Meritocracy in the Face of Group Inequality
Rohini Somanathan
March 2019
Sethi, Somanathan Meritocracy & Group Inequality March 2019 1 / 1 Introduction The 124th Constitutional Amendment
The following amendment to the constitution was passed in the two houses of parliament on January 8-9, 2019:
[The ]Constitution (One Hundred and Twenty-fourth Amendment) Bill, 2019 provides for reservation for the economically weaker sections of society in higher educational institutions, including private institutions whether aided or unaided by the State other than the minority educational institutions referred to in article 30 of the constitution and also provides for reservation for them in posts in initial appointment in services under the State
The pretext is social justice, the context is the upcoming election.
Sethi, Somanathan Meritocracy & Group Inequality March 2019 2 / 1 Introduction The politics of the new laws
August 25, 2015: Hardik Patel and the “Patidars” in Ahmedabad, India
“My sister completed her high school last year and scored very high marks. But she couldn’t get a scholarship, whereas students from the castes that are part of the quota system got it with much lower marks” (BBC, Aug 26)
Sethi, Somanathan Meritocracy & Group Inequality March 2019 3 / 1 Introduction How should we evaluate such policies?
Construct a theoretical counterfactual: how would assignment take place in their absence?
Trace the empirical evolution of educational inequality by group: who has gained from these policies?
Sethi, Somanathan Meritocracy & Group Inequality March 2019 4 / 1 Is this interpretation too narrow? Can FA be identity-contingent?
Introduction The meaning of meritocracy
Common understanding of meritocracy: selection of best qualified or fair-assignment (FA).
Narrow interpretation: Monotonicity within groups Blindness across groups
This is a reasonable theory of justice when there is equal opportunity (EO).
IfEO fails, the narrow interpretation would lead us to underrepresentation of disadvantage groups in elite positions a trade-off between increased group-representation and FA
Sethi, Somanathan Meritocracy & Group Inequality March 2019 5 / 1 Introduction The meaning of meritocracy
Common understanding of meritocracy: selection of best qualified or fair-assignment (FA).
Narrow interpretation: Monotonicity within groups Blindness across groups
This is a reasonable theory of justice when there is equal opportunity (EO).
IfEO fails, the narrow interpretation would lead us to underrepresentation of disadvantage groups in elite positions a trade-off between increased group-representation and FA
Is this interpretation too narrow? Can FA be identity-contingent?
Sethi, Somanathan Meritocracy & Group Inequality March 2019 5 / 1 Introduction
Firms do not hire workers as a reward for past achievement, but rather on the basis of prospective achievement, in the sense of making the firm more profitable. This move from merit as reward to merit as effectiveness is not innocuous as it means that merit needs to be assessed relative to the properties of the human capital production functions.
Sethi, Somanathan Meritocracy & Group Inequality March 2019 6 / 1 Introduction Profiling vs. preferential treatment
Prior achievement may not fully reveal attributes being sought:
A student who has, in some way, experienced hardship may underperform on achievement tests relative to his or her capability. By taking account of such empirically grounded differences across demographic groups, a district may be better able to determine which students are most suited to admission to the gifted program... While this profiling based on differences in distributions across racial groups is beneficial to minority students, it is not preferential treatment. Cestau, Epple, and Sieg (2017)
Sethi, Somanathan Meritocracy & Group Inequality March 2019 7 / 1 Introduction Merit is context-dependent
T.M. Scanlon (2018) in his recent book, Why does Inequality Matter?
If a position requires lifting heavy objects, then physical strength is an important form of ability. But if the job is done with a fork-lift truck then it is not. If succeeding in a particular job, or in a university course of study, requires one to understand French, then knowledge of French is a relevant ability. If everything is done in English, then it is not. This dependence on the goals that justify an institution and on the way it is organized to promote these goals are what I mean by saying that the idea of talent, or ability, that is relevant to procedural fairness is “institution- dependent.”
Sethi, Somanathan Meritocracy & Group Inequality March 2019 8 / 1 Introduction A Model
Two groups, populations shares s1 and s2
Ability levels {al , ah}, common ability distribution F (a)
Resource levels {rl , rh}, proportion with high resource in group i is qi
Training t ≤ τ(a, r), increasing in both arguments, observable
Performance p = φ(a, t) is increasing in both arguments
Elite capacity k is the share of the population to be selected
Benchmark (no underinvestment): four training levels tij = τ(ai , rj ) with
tll < tlh, thl < thh
Sethi, Somanathan Meritocracy & Group Inequality March 2019 9 / 1 Introduction
Equal size groups, 30% high ability, 10% and 30% high resources, 10% elite capacity
Sethi, Somanathan Meritocracy & Group Inequality March 2019 10 / 1 Introduction
Equal size groups, 30% high ability, 10% and 30% high resources, 10% elite capacity
Sethi, Somanathan Meritocracy & Group Inequality March 2019 11 / 1 Introduction
Equal size groups, 30% high ability, 10% and 30% high resources, 10% elite capacity
Sethi, Somanathan Meritocracy & Group Inequality March 2019 12 / 1 Introduction Representation as a function of capacity
Sethi, Somanathan Meritocracy & Group Inequality March 2019 13 / 1 Figure: Disadvantaged Group Representation as Elite Capacity Varies Introduction Implications for inequality
Group inequality in the meritocratic assignment will vary by
schooling and industry technologies
the rate of employment
demographics of disadvantaged groups
If only ability mattered for performance, we will get statistical mirroring in the meritocratic allocation, not otherwise.
Sethi, Somanathan Meritocracy & Group Inequality March 2019 14 / 1 Introduction The scheduling of caste in India
Exterior Castes Scheduled Castes Census of India, 1931 Census of India, 1961 278 castes, 11% of the population 15% of the population all religions only Hindu, Buddhist, Sikh
Sethi, Somanathan Meritocracy & Group Inequality March 2019 15 / 1 Introduction The chronology
1935-36: scheduled castes list based on the exterior castes of the 1931 census
1950-51: two separate lists for ]scheduled castes and scheduled tribes (7.5%) constitutional provisions for quotas in parliament enabling provisions for preference in education and employment population based quotas (reservations) subsequently introduced in universities and public employment 1955: first backward classes commission appointed to create a new list of other backward classes (OBCs) (2,399 groups, 50%) 1976: most territorial restrictions for scheduled castes and tribes removed within states. 1978: second backward classes (mandal) commission appointed to create a new list of other backward classes (2,399 groups, 50%) 1980: 27% quota for OBCs in federal jobs 2008: 27% quota for OBCs in university admissions 2019: 10% EWS quota
Sethi, Somanathan Meritocracy & Group Inequality March 2019 16 / 1 Introduction How homogeneous were scheduled groups?
Literacy rates among them very uniformly low in British India
Literacy rates by caste, 1891 and 1931
Priests (Brahmans) Writers (Kayastha) Traders (Lohana) Oil Pressers (Teli)
Washermen (Dhobi) 1891 Potters (Kumhar) Cobblers (Chamar)
Village Menials (Mahar, Pasi,etc.) Tribes (Gond, Santhal)
Scavengers (Megh)
Priests (Brahmans) Writers (Kayastha) Traders (Lohana) Oil Pressers (Teli)
Washermen (Dhobi)
1931 Potters (Kumhar) Cobblers (Chamar) Village Menials (Mahar, Pasi,etc.)
Tribes (Gond, Santhal) Scavengers (Megh)
0 10 20 30 40
Sethi, Somanathan Meritocracy & Group Inequality March 2019 17 / 1 Introduction The claim of homogeneity was critical for representation
Drs. Ambedkar and Solanki as witnesses before the Simon Commission on representation of the depressed classes in Bombay Chairman: Among the untouchables themselves there are degrees; there are certain among them who may be regarded as only semi-untouchable? Both witnesses: No Chairman: I will give you an example. What is the position of the Chambhar? Ambedkar: He is entirely untouchable Chairman: As much as the Mahar? Ambedkar: Yes ....
Chairman: Now, can you tell me how many reserved seats ...you would suggest..?
Ambedkar: ..I say out of 140, we claim 22 seats.
Sethi, Somanathan Meritocracy & Group Inequality March 2019 18 / 1 Introduction How homogeneous are scheduled groups?
Education attainment is now quite unequal
Mahar Bhambi Adi Dravida Pallan Mala Adi Karnataka Namasudra Dhobi All Scheduled Castes Balmiki Chamar Kori Madiga Rajbanshi Megh Mazhabi Dosadh Pasi Bauri Bagdi Musahar
0 20 40 60 80 100
primary middle grade 10 grade 12 graduate
Sethi, Somanathan Meritocracy & Group Inequality March 2019 19 / 1 Introduction Divergence in secondary schooling, 1961-2001
Divergence is both across and within geographies
KE Mahar .25 MH .25 GJ
Adi Dravida
PN+ Mala .2 .2 Paraiyan Adi Karnataka
TN
AS AS KT PN+ AP .15 Namasudra Dhobi Oraon .15 Chamar Madiga Kori
.1 KT KE DusadhBalmiki GJ OR Mina Munda MP Pan Rajbanshi MH Pasi .1 UP BR WB Gond Secondary Education, 2001 Education, Secondary 2001 Education, Secondary .05 Bhil Santal TN RJ SC SC Bagdi BR AP ORRJ ST Musahar ST .05 MP 0 WB 0 .002 .004 .006 .008 .01 0 .002 .004 .006 .008 .01 Secondary Education, 1961 Secondary Education, 1961
Sethi, Somanathan Meritocracy & Group Inequality March 2019 20 / 1 Introduction Attempts at rational reclassification: Mahadalits in Bihar
The category gradually expanded to include Dhobis, Pasis and finally Chamars in 2009
Hari etc. Dom etc. Rajwar 1% 1% 2% Bhuiya 4%
Dusadh etc. 31% Musahar 16%
Dhobi 5% Pasi 5%
Chamar etc. 31%
Sethi, Somanathan Meritocracy & Group Inequality March 2019 21 / 1 Introduction Identity within the Mahadalits
0.45 Caste Unmatched Caste Matched 0.40 0.40
1.00 0.35
median Caste group 0.30 Musahar 0.262 0.75 0.26 0.80 Chamar 0.25 Other SC 0.56 Non SC 0.20 0.176 0.17 0.50 0.16 0.52 0.15 Empirical density
0.10 0.25 0.37 0.05
0.00 0.00 All Chamar Musahar −2 0 2 4 BPL Score (standardised)
Sethi, Somanathan Meritocracy & Group Inequality March 2019 22 / 1 Introduction Conclusions
Group representation could do better or worse than psuedomeritocracy in achieving meritocratic ideals
India has favored the former, the U.S. the latter, neither has adequately dealt with stereotypes and stigma.
Disadvantage is binary, expected performance is not - should we allow more nuanced instruments?
Sethi, Somanathan Meritocracy & Group Inequality March 2019 23 / 1