Building Connections Political Corruption and Road Construction in India PDF

Journal of Development Economics 131 (2018) 62–78
Contents lists available at ScienceDirect
Journal of Development Economics

journal homepage: www.elsevier.com/locate/devec
Building connections: Political corruption and road construction in India

Jonathan Lehne a, *, Jacob N. Shapiro b, Oliver Vanden Eynde c
a
Paris School of Economics, Boulevard Jourdan 48, 75014, Paris, France
b
Princeton University, Woodrow Wilson School of Public and International Affairs, Robertson Hall, 20 Prospect Ave, Princeton, NJ, 08540, USA
c
CNRS and Paris School of Economics, France
A R T I C L E I N F O A B S T R A C T
JEL codes: Politically-driven corruption is a pervasive challenge for development, but evidence of its welfare effects are
D72 scarce. Using data from a major rural road construction programme in India we document political influence in a
D73 setting where politicians have no official role in contracting decisions. Exploiting close elections to identify the
L14 causal effect of coming to power, we show that the share of contractors whose name matches that of the winning
O18 politician increases by 83% (from 4% to 7%) in the term after a close election compared to the term before.
Keywords: Regression discontinuity estimates at the road level show that political interference raises the cost of road con-
Corruption struction and increases the likelihood that roads go missing.
Political connections
Public procurement
Kinship networks
1. Introduction extrapolating from a LATE) would imply that state-level parliamentarians

(MLAs) intervened in the allocation of roughly 1900 of the 4127 road
A growing literature documents the private returns to holding public contracts let to connected contractors, approximately $540M of the
office and the benefits from political connections in both the developed $1.2B spent on such roads and approximately 4% of the total spent on the
and developing world (e.g. Fisman et al., 2014; Cingano and Pinotti, programme. These results are broadly representative of Indian polities.
2013).1 It is often not clear from existing work to what extent the Our sample consists of 4058 electoral terms from 2001 to 2013, covering
observed gains by public officials or connected firms represent a welfare 2632 constituencies in 24 of the 28 states which existed in our sam-
loss. We document political influence over the allocation of individual ple period.
road contracts in India's major rural road development scheme, providing The allocation of contracts to those with political connections does
econometric evidence that politicians intervene in the allocation of not conclusively prove that politicians' motives are corrupt. In an envi-
contracts on behalf of members of their own network. By studying the ronment of imperfect information, MLAs could, in theory, be better
performance of firms on contracts that are likely to have been allocated informed about, and better able to monitor, contractors in their own
preferentially, our paper provides direct evidence on the welfare costs of network, and might therefore improve programme performance through
political corruption. A distinguishing feature of our findings is that we benevolent interference. RD estimation at the road level provides no
study a programme in which politicians have no formal role. evidence that is the case for PMGSY road construction. Instead, we
Specifically, we use data on more than 88,000 rural roads built under document direct negative welfare consequences for the people the pro-
the Pradhan Mantri Gram Sadak Yojana (PMGSY) programme to study gramme is supposed to serve.
how close-election victories shift spending. Using regression disconti- We find that roads allocated to politically connected contractors are
nuity (RD) estimates to identify the causal effect of coming to power we significantly more likely never to be constructed. Census data at the
show in our preferred specification that the share of contractors whose village-level, collected after road construction was officially completed,
name matches that of the winning politician increases from 4% to 7% (an reveal that a number of roads listed as having been completed in the
83% increase). The magnitude of these distortions are large relative to PMGSY monitoring data, and for which payments were made, do not
programme size. Applying our RD estimate to the full sample (i.e. appear to exist. We define a road as “missing” if a village it was meant to
* Corresponding author.
E-mail addresses: [email protected] (J. Lehne), [email protected] (J.N. Shapiro), [email protected] (O. Vanden Eynde).
1
See Eggers and Hainmueller (2009) for members of the UK House of Commons and Truex (2014) for Chinese deputies.
https://doi.org/10.1016/j.jdeveco.2017.10.009
Received 18 November 2016; Received in revised form 6 October 2017; Accepted 19 October 2017
Available online 28 November 2017
0304-3878/© 2017 Elsevier B.V. All rights reserved.
J. Lehne et al. Journal of Development Economics 131 (2018) 62–78
reach subsequently lacked “all-weather road access” (PMGSY's stated interference would not be documented. We address the first problem by
objective). The preferential allocation of roads is estimated to increase constructing a surname-based measure of proximity between candidates
the likelihood of a missing all-weather road by 86%. Assuming an for state-level legislatures and contractors. This approach follows a
extrapolation from a LATE were valid, this would imply that an addi- number of papers that use Indian surnames as identifiers of caste or
tional 497 all-weather roads are missing as a result of corrupt political religion (e.g. Hoff and Pandey, 2004; Field et al., 2008; Banerjee et al.,
intervention and, that the 857,000 people these roads would have served, 2014). Dealing with the second issue – identifying improper intervention
remain at least partially cut-off from the wider Indian economy. Political – requires isolating the variation in proximity to contractors that results
interference in PMGSY is also detrimental when road construction from the MLA coming to power. We do so with a regression discontinuity
actually takes place. Further road-level RD estimations show that roads approach that exploits the fact that in close elections, candidates who
allocated to connected contractors are more expensive to construct. barely lost are likely to have similar characteristics to those who were
These results indicate that corruption in PMGSY imposes social costs barely elected. If MLAs are intervening in the assignment of contracts,
while providing no offsetting benefits in terms of efficiency or quality. one would expect a shift in the allocation towards contractors who share
Importantly, because road locations were largely determined before the their name, and no equivalent shift for their unsuccessful opponents. This
elections we study, the impact of the corruption examined here arises approach to detect undue influence is what Banerjee et al. (2013) refer to
primarily from who is allocated a contract rather than where a road as a “cross-checking” method for identifying corruption: the comparison
is built. between (i) an actually observed outcome, and (ii) a counterfactual
Our paper's first contribution is to provide micro-evidence on measure which should be equivalent to the former in the absence of
informal channels of political influence. A growing number of papers corruption.3 In our setting, if politicians are not intervening in the allo-
document how firms benefits from political connections, (Amore and cation of road projects, they should be no ‘closer’ to contractors than
Bennedsen, 2013; Do et al., 2015). Khwaja and Mian (2005) show that their unsuccessful opponents.
banks in Pakistan lend more to politically connected firms - in spite of Our third contribution is to shed light on the social costs of political
higher default rates. Cingano and Pinotti (2013) show that connected connections. In principle, allocating contracts to connected firms could
firms in Italy benefit from a misallocation of public expenditures, which be beneficial – politicians could use private information to select higher
helps them to increase profits. Mironov and Zhuravskaya (2016) show quality firms, or they could use their social networks to discipline con-
that Russian firms who funnel money in the run-up to elections are tractors. And while the common intuition is that political influence has
significantly more likely to receive procurement contracts after the deleterious effects, few papers document the social costs of political
election. We add to these recent contributions, by showing a link connections. For example, Mironov and Zhuravskaya (2016) show that
between the election of Indian legislators and the allocation of PMGSY politically connected firms have lower average productivity.4 Our
road contracts to connected contractors. Preferential allocation in the unusually rich data allows us to examine the performance of contractors
context of PMGSY is particularly striking, because state-level legislators in the exact contracts that are likely to be preferentially allocated. We
do not have any formal role in the allocation of contracts. In fact, this introduce a particularly powerful measure of contractor under-
programme's bidding rules were designed in ways that should have performance: “missing roads”. These are roads that are complete in the
forestalled political influence at the bidding stage (NRRDA, 2015). Our PMGSY records and for which payments have been made, but that do not
evidence on preferential allocation in such a programme helps us to appear in the (independently conducted) Population Census. This
understand the economic role of local politicians, in particular in India. approach allows us to provide very clear evidence of the welfare costs of
Recent work shows how Indian state legislators have a sizable impact on undue political influence: the roads that are built by connected con-
local economic outcomes. Asher and Novosad (2017) show that tractors are more likely to go missing.
employment is higher in constituencies whose MLAs are aligned with the This finding speaks to an old debate in the corruption literature: the
state-level government. Prakash et al. (2015) find that the election of contrast between costly rent-seeking or “greasing the wheels”. Theoret-
criminal MLAs leads to lower economic growth in their constituencies. ically, corruption is typically thought of as rent-seeking. Public officials
Fisman et al. (2014) show that the assets of marginally elected MLAs use their control over the allocation of contracts or the provision of
grow more than those of runners-up, which confirms the idea that there services to ask for bribes (e.g. Becker and Stigler, 1974; Krueger, 1974;
are substantial private returns to holding office.2 Existing work has Rose-Ackerman, 1975; Shleifer and Vishny, 1993). This behaviour is
shown that MLAs have influence over the assignment of bureaucrats (Iyer most likely to arise in contexts where enforcement is weak and officials
and Mani, 2012). In further analysis, our paper documents that the are poorly remunerated.5 The so-called “greasing the wheels” hypothesis
misallocation of roads is stronger when MLAs and local bureaucrats argues that such corruption can be optimal in a second-best world, by
(District Collectors) share the same name, and weaker when local allowing agents to circumvent inefficient institutions and regulation
bureaucrats are up for promotion and subject to greater scrutiny. These (Huntington, 1968; Lui, 1985). In principle, both arguments could apply
results suggest that bureaucrats play an important role in facilitating to the preferential assignment of PMGSY roads by Indian MLAs.6
political corruption, and they help to explain how politicians can exert
influence even when they do not have any formal role. Hence, our paper
provides micro-evidence that accounts for MLAs' disproportionate impact 3
Other exponents of the “cross-checking” approach include Acemoglu et al. (2014),
on the economies of their constituencies, and for the private benefits they Golden and Picci (2005), Reinnika and Svensson (2004), Olken (2007), Fisman (2001),
derive from holding office. and Banerjee et al. (2014). Several countries conduct regular audits of local government
expenditure and make the results publicly available. Examples of research based on these
Our second contribution is to demonstrate a new approach to quan-
data include: Ferraz and Finnan (2008, 2011) and Melo et al. (2009) for Brazil; or Lar-
tifying politicians' influence over public procurement contracting. The reguy et al. (2014) for Mexico; and Bobonis et al. (2016) for Puerto Rico. In some settings,
core challenges we confront in doing so are that: (i) there is no infor- corruption can be observed directly, as in the driving license experiment conducted by
mation on actual connections between politicians and the contractors Bertrand et al. (2007) or in the trucking survey of Olken and Barron (2009).
4
Cingano and Pinotti (2013) measure the welfare costs of preferential contract allo-
active in their constituency; and (ii), to the extent that politicians inter-
cation through simulation techniques. Fisman and Wang (2015) show that politically
vene in the allocation of roads on contractors' behalf, such improper connected firms in China have higher worker death rates.
5
In the case of Indian MLAs, calculating efficiency wages (as suggested by Becker and
Stigler, 1974) may be complicated by the fact that candidates frequently need to pay their
parties significant sums for their place on the ticket. This could prompt them to engage in
2
Gulzar and Pasquale (2016) also confirm the importance of MLAs for local develop- corrupt behaviour once elected (Jensenius, 2013).
6
ment outcomes. In blocks that are split between different MLAs, the implementation of An intermediate argument is that initial corrupt allocations may not matter if there is
India's rural employment guarantee is worse than in blocks that are entirely part of one scope for Coasian bargaining. Sukhtankar (2015) finds evidence in this direction for the
MLAs constituency. allocation of the wireless spectrum in India.
63
However, the evidence we present on missing roads clearly supports the 2. Background
rent-seeking hypothesis, which is consistent with most of the recent work
that relies on the direct observation of corruption (e.g. Bertrand et al., 2.1. PMGSY
2007) or cross-checking approaches (Banerjee et al., 2013). The simple
“missing infrastructure” measure we propose, can be used in a wide In the year 2000, an estimated 330,000 Indian villages or habitations
variety of contexts. It is a particularly cost-effective alternative to phys- – out of a total of 825,000 – were not connected to a road that provided
ical road audits as conducted by Olken (2007). Improvements in remote all-weather access (NRRDA, 2005). Their inhabitants were at least
sensing techniques mean that confirming the existence of administra- partially cut-off from economic opportunities and public services (such as
tively completed projects will become a very economical way to detect health care and education). To address this lack of connectivity, the
and measure the diversion of public funds, even when census data is Indian government launched the Pradhan Mantri Gram Sadak Yojana
not available. (PMGSY) in December 2000. Its goal was to ensure all-weather access to
Our fourth contribution is to shed new light on the electoral motives all habitations with populations over 1000 by the year 2003, and to those
for corruption. A standard explanation for why politicians target with more than 500 inhabitants by 2007. In hill states, desert and tribal
patronage along in-group lines, which in India often means caste, is that areas, as well as districts with Naxalite insurgent activity, habitations
it acts as a form of vote-buying (Banerjee et al., 2014). Targeting with a population over 250 were targeted (NRRDA, 2005). The proposed
patronage could be easier within ethnic or caste groups (Chandra, 2004; network of roads was determined ex-ante in 2001, and the imple-
Horowitz, 1985). In the context of large-scale contracts like the ones we mentation of PMGSY in subsequent decades has consisted of the gradual
study, patronage could also be used to reward firms who help fund realisation of this “Core Network”.
political campaigns. Mironov and Zhuravskaya (2016) document how The programme has been described as “unprecedented in its scale and
Russian politicians allocate contracts to firms who funded their elec- scope” (Aggarwal, 2017), with roadwork for over 125,000 habitations
tion. Sukhtankar (2012) shows that political candidates in India siphon completed and another 22,000 under construction as of November
funds from sugar mills in election years. However, in our analysis, we 2016.7 A second phase of the scheme (PMGSY II), launched in 2013,
find no evidence that the preferential allocation of roads or cost infla- targets all habitations with populations over 100. According to World
tion increase immediately before or after election dates. If vote-buying Bank estimates, expenditures under PMGSY had reached 14.6 billion
is going on for this programme it must be a long-run transaction. If USD by the end of 2010, with a further 40 billion USD required for its
anything, we observe that roads built by connected contractors are less completion by 2020 (World Bank, 2014).
expensive around election periods – which could be consistent with Several studies have focused on the first-order research question that
higher scrutiny in election times. In the context of Puerto Rican arises in relation to PMGSY: its impact on habitations and the lives of
municipalities, Bobonis et al. (2016) show that financial audits are most their inhabitants. Asher and Novosad (2016) analyse the employment
effective in reducing corruption when they are conducted shortly before effects of the programme in previously unconnected villages. They find
elections. A recent paper by Bohlken (2016) argues that road comple- that a new paved road raises participation in the wage labour market
tion in PMGSY is higher when the ruling party in the state is aligned with a commensurate decrease in the share of workers employed in
with local MLAs in marginal constituencies, which suggests that voters agriculture. This translates into higher household earnings and a rise in
hold the state government accountable for PMGSY performance. As a the share of households who live in houses with solid roof and walls.
second test of electoral motives for corruption, we exploit India's 2008 Aggarwal (2017) also finds a positive effect on employment and reduced
re-drawing of electoral constituency boundaries to study the behaviour price dispersion among villages. While these studies analyse what PMGSY
of MLAs in regions that have become “politically irrelevant” after the has achieved, this paper looks at how it has been implemented.
redistricting. We find no evidence of different behaviour in these Compared to other public works programmes, the implementation of
regions. Thus, while our paper documents the preferential allocation of PMGSY stands out because of its reliance on private contractors com-
road contracts, we find no evidence linking corrupt behaviour to elec- bined with relatively strong monitoring and quality assurance provisions,
toral incentives. designed to limit the scope for undue corruption. All tenders have to
Our results are more consistent with either standard in-group follow a competitive bidding procedure, for which the rules were pre-
favouritism, or a subtler mechanism by which caste or kinship scribed by the National Rural Roads Development Agency (NRRDA) and
networks facilitate corrupt exchange under the threat of punishment. set out in the so-called Standard Bidding Document (SBD). The SBD
Corruption is illegal and therefore requires either trust among consists of a two envelope tendering process administered at the circle
collaborators, or a predictable ability to sanction defections, both of level. Each bid consists of both technical and financial volumes. The
which are more likely to exist between members of the same family, technical bids are opened first. Contractors have to fulfil eligibility
ethnic group, or network (Lambsdorff, 2002; Tonoyan, 2003). While criteria, taking into account factors such as their current workload and
we are unable to test it explicitly, this interpretation fits our findings experience. Only the financial bids of contractors whose technical bids
and the context of PMGSY. The involvement of the central govern- are found to meet the requirements are evaluated, and subject to meeting
ment in the programme guarantees a minimum level of monitoring. In the technical standards, the lowest bidder has to be selected. After the
line with the idea that contractors trade off rent-seeking and the cost contract has been assigned, administrative data on the programme is
of detection, we find no evidence that preferential allocation affects gathered, while central and state-level inspectors can carry out quality
the performance markers that are most easily observed in the inspections. In spite of these provisions, there remains clear scope for
administrative data collected at the central level: over-runs and corruption, and the financial incentives are sizeable given the scale of the
delays. We also find that preferentially allocated contracts are less project.8 A large number of newspaper reports document alleged cor-
expensive in the run-up to elections, a time when monitoring may
be greater.
The remainder of the paper proceeds as follows. Section 2 provides
context on PMGSY, the role of MLAs, and Indian surnames as identifiers
of caste or religion. Section 3 describes the dataset used in the analysis.
7
Section 4 outlines the empirical strategy. Section 5 presents the main OMMAS (Online Management, Monitoring and Accounting System), http://omms.nic.
results on re-allocation and robustness. Section 6 analyses the social costs in/, accessed in November 2016.
8
Existing work reports that the price bid of only one firm was evaluated in 95% of a
of re-allocation. Section 7 provides evidence for the intermediary role of
random sample of 190 road contracts issued between 2001 and 2006 in Uttar Pradesh; i.e.
bureaucrats. Section 8 evaluates electoral motives for corruption. Section only one bid submitted or all other bids were disqualified based on technical requirements
9 concludes. (Lewis-Faupel et al., 2016).
64
ruption in PMGSY.9 Corruption in PMGSY could take several forms, and surnames can be an indicator of caste affiliation, religion, or geographic
the possible manipulation of road allocations is one of the challenges for provenance. Naming conventions differ across India; it is common for
impact evaluations of the programme (Asher and Novosad, 2016).10 Our Indians to have multiple surnames and the same name can appear in
paper tests for a specific form of corruption: interventions by state-level different positions within the list of names. This is also true of caste
parliamentarians (MLAs) in the allocation of road contracts (but not of identifiers. Still, as a general rule, the last name will be caste or religion
the location of roads) within their constituencies. specific and follow the paternal line. This pattern is sufficiently strong
An advantage of focussing on MLAs in this context is that under the for Indian surnames to have been used as identifiers of caste or religion
programme guidelines, they should be in no way involved in the in many empirical studies (Banerjee et al., 2014; Hoff and Pandey,
tendering process or the selection of contractors. In fact, they are granted 2004; Vissa, 2011; Fisman et al., 2017; Field et al., 2008). One clear
practically no official role in the implementation of PMGSY whatso- exception to this rule is Tamil Nadu, where surnames do not exist, and
ever.11 Funding for PMGSY comes primarily from the central govern- we will document that our method does not work for Tamil Nadu.
ment. The scheme is managed by local Programme Implementation Units Surnames are a significantly more accurate predictor of connections
(PIUs), which are under the control of State Rural Roads Development when comparing individuals from the same area or linguistic region.
Agencies (SRRDA). These agencies are responsible for inviting tenders The contractors and politicians in our setting are highly likely to be
and awarding contracts. Given their lack of formal involvement, any from the same state and will in many cases be from the same district or
systematic relationship between MLAs and the contractors working in constituency.
their constituencies can therefore, in itself, be construed as evidence for Our paper treats a match between the names of a politician and a
an irregularity in the allocation of contracts. contractor as a rough overall measure of proximity, without seeking to
establish whether the individuals are of the same religion, caste, or
2.2. The role of MLAs (potentially) family. All of these types of connections are likely to
increase the probability that a contractor would approach an MLA when
MLAs, or Members of Legislative Assembly, are India's state-level bidding for a contract, and that the MLA would be receptive.
parliamentarians. They are elected for 5-year terms in a first-past-the- Name-based matching is an imperfect measure of proximity. Con-
post voting system in state-wide elections. In general, the state assem- tractors may have connections to politicians without sharing a name, or
bly elections in India's different states do not coincide. MLAs are typically equally, share a name but have no connection. Surnames that are not
nominated by the party, and each MLA represents a single constituency. caste-identifiers, former honorific titles for example, are likely to dilute
Is it plausible that these MLAs would seek to intervene on behalf of the accuracy of the measure. Hence, the estimates in this paper can be
specific contractors? While their official function is to represent their viewed as a lower bound for MLAs' true effect on contract allocation.
constituents in state legislative assemblies, surveyed MLAs over-
whelmingly report this to be a minor part of their work (Chopra, 1996). 3. Data
State assemblies meet rarely and according to Jensenius (2013), indi-
vidual legislators have little impact on political decisions: “much more The empirical strategy requires three kinds of data. Information on
important to the MLAs are all their unofficial tasks of delivering pork, contractors and agreements is available in the administrative records of
blessing occasions, and helping people out with their individual prob- the PMGSY project, at the road level. Data on political candidates and
lems”. Qualitative accounts suggest that MLAs spend much of their time elections are at the level of the assembly constituency. These two are
receiving requests from their constituents. Describing such meetings linked using the population census, which allows for habitations to be
Chopra (1996) writes “constituents came to ask for favours that clearly matched to constituencies, as well as providing additional covariates
contravened rules and laws”. MLAs often respond to requests by passing used in the analysis.
them on to ministers or high-ranking officials, and they are also known to
put pressure on bureaucrats by threatening them with reassignment (Iyer 3.1. PMGSY data
and Mani, 2012; Bussell, 2015).
The administrative records of projects sanctioned under PMGSY are
2.3. Surnames as a measure of interpersonal proximity in India publicly available in the Online Management, Monitoring, and
Accounting System (OMMAS). The dataset used for this paper contains
To measure proximity between MLAs and contractors we construct a the agreement details of 110,185 roads serving 188,394 habitations.
proxy based on politicians' and contractors' surnames.12 Indian This information includes: the date of contract signing, sanctioned cost,
proposed length, proposed date of completion, name of the contracting
company, and – crucially for this analysis – the name of the winning
9
Examples include articles in “The Hindu” on April 11, 2012, “The Economic Times” on contractor. In addition to the agreement details, which precede road
March 8, 2013, “The Arunachal Times” on March 6, 2013, the online news-platform
construction, the OMMAS also contains later data on the physical
“oneindia” on July 31, 2006, and “Zee News” on 30 August 2014. For example, the
“oneindia” article reports that the former Chief Minister of Sikkim accused the current progress of work, data on completed roads, and reports from subse-
administration of “widescale corruption” in the implementation of PMGSY and “alleged that quent quality inspections. These are used in section 6 to evaluate the
the works were awarded to relatives of Chief Minister, Ministers and MLAs of the state”. effect of political interference on the efficiency and quality of road
10
These authors find that the habitation population figures reported to PMGSY had been construction.
manipulated, particularly around the 1000 and 500 population cut-offs used to target the
programme.
11
MLAs are mentioned in the PMGSY guidelines, but only in reference to the initial 3.2. Assembly election data
planning stage. Intermediate panchayats and District panchayats were responsible for
drawing up a planned “Core Network” which encompasses all future roadwork to be The Election Commission of India (ECI) publishes statistical reports
carried out under PMGSY. These plans were to be circulated to MPs and MLAs, whose
on assembly elections that record each candidate's name, party, gender
suggestions were to be incorporated. MLAs could therefore have influenced which habi-
tations were targeted ex-ante through official channels. However, this role is irrelevant for and vote share. Since 2003, candidates have moreover been required to
the timing of the construction work and assignment of road contracts, on which MLAs submit sworn affidavits to the ECI with information on their assets, lia-
have no formal influence. Moreover, these consultations took place prior to our sample bilities, educational attainment, and any pending criminal cases. Both the
period. The majority of MLAs in our sample were not in office at the time and therefore election reports and affidavits are publicly available from the ECI in pdf
had no opportunity to review the planned network. Our results are unchanged when we
drop MLAs who were in office prior to 2001 from the sample (see Appendix Table A10).
format. This paper draws on digitised versions of this information from
12
Angelucci et al. (2010), and Mastrobuoni and Patacchini (2012) also use name-based four separate sources. Table A1 of the online appendix lists these sources
matching to study social networks. – which cover different time periods and variables – and describes which
65
variables from each source are used in the analysis (all these secondary The results are, however, robust to broader definitions of matches.16 To
sources are based on the ECI).13 account for different spellings of the same name, we implement a fuzzy
Assembly elections operate on a plurality rule. While the median matching algorithm optimised for Hindi names.
number of candidates per election is eight, typically only the top-2 can- Matches are aggregated at the electoral term level as follows. The
didates are competitive: the third placed candidates average 7% of the variable matchnijt takes the value of 1 if the contractor for a road agree-
vote, the fourth placed candidates average 3%, the fifth 1.6% and the rest ment n, signed in constituency j in term t, shares a name with candidate i,
less than 1%. To estimate the RD we restrict attention to elections in and 0 otherwise. This variable is determined for the N road agreements
which there are PMGSY contracts issued in the term before and after the signed in the constituency during an electoral term. shareijt is defined as
election and focus on the winner and runner-up. This gives us a sample of the share of contracts in term t allocated to contractors who share a
8116 candidates in 4058 elections from 2001 to 2013, covering 2632 candidate's name. shareijt1 provides the equivalent share for contracts
constituencies. In our preferred specification we estimate on the resulting in the term prior to the election in which a candidate took part.
sample of 8116 candidate-terms. In a placebo test we show that the Pn Pn
effects are not present for the contrast between the runner-up and third- n¼1 matchnijt n¼1 matchnijt1
shareijt ¼ shareijt1 ¼
placed candidate. Map 1 (of the online appendix) shows the constitu- Njt Njt1
encies included in the sample which cover 24 of the 28 states that existed
The dependent variable in the main regressions is the difference be-
during the timeframe under analysis.14 Map 2 shows the constituencies
tween these two:
which had at least one close election, the sub-sample for our local linear
RD estimation. Δshareijt ¼ shareijt shareijt1 ;
3.3. Matching roads and electoral terms using census data which we calculate for all candidates i 2 fwinner; runner upg.
A complication arises in elections where winning and losing candi-
The Population Census of India 2001 contains village-level data on dates have the same surname. It is not possible to estimate the effect of
demographic and socio-economic variables used as controls in the winning an election in this situation, as candidates who lost will see their
analysis. We use the Village Amenities part of the 2011 census, to identify proximity to contractors evolve in parallel to that of the elected politi-
the ‘missing roads’ evaluated in section 6. cians. In the main regressions, we therefore exclude candidates from
The 2001 census is also the source for habitation-level data, which elections where this issue arises.
was collected by the PMGSY in order to determine the prioritisation of
roads. This includes information on the size of the population (the project
guidelines stipulate that habitations above certain population thresholds 3.5. Descriptive statistics
are to be prioritised), whether or not it was connected to a road in 2001,
and if so, whether this road provided all-weather access. Moreover, it Table 1 reports descriptive statistics for the sample of candidates used
reports the MLA constituency in which it each habitation was situated in the main regressions. For the average term in the sample, the number
in 2001. of road contracts signed is 28. The average value of shareijt1 – which can
Using this information, it is possible to match PMGSY roads (at the be construed as a baseline measure of the frequency of surname-matches
habitation-level) to the assembly election data described in the previous – is 4%. There is however, significant geographic variation in the
sub-section. However, changes in the delimitation of MLA constituencies frequency of matches, ranging from a mean of 0% in Mizoram to a mean
– which took effect in mid-2008 – led to changes in boundaries, the of 13% in Andhra Pradesh (Map 3 of the online appendix shows this
abolition of some constituencies, and the creation of new ones. For roads variation at the constituency-level).17 However, these means do not
built in electoral terms after the new delimitation we use the coordinates distinguish between winning and losing candidates – the variation
of habitations and match these to GIS data on constituency boundaries. exploited in the empirical strategy below.
While the census data allows for spatial matching of roads and con-
stituencies, it is also necessary to match them in time. Road contracts are 4. Empirical strategy
allocated to electoral terms based on the date of the agreement, as
recorded in the PMGSY data. In order to precisely assign road contracts, it A natural control group for elected politicians are those who aspire to
is necessary to set an exact date that marks the end of one term and the the same office. If being an MLA is associated with the power to intervene
beginning of the next. We define this as the date on which the results of in the allocation of roads in one's constituency, one would expect the
an election are announced.15 share of contractors with the same name as a winning candidate to be
higher than the corresponding share for losing candidates.
As our main outcome of interest, we use the first difference of shareijt .
3.4. Matching politicians and contractors using surnames
Taking the difference should remove unobservable, time-invariant
characteristics of an individual candidate that may be correlated with
In the electoral terms that preceded and followed the elections in the
the number of matches with contractors. In our context, this is primarily a
sample, 88,020 road agreements were signed. For each political candi-
way of controlling for specificities that individual names may have
date, we assess whether they share a surname with the contractors who
within certain constituencies. Some candidates' names will be more
received projects in their constituency in the term after the election. For
common than others. Some may be more prevalent among certain pro-
every politician-contractor pair, we exclude all names except for each
fessions (e.g. contractors) for historical reasons. Under the assumption
individual's final name and then look for matches among these surnames.
that winning and losing candidates had a common trend in their share of
matches with contractors, a simple difference-in-differences (DiD)
13
The matching process is complicated by discrepancies in the spelling of constituency
approach would be sufficient for identification. However, given that
and candidate names. These occur not only across datasets but also across time within
datasets. Using different secondary sources helps us to construct a consistent data set. In a
16
small number of cases, multiple constituencies within the same state have the same name. The results are robust to considering all matches among individuals' names (excluding
We drop all of these constituencies from our sample, to prevent false matches between their first name) or only matches based only on the last two names.
17
election datasets and to avoid the risk of assigning roads to the wrong constituency. It is likely that these baseline frequencies lead to heterogeneous treatment effects. In
14
Goa, Meghalaya, Nagaland and Sikkim are not part of our sample. states or constituencies, where the distribution of names is such that matches are relatively
15
These dates were collected from the website www.electionsinindia.com (accessed in rare, a politician who is elected may not have many potential contractors of the same
2015). name to allocate roads to.
66
Table 1A
Descriptive statistics (candidate/constituency).
Variable Observations Mean Std. Dev. Min Max
Panel A: Roads allocated to contractors of the same name

Sharet-1 8116 0.037 0.141 0.000 1.000
Sharet 8116 0.035 0.134 0.000 1.000
ΔShare 8116 0.002 0.151 1.000 1.000
Panel B: Candidate characteristics
Vote share 8116 0.275 0.097 0.020 0.837
Margin 8116 0.000 0.102 0.695 0.695
Incumbent 8116 0.282 0.450 0.000 1.000
Runner-up previous election 8116 0.147 0.354 0.000 1.000
Age 7580 49.179 10.191 23 87
Female candidate 8116 0.062 0.242 0.000 1.000
Candidate with criminal charges 5990 0.196 0.397 0.000 1.000
Total assets (1000000s of INR) 5009 106 4740 0.000 300000
Liabilities (1000000s of INR) 5286 1.841 17.800 0.000 644
University graduate 5286 0.596 0.491 0.000 1.000
Postgraduate degree 5286 0.192 0.394 0.000 1.000
Congress candidate 8098 0.290 0.454 0.000 1.000
BJP candidate 8098 0.192 0.394 0.000 1.000
Named Kumar 8116 0.058 0.234 0.000 1.000
Named Lal 8116 0.022 0.145 0.000 1.000
Named Patel 8116 0.009 0.094 0.000 1.000
Named Ram 8116 0.018 0.133 0.000 1.000
Named Reddy 8116 0.016 0.124 0.000 1.000
Named Singh 8116 0.112 0.316 0.000 1.000
Named Yadav 8116 0.014 0.117 0.000 1.000
Panel C: Constituency characteristics
Reserved seat 8116 0.275 0.446 0.000 1.000
Road countt 8116 27.691 30.822 1.000 479
Road countt-1 8116 22.086 25.744 1.000 388
Mean road lengtht 8116 5.833 3.999 0.350 42.654
Mean road lengtht-1 8116 4.963 3.838 0.410 53.985
Mean population 7822 961.697 633.986 30.000 7230
Mean SC/ST population 7822 244.078 193.401 0.000 2283
Mean connectivity 7822 0.561 0.308 0.000 1.000
Note: The number of observations varies due to missing values. Reserved seat refers to constituency reserved for MLAs from scheduled castes or tribes.
Road countt is computed at the term-level by counting the number of road contracts signed in a constituency within a term. Mean road length is the
average length of roads (in km) built in a constituency and term. Mean population and mean SC/ST population are averages of 2001 census data for all of
a constituency's villages. Mean connectivity is the share of a constituency's villages that had all-weather road access at the time of the 2001 census.
Table 1B
Descriptive Statistics (Roads built by same name contractors).
Variable Observations Mean Std. Dev. Min Max
Panel A: Road construction

Length of road 4921 3.997 3.654 0.01 41
Bridge 4921 0.008 0.091 0 1
Cost (1000000s of INR) 4921 133.7 149.625 3 2730.56
Log of (sanctioned cost/km) 4921 3.347 0.651 0.0312 11.24
Days overrun 3202 588.0 543.8 1750 3932
Actual cost/sanctioned cost 3867 0.954 0.257 0 9.35
Failed Inspection 1513 0.365 0.482 0 1
Missing all-weather road 1815 0.260 0.438 0 1
Missing any road 1815 0.025 0.156 0 1
Panel B: Local Geography
Altitude 4921 588.707 805.450 0 4864.6
Ruggedness 4921 0.243 0.657 0 8.506
Forest Cover 4494 0.398 0.110 0 0.861
Distance to nearest town 4482 22.622 20.994 0 244
Panel C: Village Demographics
Total population 4464 2416.8 3425.2 1 49192
Number of households 4464 449.2 707.8 1 9851
Village area 4482 722.7 1772.0 0 67737
Sex ratio 4462 1.062 0.101 0.583 2.1
Population under 6 4464 0.179 0.042 0 0.335
SC share 4464 0.173 0.154 0 1
ST share 4464 0.106 0.238 0 1
Panel D: Village Socioeconomic Characteristics
Literacy 2952 54.019 16.546 0 100
Employment 2952 39.166 12.141 12.938 100
Male employment 2952 0.499 0.075 0.227 1
Female employment 2952 27.431 19.772 0 82.304
Drinking water 4464 0.998 0.047 0 1
Power supply 4461 0.767 0.423 0 1
Phone connections 4,482 10.576 79.597 0 1713
Road connectivity in 2001 4921 0.311 0.463 0 1
Note: Table 1b provides descriptive statistics for the sample used in Tables 4 and 5 and columns 4–6 of Table 6. Cost, delays, cost overruns, and quality
inspections are based on the PMGSY data. Quality is a dummy variable equal to one if the road is “unsatisfactory” or “in need of improvement” in latest
inspection. “Missing roads” are defined on the basis of the 2011 census and the PMGSY data. The dummy for the all-weather road missing variable takes the
value of one if any village on the route of an officially completed road lacks all-weather road access according to the 2011 census. For the missing-any-road
definition, we set the missing dummy equal to one if all villages on the completed PMGSY road had no road of any type (tarmac, gravel, or water bound
macadam) according to the 2011 census.
Table 2
Randomization test for MLA level local linear regression.
Margin of victory: 6.2% 3%
Obs. Winner St. Error Obs. Winner St. Error
Panel A: Candidate characteristics

Sharet-1 4396 0.0072 (0.0084) 2472 0.0114 (0.0118)
Incumbent 4396 0.0366 (0.0296) 2472 0.0644 (0.0410)
Runner-up in prev. election 4396 0.0119 (0.0214) 2472 0.0212 (0.0295)
Age 4127 0.2740 (0.5990) 2314 0.3868 (0.8370)
Female candidate 4396 0.0004 (0.0129) 2472 0.0086 (0.0170)
Cand. faces criminal charge 3352 0.0075 (0.0264) 1889 0.0562 (0.0370)
Assets (INR millions) 2770 261.47 (719.00) 1554 365.99 (1.1685)
Liabilities (INR millions) 3049 0.1260 (0.5146) 1747 0.0494 (0.473)
University degree 3049 0.0139 (0.0328) 1747 0.0262 (0.0447)
Post-grad. degree 3049 0.0047 (0.0271) 1747 0.0150 (0.0373)
BJP candidate 4387 0.0147 (0.0256) 2465 0.0271 (0.0361)
Congress candidate 4387 0.0257 (0.0297) 2465 0.0468 (0.0412)
Panel B: Share of roads built by contractors of same name in term prior to election
Share 5 yrs before election 2502 0.0078 (0.0111) 1416 0.0138 (0.0155)
Share 2 yrs before election 1688 0.0037 (0.0133) 942 0.0319* (0.0189)
Share 1 year before election 1866 0.0131 (0.0152) 1070 0.0203 (0.0207)
Panel C: Prevalence of most common names
Named Kumar 4396 0.0119 (0.0136) 2472 0.0461** (0.0188)
Named Lal 4396 0.0019 (0.0084) 2472 0.0114 (0.0118)
Named Patel 4396 0.0026 (0.0061) 2472 0.0053 (0.0079)
Named Ram 4396 0.0021 (0.0076) 2472 0.0017 (0.0102)
Named Reddy 4396 0.0074 (0.0054) 2472 0.0036 (0.0076)
Named Singh 4396 0.0195 (0.0171) 2472 0.0203 (0.0244)
Named Yadav 4396 0.0052 (0.0076) 2472 0.0067 (0.0104)
Note: Coefficients are estimated by regressing the row variables on winner, the vote margin, and the vote margin interacted with winner in OLS regressions
Standard errors are clustered at the election level. The bandwidth of 6.2% is derived from the optimal bandwidth choice rule of Imbens and Kalyanaraman
(2012). ***p < 0.01, **p < 0.05, *p < 0.1.
winners are likely to be systematically different from losing candidates in required for identification.19 Because we have the top-two candidates in
many respects, it is possible that they may face divergent trends in shareijt each election we cluster standard errors at the election level.20 The
that are not determined by election outcomes. This suggests the use of a results of non-parametric RD estimations can be sensitive to the choice of
regression discontinuity (RD) design.18 bandwidth, and there is trade-off between bias and efficiency inherent in
In order to identify whether there is a causal relationship between the this choice (Lee and Lemieux, 2010). For our main results we apply two
election of politicians and the allocation of road contracts in their con- bandwidths: 6.2% (derived from the optimal bandwidth choice rule of
stituencies, we exploit the fact that in close elections, the assignment of Imbens and Kalyanaraman, 2012) and a more conservative bandwidth of
victory can be considered conditionally independent of subsequent 3%. In later sections, when we show results for other dependent variables
contracting patterns. The underlying assumption is that candidates who with different optimal bandwidths, we continue to report the 3% band-
won an election by a very small margin are comparable to those who width for consistency.21
narrowly lost (Lee, 2008). We evaluate whether this assumption holds in While using an RD to identify the effects of electoral outcomes is
our sample by running balance checks on observable characteristics (see standard, our setting is different from many applications in that we
below). In order to determine how close elections were, we define the exploit within-constituency variation. We compare winning and losing
variable marginijt for candidate i in constituency j and term t: candidates (and the contractors sharing their surnames), rather than the
electoral constituencies which were narrowly won or lost by a particular
vote shareðwinnerÞjt vote shareðrunnerupÞjt if winnerijt ¼ 1 type of candidate. Recent work using similar candidate-level RDs in-
marginijt ¼
vote shareðrunnerupÞjt vote shareðwinnerÞjt if winnerijt ¼ 0 cludes Do et al. (2015) and Fisman et al. (2014).
We estimate Equation (1) in a non-parametric RD for a range of 5. Reallocation result

bandwidths μ, controlling for the assignment variable marginijt and its
interaction with winnerijt to allow for a different relationship between 5.1. Continuity test
Δshareijt and marginijt among winning and losing candidates:
Our identification strategy is based on the premise that all con-
Δshareijt ¼ α þ βwinnerijt þ δmarginijt þ ρwinner*marginijt þ εijt founding factors are smooth at the treatment threshold, so that any dis-
(1)
∀ i where marginijt 2 ½μ; μ and i 2 fwinner; runner upg continuities capture the causal effect of a politician winning his/her MLA
In order to improve the efficiency of the estimates, we introduce election. Table 2 presents the results of a continuity test for the optimal
constituency-level controls, individual-level controls, state fixed-effects,
and year fixed-effects in most specifications although these are not
19
Legislative assembly terms are not synchronised across Indian states. In each year in
our sample window, there were elections in multiple states.
20
The main results are robust to clustering standard errors at the state-year level to
18
We obtain similar results when estimating the RD using levels with shareijt as the account for within-state-political-season correlations in the errors.
dependent variable (Table A2). Taking the first difference of shareijt in our RD estimation 21
Applying the optimal bandwidth choice rule of Imbens and Kalyanaraman (2012) for
contributes to precision and allows for a direct comparison of our RD results with the DiD each RD estimation in our paper, the mean optimal bandwidth is 3.8% and the median is
results for the full sample (see discussion of LATE in section 5). 3.5%.
68
Table 3
Local linear regression RD.
Dependent Variable: ΔShare of same name contractorst Whole Sample Margin of Victory <6.2% Margin of Victory <3%
(1) (2) (3) (4) (5) (6)
Winner 0.0094* 0.0099* 0.0252*** 0.0242** 0.0319** 0.0323**

(0.0052) (0.0054) (0.0092) (0.0097) (0.0128) (0.0138)
Margin 0.0001 0.0114 0.3153* 0.2697 1.0165** 1.0816**
(0.0299) (0.0337) (0.1856) (0.1839) (0.5049) (0.5481)
Margin*winner 0.0059 0.0353 0.0835 0.0577 0.8541 0.9464
(0.0414) (0.0445) (0.2724) (0.2671) (0.6269) (0.6683)
Incumbent 0.0014 0.0034 0.0053
(0.0045) (0.0065) (0.0086)
Runner-up in previous election 0.0068 0.0074 0.0069
(0.0055) (0.0079) (0.0110)
Female candidate 0.0012 0.0130 0.0014
(0.0066) (0.0107) (0.0117)
Age 0.0002 0.0000 0.0004
(0.0002) (0.0003) (0.0003)
AC controls X X X
State fixed effects X X X
Election year fixed effects X X X
N 8116 7290 4396 4012 2472 2248
Control group mean sharet-1 0.0342 0.0344 0.0355 0.0362 0.0383 0.0390
Note: Local linear regression estimates. Standard errors are clustered at the election-level. Variables are defined either in the text or in the note for Table 1. AC controls
include: Reserved seat, Road countt-1, Mean population, Mean SC/ST population, Mean connectivity, Mean road lengtht-1. The bandwidth of 6.2% is derived from the
optimal bandwidth choice rule of Imbens and Kalyanaraman (2012). ***p < 0.01, **p < 0.05, *p < 0.1.
bandwidth of 6.2% and the 3% bandwidth. At the optimal bandwidth Reassuringly, the results are consistent across a wide range of band-
none of the MLA characteristics display a discontinuity when the vote widths. Fig. 2 plots the coefficient on winnerijt for the main specification
margin exceeds zero. At 3%, only two variables are discontinuous at the for different bandwidths. As the samples get smaller the estimates are less
10% significance level.22 precise but the coefficient is relatively stable for all but very small
bandwidths (less than 1%).
Relative to the total number of roads – most of which are allocated to
5.2. Main results contractors whose name does not match the MLA's – the absolute value of
the coefficient implies a small effect. Yet as explained in section 3.3, these
The results of local linear regression RD estimation are presented in estimates can be considered a lower bound on MLAs' true intervention in
Table 3 and Fig. 1. For each bandwidth there are two columns in the PMGSY contract allocation.25
table. The first corresponds to the basic RD in Equation (1). The second If the results are interpreted as evidence of improper political
adds state fixed effects, year fixed effects and additional controls. These involvement in the assignment of roads, it raises the question whether
include whether or not a constituency is reserved for candidates from this improper involvement only occurs on behalf of individuals with the
scheduled castes (SC) or scheduled tribes (ST), characteristics of the same surname.26 In this sense the sign and significance of the coefficient
PMGSY roads built in the constituency prior to the election, and might be seen as more important than the magnitude. Secondly, given
candidate-level controls. The latter set of variables includes a candidate's the scale of PMGSY, even a relatively small fraction can translate into
vote share, their age, gender, and whether they were an incumbent or a what can be considered a sizeable number of affected roads and sub-
former runner-up. stantial financial expenditure. This is illustrated by the following, back-
For both bandwidths, the effect of winning an election on the change of-the-envelope calculation. In our dataset (including the first electoral
in shareijt is consistently positive and significant.23 In our preferred term), 4127 road projects were allocated to contractors sharing a name
specification including fixed effects and the full set of controls the coef- with the MLA. The total sanctioned cost of these projects was 56 billion
ficient is around 0.024 at the 6.2% bandwidth and 0.032 at the 3% INR, or around 1.2 billion USD.27 Applying our preferred RD estimate
bandwidth. Relative to the baseline, pre-election level of matches, the (3% bandwidth) to the full sample, would imply that MLAs had inter-
latter estimate implies that the effect of a candidate coming to power is vened in the allocation of roughly 1900 road contracts worth around 540
an 83% increase in the share of roads allocated to contractors who share million USD.28
their surname. 24
25
Spurious matches in names will bias the coefficient towards zero. In Appendix
22
Our results are robust to controlling for the name Kumar, and the overall pattern of Table A6 we drop all elections where either candidate has a name that occurs with a
results in Panel B does not suggest differential pre-trends. A simple comparison of means frequency of more than 10% in their state. This reduces the mean within-state name
for winners and losers within the 3% bandwidth shows that the two samples are balanced frequency by around 50%, significantly reducing the likelihood of spurious matches.
on all variables, including the name Kumar and the share two years before the election Compared to our main results the implied proportional increase in the share of matches
(Appendix Table A3). rises from 83% to 108%.
23 26
In Appendix Table A4 we report results for a fully non-parametric RD, excluding the Our name based approach will provide a more accurate measure of connections in
controls for the running variable and its interaction with the treatment. The results are areas where there is a strong association between surnames and caste. This is more likely
similar in magnitude and significant for all bandwidths. to be the case in Northern states, where we find a somewhat stronger effect. When we
24
Fafchamps and Labonne (2017) provide evidence from the Philippines that politicians focus on Tamil Nadu, where naming conventions imply that surnames do not provide a
punish family members of losing candidates. In theory, our results could be partly driven measure of connections, the coefficient is statistically insignificant and very close to zero.
27
by reductions for contractors connected to losing candidates. Our identification strategy Applying the average exchange rate over the period (December 2000 to December
does not allow us to test this directly. However, we analyse the simple change in shareijt for 2013): 1 INR ¼ 0.021 USD.
28
losing candidates comparing the terms before and after the election. Excluding losing The estimated impact in the RD with a full set of controls on a 3% bandwidth is an
incumbents (for whom there is a significant reduction after they leave office), we find no 83% increase. This implies that 45.3% of roads allocated to contractors with the same
significant decline in shareijt for losing candidates (see Table A5). name as the politicians would otherwise have gone to another contractor.
69
results as causal, one possible interpretation of this discrepancy is that

electoral competition intensifies political corruption in this setting. The
results of this section lend support to qualitative accounts on favouritism
in the allocation of PMGSY contracts. Only recently, BJP leader Munna
Singh Chauhan accused the Uttarkhand State Government of such
misallocations:
“There is a huge scam in tender allotment in Pradhan Mantri Gram

Sadak Yojana (PMGSY) in Bahuguna government. Of a total of 113
mega road construction projects, 75 contracts were awarded to cho-
sen ones close to the echelons of power on a single bid basis. […]
Coincidentally, one of the contractors awarded the project is also the
brother-in-law of state rural development minister Pritam Singh,”
(Quoted in Zee News, 30 August 2013).
Our analysis suggests that episodes of suspected favouritism in

particular states match a wider pattern of corruption that shows up in our
sample covering the whole of India.
5.3. Validity of the RD approach
The RD design requires that no variables other than the dependent

variable exhibit discontinuities at the cut-off. The test in Table 2 con-
stitutes the first assessment of the continuity of observable characteristics
across the win threshold. Fig. A2 of the appendix provide graphical ev-
idence for a number of candidate characteristics that there are no dis-
continuities across the cut-off.
Close elections can only be considered to provide quasi-random
treatment assignment when the probability density function of candi-
dates' vote shares is continuous (Lee, 2008). This will not be the case if
candidates are able to strategically manipulate their vote share.30 The
standard test for strategic manipulation of the running variable in a RD
design was formulated by McCrary (2008). Applying the McCrary test to
the assignment variable in this analysis (marginijt ), would not make sense
because the density is continuous by construction. For every winner with
a positive marginijt , there is a runner-up with the equivalent negative
value of marginijt . We therefore test for manipulation in the vote share
Fig. 1. Graphical depiction of RD. Note: Lines fitted separately on the samples left and based on an alternative variable: the margin of victory/defeat for the
right of the cut-off. Each dot represents the mean for a bin. The number of observations in candidate in the constituency with the higher value of shareijt1 . The
a bin varies, based on the density of candidates at a given margin. 90% confidence in- McCrary test does not reject the continuity of this variable at the
tervals plotted in grey. The first panel shows a linear fit within the 3% bandwidth used in
threshold. Fig. A1 in the online appendix presents a graphical depiction
the RD estimation. The second shows a quadratic fit for a wider margin of 20%.
of the test.
The online appendix provides further robustness checks, including
results of a parametric RD regression estimated on the full sample (in
Table A8), as well as the main results in levels instead of first differences
These estimates serve to illustrate the economic significance of even (Table A2). The main result is robust to these alternative specifications.
proportionately small misallocations in PMGSY contracts but they rely on To evaluate whether our results can be interpreted as the causal impact of
an extrapolation from a LATE. To what extent can the results of our RD gaining public office, as opposed to the information revealed by per-
estimation be considered informative for the programme as a whole? The forming well in an election, we conduct a placebo test comparing
constituencies with close elections in our sample are characterised by a runners-up to third-placed candidates (Table A9). If shifts in allocation
different political equilibrium. They are typically contested by more somehow reflect individuals' increased status following a strong electoral
candidates and parties, with each earning a lower vote share.29 Ex-post, performance, rather than their official position, second-placed candi-
the winner of a close election may be more concerned about re-election, dates might experience gains relative to third-placed candidates. We find
may need to raise greater resources for the next campaign and may no such effect; the coefficient on coming second is close to zero across all
choose different strategies with regard to patronage and political cor- specifications.
ruption. One way to evaluate this concern is to compare our RD estimate
to the equivalent DiD estimate for the full sample (column 2 of Table 3). 6. Social costs of misallocation
The coefficient is lower, and the implied post-election increase in con-
tractors of the same name falls to 28%. This estimate would indicate that PMGSY roads have been shown to deliver significant benefits in
‘only’ around 910 road contracts were diverted to contractors who targeted villages: improved labour market access, higher incomes, and
shared the MLAs surname. While we are less inclined to accept the DiD better living conditions (Asher and Novosad, 2016). In this section we
evaluate whether political interference in the allocation of PMGSY
29
The correlation between the number of candidates and the margin of victory is 0.22.
30
The correlation between the winner's vote share and the margin of victory is 0.59. We find Using data on close US house races, Caughey and Sekhon (2011) provide evidence of
no heterogeneity in our main result based on these observable measures of competition such strategic sorting. Eggers et al. (2015) examine over 40,000 close elections from a
(see Table A7). range of countries (including India) and find no other country that exhibits sorting.
70
Fig. 2. Main effect by bandwidth. Note: The chart plots the coefficient for winner in our main specification (equivalent to Table 3, columns 2, 4, 6, and 8) with the full set of candidate and
constituency controls as well as state and year fixed effects.
contracts undermines these benefits, or whether it is in fact Road Characteristicnsy ¼α þ βSame name as MLAnsy þ δmarginijt
welfare-enhancing. Theoretical work has contended that corruption þ ρSame name as MLA*marginijt þ γXnsy
could be beneficial by 'greasing the wheels' and allowing agents to
þ θs þ ϑy þ εnsy ; marginijt 2 ½μ; μ
circumvent inefficient bureaucracy (Leff, 1964). In a low information
environment with a potential for adverse selection and moral hazard, (3)
political connections may be associated with better information ex-ante While this RD-design is likely to be an improvement on a naïve OLS
or greater sanctioning power ex-post. Improved outcomes under pref- approach, it does not address one potential source of selection bias. To
erential allocation would constitute evidence in support of these the- the extent that politicians only intervene on behalf of their network for a
ories. Given that the location of PMGSY roads is officially subset of roads, and this selective intervention is not random, the ex-ante
predetermined, politicians are unlikely to influence where a road is built, characteristics of the roads in the treatment group may differ from those
but their informal control over who is awarded a contract may alter the in the control group. For example, politicians might try to ensure that
welfare impacts. To estimate the social costs (or benefits) of corruption more difficult projects are allocated to contractors from their network
we analyse projects at the road level, distinguishing between those whom they trust. Given that the available PMGSY data are predominantly
allocated to connected and unconnected contractors. Our principal determined ex-post – at the time of the contract or during construction –
outcome of interest is whether roads listed as completed – and for which this possibility cannot be ruled out. We address this concern in two ways.
funds were disbursed – exist in practice. Clearly, the employment Firstly, we check whether pre-determined characteristics of the roads
opportunities and resultant welfare gains associated with PMGSY are (and the villages they serve) exhibit discontinuities at the cut-off in our
contingent on construction actually taking place. We also consider the running variable. The variables we consider are the length of the road,
impact of MLA's interventions on the cost, timeliness, and quality of road whether the project involved the construction of a bridge,32
construction. demographic, socioeconomic, and infrastructure indicators from the
Identifying the causal impact of corruption on road-level outcomes 2001 Demographic Census as well as a set of geographic variables likely
poses several selection problems. 31 As before, we adopt an RD-approach to affect the difficulty of road construction: altitude, ruggedness, forest
that exploits close elections. We drop all roads from the sample that cover and distance from the nearest town. Appendix Table A11 does not
were not built either by a contractor who shares a name with the current indicate systematic differences between the locations of roads built by
MLA, or by a contractor who shares a name with the runner-up in the connected and unconnected contractors, based on their observable
most recent election. Once this sample is restricted to close elections, the characteristics. Two of 30 variables exhibit a statistically significant
latter set of roads can be considered a more appropriate control group, discontinuity: forest cover and bridge construction. 33 We control for
since they were assigned to contractors who are connected to a politi- these variables in subsequent road level regressions. Secondly, we show
cian who could have won. Once again we control for the margin of that all road level results are robust to the inclusion of local geographic
victory (the assignment variable) and its interaction with whether the controls, and that our coefficient of interest remains stable.
candidate of the same name won. The equation for this non-parametric Does political corruption result in more roads ‘going missing’? This
RD is given by: can be evaluated by comparing PMGSY's administrative records to data
from the 2011 Demographic Census. When PMGSY lists a road as having
been completed prior to the census, one would expect the villages on that
31
One way to approach this question empirically would be to run regressions of road
characteristics on a dummy variable that takes the value of one if the MLA and the
32
contractor for road have the same name. However, this approach would fail to control for The length of the road and the requirement for a bridge should have been established
two important sources of unobserved variation. Firstly, contractors who have the same in the planning stage as part of the ‘Core Network’. As such they can be seen as a pre-
name as politicians may have systematically different characteristics from other contrac- determined characteristics rather than an outcome of the contracting process.
33
tors. Secondly, the locations where contractors of the same name as the MLA operate could The coefficient on bridge construction is negative, which is contrary to what one
be systematically different from other areas targeted by PMGSY. would expect if MLAs were allocating harder projects to members of their network.
71
Table 4
Road-level regression discontinuity – Missing Roads.
Dependent Variable: Missing all-weather road Missing road
Margin of victory: <3% Margin of victory: <3%
(1) (2) (3) (4)
Same name as MLA 0.2664*** 0.2273*** 0.1359** 0.1349**

(0.0951) (0.0862) (0.0645) (0.0594)
Margin 2.9287 1.0865 2.1717 2.1586
(4.2918) (4.0746) (2.0146) (1.7037)
Margin*same name as 5.6897 7.9904 1.8635 1.9458
MLA (5.4173) (5.1932) (2.8181) (2.6496)
Bridge 0.1255 0.0027
(0.0984) (0.0363)
Altitude 0.0001 0.0000
(0.0000) (0.0000)
Ruggedness 0.0041 0.0534***
(0.0280) (0.0176)
Forest cover 0.6860** 0.0858
(0.2962) (0.1157)
Power supply in 2001 0.1938*** 0.0631*
(0.0664) (0.0336)
Road-level controls X X X X
State fixed effects X X X X
Agreement year fixed X X X X
effects
N 581 581 581 581
Control group mean dep, 0.2639 0.2639 0.0260 0.0260

var
Note: Standard errors clustered at the contractor level to account for intra-contractor
correlation of the error term at the road level. All regressions include the following set of
road-level controls: ln(length) (to account for non-linear relationship between cost and
distance), whether the constituency is a reserved seat, the mean population of habitations
on the road, the mean population share of Scheduled Castes and Scheduled Tribes of
habitations on the road, and the mean connectivity of those habitations in 2001. To ensure
comparability of coefficients, the sample for columns (1) and (3) is restricted to observa-
tions for which all additional controls are available. Appendix Table A12 presents results
for the respective optimal bandwidths of 2.5% and 4.4% derived from the optimal band-
width choice rule of Imbens and Kalyanaraman (2012). ***p < 0.01, **p < 0.05, *p < 0.1.
road to have all-weather road access according to the census. We define

roads that do not meet this criteria as ‘missing all-weather roads’. By this
measure, around 26% of roads listed as completed prior to the census are
missing.34 Estimating Equation (3), we find that preferential allocation
Fig. 3. Graphical depiction of RD. Note: Lines fitted separately on the samples left and
has a large, statistically significant impact on the likelihood of a road right of the cut-off. Each dot represents the mean for a bin. The number of observations in
going missing (see Table 4 and Fig. 3). This result is robust to the in- a bin varies, based on the density of candidates at a given margin. 90% confidence in-
clusion of additional observable characteristics of the location.35 The tervals plotted in grey. The first panel shows a linear fit within the 3% bandwidth used in
coefficient for the 3% bandwidth implies that this probability increases the RD estimation. The second shows a quadratic fit for a wider margin of 20%.
by 86% when the contractor shares a name with the constituency's MLA.
Applying this estimate to our whole sample in a back-of-the-envelope
calculation, suggests that preferential allocation accounts for 497 addi-
tional missing all-weather roads that would have served around bound macadam road, or a gravel road, according to 2011 census data.
860,000 people.36 This definition yields a much smaller number of missing roads (2.6%)
As a robustness check, the third and fourth columns of Table 4 pro- which is likely to be an underestimate.37 While the coefficient in Table 4
vide the equivalent results for an alternative, more conservative, defi- is correspondingly smaller, the implied effect on the probability of a road
nition of missing roads. Instead of applying PMGSY's stated objective (all- not being constructed is significantly higher: 519%. Performing the same
weather road access), we define roads as missing if none of the villages back-of-the-envelope calculation for this more conservative measure
located on the planned road had either a black-topped road, a water implies that preferential allocation accounted for 87 additional missing
roads with around 150,000 people affected. 38 In short, the finding that
34
There are two reasons why a road could appear as missing, both of which are
37
indicative of corruption. Firstly, roads may be listed as completed without ever being built. Given that some existing roads (gravel roads in particular) would not have met
Secondly, roads could be built with sub-standard materials leading to complete or partial PMGSY's objective of all-weather access, villages that have such roads in the census may
deterioration by the time of the 2011 census. still never have received the PMGSY road they were supposed to. Moreover, it is possible
35
Performing the test proposed by Oster, forthcoming yields a value of δ – the pro- for villages to be on more than one planned PMGSY road (if they are on through roads), so
portionality of selection – of 3.612 when the maximum R2 is set following Oster's proposed the presence of a road in that village need not indicate that all scheduled PMGSY roads
criterion ( Rmax ¼ minf1:3R* ; 1g ). Oster suggests that values above 1 can be typically be were built.
38
considered indicative of robust treatment effects. 4127 roads in our sample were built by connected contractors. Of these 2.6% are
36
4127 roads in our sample were built by connected contractors. Of these 26% are deemed to be missing. Our estimates imply that the share of these missing roads due to
missing all-weather road access. Our estimates imply that the share of these missing roads preferential allocation is 81%. This yields an estimate of 87 roads. The average road in our
due to preferential allocation is 46% (1-1/(1þ0.86), or 497 roads. Multiplying this by the sample serves villages with a total of 1726 inhabitants, giving an estimate of 149,508
average number of inhabitants on a road, gives an estimate of 857,018 people affected. people left unconnected.
72
Table 5
Road-level regression discontinuity – efficiency and quality.
Dependent Variable: Log of (sanctioned cost/km) Days overrun Actual cost/sanctioned Failed inspection
cost
Margin of victory: <3% Margin of victory: <3% Margin of victory: <3% Margin of victory: <3%
(1) (2) (3) (4) (5) (6) (7) (8)
Same name as MLA 0.1163** 0.1027** 88.9 78.1 0.0518 0.0547 0.0881 0.1018
(0.0503) (0.0476) (91.6) (95.1) (0.0528) (0.0520) (0.0975) (0.0973)
Margin 7.6807*** 6.7278*** 472.1 1176.1 1.8696 2.0243 1.9865 1.7221
(2.5758) (2.4074) (4423.1) (4548.7) (2.7133) (2.6636) (4.1406) (4.2063)
Margin*same name as MLA 10.4191*** 9.2084*** 1076.6 1784.4 2.1774 2.3181 7.0371 6.7129
(3.3424) (3.1324) (5546.3) (5551.0) (2.9985) (2.9145) (5.7334) (5.7120)
Bridge 0.0545 32.4632 0.0181 0.0453
(0.3156) (94.5708) (0.0432) (0.2325)
Altitude 0.0000 0.0575 0.0000 0.0000
(0.0000) (0.0554) (0.0000) (0.0001)
Ruggedness 0.0146 10.3473 0.0116 0.0870**
(0.0141) (30.2280) (0.0090) (0.0410)
Forest cover 0.4163*** 398.4231 0.0714 0.0028
(0.1424) (320.2260) (0.1103) (0.3220)
Power supply in 2001 0.0076 22.0734 0.0158 0.0539
(0.0185) (38.9772) (0.0179) (0.0548)
Road-level controls X X X X X X X X
State fixed effects X X X X X X X X
Agreement year fixed effects X X X X X X X X
N 1470 1470 970 970 1161 1161 468 468
Control group mean dep, var 3.325 3.325 538.5 538.5 0.974 0.974 0.373 0.373
Note: Standard errors clustered at the contractor level to account for intra-contractor correlation of the error term at the road level. All regressions include the following set of road-level
controls: ln(length) (to account for non-linear relationship between cost and distance), whether the constituency is a reserved seat, the mean population of habitations on the road, the mean
population share of Scheduled Castes and Scheduled Tribes of habitations on the road, and the mean connectivity of those habitations in 2001. To ensure comparability of coefficients, the
samples for columns (1), (3), (5), and (7) are restricted to observations for which all additional controls are available. Appendix Table A12 reports results for the respective optimal
bandwidths derived from the optimal bandwidth choice rule of Imbens and Kalyanaraman (2012). ***p < 0.01, **p < 0.05, *p < 0.1.
political intervention reduces the number of roads actually constructed is find no significant difference between roads constructed by contractors
robust to widely different definitions of what constitutes a missing road. whose name matches the MLA's and those whose name matches the
Assuming that construction does take place, its efficiency and quality runner-up. Our results on quality inspections are less conclusive. At the
may depend on whether contractors were selected for political reasons. 3% bandwidth the coefficient is positive but not statistically significant.
Using PMGSY's administrative data, we therefore analyse four additional However, at the optimal bandwidth the effect is significant (see Appendix
measures of quality using the same RD approach: (i) the initial cost of the Table A12), which would imply that preferentially allocated contracts are
project (per kilometre), (ii) the number of days between the completion more likely to result in sub-standard construction.
date specified in the contract and the actual date of completion; (iii) the Even though the set of indicators on which we have data is necessarily
ratio between the actual cost of the project and the cost sanctioned in the limited, preferential allocation appears to reflect costly corruption with
agreement; and (iv) a dummy variable for whether a road was deemed no mitigating improvements in the efficiency of road construction.
“unsatisfactory” or “in need of improvement” in either the latest state Indeed, we find suggestive evidence that political intervention leads to
quality inspection or the latest national quality inspection.39 roads that are not only more expensive, but also more likely to be either
If rent-seeking politicians are putting pressure on bureaucrats to of poor quality or never to have been built at all. In principle, these costs
reject the lowest bidder (or the most qualified bidder) in favour of their could be offset if political intervention allowed new firms to enter the
preferred contractor, we would expect to see a rise in costs (or a deteri- market, leading to greater competition and better performance over time.
oration in quality). Table 5 40 shows that roads built by contractors who There is no evidence in the data that preferential allocation dispropor-
share a name with an elected official are more expensive (per kilometre). tionately targets small firms or facilitates entry.42 As such, one inter-
The inclusion of additional geographic controls (column 2) does not pretation consistent with our results is that the benefits of corruption are
significantly alter the coefficient.41 For delays and cost discrepancies we confined to the private rents for politicians (see Fisman et al., 2014),
contractors and potentially, intermediary bureaucrats.
39
The quality data available on the OMMAS has some shortcomings for the purpose of 7. Mechanisms
this analysis. Data is available on national and state quality inspections, and a single road
may have multiple inspections in each category. However, only the grade assigned in the Our findings are consistent with recent work showing that MLAs have
latest inspection is provided (for each category). The data therefore do not allow us to
a significant impact on economic outcomes in their constituency (Asher
distinguish between roads that were satisfactory at the outset, and roads that initially did
not pass inspection but were improved prior to subsequent inspections. Moreover, only a and Novosad, 2017; Prakash et al., 2015), even though the parliamen-
fraction of the roads in our sample appear in the quality data, and many of these only had tarians in question have limited official power. In the literature this
one of the two inspection types (national or state). Pooling the two inspections is not ideal, discrepancy between de jure responsibilities and de facto influence is
but it provides the best available measure of initial road quality. often explained by politicians' informal control over the assignment of
40
The results in Table 5 are reported for the common bandwidth of 3%. Equivalent
results for the 5% and 2.5% bandwidths and the optimal bandwidth derived from the
optimal bandwidth choice rule of Imbens and Kalyanaraman (2012) are reported in
Table A12. There are some outliers in the road level data for cost, delays and cost overruns
42
(see Table 1B). All results are robust to dropping the top 1% and bottom 1%, or the top 5% For contractors with no prior PMGSY experience, the share of contracts where the
and bottom 5% (see Table A13). MLA shared the contractors name is 4.53%. The equivalent share for experienced con-
41
In this case the δ resulting from Oster's, forthcoming test is 1.38, i.e. above the value tractors (with between 10 and 40 prior road contracts) is 4.52%. There is thus no obvious
of 1 indicative of robust treatment effects. evidence that preferential allocation favours small firms or promotes entry.
73
Table 6
Bureaucrat characteristics and contract allocation.
Dependent variable: ΔSharet Margin of victory: <6.2% Margin of victory: <3% Margin of victory: <6.2% Margin of victory: <3%
(1) (2) (3) (4) (5) (6) (7) (8)
Winner 0.0167* 0.0147 0.0249** 0.0275** 0.0227** 0.0204** 0.0304** 0.0345***

(0.0097) (0.0103) (0.0125) (0.0128) (0.0096) (0.0100) (0.0125) (0.0127)
Margin 0.0793 0.0017 0.8301* 0.9857* 0.0694 0.0052 0.7915* 0.9514*
(0.1933) (0.2146) (0.4574) (0.5023) (0.1930) (0.2138) (0.4593) (0.5047)
Margin*winner 0.4108 0.4531 0.3327 0.3723 0.4253 0.4615 0.3087 0.3579
(0.2970) (0.3247) (0.6445) (0.6649) (0.2963) (0.3246) (0.6416) (0.6643)
Same name as DC 0.0512 0.0293 0.1108 0.1037
(0.0569) (0.0542) (0.0902) (0.0949)
Same name as DC*winner 0.1670** 0.1287* 0.1385 0.1404
(0.0701) (0.0667) (0.0890) (0.0966)
DC in promotion year 0.0089 0.0113 0.0233** 0.0262**
(0.0080) (0.0090) (0.0110) (0.0120)
DC in promotion year*winner 0.0313** 0.0318** 0.0342* 0.0441**
(0.0136) (0.0144) (0.0180) (0.0197)
AC controls X X X X
State fixed effects X X X X
Election year fixed effects X X X X
N 7342 6613 4146 3732 7336 6607 4144 3730
Note: Local linear regression estimates. The unit of observation is the constituency-term-DC so each term is split into a separate observation for every DC in office during the term. Standard
errors are clustered at the district-level. Same name as DC is a dummy that takes the value of 1 if the DC and MLA have the same surname. DC in promotion year is a dummy that takes the
value of 1 if the DC is in their 13th or 16th year of service during their tenure in the district. The bandwidth of 6.2% is derived from the optimal bandwidth choice rule of Imbens and
Kalyanaraman (2012). ***p < 0.01, **p < 0.05, *p < 0.1.
bureaucrats (Iyer and Mani, 2012).43 Bureaucrats have many ways to prove that this result is causal.
sway the allocation of contracts. The technical evaluation of bids allows Our second test exploits variation in the incentives which bureaucrats
for discretion. In conversations with bureaucrats involved managing face over their career. IAS officers are screened for promotion in their
PMGSY tenders, we were told that MLAs will often get technical 13th and 16th years of service. The prospect of higher positions and
requirements from bureaucrats and then share that information with increased pay may cause them to improve their performance during these
favoured contractors who then have a better chance of winning the periods. For example, Nath (2016) finds that DCs close to these promo-
tender (Author interviews, April 12, 2016). In this section, we provide tion dates are faster at approving local public good projects. To assess
evidence for role of bureaucrats as intermediaries. In particular, if whether this dynamic applies here, we split each constituency-term into
bureaucrats are part of the corruption chain, we expect that (i) “con- separate observations for each DC who held office during the term. DC's
nected” bureaucrats facilitate corruption; and that (ii) bureaucrats who are reassigned frequently with a median tenure of 17 months. We define
are under more scrutiny are less inclined to facilitate corruption. a DC as being subject to promotion screening if their 13th or 16th year of
PMGSY tendering is generally administered at the district level. The service overlap with their tenure in a district.
highest ranking official in a district is the District Collector (DC) – a The increase in connected contracts is, attenuated during such
member of the Indian Administrative Service (IAS).44 It is likely that periods, as one would expect. Columns (5) to (8) of Table 6 show that
MLAs who have a connection to their district's collector would find it under a DC in their 13th or 16th year of service, preferential allocation
easier to influence the allocation of contracts. Applying our surname- towards contractors who share the MLA's name is much lower than when
matching approach we therefore test for heterogeneity in our main the DC is not in those years and the point estimate is indistinguishable
result based on whether or not the DC and politician have the same from zero.45 This could reflect bureaucrats' reluctance to participate in
surname. We see such an alignment in 1.4% of the bureaucrat-election risky corruption, or greater monitoring of improprieties on the part of
observations. As bureaucrats are reassigned frequently, we split each their subordinates. Both channels suggest that political interference
constituency-term into separate observations for each DC. Columns (1) to involves the cooperation of bureaucrats.
(4) of Table 6 show that the interaction between winning an election and In Table 7 we apply the same sources of variation to test for
sharing the DC's name is positive and significant at the optimal band- heterogeneity in our road-level results. There is no evidence of a
width. The RD estimates lose precision in the smaller sample for the 3% differential effect on the likelihood of a missing road or the cost of the
bandwidth, but the magnitude of the coefficients is comparable. These road for contracts allocated to contractors who share a surname with
results suggest that MLAs may engage in more preferential allocation both the MLA and the DC. However, when the DC is close to their
when the highest ranking bureaucrat is also a member of their network. promotion screening date, our results for missing roads and for cost
However, as the assignment of DCs may not be exogenous we cannot inflation are weaker. This is consistent with the idea that bureaucrats
up for promotion are less amenable to corrupt activities within
their district.
43
The state's chief minister (generally the head of the ruling party) is responsible for
assigning top-level bureaucrats to districts and therefore has the most direct control over
the bureaucracy. Asher and Novosad (2017) find that MLAs aligned with the party in
45
power are able to influence the implementation of regulation in favour of local firms. In Appendix Table A15 shows that this result is robust to controlling for officers' years of
Appendix Table A14 we evaluate whether such MLAs have a greater influence over service and the interaction of years of service with winning the election. Columns (5) to
PMGSY contract allocation, with inconclusive results. The differential effect for aligned (8) compare the effects in the promotion screening year to effects in prior and subsequent
MLAs is consistently positive, but not statistically significant. years. There is a weak negative effect in the year before promotion screening which may
44
IAS officers' executive record sheets provide data on the location and duration of their reflect anticipation of the screening process. The result is thus not due to some linear trend
postings, allowing us to match DCs to districts, and by extension, to the constituencies in in DC behaviour; the attenuation in the effect of winning on the share of connected
our sample. This data is available on the website of the Department of Personnel and contractors appears to be specific to having a DC under the scrutiny that comes with
Training: dopt.gov.in (accessed in December 2016). evaluation for promotion.
74
Table 7
Bureaucrat characteristics and road outcomes.
Dependent Variable: Missing all-weather road Missing road Log of Missing all-weather road Missing road Log of
(sanctioned cost/km) (sanctioned cost/km)
(1) (2) (3) (4) (5) (6)
Same name as MLA 0.2895*** 0.1504** 0.1126** 0.3885*** 0.1676** 0.1479***

(0.0978) (0.0649) (0.0467) (0.1068) (0.0779) (0.0569)
Margin 3.8176 2.5438 7.3982*** 7.4694 1.8026 7.8830***
(4.1530) (1.8993) (2.4709) (5.4338) (2.6045) (2.8675)
Margin*same name as MLA 4.2841 1.3217 10.0164*** 1.9472 2.7491 10.9856***
(5.2783) (2.7645) (3.2321) (6.5345) (3.3937) (3.6501)
Same name as DC 0.1144 0.0480 0.0692
(0.0805) (0.0371) (0.0463)
Same name as DC*winner 0.0794 0.0506 0.0148
(0.0961) (0.0457) (0.0587)
DC in promotion year 0.0262 0.0181 0.0261
(0.0862) (0.0253) (0.0393)
DC in promotion year*winner 0.0911 0.0990** 0.1303**
(0.1160) (0.0472) (0.0603)
Road-level controls X X X X X X
State fixed effects X X X X X X
Agreement year fixed effects X X X X X X
N 581 581 1471 459 459 1057
Note: Local linear regression estimates. Standard errors clustered at the contractor level to account for intra-contractor correlation of the error term at the road level. All regressions include
the following set of road-level controls: ln(length), whether the constituency is a reserved seat, the mean population of habitations on the road, the mean population share of Scheduled
Castes and Scheduled Tribes of habitations on the road, and the mean connectivity of those habitations in 2001. Same name as DC is a dummy that takes the value of 1 if the DC and MLA have
the same surname. DC in promotion year is a dummy that takes the value of 1 if the DC is in their 13th or 16th year of service during their tenure in the district. The bandwidth of 6.2% is
derived from the optimal bandwidth choice rule of Imbens and Kalyanaraman (2012). ***p < 0.01, **p < 0.05, *p < 0.1.
The results of this section provide support for the hypothesis that explanations: vote-buying and particularised trust. In this section we
local bureaucrats are a crucial part of the chain that links politicians to attempt to shed light on which is more applicable to corruption in PMGSY.
the preferential allocation of contracts. If road contracts are awarded in exchange for political contributions
or political support, one could expect the bias towards connected con-
8. Electoral incentives for corruption tractors to increase in election periods. To test for this we construct more
disaggregated measures of proximity: the share of contractors with a
We implicitly assume that kinship ties to politicians are relevant candidate's name in the first 12 months after an election
connections in the structure of local political corruption in India, and our (start of termijt .), the equivalent share for the last 12 months before the
results appear to validate this assumption. But why should patronage be subsequent election (end of termijt ), and finally the share for the inter-
targeted along caste or familial lines? The literature offers two main mediate, mid-term, period. The first three columns of Table 8 shows the
Table 8
Three tests for electioneering.
Dependent variable: ΔSharet in cols 1–3 and Electoral cycles: allocation Electoral cycles: cost Effect in
7 Log of (sanctioned cost/km) in cols 4-6 “politically
Start of term End of term Start and end of Start of term End of term Start and end of
irrelevant” areas
heterogeneity heterogeneity term heterogeneity heterogeneity term
heterogeneity heterogeneity
(1) (2) (3) (4) (5) (6) (7)
Winner (Same name as MLA in cols 4–6) 0.0328** 0.0350** 0.0344** 0.1166** 0.1359*** 0.1433*** 0.0250*
(0.0155) (0.0152) (0.0160) (0.0518) (0.0525) (0.0540) (0.0146)
Margin 1.1660* 1.1466 1.1591* 7.0964*** 7.6324*** 7.0332*** 0.2272
(0.6971) (0.6969) (0.6971) (2.5415) (2.6163) (2.5659) (0.5518)
Margin*winner (Margin*same name as 0.8036 0.7620 0.7868 9.8967*** 10.3096*** 9.7854*** 0.0918
MLA in cols 4–6) (0.7374) (0.7388) (0.7390) (3.2878) (3.3607) (3.2915) (0.7128)
Start of term 0.0091 0.0119 0.1270*** 0.1420***
(0.0083) (0.0086) (0.0422) (0.0444)
Start of term*winner (Start of term*same 0.0029 0.0013 0.0746 0.0973*
name as MLA in cols 4–6) (0.0111) (0.0112) (0.0546) (0.0552)
End of term 0.0040 0.0089 0.0521 0.0796
(0.0074) (0.0075) (0.0568) (0.0581)
End of term*winner (End of term*same 0.0050 0.0044 0.1303** 0.1480***
name as MLA in cols 4–6) (0.0114) (0.0114) (0.0551) (0.0553)
Politically irrelevant*post 0.0273
announcement*winner (0.0218)
Constituency controls X X X X X X X
Candidate controls X X X X
Road-level controls X X X
State and agreement year fixed effects X X X X X X X
N 3201 3201 3201 1470 1470 1470 2707
Note: Terms in the sample for columns 1–3 are disaggregated by time: first year, last year, and remaining period. Terms in the sample for column 7 are disaggregated by time (pre- and post-
the announcement of delimitation) and spatially (based on intra-constituency boundaries introduced by delimitation). The RD bandwidth for all columns is 3%. Appendix Tables A16-18
reports the equivalent results for the respective optimal bandwidths derives from the optimal bandwidth rule of Imbens and Kalyanaraman (2012). Standard errors are clustered at the
election level in columns 1–3 and 7 and at the contractor level in columns 4–6. Controls for columns 1–3 and 7 are the same as in Table 3. Controls for columns 4–6 are the same as in Tables 4
and 5. Regression in column 7 includes all lower-order interactions (not reported). All regressions include a constant. ***p < 0.01, **p < 0.05, *p < 0.1.
75
results of applying our main estimation approach to this disaggregated that we find no variation based on political incentives is further
sample and interacting dummies for start of termijt and end of termijt evidence to suggest that corrupt contract allocations are not motivated
with winnerijt . The overall effect of winning the election is comparable to by vote-buying.
the term-level results, and we find no differential effects in election years. In the absence of clear evidence for vote buying, it is possible that
Although the bias towards connected contractors does not increase in corruption arises within kinship networks because these provide the
election periods, there could be different patterns for the within-term “particularised trust” needed to engage in risky collusive behaviour
variation on the cost margin. Politicians might need to extract rents, (Tonoyan, 2003). While we are unable to test this explanation explicitly,
buy support, or reward supporters more before or after elections. it fits the context of PMGSY in that the involvement of the central
Including the interactions between same name as MLAnjt and government provides a minimum level of monitoring. Given this
start of termijt and end of termijt in the road-level regressions, we again supervision, it is notable that we only find evidence of corruption in
find no evidence of a political cycle in which election periods see certain aspects of the programme. Our results suggest that preferential
increased corruption (columns 4–6 of Table 8). The observed negative allocation adversely affects every performance marker except the two
effect for both the start and end of term is more consistent with increased that are most easily observed in the central administrative data:
scrutiny in the run-up to elections acting as a deterrent to corruption. over-runs and delays. This is in line with the idea that contractors face a
Changes to the delimitation of parliamentary constituencies allow for trade-off between potential rents and the cost of detection. Similarly, we
an additional test of the vote-buying hypothesis. The changes proposed find that preferentially allocated contracts are less expensive in the
by the delimitation commission of 2002 were approved in February period before elections, when scrutiny may be higher and detection is
2008. Subsequent assembly elections, starting with Karnataka in May more damaging to politicians. The overall pattern of our results is
2008, were carried out under the new delimitation. After the reform had therefore consistent with a setting in which politicians, bureaucrats, and
been announced and approved, the majority of MLAs elected under the contractors are constrained by monitoring and operate on the least risky
old delimitation continued to hold office for several years until the next margins and with the least risky collaborators – the members of their
election. In constituencies where the boundaries were redrawn, this family or caste network.
meant that only some areas would remain part of the constituency at the The introduction of e-procurement for PMGSY also allows us to
next election, while others would be of no consequence to the MLA's evaluate the effect of an intervention that increased the monitoring
chances of re-election. We identify such areas with a dummy variable capacity of higher level officials. Lewis-Faupel et al. (2016) find that
politically irrelevantijt and also disaggregate temporally, splitting the e-procurement improved the quality of PMGSY road construction. They
applicable electoral term into the period before the announcement, and identify the entry of higher quality contractors from outside regions as a
the period between February 2008 and the next election (the variable key mechanism behind the quality improvement. However, we find no
post announcementijt denotes the latter). Given that the boundaries were evidence that e-procurement has helped to prevent the specific form of
defined by an independent commission following objective pre-set corruption we document in this paper.48 This null result is consistent
guidelines, the reform could provide plausibly exogenous variation in with corrupt practices being relatively sophisticated in the context of
the incentive for vote-buying.46 The final column of Table 8 presents the PMGSY, even before e-tendering.
results of our main specification for the disaggregated sample and A distinguishing feature of PMGSY is that the programme has been
interaction terms. The coefficient of interest is the triple interaction term: subject to a central monitoring system from its inception. However, by
explicitly prohibiting political involvement, it foregoes a mechanism that
winnerijt *politically irrelevantijt *post announcementijt could provide local accountability. If voters held their MLAs responsible
for the services delivered under PMGSY, the latter would have an
A negative and significant coefficient would suggest that political
incentive to limit corruption. By contrast, a scheme in which local poli-
corruption is weaker in areas where politicians have no incentive to buy
ticians have no formal role but over which they still retain influence
votes. The coefficient on the triple interaction is negative at the 3%
through informal channels, can be seen as an ideal vehicle for rent-
bandwidth, positive at the optimal bandwidth (see Appendix Table A18),
seeking. Our results provide mixed evidence on electoral account-
and statistically insignificant throughout. Hence, we find little evidence
ability. While cost overruns are weaker around elections, the preferential
of strategic vote-buying. This result is also consistent with recent work by
allocation does not show any electoral cycle. One interpretation of these
Chhibber and Jensenius (2016), who use the same delimitation experi-
findings is that PMGSY limited political control in the allocation of
ment and find that MLAs from “ethnic” or “well-organized” parties tend
contracts, but that this came at the expense of reduced political
to target existing loyalists rather than the electorally valuable voters who
accountability. While we cannot test this trade-off directly in our context,
joined their constituencies post-delimitation.
we believe it is an interesting avenue for future research.
In two final tests for the role of electoral incentives, we evaluate
whether preferential allocation is more likely when (a) the road in
question targets politically valuable villages or (b) when the party of 9. Conclusion
the relevant MLA has only tenuous control over parliament.47 In
Appendix Tables A19 and A20 we find no evidence that the likelihood This paper provides direct empirical evidence that local politicians in
of MLA interference depends on the political significance of destination India abuse their power to benefit members of their own network. We
villages (as proxied by having the Panchayat headquarters, which may exploit the variation in political leadership due to the electoral cycle, to
make them more valuable, or majority SC/ST population, which can identify systematic distortions in the allocation of contracts for a major
make them less so, particularly in unreserved constituencies) or that rural road construction programme (PMGSY). By matching contractors'
aligned MLAs engage in less corruption when their party's control of and political candidates' surnames, we generate a measure of proximity
parliament is more precarious (as measured by its seat share). The fact which evolves as the pool of contractors changes. A regression discon-
tinuity design based on close elections, suggests that the causal impact of
a politician coming to power is an 83% increase in the share of roads
allocated to contractors who share their surname.
46
According to the Electoral Commission of India's Guidelines and Methodology for This result withstands a series of alternative specifications and
Delimitation, “the delimitation of the constituencies in a district shall be done starting robustness checks. Further regression discontinuity estimates at the road
from North to North-West and then proceeding in a zig-zag manner to end at the Southern
side.” Constituencies were to have equal populations, as far as possible, with maximum
level, indicate that corruption is welfare-reducing in this context.
deviations of 10% from the State average, based on the 2001 Census.
47
Which Bohlken (2016) suggests would create incentives at the party level to curb
48
corruption. These results are reported in Appendix (Table A21).
76
Political interference is associated with higher costs and a greater like- References
lihood that roads go missing. While constituencies with close elections
are clearly in a different political equilibrium (close elections are con- Acemoglu, Daron, Hassan, Tarek, Tahoun, Ahmed, 2014. The Power of the Streets:
Evidence from Egypt's Arab Spring. Mimeo MIT.
tested by more candidates, for example), a difference-in-differences es- Aggarwal, Shilpa, 2017. Do rural roads create pathways out of poverty? Evidence from
timate for the whole sample also shows a positive and statistically India. Mimeo.
significant increase in the share of connected contacts, albeit a smaller Amore, Mario Daniele, Bennedsen, Morten, 2013. The value of local political connections
in a low-corruption environment. J. Financ. Econ. 110 (2), 387–402.
one than in the RD specification. This finding suggests that the LATE we Angelucci, Manuela, De Giorgi, Giacomo, Rangel, Marcos A., Rasul, Imran, 2010. Family
estimate in the RD has external validity. networks and school enrolment: evidence from a randomized social experiment.
A distinguishing feature of our analysis, is that we identify the effect J. Public Econ. 94 (3), 197–221.
Asher, Sam, Novosad, Paul, 2016. Market Access and Structural Transformation: Evidence
of political connections to state-level legislators who have no official from Rural Roads in India. Manuscript: Department of Economics, University of
involvement in the road construction programme. Our results therefore Oxford.
not only indicate preferential treatment of the politically connected, they Asher, Sam, Novosad, Paul, 2017. Politics and local economic growth: evidence from
India. Am. Econ. J. Appl. Econ. 9 (1), 229–273.
also provide indirect evidence that local politicians' power over pur-
Banerjee, Abhijit, Mullainathan, Sendhil, Hanna, Rema, 2013. Corruption. In: Handbook
portedly neutral bureaucrats is sufficient to coerce them into corruption. of Organizational Economics. Princeton University Press.
From a policy perspective, these findings indicate that more could be Banerjee, Abhijit, Green, Donald P., McManus, Jeffery, Pande, Rohini, 2014. Are poor
done to insulate the officials implementing government programmes at voters indifferent to whether elected leaders are criminal or corrupt? A vignette
experiment in rural India. Polit. Commun. 31 (3), 391–407.
the local level, including those involved in PMGSY. The result that Becker, Gary S., Stigler, George J., 1974. Law enforcement, malfeasance, and
misallocation is lower and contract performance is better for bureaucrats compensation of enforcers. J. Leg. Stud. 3 (1), 1–18.
who face promotion incentives, points in this direction. Bertrand, Marianne, Djankov, Simeon, Hanna, Rema, Mullainathan, Sendhil, 2007.
Obtaining a driver's license in India: an experimental approach to studying
A striking feature of PMGSY is that the bidding rules were designed corruption. Q. J. Econ. 122 (4), 1639–1676.
explicitly to limit political influence. However, such anti-corruption Bobonis, Gustavo J., Fuertes, Luis Camara, Schwabe, Raier, 2016. Monitoring corruptible
measures could face a trade-off between political accountability and politicians. Am. Econ. Rev. 106 (8), 2317–2405.
Bohlken, Anjali Thomas, 2016. Development or Rent-seeking: How Political Influence
political control. If voters believe the limitations work, then a logical Shapes Infrastructure Provision in India. Mimeo.
conclusion would be that their MLAs are not responsible for poor pro- Bussell, Jennifer, 2015. Clients or Constituents? Distribution between the Votes in India.
gramme performance. While we find some evidence that cost inflation is Mimeo, University of California Berkeley.
Caughey, Devin, Sekhon, Jasjeet S., 2011. Elections and the regression discontinuity
weaker around election times, our preferential allocation results remain design: lessons from close us house races, 1942–2008. Polit. Anal. 19 (4), 385–408.
very stable throughout the electoral cycle. These findings are consistent Chandra, Kanchan, 2004. Why ethnic parties suceed. In: Patronage and Ethnic Head
with the idea that the limited control of politicians over PMGSY could Counts in India.
Chhibber, Pradeep, Jensenius, Francesca R., 2016. Privileging One's Own? Voting
have come at the expense of reduced accountability to voters.
Patterns and Politicized Spending in India. Mimeo.
While this paper is primarily about the measurement and mecha- Chopra, Vir K., 1996. Marginal Players in Marginal Assemblies: the Indian MLA. Orient
nisms of corruption, its findings have significance beyond the potential Longman.
number of misallocated roads or the amount of misdirected funds. If Cingano, Federico, Pinotti, Paolo, 2013. Politicians at work: the private returns and social
costs of policial connections. J. Eur. Econ. Assoc. 11 (2), p.433–465.
corrupt arrangements were made based on random matching between Do, Quoc-Anh, Teik Lee, Yen, Nguyen, Bang Dang, 2015. Political Connections and Firm
individuals, the empirical strategy would have revealed nothing. Our Value: Evidence from the Regression Discontinuity Design of Close Gubernatorial
results provide further evidence for the role of networks in enabling Elections. CEPR Discussion Paper No. 10526.
Eggers, Andrew C., Hainmueller, Jens, 2009. MPs for sale: returns to office in post-war
corruption and point towards theories in which kinship networks British politics. Am. Political Sci. Rev. 103 (4), 1–21.
facilitate corruption through trust or the ability to impose social sanc- Eggers, Andrew C., Fowler, Anthony, Hainmueller, Jens, Hall, Andrew B., Snyder, James
tions. The irony is, that the setting for the analysis – PMGSY – is M., 2015. On the validity of the regression discontinuity design for estimating
electoral effects: new evidence from over 40,000 close races. Am. J. Political Sci. 59
conceptually a profoundly inclusive programme, facilitating the inte- (1), 259–274.
gration of over 100 million people into the Indian economy (Aggarwal, Fafchamps, Marcel, Labonne, Julien, 2017. Do politicians' relatives get better jobs?
2017). This paper suggests that allowing them to compete equally for Evidence from Municipal Elections. J. Law, Econ. Organ. 33 (2), 268–300.
Ferraz, Claudio, Finan, Federico, 2008. Exposing corrupt politicians: the effects of
jobs, permits, licenses, or government procurement contracts, may Brazilians publicly released audits on electoral outcomes. Q. J. Econ. 123 (3),
require building more than roads. 703–745.
Ferraz, Claudio, Finan, Frederico, 2011. Electoral accountability and corruption: evidence
from the audits of local government. Am. Econ. Rev. 101 (4), 1274–1311.
Acknowledgements
Field, Erica, Levinson, Matthew, Pande, Rohini, Visaria, Sujata, 2008. "Segregation, rent
control, and riots: the economics of religious conflict in an Indian city. Am. Econ. Rev.
We would like to thank the International Growth Centre (IGC) for 98 (2), 505–510.
Fisman, Raymond, 2001. Estimating the value of political connections. Am. Econ. Rev. 91
supporting the data collection of this project through the grants
(4), 1095–1102.
“Building Bharat Nirman” and “Bidding for Roads”. We would also like Fisman, Raymond, Wang, Yongxiang, 2015. The mortality cost of political connections.
to thank CEPREMAP for supporting additional data collection. We are Rev. Econ. Stud. 83 (No. 4), 1346–1382.
grateful to Nawal Agrarwal, Ashish Modi, Anukriti Ranjan, Shrenik Fisman, Raymond, Schulz, Florian, Vig, Vikrant, 2014. The private returns to public
office. J. Political Econ. 122 (4), 806–862.
Sanghvi, Radha Sarkar, and Paolo Santini for excellent research assis- Fisman, Raymond, Paravisini, Daniel, Vig, Vikrant, 2017. Cultural proximity and loan
tance. This paper has benefited from discussions with Sam Asher, Anjali outcomes. Am. Econ. Rev. 107 (2), 457–492.
Bohlken, Tarek Ghani, Guy Grossman, Dan Keniston, Ariane Lambert- Golden, Miriam A., Picci, Lucio, 2005. Proposal for a new measure of corruption,
illustrated with Italian data. Econ. Polit. 17 (1), 37–75.
Mogilianski, Karen Macours, Paul Novosad, Alexander Plekhanov, Gulzar, Saad, Pasquale, Ben, 2016. Politicians, bureaucrats, and development: evidence
Akiko Suwa-Eisenmann, Liam Wren-Lewis, and Maiting Zhuang, as well from India. Am. Political Sci. Rev. 111 (1), 162–183.
as excellent feedback from our anonymous referees and the editor. We Hoff, Karla, Pandey, Priyanka, 2004. Belief System and Durable Inequalities: an
Experimental Investigation of Indian Caste. World Bank, Policy Research Paper: No.
also thank seminar participants in PSE's IRG and India-China Work- 3351.
shops, the Namur Indian Political Economy Workshop, the Graduate Horowitz, Donald L., 1985. Ethnic Groups in Conflict. Univ of California Press.
Institute (Geneva), and the Sussex Development Workshop. All Huntington, S., 1968. Modernisation and Corruption ‘in Political Order in Changing
Societies. Yale University Press, New Havern.
remaining errors are our own.
Imbens, Guido, Kalyanaraman, Karthik, 2012. Optimal bandwidth choice for the
regression discontinuity estimator. Rev. Econ. Stud. 79 (3), 933–959.
Appendix A. Supplementary data Iyer, Lakshmi, Mani, Anandi, 2012. Traveling agents: political change and bureaucratic
turnover in India. Rev. Econ. Stat. 94 (3), 723–739.
Jensenius, Francesca Refsum, 2013. Power, Performance and Bias: Evaluating the
Supplementary data related to this article can be found at https://doi. Electoral Quotas for Scheduled Castes in India. PhD Thesis. University of California
org/10.1016/j.jdeveco.2017.10.009. Berkeley.
77
Khwaja, Asim Ijaz, Mian, Atif, 2005. Do lenders favor politically connected Firms? Rent NRRDA, 2005. Pradhan Mantri Gram Sadak Yojana Operations Manual.
provision in and emerging financial market. Q. J. Econ. 120 (4), 1371–1411. NRRDA, 2015. Standard Bidding Document for PMGSY.
Krueger, Anne O., 1974. The political economy of the rent-seeking society. Am. Econ. Olken, Benjamin A., 2007. Monitoring corruption: evidence from a Field experiment in
Rev. 64 (3), 291–303. Indonesia. J. Political Econ. 115 (2), 200–249.
Lambsdorff, Johann Graf, 2002. Corruption and rent-seeking. Public Choice 113 (1–2), Olken, Benjamin A., Barron, Patrick, 2009. The simple economics of extortion: evidence
97–125. from trucking in aceh. J. Political Econ. 117 (3), 417–452.
Larreguy, Horacio A., Marshall, John, Snyder Jr., James M., 2014. Revealing Malfeasance: Emily Oster, Unobservable selection and coefficient stability: theory and evidence. J. Bus.
How Local Media Facilitates Electoral Sanctioning of Mayors in Mexico. No. W20697. Econ. Stat., forthcoming
National Bureau of Economic Research. Prakash, Nishith, Rockmore, Marc, Uppal, Yogesh, 2015. Do Criminally Accused
Lee, David S., 2008. Randomized experiments from non-random selection in US House Politicians Affect Economic Outcomes? Evidence from India. Mimeo.
elections. J. Econ. 142 (2), 675–697. Reinikka, Ritva, Svensson, Jakob, 2004. Local capture: evidence from a central
Lee, David S., Lemieux, Thomas, 2010. Regression discontinuity designs in economics. government transfer program in Uganda. Q. J. Econ. 3 (2–3), 679–705.
J. Econ. Lit. 48, 281–355. Rose-Ackerman, Susan, 1975. The economics of corruption. J. Public Econ. 4 (2),
Leff, Nathaniel H., 1964. Economic development through bureaucratic corruption. Am. 187–203.
Behav. Sci. 8 (3), 8–14. Shleifer, Andrei, Vishny, Robert W., 1993. Corruption. Q. J. Econ. 108 (3), 599–617.
Lewis-Faupel, Sean, Neggers, Yusuf, Olken, Benjamin A., Pande, Rohini, 2016. Can Sukhtankar, Sandip, 2012. Sweetening the deal? Political connections and sugar mills in
electronic procurement improve infrastructure Provision? Evidence from public India. Am. Econ. J. Appl. Econ. 4 (3), 43–63.
works in India and Indonesia. Am. Econ. J. Econ. Policy 8 (3), 258–283. Sukhtankar, Sandeep, 2015. The impact of corruption on consumer markets: evidence
Lui, Francis T., 1985. An equilibrium queuing model of bribery. J. Political Econ. 93 (4), from the allocation of second-generation wireless spectrum in India. J. Law Econ. 58
760–781. (1), 75–109.
Mastrobuoni, Giovanni, Patacchini, Eleonora, 2012. Organized crime networks: an Tonoyan, Vartuhí, 2003. Corruption and entrepreneurship: impact of trust and civic
application of network analysis techniques to the American mafia. Rev. Netw. Econ. society; a cross cultural comparison of 21 countries. In: Paper Presented at the
11 (3), 1–43. International Conference on" Trust and Entrepreneurial Behavior in East and West
McCrary, Justin, 2008. Manipulation of the running variable in the regression European Economies: Concepts, Developments and Comparative Aspects. University
discontinuity design: a density test. J. Econ. 142 (2), 698–714. of Bremen 2003 ifm.
Melo, Marcus Andre, Pereira, Carlos, Figueiredo, Carlos Mauricio, 2009. Political and Truex, R., 2014. The returns to office in a ‘rubber stamp’ parliament. Am. Political Sci.
institutional checks on corruption: explaining the performance of Brazilian audit Rev. 108 (2), 235–251.
institutions. Comp. Polit. Stud. 42 (9), 1217–1244. Vissa, Balagopal, 2011. A matching theory of entrepreneurs' tie formation intentions and
Mironov, Maxim, Zhuravskaya, Ekaterina, 2016. Corruption in procurement and the initiation of economic exchange. Acad. Manag. J. 54 (1), 137–158.
political cycle in tunneling: evidence from financial transactions data. Am. Econ. J. World Bank, 2014. PMGSY Rural Roads Project : P124639-implementation Status Results
Econ. Policy 8 (2), 287–321. Report.
Nath, Anusha, 2016. Bureaucrats and Politicians: How Does Electoral Competition Affect
Bureaucratic Performance? Institute for Economic Development (IED) Working Paper
269.
78

Building Connections Political Corruption and Road Construction in India PDF

Uploaded by

Copyright:

Available Formats

Building Connections Political Corruption and Road Construction in India PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Building Connections Political Corruption and Road Construction in India PDF

Uploaded by

Copyright:

Available Formats

Journal of Development Economics 131 (2018) 62–78

Contents lists available at ScienceDirect

Journal of Development Economics

Building connections: Political corruption and road construction in India

1. Introduction extrapolating from a LATE) would imply that state-level parliamentarians

Variable Observations Mean Std. Dev. Min Max

Panel A: Roads allocated to contractors of the same name

Variable Observations Mean Std. Dev. Min Max

Panel A: Road construction

Margin of victory: 6.2% 3%

Obs. Winner St. Error Obs. Winner St. Error

Panel A: Candidate characteristics

We estimate Equation (1) in a non-parametric RD for a range of 5. Reallocation result

(1) (2) (3) (4) (5) (6)

Winner 0.0094* 0.0099* 0.0252*** 0.0242** 0.0319** 0.0323**

results as causal, one possible interpretation of this discrepancy is that

“There is a huge scam in tender allotment in Pradhan Mantri Gram

Our analysis suggests that episodes of suspected favouritism in

5.3. Validity of the RD approach

The RD design requires that no variables other than the dependent

Dependent Variable: Missing all-weather road Missing road

Margin of victory: <3% Margin of victory: <3%

(1) (2) (3) (4)

Same name as MLA 0.2664*** 0.2273*** 0.1359** 0.1349**

Control group mean dep, 0.2639 0.2639 0.0260 0.0260

road to have all-weather road access according to the census. We deﬁne

(1) (2) (3) (4) (5) (6) (7) (8)

(1) (2) (3) (4) (5) (6) (7) (8)

Winner 0.0167* 0.0147 0.0249** 0.0275** 0.0227** 0.0204** 0.0304** 0.0345***

(1) (2) (3) (4) (5) (6)

Same name as MLA 0.2895*** 0.1504** 0.1126** 0.3885*** 0.1676** 0.1479***

(1) (2) (3) (4) (5) (6) (7)

You might also like

Winner 0.0094* 0.0099* 0.0252* 0.0242 0.0319 0.0323

Same name as MLA 0.2664* 0.2273* 0.1359 0.1349

Winner 0.0167* 0.0147 0.0249 0.0275 0.0227 0.0204 0.0304 0.0345*

Same name as MLA 0.2895* 0.1504 0.1126 0.3885* 0.1676 0.1479*