«SUMMARY | APRIL 2 WORKING PAPER No. 030 | May 2015 Religion, Land and Politics: Shrines and Literacy in Punjab, Pakistan Adeel Malik and Rinchan Ali ...»
zamindari, pattidari, bhayachara, imperfect bhayachara and Government-owned lands. Our main interest is in the proportion of the total number of villages governed by zamindari contracts.76 Land held in zamindari contracts SUMMARY | APR recognized proprietary rights of individual owners and entailed a direct payment. 77 Effectively, the contracts recognized the de facto ownership and influence of large landlords. Adding the zamindari variable in column (2) results in a weakly significant coefficient and suggests a negative impact of historical land inequality on contemporary literacy. Importantly, however, the interaction between shrines and politics retains its explanatory power.
The inclusion of historical literacy rate, compiled from the 1931 Punjab Census Report, does not change the results. Its own coefficient is not significantly different from zero. The next column (3) adds literacy rate in 1951 of migrants who crossed over from India in the wake of the 1947 partition. If tehsils with riverine shrines attracted relatively less literate migrants than this could be a potential confounding factor. Results in column 3 dispel this concern: the migrant literacy rate enters with a statistically insignificant coefficient. Together, these results suggest that our findings are not driven by the exclusion of historical literacy rate.
We experimented with a range of other historical characteristics that might have a bearing on contemporary literacy and compete with our political explanation. Prominent amongst these are: the share of total public spending dedicated to education in 1911 (column 4); the proportion of total cultivated area that is irrigated (column 5)78; and military recruitment as a percentage of total men of military age (column 6). Apart from military recruitment, which has beneficial impact on contemporary literacy, none of the other dimensions were significant. Interestingly, with the addition of military recruitment, the dummy variable for north Punjab becomes unimportant.
This is easy to reconcile: north Punjab has historically been a recruitment centre for the British and, later, Pakistani military.79 We experimented with a range of other variables, none of which had a statistically significant impact (not reported for shortage of space); these included measures of religious diversity 80; and the number of zaildars per capita (local revenue collectors).81 Overall, models estimated on the historical sample explain around 70 percent of the variation in literacy rate. Importantly, our emphasis on the political economy of shrines is maintained even after controlling for a variety of historical explanations. The coefficient on shrines-politics interaction is statistically significant at the 5 percent level in all specifications.
Determinants of Political Selection If the impact of riverine shrines on literacy is mediated through politics, it is worth asking: What factors determine the selection of shrines into politics? Which tehsils are more likely to have political shrines? As argued in the second section, the local influence of pīrs, their capacity to act as intermediaries and their control of landed property act as crucial political assets. The pīrs established an early foothold into politics, dating back the pre-partition era when noted shrine families participated in the 1937 and 1946 elections. As Table 1 shows, many of these families have persisted in electoral politics after independence.
In probing the determinants of political selection, we emphasize factors embedded in the political history of colonial Punjab. Our main prior is that shrines that enjoyed greater influence in the colonial era are more likely to have entered into politics in post-partition Punjab. To capture this, we draw upon the historical database described in the third section that records the number of shrines in a tehsil that were mentioned in District Gazetteers.
Typically, the Gazetteers recorded shrines that wielded greater influence over local political economy. These shrines were more likely, in turn, to receive colonial patronage and enter into electoral politics. Qualitative evidence offers support for this hypothesis. For example: the Kirmani Syeds of Shergarh, who entered into politics in 1920s and survive in the parliament to this day, were recognized by the British through a sizeable land grant (1,168 acres).82 The Kirmanis and their shrine, Daud Bandagi, were similarly recognized in the District Gazetteer.
76 Ideally it would be more appropriate to estimate the total area under zamindari contracts, rather than counting the number of villages.
However, the coverage of data on land tenures by area is very sparse.
77 In Pattidari, both the land and revenue were divided in accordance with ancestral or customary shares as per the laws of inheritance.
Under Bhayachara contracts, possession was the measure of right of land. Through both Pattidari and Bhayachara the de facto land tenure arrangements that pre-existed the British rule were recognized. Another category, imperfect bhayachara and pattidari, denoted a situation where the land was held partly in severalty and partly in common; the measure of right in common land was the amount of the share or the extent of land held in severalty.
78 Specifically, proportion of cultivated area characterized as Nahri and Chahi Nahri in District Gazetteers and Census Reports.
79 The role of military recruitment in human capital accumulation in colonial Punjab has been discussed in greater detail in Eynde (2011).
The variable combines census information on the proportion of population identified as Muslim, Hindu, and Sikh.
81 Zaildars were effectively village headmen in charge of tax collection from their respective zails (revenue extraction units – usually a collection of villages). Typically, zaildars were locally influential landlords.
82 This was, in fact, one of the largest land grants given to a Muslim shrine in the Montgomery District. For further details, see the Final Report on Settlement of the Montgomery District.
In fact, this is true for all politically resilient shrine families mentioned in Table 1. We explore this historical connection in the empirical domain.
SUMMARY | APR Drawing on these two unique databases, we now examine our claim that historically important shrines are more likely to select into electoral politics in post-independence Punjab. To investigate this we run probit models for the dichotomous dependent variable, coded one for tehsils where shrines have a direct electoral linkage, zero otherwise (see the third section for a detailed discussion of the politics database). Our main explanatory variable for political selection is a dummy variable selecting tehsils containing at least one shrine mentioned in the historical District Gazetteer. The complete list of shrines mentioned in PDGs is available in Appendix 1.
Figures 4-5 visually represent the spatial distribution of shrines across these two metrics: history and politics. All across Punjab there is evidence of historically influential shrines, but south-western parts of Punjab have a greater concentration of such shrines (as indicated by bigger circles on the map in Figure 4). There is also a larger presence of politically influential shrines in south-west Punjab (Figure 5).
Figure 4: Shrines Mentioned in District Gazetteers (Number) Figure 5: Politically Influential Shrines (Number)
Table 6 presents results for this investigation. For ease of interpretation, we report marginal probability effects for different variables.83 Column 1 provides an initial test of our prediction. As expected, the coefficient on the binary PDG variable is positive and statistically significant at the 1 percent level. This suggests that as the variable for shrines mentioned in PDG takes the value 1, the probability that a tehsil has politically affiliated shrine(s) increases by 31 percentage points. Column 2 includes the usual set of geographic controls: latitude, longitude, elevation, and distance from river.84 While the coefficient on the log of elevation is negative and significant, the effect of our historical shrine variable remains fairly robust. This impact also holds up to the inclusion of regional fixed effects in column 3; the coefficients on central and south Punjab are positive and statistically significant. If the dummy variable for central Punjab takes the value one, the probability that the tehsil has a politically affiliated shrine increases by 44 percentage points relative to western Punjab. The inclusion of average rainfall (col. 4) does not alter the results.
It is possible that shrines have a greater political role in tehsils with a more unequal land distribution. We checked this by entering Proportion Muzaara in the model (results not shown). It was neither significant on its own nor did it dislodge the impact of the historical shrine measure.85 The preceding analysis provides strong support for our argument that the likelihood of a tehsil containing a politically linked shrine increases as the average number of shrines mentioned in PDGs rises. What it does not test, however, is the role of historical literacy rate on the probability of political selection. This may be important to control, since tehsils with a historically lower literacy rate may also have a greater concentration of PDG shrines. We test this possibility in columns 5-6 that report probit estimates on the historical sample consisting of 69 tehsils.86 83 Marginal probability effects are the partial effects of each explanatory variable on the probability that the observed dependent variable equals 1. These are estimated at the sample mean values of regressors. The ‘dprobit’ command in stata was used to compute the maximum likelihood estimates of these effects.
84 The last three variables are expressed in natural log.
Experimenting with an alternative, district-level, measure of land inequality (the proportion of total cultivated area in a district that is 25 acres and above, obtained from the Agricultural Census of Pakistan) yields a slightly more informative result. While the coefficient on this alternative land inequality measure was positive and significant at the 10% level, the PDG variable retained its robust influence, indicating that it is not simply capturing underlying land inequality.
86 Please refer to Appendix 4 for a complete list of tehsils included in the historical sample.
Robust standard errors in parentheses: *** p0.01, ** p0.05, * p0.1 The sample consists of 115 tehsils (see Appendix 2 for a complete list). In columns 5-6, the historical sample consists of 69 tehsils (see Appendix 4 for the complete list).
The initial pattern of results also holds in the historical sample. Replicating the model in column 4 on the historical sample preserves the basic pattern of results (column 5). In column 6 we add the historical literacy rate (based on the 1931 census of Punjab), which has a statistically significant negative impact. Calculated at the mean values of control variables, a small increase in the average historical literacy rate decreases the probability that a tehsil has a political shrine by 57 percentage points.87 Importantly, however, the coefficient on the PDG dummy variable is still positive and statistically significant at the 1 percent level. This means that if the PDG dummy changes from zero to one, the probability that a tehsil has a political shrine increases by 42 percentage points. The diagnostic chi-square tests suggest that our model fits the data reasonably well. 88 In short, tehsils with shrines considered as historically important, as measured by their mention in historical district gazetteers, predict their selection into politics in post-independence Punjab. Importantly, the underlying pattern is not simply driven by lower initial historical literacy rates in such tehsils or regional fixed effects. Taken together, the above results indicate a strong pattern of persistence.
THE POLITICAL ECONOMY OF SCHOOLINGGiven the primacy of politics in our empirical analysis, it is worth probing how politics might shape literacy in Punjab. Globally, there is a well-established literature on the political economy of public service provision. Recent 87 Including a measure of historic land inequality (the proportion of zamindari villages) together with the 1931 total literacy rate doesn’t alter the results.
88 The Pearson chi-square test, which is a test of the observed against expected number of responses using cells defined by the covariate
patterns, is estimated using the estat gof command in stata.
empirical evidence has also highlighted the relevance of politics in explaining both variation in educational outcomes and administrative arrangements for educational provision (Ansell and Lindvall 2013; Gallegho 2010).
Given that literacy also serves as a political resource, there can be multiple pathways linking politics to education.
SUMMARY | APR This section furnishes some pertinent evidence from Punjab. For brevity, we focus only on two inter-related dimensions: the modes of governance affecting the educational sector and the mechanisms behind elite capture of resources for schooling.
Modes of Governance Punjab’s political elites have historically enjoyed significant control over public policy and administrative structures for education. Whether administered through the centre, province or a decentralized dispensation, educational provision has largely remained a centralized affair, where interests of political elites are routinely accommodated. Since colonial rule, institutional arrangements for public provision of education have privileged the role
of local elites and the bureaucratic class. Under the British, education was administered by local governments:
district boards in rural areas and municipal committees in urban centres. 89 Local governments were usually composed of members that were elected, nominated or appointed. The nominated members were typically drawn from the locally influential elite who owed their allegiance to colonial authorities. Appointed members were officers of the colonial bureaucracy, such as the District Commissioner or the Magistrate. Elected members usually consisted of middle class professionals (e.g., lawyers and doctors) primarily belonging to major towns of the district.