Women s Education and Women s Political Participation

2014/ED/EFA/MRT/PI/23 Background paper prepared for the Education for All Global Monitoring Report 2013/4 Teaching and learning: Achieving quality for all Women s Education and Women s Political Participation Sonia Bhalotra, Irma Clots, Lakshmi Iyer 2013 This paper was commissioned by the Education for All Global Monitoring Report as background information to assist in drafting the 2013/4 report. It has not been edited by the team. The views and opinions expressed in this paper are those of the author(s) and should not be attributed to the EFA Global Monitoring Report or to UNESCO. The papers can be cited with the following reference: Paper commissioned for the EFA Global Monitoring Report 2013/4, Teaching and learning: Achieving quality for all For further information, please contact efareport@unesco.org

Women s Education and Women s Political Participation Sonia Bhalotra, Irma Clots, Lakshmi Iyer 28 January 2013 Abstract This paper presents the first estimates of the relationship between women s political participation and women s literacy conditional upon and relative to that of men. Using electoral data on electoral outcomes matched to census data on literacy rates at the district and year level for the period 1980-2008 which covers five or six elections per state in India, we find significant impacts of the literacy of women relative to men on women s candidacy, competitiveness and turnout. 1. Introduction There is a long-standing view that education is a pre-requisite for democracy (Dewey 1916, Lipsett 1959). However recent research finds no evidence that improvements in education have played a causal role in the transition to democracy (Acemoglu et al. 2005). With many developing countries having transitioned to democracy in recent decades, a pertinent question concerns the role of education in promoting the quality or functioning of democracy. There is some evidence that educated citizens are more likely to monitor corruption and, by this route, education promotes the quality of governance (Botero et al. 2012). Similarly, democratic governments are more responsive in areas with higher newspaper circulation (Besley and Burgess 2002) and these are likely to be areas with more education. Alongside, there is laboratory and other evidence that women are fairer or less corrupt in general (Andreoni and Vesterlund 2001, Swamy et al. 2001) and in government (Brollo and Troiano 2012, Dollar et al. 2001, Swamy et al. 2001). We link these findings by asking whether education and in particular the education of women influences the participation of women in politics. 2. Data and Methods Electoral data including information on the gender of and the votes obtained by every political candidate in every constituency in India for elections to state legislative assemblies that occurred during the years 1980-2008 are available from the Election Commission of India and were digitized by the authors. The data also contain male and female turnout as a fraction of the constituency population. The electoral data were aggregated to the district level for each election year and then merged by district and year with census data on district-specific literacy rates for men and women since time series of constituency level data on education are not available. The analysis sample contains 2255 observations and includes 346 districts and six to seven election years. The data contain one observation for every election period, so the observations are roughly quinquennial. The precise number of elections per district depends upon the state since state elections are staggered, creating some variation in timing and hence in frequency within the sample window. We use the district-year panel to estimate equations of this form (1) P dt = α d + λ t + β E dt + X dt γ + e dt P is an indicator for an electoral outcome. We analyse a range of measures of women s participation in politics, namely, the fraction of all candidates that are women from all parties and from major parties, the share of all votes won by women, the probability that a woman is among 1

the top-two vote winners, female turnout and, also, male turnout. Candidacy and voter turnout are classical measures of elite and mass participation and votes are a measure of competitiveness. Candidacy and votes are measured as shares for women relative to men and women. Since female turnout is measured as a share of the female population, we also model male turnout. E is an indicator of the education of women conditional upon or relative to that of men. We focus on literacy as we have systematic district and year variation in this variable and it seems plausible that it influences the political outcomes of interest. We investigate different specifications, which are the literacy rates of women and men (each conditional upon the other), the ratio of these rates (which adjusts for level differences in rates across districts, allowing us to focus on the gender gap), and the log of the ratio of these rates (which captures proportional changes). We include district fixed effects (α d ), election year fixed effects (λ t ), and district demographic composition controls, which are the fractions urban, Muslim, low-caste and female (the vector X dt ). Standard errors are clustered at the district level to allow for within district correlation of the errors over time. In order to assess the extent to which the raw association of literacy and the political outcomes we consider is driven by cross-sectional variation in district-specific time-invariant factors, and to allow similarly for the correlation of literacy and demographic characteristics, we display estimates in which the district fixed effects and the vector of controls are added in sequential steps. In a similar spirit, we initially display the results of estimating equation (1) on the crosssection of 346 districts : (2) P d = α 1 + β 1 E d + X d γ 1 + e 1d 3. Descriptive Statistics Summary statistics for all variables in the analysis are in Table 1. On average over the sample period, 1980-2008, only 4.3% of candidates for election to India s state assemblies are women. The corresponding share of women candidates fielded by major parties (parties that at some time in the sample period won at least 5% of seats) is 5.5%, about 25% larger than the share from all parties, which suggests that when women are fielded then they are serious candidates rather than token candidates. The vote share of all women candidates is 5.1%, slightly larger than proportional to their candidate share and the chances that a woman ranks in the top-two vote winners in an electoral competition is 10.1%. The share of women who win and hold state seats as legislators is 5.5% which, given that women are only 4.3% of all candidates, implies that their chances of winning conditional upon standing are stronger than those of men (Bhalotra, Clots, Iyer 2013). Overall, these figures suggest that women candidates, while scarce, are competitive. We also analyse the wider participation of men and women as voters and note that, on average, 56.7% of women and 65.3% of men turnout to vote in state elections. These indicators of women s political candidacy, competitiveness and turnout all vary considerably across the Indian districts. For instance, the average share of women candidates from major parties ranges from zero to 40.7%, the average (unconditional) chances that a woman is in the top two candidates ranges from zero to 75% and women s turnout ranges from 2% to 88.8%. The small means highlight the importance of understanding the low levels of participation, and the large standard deviations indicate the potential for finding an explanation in analysis of district and year varying data. 2

The explanatory variable that we focus upon in this study is female literacy, conditional upon and relative to male literacy. On average, only 33.9% of Indian women are literate and the district average ranges from 3.4% to 84.7%. The ratio of male to female literacy rates is, on average, 1.9. In other words, men are almost twice as likely as women to be literate. Plots of trends in the main variables are in Figures 1 to 4. There is a secular trend in the share of women candidates between 1980-2000, with a particularly sharp rise in the 1990s, and an uncertain trend after 2000. Overall, the share of women has climbed from just more than 2% to almost 8% (Figure 1). The share of votes received by women displays a similar pattern (Figure 2). It is interesting that trends in female and male literacy rates are broadly similar to trends in women s candidacy: they show a secular increase that is sharper in the 1990s than in the 1980s or 2000s (Figure 3). The ratio of male to female literacy rates declines secularly, reflecting stronger trends in women s than in men s literacy and, as with the other trends, this ratio declines most rapidly in the 1990s (Figure 4) 4. Results 4.1. Cross-District Estimates There is a strikingly similar pattern in the association of female and male literacy and the different indicators of women s role in politics (see Table 1 which reports estimates of equation 2). Women s Political Candidacy: In the cross-section, which may be interpreted as a long run average, there is a significantly positive association of female literacy rates and female candidate share conditional upon the male literacy rate. In fact male literacy exhibits a negative association with female candidate share. The estimates indicate that if, other things equal, two districts were to differ in their female literacy rates by 10% points then the share of female candidates in these districts would differ by 1 percentage point. This is about 23% of the mean female candidate share in the sample. The impact of literacy is larger when we focus upon the share of female candidates fielded by major parties, which we define as parties that, at some point in the sample period, achieved at least 5% of seats in the state. A 10% female literacy differential changes the share of female candidates from major parties by 1.5% points, which is about 28% of the sample mean. The negative association of female candidacy with male literacy given female literacy is of a similar magnitude. The share of female candidates is lower in districts with a higher share of Muslims and it is higher in districts that have a larger share of state assembly seats reserved for scheduled caste and tribe candidates (which is a function of the share of these low caste groups in the state population). There is no statistically significant association with the population sex ratio or with the share of the district population that is designated urban. When we isolate the women among candidates from major parties then the influence of religion and caste composition is poorly determined but there is a significant positive association with the ratio of females to males in the population, which is widely regarded as an indicator of gender-equal preferences (Miller 1981, Sen 1991). Women s Competitiveness: Women s performance as political candidates, indicated by the share of all votes cast that are cast in favour of women candidates, is positively associated with women s literacy. A 10% point differential in literacy rates is associated with a 1.6% point increase in vote share (close to 25% of the sample mean) and a 3.2% point increase in the probability that a woman ranks amongst the top two in an electoral race (just more than 30% of the mean). Similar 3

to the results for candidacy, for a given level of female literacy, male literacy exhibits a negative association with women s competitiveness of roughly similar magnitude. Electoral Turnout: Both female and male turnout are positively associated with female literacy and, conditional upon this, negatively associated with male literacy, and the positive influence of female literacy is larger for female than for male turnout, that is, it raises female relative to male turnout. A 10% point difference in district literacy rates, other things equal, is associated with a 7.8% point increase in female turnout (almost 14% of the sample mean) and a 4% point increase in male turnout (6% of the mean). 4.2. Estimates from the District Panel 4.2.1. Female and Male Literacy Table 2 reports estimates of equation (1) in column 3. Column 1 shows the unconditional association in the panel which is estimated from pooled variation across districts and over time within districts. Column 2 isolates the within-district variation over time by introducing district fixed effects and column 3 differs only in adding district demographic variables as controls. The estimates in columns 2 and 3 may be interpreted as quasi-causal under standard panel data identifying assumptions, which seem plausible since adult literacy rates are pre-determined with respect to the electoral variables we consider. The estimates in column 1, which are obtained from data that include cross-district variation show, like the cross-sectional estimates, that female literacy is positively associated with every one of the electoral outcomes and male literacy is negatively associated (although the negative associations of male literacy are smaller or less robust than in the cross-section). However the within-district estimates indicate a broadly reversed pattern. For a given level of female literacy, improvements in male literacy over time produce significant increases in female candidacy and in female and male turnout, with no consistently significant impact on the competitiveness of female candidates (although before we control for demographic composition, there is a large positive impact on women s vote share). Conditional upon male literacy, female literacy shows no significant association with female candidacy or competitiveness and it acts to lower female and male turnout. A 10% point increase in the male literacy rate within district over time predicts a 1.4% point increase in the share of females amongst all candidates (32% of the sample mean), a 2.35% point increase in the share of females amongst major party candidates (43% of the mean), a 10.9% point increase in female turnout (19% of mean) and a 7.9% increase in male turnout (12% of mean), raising female relative to male turnout. A 10% point increase in female literacy conditional upon male literacy lowers female and male turnout to a similar degree, by about 4.5% points. In a reversal of the result that emerges from cross-sectional variation, the share of the population that is Muslim is positively associated with the overall share of female candidates. As in Table 1, the share of state assembly seats reserved for scheduled castes and tribes is also positively associated with the share of female candidates. However both of these associations- and also the association with the population sex ratio that we noted in studying cross-district variation- are rendered insignificant upon introduction of district fixed effects. 4.2.2. Ratio of Male to Female Literacy Estimates of equation (1) in which we replace female and male literacy rates with the ratio of the male to the female rate are presented in Table 3. The column format is the same as in Table 2. 4

The advantage of restricting the literacy variables to enter as a ratio rather than independently is that this adjusts for district and year differences in overall (male plus female) literacy levels, focusing in on gender inequality in literacy. It also sits nicely with our use of shares for women in the candidacy and vote share models. A higher ratio indicates greater gender inequality in literacy. With this specification, the estimates are not sensitive to whether or not we isolate within-district variation from the pooled variation. Every column of Table 3 shows a significant negative impact of the male-female ratio of literacy on every electoral outcome that we model. A 10% point increase in female literacy from the mean level, with male literacy held constant is equivalent to a decrease in the male/female literacy ratio of 0.4 (a change in the mean of the ratio from 1.9 to 1.5). This change is predicted to raise the share of female candidates by 0.84% points (20% of the sample mean), the share of female candidates from major parties by 0.96% points (18% of the mean), female vote share by 0.76% points (15% of the mean), the probability that a woman ranks among the top two in an electoral race by 1.32% points (13% of the mean), female turnout by 3.2% points (5.6% of mean) and male turnout by 1.6% points (2.5% of mean). 5. Conclusions Using a large sample of electoral data matched to census data for Indian districts that is broadly representative for the country, this paper finds that narrowing of the gender gap in literacy has substantial impacts on women s political participation. The relationship is strikingly consistent across measures, including the probability that women stand as candidates for election, their competitiveness in elections relative to men and female relative to male voter turnout. A 10 percentage point increase in women s literacy which, at the mean levels of women s and men s literacy in India over the sample period results in a contraction of the male-female literacy ratio by 40 percentage points, is associated with increases in female candidacy of 20%, of the chances that a woman is in the top two vote winners of 13% and in female turnout of close to 6%, in every case, relative to the sample mean. We know of no similar evidence for India or any other context. The results suggest that educating women, even at the most basic level, has the potential to raise women s participation in democracy both as voters and as legislators. As discussed, to the extent that women are less corrupt, this stands to improve the quality of democratic government. In addition, it improves the representation of women s interests in policy making. Recent research shows that raising the share of women in government tends to improve, for instance, education, health, provision of drinking water and the reporting of crime against women (Clots-Figueras 2012. Bhalotra & Clots-Figueras 2011, Brollo & Troiano 2012, Chattopadhyay and Duflo 2004, Iyer et al 2012). There is evidence that a woman being elected to office fuels women s candidacy, for instance, through lowering party bias against women or possibly through lowering voter bias against women (Bhalotra et al. 2013, Beaman et al. 2010). Of particular note is evidence of feed back in the estimated relationship. Recent work on India shows that quotas for women in headships of village councils in India that institutionalize women s participation in politics result in aspirations for and investments in girls education increasing (Beaman et. al. 2012). This suggests a virtuous cycle and implies that the long run impacts of investing in women s literacy are likely to be larger than the impacts we estimate. 5

References Acemoglu, Daron, Simon Johnson, James Robinson and Pierre Yared. 2005. From education to democracy? NBER Working Paper 11204. Andreoni, James, and Lise Vesterlund (2001). Which is the Fair Sex? Gender Differences in Altruism. The Quarterly Journal of Economics, 2001, 116, 293-312. Besley, Timothy, and Robin Burgess. 2002. The Political Economy of Government Responsiveness: Theory and Evidence from India. Quarterly Journal of Economics 117: 1415-1451. Bhalotra, Sonia and Irma Clots-Figueras 2011. Health and the Political Agency of Women Bhalotra, Sonia, Irma Clots-Figueras and Lakshmi Iyer. 2013. Path-Breakers: How Does Women s Political Participation Respond to Electoral Success? Mimeograph. Chattopadhay,R., and E. Duflo (2004). Women as Policy Makers: Evidence from a India-Wide Randomized Policy Experiment. Econometrica, 72, 1409 1443. Clots-Figueras, I., (2011). Women in Politics. Evidence from the Indian States. Journal of Public Economics, 95, 664-690. Clots-Figueras, Irma (2012). Are Female Leaders Good for Education? Evidence from India. American Economic Journal: Applied Economics. 212-44. Vol 4, nº1. Beaman, Lori, Esther Duflo, Rohini Pande and Petia Topalova. 2010. Political Reservation and Substantive Representation: Evidence from Indian Village Councils. India Policy Forum 7, 2010-11 Beaman, Lori, Esther Duflo, Rohini Pande and Petia Topalova. 2012. Female Leadership Raises Aspirations and Educational Attainment for Girls: A Policy Experiment in India. Science: January 12, 2012. Botero, Juan Alejandro Ponce, and Andrei Shleifer. 2012. Education and the Quality of Government. NBER Working Paper No. 18119 June. Brollo, Fernanda and Ugo Troiano. 2012. What happens When a Woman Wins a Close Election? Evidence from Brazil. Mimeograph. Universities of Alicante and Harvard. Clots-Figueras, Irma. 2011. Women in Politics: Evidence from the Indian States. Journal of Public Economics 95: 664-690. Dewey, John (1916) Democracy and Education, New York; The Macmillan Company. Dollar, David, Raymond Fisman, Roberta Gatti. 2001. Are women really the fairer sex? Corruption and women in government. Journal of Economic Behavior & Organization. Iyer, Lakshmi, Anandi Mani, Prachi Mishra and Petia Topalova. 2012. The Power of Political Voice: Women s Political Representation and Crime in India. American Economic Journal: Applied Economics, 4(4): 165-193. Lipset, Seymour Martin (1959) Some Social Requisites of Democracy: Economic 6

Development and Political Legitimacy, American Political Science Review, 53, 69-105 Swamy, Anand, Stephen Knack, Young Lee, Omar Azfar. 2001. Gender and corruption. Journal of Development Economics. 7

Table 1: Cross Section of Districts in India Table 1a: Female candidate share a3 fraclit_female 0.090*** 0.099*** a4 [0.025] [0.025] fraclit_male -0.099*** -0.116*** [0.032] [0.035] fracurban 0.012 [0.008] fracmuspop -0.027*** [0.010] res_scst 0.016** [0.008] fracfemalepop 0.051 [0.079] r2 0.06 0.13 N 346 346 Table 1b: Female candidate share from major parties a3 a4 fraclit_female 0.141*** 0.150*** [0.037] [0.038] fraclit_male -0.181*** -0.207*** [0.048] [0.051] fracurban 0.017 [0.014] fracmuspop -0.034 [0.021] res_scst 0.016 [0.012] fracfemalepop 0.219* [0.131] r2 0.05 0.08 N 346 346 Table 1c: Female vote share a3 fraclit_female 0.140*** 0.155*** a4 [0.039] [0.042] fraclit_male -0.179*** -0.206*** [0.051] [0.055] fracurban 0.01 [0.014] fracmuspop -0.037* [0.020] res_scst 0.017 [0.011] fracfemalepop 0.143 [0.138] r2 0.04 0.07 N 346 346 8

Table 1d: Female in top 2 vote winners a3 fraclit_female 0.288*** 0.322*** a4 [0.077] [0.086] fraclit_male -0.372*** -0.443*** [0.098] [0.109] fracurban 0.027 [0.028] fracmuspop -0.084** [0.042] res_scst 0.021 [0.024] fracfemalepop 0.297 [0.280] r2 0.05 0.07 N 346 346 Table 1e: Female turnout a3 fraclit_female 0.742*** 0.783*** a4 [0.115] [0.120] fraclit_male -0.448*** -0.391** [0.157] [0.162] fracurban -0.206*** [0.037] fracmuspop 0.094 [0.059] res_scst -0.065** [0.032] Fracfemalepop 0.377 [0.399] r2 0.31 0.4 N 346 346 Table 1f: Male turnout a3 fraclit_female 0.363*** 0.395*** a4 [0.089] [0.093] fraclit_male -0.261** -0.225* [0.115] [0.117] fracurban -0.148*** [0.027] fracmuspop 0.063 [0.046] res_scst -0.046** [0.020] fracfemalepop 0.304 [0.297] r2 0.11 0.21 N 346 346 9

Table 2: District level panel Table 2a Fraction of female candidates fraclit_female 0.079*** 0.047 0.045 [0.023] [0.055] [0.057] fraclit_male -0.021 0.149** 0.142** [0.028] [0.068] [0.069] fracurban -0.013 [0.030] fracmuspop 0.205** [0.079] res_scst 0.326** [0.162] fracfemalepop 0.212 [0.254] r2 0.09 0.46 0.46 N 2255 2255 2255 Table 2b: Fraction of female candidates from major parties fraclit_female 0.117*** -0.028-0.022 [0.036] [0.097] [0.099] fraclit_male -0.078* 0.241** 0.235* [0.045] [0.121] [0.122] fracurban -0.013 [0.050] fracmuspop -0.031 [0.152] res_scst 0.423 [0.531] fracfemalepop 0.452 [0.382] r2 0.03 0.4 0.4 N 2255 2255 2255 Table 2c: Female vote share fraclit_female 0.118*** -0.003 0.001 [0.036] [0.082] [0.084] fraclit_male -0.089** 0.181* 0.172 [0.044] [0.104] [0.105] fracurban -0.022 [0.050] fracmuspop 0.151 [0.132] res_scst 0.13 [0.305] fracfemalepop 0.185 [0.400] r2 0.03 0.46 0.46 N 2255 2255 2255 10

Table 2d: Probability that a woman is in the top two vote winners fraclit_female 0.249*** 0.042 0.075 [0.071] [0.172] [0.177] fraclit_male -0.209** 0.263 0.231 [0.086] [0.216] [0.219] fracurban -0.099 [0.119] fracmuspop 0.179 [0.304] res_scst 0.156 [0.699] fracfemalepop 0.236 [0.911] r2 0.02 0.42 0.42 N 2255 2255 2255 Table 2e: Female turnout fraclit_female 0.578*** -0.473*** -0.473*** [0.102] [0.102] [0.112] fraclit_male -0.215 1.071*** 1.089*** [0.134] [0.124] [0.127] fracurban -0.012 [0.068] fracmuspop 0.021 [0.303] res_scst -1.940* [0.998] fracfemalepop -2.790*** [0.704] r2 0.27 0.74 0.75 N 2252 2252 2252 Table 2f: Male turnout fraclit_female 0.273*** -0.412*** -0.448*** [0.079] [0.092] [0.104] fraclit_male -0.109 0.758*** 0.792*** [0.100] [0.108] [0.112] fracurban 0.106 [0.094] fracmuspop -0.101 [0.300] res_scst -2.037* [1.064] fracfemalepop -0.877 [0.690] r2 0.1 0.61 0.62 N 2252 2252 2252 11

Table 3: District level panel, ratio of male to female literacy rate Table 3a: Female candidate share ratiolit -0.016*** -0.027*** -0.020*** [0.001] [0.002] [0.003] fracurban 0.111*** [0.036] fracmuspop 0.440*** [0.108] res_scst 0.336** [0.170] fracfemalepop 0.723** [0.312] r2 0.09 0.37 0.4 N 2255 2255 2255 Table 2b: Female candidate share from major parties ratiolit -0.016*** -0.027*** -0.023*** [0.002] [0.004] [0.004] fracurban 0.105** [0.048] fracmuspop 0.197 [0.154] res_scst 0.432 [0.530] fracfemalepop 1.001** [0.408] r2 0.03 0.37 0.38 N 2255 2255 2255 Table 3c: Female vote share ratiolit -0.014*** -0.024*** -0.019*** [0.002] [0.003] [0.004] fracurban 0.079* [0.044] fracmuspop 0.344** [0.137] res_scst 0.137 [0.302] fracfemalepop 0.638 [0.415] r2 0.03 0.44 0.44 N 2255 2255 2255 12

Table 3d: Female in top 2 vote winners ratiolit -0.025*** -0.040*** -0.033*** [0.004] [0.007] [0.008] fracurban 0.105 [0.095] fracmuspop 0.568* [0.314] res_scst 0.173 [0.708] fracfemalepop 1.072 [0.936] r2 0.02 0.4 0.4 N 2255 2255 2255 Table 3e: Female turnout ratiolit -0.105*** -0.083*** -0.077*** [0.006] [0.006] [0.006] fracurban 0.169* [0.091] fracmuspop 0.269 [0.320] res_scst -1.937* [1.052] fracfemalepop -1.386* [0.790] r2 0.28 0.72 0.73 N 2252 2252 2252 Table 3f: Male turnout ratiolit -0.050*** -0.041*** -0.036*** [0.005] [0.005] [0.006] fracurban 0.168* [0.097] fracmuspop 0.063 [0.311] res_scst -2.038* [1.084] fracfemalepop -0.175 [0.731] r2 0.11 0.59 0.61 N 2252 2252 2252 13

Appendix Table: Descriptive Statistics Variable Obs Mean Std. Dev. Min Max Female candidates 2255 0.0429651 0.0357834 0 0.2833333 Female candidates major 2255 0.0546544 0.0589729 0 0.4 Female vote share 2255 0.0508448 0.0541107 0 0.4066014 Female in top-two parties 2255 0.1009562 0.1141907 0 0.75 Female turnout 2252 0.5668884 0.1369804 0.0203457 0.8878397 Male turnout 2252 0.6532055 0.1055115 0.0375666 0.9076594 Urban share 2255 0.2191843 0.1603143-0.044965 1 Muslim share 2255 0.1085515 0.1039515 0.0004653 0.689937 Reserved seats 2255 0.2422921 0.1990159 0 1 Female/male population ratio 2255 0.4823118 0.0161022 0.4195847 0.5660195 Female literacy 2255 0.3391054 0.1711246 0.0339243 0.8465639 Male-female literacy rate 2255 0.2070497 0.0599993 0.0178998 0.3972896 Log male-female literacy rate 2255 0.5865876 0.3230304 0.0209236 1.767161 Male/female literacy rate ratio 2255 1.900943 0.6933078 1.021144 5.85421 Notes: District level data for 1980-2008. Reserved seats refer to the share of seats reserved for members of scheduled castes and tribes in the state assembly. 14

Figure 1: Trend in Female Share of Candidates for State Legislative Assemblies in India Fraction of candidates that are female.02.04.06.08 1980 1990 2000 2010 year Figure 2: Trend in Female Vote Share in State Legislative Assembly Elections in India Share of votes received by women.02.04.06.08.1 1980 1990 2000 2010 year 15

Figure 3: Trends in Female and Male Literacy Rates in India Figure 4: Trend in Male Relative to Female Literacy Rate in India 16