the bureaucracy that have been subject to the quota system for three
decades, the fit between the quota's target and actual representation in
the bureaucracy is close. That is, the quota works.
THE SYSTEM'S FIXED COSTS
But if the operation of the quota has been successful in promoting
regional equality, it has also resulted in several unhappy consequences.
The first and most often mentioned cost concerns the systematic frustration of the merit principle of selection for bureaucratic posts. For posts
subject to competition, some candidates are recruited into the bureaucracy who would not be eligible in the absence of the quota, and others
are denied positions for which they would be eligible on the sole basis of
merit. For posts subject to interview, prospective candidates with requisite qualifications for particular posts are systematically excluded from
competition because of regional considerations. In both instances the
quota selects candidates who may not be the "best" available for particu34. Charles H. Kennedy, "Context, Content and Implementation of Bhutto's Administrative Reforms" (Ph.D. thesis, Duke University, 1979), pp. 188-223.
35. Ibid., p. 212.

700

ASIAN SURVEY, VOL. XXIV, NO.

6,

JUNE

1984

lar jobs. It can be argued that such a selection policy makes the bureaucracy of Pakistan, seldom praised for its efficiency, even less efficient.
The process of implementing the quota also often results in significant
delays in filling vacant posts. For example, if a post falls vacant in a
federal department subject to FPSC recruitment rules, the notification of
the vacancy is passed on to the FPSC both for advertisement and for the
assignment of the quota. The post is then listed, candidates with the
necessary qualifications and domicile are interviewed, and ideally an
individual is recommended for appointment. If all runs smoothly, this
process takes approximately six months. However, if finding an individual with both the requisite qualifications and domicile certificate proves
difficult, as often happens, the process can be delayed considerably. It is
not unusual for a case of this type to drag on for two to three years, with
the post left vacant or filled on an ad hoc basis by the relevant department. It is also not uncommon for vacancies to go unfilled because no
candidates with the requisite qualifications and domicile apply for the
given job. In recent years this latter phenomenon has occurred with
increasing frequency, particularly for technical vacancies.36 Also, since
1977 the FPSC has had trouble filling vacancies through the competitive
examination process. Indeed, in 1982 the FPSC was forced to conduct a
special competitive examination limited to candidates from urban Sind,
rural Sind, Baluchistan, the Northern Areas, and Azad Kashmir to address the shortfall.37
Third, the complexities of the quota may discourage prospective candidates, particularlyfrom urban areas, from standing for the competitive
examination and/or from seeking other governmental employment. The
opportunity costs of participating in the recruitment process to the
Central Superior Services are great. A candidate must first prepare and
stand for the examination, then wait for the results before knowing
whether one is successful or, often more important, whether one has
gained entry to a preferred occupational grouping. This process takes
anywhere from fifteen to eighteen months.38 Such costs are compounded
by the uncertainty of the operation of the quota. From the perspective of
36. Pakistan, FPSC, Annual Report, 1975, 1976, 1977, 1978, 1979, 1980, 1981
(Karachi).
37. There were 73 vacancies in all. Urban Sind 21, rural Sind 36, Baluchistan 5,
Northern Areas 9, and Azad Kashmir 2. For the special exam, age limits were lifted from
25 to 35 years, and Third Division candidates, normally barred from participation in the
CSS, were allowed to compete. The results of the exam had not been announced by
October 1982, although 2888 candidates took the exam. Source: Interviews.
38. From 1974 to 1979, assignment to occupational groups was delayed from nine
months to one year longer until after the Final Passing Out Examination from the program
of joint training at the Academy for Administrative Training.

CHARLES H. KENNEDY

701

an individual candidate, there are two avenues to assure assignment to

one's preferred occupational grouping: (1) scoring in the top 10% of the
successful recruits (i.e., within the merit range), or (2) scoring relatively
high in comparison with one's regional grouping. However, it is impossible to assess accuratelythe probabilities of either of these outcomes given
the limited amount of information available to candidates. Until group
assignment takes place, candidates do not know how they stand in relation
to other candidates. Similarly,as mentioned above, recruitment by interview is also a slow and uncertain process. Given such considerations,
candidates are in effect encouraged to seek alternative careers.
Of course, the likelihood of finding satisfactory alternative employment is a function of the availability of such employment. And since
relatively greater numbers of such jobs are available in urban areas, if our
hypothesis is correct, it follows that there would be a tendency for the
number of candidates standing for the competitive examination from
urban areas to decrease. Our data confirm this expectation. The number
from the Punjab has declined 22% and from urban Sind 39% over the
period 1969-74, while there has been no comparable decrease in the
number standing from other provinces.39 Before 1974 such urban candidates typically sought employment in corporations based in urban areas,
particularlyKarachi-hence the overrepresentation of urban Sind in autonomous corporations as demonstrated in Table 1. Since 1974, when
the quota was applied to the nationalized autonomous corporations, such
channels have been closed and the resultant glut of prospective urban
employees has swelled the ranks of those seeking employment in the
Middle East. Indeed, perhaps not unexpectedly, demands for the application of regional quotas for granting exit visas to Middle Eastern states
have recently surfaced.40
Finally, the quota system, though demonstrably increasing provincial
representativeness in the bureaucracy, has also reinforced invidious distinctions between provinces. The current system's reliance on a small
merit quota, coupled with the widely publicized and oft-repeated theme
of provincial-based distinctions of performance on the examinations or of
distinctions in relevant qualificationsof levels of skill, leads to the perception that job seekers from the more-favored provinces bear the brunt of
the government's attempt to equalize access to the bureaucracy. Indeed,
hardly a day passes when one of the major Pakistani dailies does not run
an editorial by a disaffected individual, usually a Punjabi, calling for
greater reliance on merit in some phase of federal personnel policy, and
implicitly claiming reverse discrimination. Conversely, individuals from
39. See C. H. Kennedy, "Context, Content," p. 214.
40. Interviews, 1982.

702

ASIAN SURVEY, VOL. XXIV, NO. 6, JUNE

1984

less-favored provinces must live with the phenomenon that they are
perceived as less qualified than their counterparts from the Punjab. Such
distinctions are given added credence by the quota's implicit paternalism,
and reinforce the already formidable ethnic, cultural, racial,and linguistic
cleavages between provinces. Indeed, the very existence of the quota
provides statutory verification of the hypothesis that some regions or
peoples are superior or inferior to others. Such perceptions are the stuff
of secessionist movements.

Lessons
It may be dangerous to attempt to draw lessons from the admittedly
unique particularsof Pakistan'sexperience with policies of ethnic preference. However, I cautiously maintain that three themes relevant to the
Pakistani experience are likely to be replicated in any state adopting
similar policies.4'
First, policies of preferencetend to spread. The quota system in Pakistan
had relatively limited origins, with an initial application to barely one
hundred vacancies per year. But once established, the logic of its spread
became irresistible. If the quota was necessary to redress inequality of
representation through the competitive examiniation process for bureaucratic posts, was it not also necessary for it to be applied to the arguably
more aribitrary selection of candidates through interview? And, if the
quota was necessary to redress inequalities of qualifications in central
governmental departments, why was it not also applicable to provincial
governmental posts? And finally, if it was necessary to redress distinctions
in qualifications, was it not also necessary to get at the fount of such
inequalities-the educational institutions themselves?
The second lesson is thatpolicies of preferenceare hard to terminate.The
quota system was introduced in Pakistan with the expectation by its
framers that the need for its operation would end five years after its
commencement. And since its introduction, virtually every major administrative report, numerous service association demands, and innumerable editorials have called for the quota's dismantlement. But even modest proposals calling for marginalmodifications in the terms of the quota's
operation have been ignored with impunity by successive governments.42
The reason, of course, is that reforming the quota is very dangerous to
41. Cf. Myron Weiner, Mary Katzenstein, and K. V. Narayana Rao, India's Preferential
Policies (Chicago: University of Chicago Press, 1982).
42. For example, the Pay and Services Commission of 1978 examined the question of
terminating the quota and recommended a modification calling for 20% merit, instead of
the current quota of 10%. The government has not accepted the proposal. The author
testified before this commission on this issue in 1978.

CHARLES H. KENNEDY

703

the potential reformers. Any modification of the quota terms would

likely be interpreted as favoring some groups at the expense of others,
and would be attributed to the ulterior motives of clever politicians. For
this reason, prudent policy makers in Pakistanhave given the quota a wide
berth.
The third lesson may be the most important. No matter how
thoroughly implemented, policies of preferencehavsefixed costs. The more
obvious costs include the effect of abandoning the merit principle on
administrativeefficiency, and the attendant procedural entanglements associated with implementing a quota policy. Pakistan's system has suffered
enormously from both types of costs. Hypothetically, one could imagine
a set of policies that minimizes such costs, but one cannot finesse the cost
of the inevitable tradeoff between the goal of equality of outcome between groups and the consequences of inequitable laws that is provoked
by the introduction of a quota system. Pakistan's quota system works to
increase equality of representation between provinces, but coincidentally
it also exacerbates perceptions of inequities between provinces. Lurking
in the shadows of all quota policies is the specter of reverse discrimination, and perhaps more important, the cancer of invidious comparisons of
superiority or inferiority between regions or groups that such considerations imply.