Political Organizations and Political Scope Research Part 3

32 Table 6 Payment is Similar for Winning and Losing Candidates Dry Season Full Year (1) Winner (2) Loser (3) Winner (4) Loser (5) Winner (6) Loser (7) Winner (8) Loser RD Estimate 1.764∗∗ (0.757) 1.778∗∗ (0.778) 1.822∗∗∗ (0.692) 1.798∗∗ (0.729) 3.726∗ (1.957) 3.739∗∗ (1.777) 4.098∗∗ (1.750) 4.147∗∗∗ (1.598) Obs in BW Clusters in BW Bandwidth Robust p-val Metric District FEs Constituency FEs 3690 1230 0.590 0.033 1-Norm 3276 1092 0.427 0.053 1-Norm 3690 1230 0.590 0.353 1-Norm X X 3276 1092 0.427 0.032 1-Norm X X 3474 1158 0.451 0.069 1-Norm 3471 1157 0.547 0.064 1-Norm 3474 1158 0.451 0.326 1-Norm X X 3471 1157 0.547 0.020 1-Norm X X Note: We estimate Equation 9 for average payments to households of winning AITC candidates and losing AITC candidates. “Metric” gives the distance metric (to AITC absolute majority) used as the running variable. Bandwidths are calculated using the Calonico et al. (2014) method. Standard errors are clustered within panchayat. *p=0.10 **p=0.05 ***p=0.01 6.3 Candidates Who Muster More Votes Receive Bigger Post-Election Payouts Fixed payments would be sufficient to sustain the cooperation of local officials, but can they induce effort? If state officials cannot observe some key features of the local official’s electioneering— whether they are targeting jobs to the right voters, for example—a fixed payment may no longer be optimal. Organization theory predicts the optimal contract will include some incentive pay (Hölmstrom, 1979). Recent anecdotal and historical evidence suggests party bosses may behave similarly.16 We test this hypothesis by estimating the difference-in-discontinuities (Equation 10) on the village-level average payments to AITC candidates in the 4 weeks immediately after the 2014 election. We interact the discontinuity with the AITC vote share in the village. Column 1 of Table 7 shows that, after controlling for panchayat fixed effects, AITC candidates receive an extra 1 day of labor for every additional 6.6 percentage points returned by their village for the AITC’s national candidate.17 Column 2 shows that the 2014 vote share is not simply a proxy for whether the vil16 For example, Novaes (2018) describes how parties in Brazil maintain spreadsheets that record the vote returns in each local candidate’s territory alongside the payments they will receive. A recent analysis of a similar set of accounts maintained by a 1950s-era Brazilian congressman shows a positive correlation between payments and 17 the number of votes delivered in excess of prior performance (Gingerich, 2020). 6.6 = 1/15.198 × 100 33 lage elected an AITC candidate. We simultaneously control for the difference-in-discontinuity between villages that do and do not have at least one “Winning AITC Cand.” by estimating Equation 11. Our coefficient of interest is unchanged. Table 7 Candidates Whose Villages Return More Votes Receive More Pay in the 4 Weeks After the 2014 Polling Date AITC Majority X 2014 Vote Share (1) 2014 (2) 2014 (3) 2014 (4) 2014 (5) 2014 (6) 2014 (7) 2014 (8) 2015 (Placebo) 15.198∗∗∗ (4.723) 14.685∗∗∗ (4.848) 0.248 (0.764) 14.537∗∗∗ (5.020) 0.318 (0.891) -0.557 (4.452) 6.905 (6.100) 0.876 (0.892) 10.128∗ (6.090) 0.849 (0.887) 14.490∗∗∗ (5.190) 0.215 (0.774) 12.465∗∗∗ (4.815) 0.100 (0.767) -0.149 (3.408) 0.007 (5.704) -0.760 (5.059) 0.250 (0.242) 3590 915 0.498 X 3521 898 0.508 X X Winning AITC Cand. X 2013 AITC Share X 2011 AITC Share 8.022 (5.961) X 2009 AITC Share 0.638 (5.605) X Avg. Past Share X Non-Cand. Days Villages in BW Panchayats in BW Bandwidth Panchayat FEs 3523 898 0.471 X 3552 905 0.483 X 3602 920 0.509 X 2774 787 0.606 X 2777 787 0.592 X 3660 936 0.552 X Note: Estimates are from within-panchayat regressions where the unit of observation is a village. The outcome is the average NREGA payments to all AITC candidates in the village in the 4 weeks after the date of polling in the 2014 election. The key regressor of interest is the village’s 2014 AITC vote share. Columns 1—4 are OLS regressions restricted to panchayats within 0.1 of the cutoff based on the 1-norm. Column 5 is a difference-in-discontinuities regression where the bandwidth is chosen by applying the method of Calonico et al. (2014) to a basic RDD framework with village-level observations. Standard errors are clustered by panchayat. See text for description of each specification. *p=0.10 **p=0.05 ***p=0.01 Columns 3—6 add similar controls for the average vote share of AITC candidates in the 2013 local election, the vote share in the 2011 state election, the 2009 national election, and the average across all three elections. The results are qualitatively similar, though the estimate in Column 4 is somewhat smaller and it loses significance. That may be driven by the reduction in sample size (as we are unable to link some 2011 polling stations to a village, and vice-versa). It may also suggest the payment is partly backward-looking, as might be true if the candidate is paid in part for the effort of maintaining an existing bloc of voters. 34 Could these patterns arise because the entire village is rewarded for favoring the AITC, and some of the largess passes through to local candidates? One very basic test for this hypothesis is to add a difference-in-discontinuity control for the average days of labor awarded to noncandidates.18 Column 7 shows that the coefficient of interest remains largely unchanged. Are these correlations actually related to the election, or could they reflect some other characteristic of the village or the candidates who live there? One very basic placebo test is to estimate (10) on the average NREGA allocations to AITC candidates during the same 4 calendar weeks in 2015, when there was no election. Column 8 shows that there is no similar effect on 2015 allocations, supporting the idea that the payments are linked specifically to the 2014 election. 7 7.1 Horizontal Expansion: Recruiting Majorities in Local Councils Candidates Systematically Switch to the Ruling Party If control of state and local government are complements, the state ruling party has a strong incentive to take control of more local councils. The party would gain the power to target and claim credit for NREGA jobs. It could of course win control of these councils in the next local election. But a quicker approach is to recruit members of the opposition. Complementary control also implies that the ruling party has more to offer an elected local councilor than the opposition through its control of NREGA projects. Is there evidence that opposition candidates defect to the ruling party? There is no official record of whether a local candidate has changed her allegiance. Instead we link candidates for the 2013 local election by name and panchayat to candidates in the subsequent 2018 local election. Within the subset of candidates who stood for office in both years, we infer that a candidate has “switched” allegiance if they register under a different party. For example, if we observe Abhishek Chatterjee contesting in 2013 as a member of the Communist Party-Marxist (CPIM) but in 2018 as a member of the AITC, we infer he has switched from the CPIM to the AITC. Likewise, we infer that a candidate has “stayed” with their party if they register 18 To be precise we now include both the average vote share across past elections and the average non-candidate allocation to the vector X. 35 under the same party in both elections. Figure 3 Candidates are Far More Likely to Leave Opposition Parties than the Ruling Party Conditional on switching, fraction that defects to... Fraction who switch Other Fraction who stay INC .4 .6 .8 1 Fraction who switch parties CPIM AITC 0 .2 BJP AITC BJP CPIM Party in 2013 INC AITC BJP CPIM INC Party in 2013 Note: Among the set of candidates standing for election in both 2013 and 2018, the left-hand panel shows the fraction who “stay” registered with the same party in 2018 as in 2013 versus those who “switch” their 2013 party to register with a different party (or as an independent) in 2018. Each fraction is calculated among candidates registered in 2013 with the ruling party (AITC), the Bharatya Janata Party (BJP), the Communist Party-Marxist (CPIM), and the Indian National Congress (INC). The right-hand panel shows which parties candidates are switching into (conditional on switching). “Other” refers to minor parties and independents. The left-hand panel of Figure 3 shows the probability of switching versus staying conditional on the candidate’s 2013 party registration. Over 80 percent of the AITC’s candidates remain with the AITC—by far the highest retention rate of any major party in West Bengal. By contrast, both the CPIM and the Indian National Congress (INC) lose the majority of their candidates. The INC in particular loses over 70 percent of its candidates to other parties. The right-hand panel of Figure 3 shows where, conditional on switching, these candidates choose to go. The AITC’s own relatively small pool of defectors tend to switch to either a minor party or the Bharatiya Janata Party (BJP). That is not surprising because after 2014 only the BJP has the resources to compete with the AITC.19 But among all other parties the switchers overwhelmingly join the AITC, suggesting it is stealing many of these parties’ candidates. In un19 Outside of West Bengal, the BJP won a massive victory in the 2014 national election. This victory, largely at the expense of the INC, gives the BJP resources unavailable to the other parties. 36 reported results we also find that defectors from the opposition parties tend to be their most popular candidates, meaning those who earned the most votes in the 2013 local election. The opposition is thus left with fewer and less effective organizers. 7.2 Candidates who Switch to or Stay With the Ruling Party Get Bigger Payments What induces these opposition candidates to forswear their old allegiances? Though it is possible they are persuaded by ideology or a simple desire to be on the winning side, anecdotal accounts suggest money may be part of the answer. There are countless reports of politicians at all levels of government taking bribes to switch parties.20 We can look for direct evidence of such payments by measuring the average 2015 NREGA allocations to all AITC candidates who live in the village. We focus on 2015 because, as the next section implies, the year between the 2014 national election and the 2016 state election is crucial. Figure 4.a shows that candidates who stay with the AITC receive far larger NREGA allocations in 2015 than candidates who stay with any other party (the gap ranges from 8 to 13 days). More importantly, candidates who switch from the AITC between 2013 and 2018 receive far smaller allocations than those who remain (roughly 8.5 days fewer). By contrast, payments to candidates who leave the two main opposition parties (the CPIM and the INC) to join the AITC are higher than those to candidates who either stay or who switch to a party other than the AITC.21 If our hypothesis is true—that the ruling party is seeking recruits who can help it take control of a local council—we would expect the extra payments would accrue mainly to candidates who actually won their races and thus have a vote on the council. Figure 4.b shows that, at least for the two main opposition parties, the gap in payments between stayers and switchers appears only for winning candidates. By themselves these results do not prove that the payments induced candidates to switch. 20 A few recent examples: ‘BJP Offered Bribe to Congress MLAs to Switch Sides in Gujarat: Congress’, Hindustan Times, October 18, 2020; ‘On MLA’s “35-Crore Bribe” Claim, Sachin Pilot Says “Vexatious, Concocted”’, NDTV, July 20, 2020; ‘Why Minister Eshwarappa’s Letter Has Dealt ‘Unprecedented’ Blow to Already Troubled Yediyu- 21 rappa’, The Print, April 2, 2021. The differences between payments to stayers in the three opposition parties versus stayers in the AITC are all highly significant (at the 1 percent level). The difference between switchers and stayers in the AITC is also significant at the 1 percent level. The difference between members of the INC and CPIM who stay versus those who switch to the AITC is also significant. 37 It is possible they switched for other reasons and simply started receiving larger payments as part of the package. But the pattern is at least consistent with a higher return to staying in the AITC or switching to the AITC (at least for candidates originally aligned with the INC). The gap between stayers and leavers in the AITC, for example, translates to a difference of over 1400 rupees, roughly 30 percent of the monthly consumption expenditure of the median household in rural West Bengal.22 7.3 Switching to the Ruling Party is Most Common in Panchayats where the Ruling Party Falls Short of a Majority Does the party switching represent actual strategic recruitment by the ruling party, or mere onesided opportunism by candidates who are switching party without necessarily being invited or courted by the ruling party? Figure 4.b offers some evidence that only switching candidates with something to offer (a vote on the council) get a premium through NREGA payments. But a more direct test is to see whether the net inflow of candidates switching to the AITC is highest in the areas where it actually needs more votes on the council to gain the absolute majority. Figure 5.a graphs the average net influx of candidates to the AITC—that is, the number of opposition candidates switching to the AITC minus the number of AITC candidates switching to a different party—against the margin of seats it won in the 2013 election. Where the margin is negative the AITC needs additional votes for the absolute majority. Where it is zero or positive the AITC won enough seats for the majority without any need for post-election recruitment. The figure shows that the average net influx is on average positive only in areas where the party needs seats. In areas where it has enough seats the average is close to zero. This pattern is striking because it is in the areas where the AITC has control that an opposition candidate has the most to gain from a unilateral switch to the AITC. But these are also the areas where the hypothetical recruit has relatively little to offer in return. 22 According to the 2014-2015 National Survey Sample dataset. 38 7.4 By 2016, Panchayats Barely Won and Barely Lost in 2013 Vote for the Ruling Party at Similar Rates If the ruling party is using recruitment to systematically take over councils it did not previously control, it would be better positioned in those areas to contest future elections. Such changes would be most obvious in the neighborhood of the cutoff for an absolute majority as measured with the running variable, which was defined based on the 2013 local election without adjustment for subsequent recruitment. The left-hand panel of Figure 5.b is similar to 5.a, except it plots the net influx as a function of the running variable. We divide the running variable into bins of width 0.15 (the bins must be relatively wide because the outcome is noisy). We calculate the average net gains for the AITC across all panchayats within the bin. We restrict our sample to the panchayats within a distance of 0.6 in the 1-norm metric. The figure shows that the AITC is most likely to gain members on the “control” (negative) side of the cutoff. On this side of the cutoff, the size of the average gain is proportional to the distance from the cutoff. 23 By contrast, there is no obvious pattern on the “treated” side of the cutoff. The pattern is not surprising given Panel a, but plotting the net influx against the running variable clarifies that the party’s recruitment over the period from 2013 to 2018 effectively converts “control” panchayats into “treated” panchayats. The recruitment did not happen immediately after the 2013 local elections, for if they had then the regression discontinuity estimates in Table 2 would have found that winning an absolute majority in 2013 would have had no impact on the national election of 2014. But we can get some suggestive evidence of the medium-run impact of recruitment by applying the same test to the returns from a state election in 2016. The 2016 election pitted the same major parties in a contest for control of the state assembly of West Bengal, making the stakes even higher for the ruling party. Much as we did for the 2014 national election, we match polling stations to panchayats and measure the percentage of votes received by the AITC’s state 23 This is only a rough intuition because the taxi metric depends not only on the number of seats but the margin by which the seats were lost. The party would presumably care about both when making recruitments (all else equal it would prefer to recruit popular candidates with many followers), but it is not obvious how it would weigh the quality of new recruits against the simple calculus of how many seats are needed for a majority. The taxi metric itself effectively treats two seats lost by 10 percent and 1 seat lost by 20 percent as equally distant from the cutoff even though the party may not. 39 assembly candidate (relative to the candidate’s overall share in the whole constituency for that assembly seat). The right-hand panel of Figure 5 plots the regression discontinuity in the AITC lean for both the 2014 and 2016 election. To put the elections on a similar axis we first strip out district and parliamentary constituency fixed-effects, making this the same specification as Column 2 of Table 2.24 For each election we restrict observations to the optimal bandwidth and plot the linear prediction alongside the average of the outcome within 12 equally spaced bins. The figure shows that the discontinuity in 2014 election returns has vanished by 2016. Panchayats where the AITC barely won an absolute majority in the 2013 local elections return no more votes than those where it barely lost. The figure shows that the discontinuity closes largely because of improvement in the performance of “control” panchayats. This result is consistent with the earlier result that the ruling party is incrementally converting “control” panchayats into “treated” panchayats by recruiting opposition candidates. One interpretation is that the recruitment was not yet complete in 2014, but had largely been completed by 2016. The figure may suggest the ruling party has expanded the frontier of its power deeper into the opposition’s strongholds. The pattern is only suggestive—there is no way to test whether the discontinuity would have remained had the recruitment never happened. But this interpretation does offer a coherent explanation that reconciles the pattern in the figure with the results of the preceding sections. 8 Conclusion In their seminal paper on party organization in Western democracies, Katz and Mair (1995) note that parties have become semi-state agencies: “[W]inning or losing may make less difference to a party’s political objectives because of the absence of great policy battles, but could make a good deal of difference to its sheer survival, since the resources for its sustenance now come increasingly from the state.” On India, Singh (1997) writes that “In the last few years the minority and coalition governments—at the centre and in the states—have used funds either to 24 The results are similar if we add assembly constituency fixed-effects (Appendix B.3) or use the raw vote share (Appendix B.4). 40 build patronage-based support networks or to bribe opposition MPs [Member of Parliament] at critical junctures.” Our paper shows that government resources can play a key role in a political party’s strategy to tighten and extend its grip on power across three tiers of government. This occurs because there are important complementarities of power between the tier of government that determines the allocation of resources and the tier of government that controls the program implementation. Within India, West Bengal’s local government institutions are unusual in the sense that politicians are explicitly members of a political party. Most other states require local politicians to run without party affiliations. While the party preferences of a candidate may be known or implied, this makes them unobservable for researchers and increases the difficulty of exploiting the complementarities of office for political parties. More generally, however, our study context is not atypical in developing countries. The AITC as a political party is primarily built as a vehicle for the political ambitions of its leader, which is a common feature among many political parties. And many countries and international organizations have pushed to decentralize power in recent years through transferring control over government program implementation to lower tiers of government. An interesting avenue for future research is to explore what happens once a party like the AITC loses power, and what conditions enable a change in government in this situation. Our proposed mechanism suggests that control over one office makes extending power to other positions easier, strengthening a party’s position further. Reversing this development could require a miscalculation from the incumbent or a large enough external shock, as is often the case for authoritarian regimes. Potentially there are unexplored strategies that the political opposition could take to curb the process. 41 Figure 4 Payments Conditional on 2013-to-2018 Transition 0 NREGA Days of Labor, 2015 (Full Year) 10 20 30 a) Overall: AITC BJP CPIM Stay Switch to Other INC Switch to AITC b) Winners vs. Losers, Main Opposition: NREGA Days of Labor, 2015 (Full Year) 10 20 Lost in 2013 Stay Won in 2013 Switch to AITC 0 0 NREGA Days of Labor, 2015 (Full Year) 10 20 30 INC 30 CPIM Lost in 2013 Stay Won in 2013 Switch to AITC Note: Each bar shows the average 2015 NREGA allocation to candidates conditional on whether they “stay” with or “switch” from the party they were registered with in 2013 (which is labeled on the horizontal axis). For example, candidates who “stay” with the AITC get roughly 26 days of labor while those who “switch” get roughly 18 days. See Footnote 21 for the statistical significance of these differences. 42 Figure 5 Seats Gained Versus the Running Variable Average Net Influx of Candidates into AITC 0 .5 1 1.5 a) Net Switching to AITC Versus Seat Margin AITC Needs More Seats -.5 AITC Has Enough Seats -10 -5 0 Margin of Seats Held by AITC 0 = Number Needed for Majority 5 10 b) Net Switching to AITC Versus Running Variable .04 Relative AITC Vote Share, Minus FEs -.02 0 .02 2016 2014 -.2 -.06 0 -.04 Net Influx of Candidates into AITC .2 .4 .8 ...Closing the Gap in Returns in the Next Election .6 AITC Gains New Members in GPs They Barely Lost... -.5 0 1-Norm (Taxi) Distance .5 -.4 -.2 0 1-Norm (Taxi) Distance .2 .4 Note: The left-hand panel restricts the sample to panchayats that lie within 0.6 of the cutoff for an absolute majority (based on the 1-norm). We divide the distance into bins of width 0.15 and calculate the average net gain of candidates for the AITC. The net gain is calculated as the number who switch from another party (or independent status) into the AITC minus the number who switch from the AITC. We average across panchayats within the bin. The blue dashed lines show the overall average on either side of the cutoff. The right-hand panel plots the regression discontinuity of the relative AITC vote share (share within the panchayat minus overall share in the constituency) for both the 2014 national election and the 2016 state election. We remove district and parliamentary constituency fixed effects to homogenize the two outcomes. For each year we restrict the sample to the optimal bandwidth (Calonico et al., 2014) and divide the running variable on either side of the cutoff into 12 equally spaced bins. 43 References Aldrich, John H., Why Parties? The Origin and Transformation of Political Parties in America, Chicago: University of Chicago Press, 1995. Anesi, Vincent, Philippe De Donder, and Toulouse School of Economics, “Party Formation and Minority Ideological Positions,” Economic journal, October 2009, 119 (540), 1303–1323. Asher, Sam and Paul Novosad, “Politics and Local Economic Growth: Evidence from India,” American Economic Journal: Applied Economics, 2017, 9 (1), 229–273. Bardhan, Pranab and Dilip Mookherjee, “Pro-Poor Targeting and Accountability of Local Governments in West Bengal,” Journal of Development Economics, 2006, 79 (2), 303–327. Baskaran, Thushyanthan, Brian Min, and Yogesh Uppal, “Election Cycles and Electricity Provision: Evidence from a Quasi-experiment with Indian Special Elections,” Journal of Public Economics, 2015, 126, 64–73. Brollo, Fernanda and Tommaso Nannicini, “Tying Your Enemy’s Hands in Close Races: The Politics of Federal Transfers in Brazil,” The American political science review, November 2012, 106 (4), 742–761. Calonico, Sebastian, Matias D Cattaneo, and Rocio Titiunik, “Robust Nonparametric Confidence Intervals for Regression-discontinuity Designs,” Econometrica, 2014, 82 (6), 2295– 2326. Chandra, Kanchan and Cillane Boulet, “A Language for Thinking about Ethnic Identity Change,” in Kanchan Chandra, ed., Constructivist Theories of Ethnic Politics, New York: Oxford University Press, 2012. Coase, Ronald Harry, “The Nature of the Firm,” Economica, 1937, 4 (16), 386–405. Cole, Shawn, “Fixing Market Failures or Fixing Elections? Agricultural Credit in India,” American Economic Journal: Applied Economics, 2009, 1 (1), 219–250. Curto-Grau, Marta, Albert Solé-Ollé, and Pilar Sorribas-Navarro, “Does Electoral Competition Curb Party Favoritism?,” American economic journal. Applied economics, October 2018, 10 (4), 378–407. Dey, Nikhil, Jean Dreze, and Reetika Khera, Employment Guarantee Act: A Primer, Delhi: National Book Trust, India, 2006. 44 Dey, Subhasish and Kunal Sen, “Is Partisan Alignment Electorally Rewarding? Evidence from Village Council Elections in India,” 2016. IZA Discussion Paper No. 9994. Downs, Anthony, An Economic Theory of Democracy, New York: Harper and Row, 1957. Dutta, Puja, Rinku Murgai, Martin Ravallion, and Dominique van de Walle, “Does India’s Employment Guarantee Scheme Guarantee Enployment?,” 2012. World Bank Policy Research Working Paper 6003. Feigenbaum, James J., Alexander Fouirnaies, and Andrew B. Hall, “The Majority-Party Disadvantage: Revising Theories of Legislative Organization,” Quarterly Journal of Political Science, 2017, 12 (3), 269–300. Gibbons, Robert S and John Roberts, The Handbook of Organizational Economics, Princeton University Press, December 2012. Gingerich, Daniel W., “Buying Power: Electoral Strategy before the Secret Vote,” American Political Science Review, 2020, 114 (4), 1086–1102. Golden, Miriam A and Lucio Picci, “Pork-barrel politics in postwar Italy, 1953–94,” American journal of political science, April 2008, 52 (2), 268–289. Golden, Miriam and Brian Min, “Distributive Politics around the World,” Annual Review of Political Science, 2013, 16, 73–99. Government of India, “Mahatma Gandhi National Rural Employment Guarantee Act, 2005 Operational Guidelines,” Technical Report, Ministry of Rural Development 2013. , “The Mahatma Gandhi National Rural Employment Guarantee Act, 2005,” Technical Report, Government of India 2018. Hicken, Allen, “Clientelism,” Annual Review of Political Science, 2011, 14, 289–310. Hölmstrom, Bengt, “Moral Hazard and Ebservability,” The Bell journal of economics, 1979, pp. 74–91. Imbert, Clement and John Papp, “Labor Market Effects of Social Programs: Evidence of India’s Employment Guarantee,” American Economic Journal: Applied Economics, 2015, 7 (2), 233– 263. Katz, Richard S. and Peter Mair, “Changing Models of Party Organization and Party Democracy,” Party Politics, 1995, 1 (1), 5–28. Levy, Gilat, “A Model of Political Parties,” Journal of economic theory, April 2004, 115 (2), 250–277. 45 Lipset, Seymour Martin and Stein Rokkan, Party Systems and Voter Alignments: Cross-National Perspectives, New York: Free Press, 1967. McCrary, Justin, “Manipulation of the Running Variable in the Regression Discontinuity Design: A Density Test,” Journal of Econometrics, 2008, 142 (2), 698–714. Mitra, Dola, Decoding Didi: Making Sense of Mamata Banerjee, Rupa Publications, 2014. Muirhead, Russell and Nancy L Rosenblum, “The Political Theory of Parties and Partisanship: Catching Up,” Annual Review of Political Science, May 2020, 23 (1), 95–110. Mukhopadhyay, Abhiroop, Himanshu, and M R Sharan, “The National Rural Employment Guarantee Scheme in Rajasthan: Rational Funds and their Allocation across Villages,” Economic and Political Weekly, 2015, 50 (6), 1409—-1443. Muralidharan, Karthik, Paul Niehaus, and Sandip Sukhtankar, “Building State Capacity: Evidence from Biometric Smartcards in India,” American Economic Review, 2016, 106 (10), 2895–2929. Novaes, Lucas M., “Disloyal Brokers and Weak Parties,” American Journal of Political Science, 2018, 62 (1), 84–98. Paul, Shutapa, Didi - The Untold Mamata Banerjee, India Penguin, 2018. Persico, Nicola, J.C. Rodriguez-Pueblita, and Dan Silverman, “Factions and Political Competition,” Journal of Political Economy, 2011, 119 (2), 242–288. Singh, Gurharpal, “Understanding Political Corruption in Contemporary Indian Politics,” in Paul Heywood, ed., Political Corruption, Oxford: Blackwell Publishers, 1997, pp. 210–222. Stokes, Susan C, “Political Parties and Democracy,” Annual Review of Political Science, November 2003, 2, 243–267. , Thad Dunning, Marcelo Nazareno, and Valeria Brusco, Brokers, Voters, and Clientelism: The Puzzle of Distributive Politics, Cambridge University Press, 2013. Susewind, Raphael, “Data on Religion and Politics in India,” 2016. Published under an ODbL 1.0 license. Available from https://github.com/raphael-susewind/india-religion-politics. Zimmermann, Laura, “Why Guarantee Employment? Evidence from a Large Indian PublicWorks Program,” 2021.