In my last post I discussed how elites might weaken bureaucratic institutions in anticipation of losing electoral power, a phenomenon I call “hollowing out the state.” In this post I want to tackle a more fundamental chicken or egg question: do historic institutions even matter? If we believe institutions matter, then they shape political and economic trajectories, in addition to norms and beliefs. If they are endogenous, then what they achieve are merely a function of the conditions that created them. As Przeworski notes “the embarrassingly obvious observation is that if endogeneity is strong, then institutions cannot have a causal efficacy of their own.” How then do we show that institutions matter?
The question — do institutions matter? — presents two types of challenges. The first challenge lies in identifying conditions under which institutions are assigned somewhat randomly to societies, and in this way avoiding the issue that Przeworski raises of context shaping institutional development. Studies in historical political economy have rallied to this particular challenge. Scholars studying the effects of colonial institutions have relied on pain staking archival research, historical data, and, importantly, natural experiments — through settler mortality rates, boundary changes, or trading routes — to show that extractive historic institutions have shaped long-run economic growth, and the quality of institutions in the present day. Together these studies have definitively shown us that institutions indeed matter, long after they have ceased to exist. But they have now raised a second challenge – what are the causal pathways that connect historic institutions with present-day outcomes. What gives these colonial institutions causal efficacy?
Take for instance debates on the effects of colonialism in India — the “jewel in the crown” of the British empire. Banerjee and Iyer’s (BI) paper on the effects of land-tenure systems in colonial India is an influential example of how “institutions matter.” Using both a district border design and an instrument for the British date of conquest of a district, BI show that districts where the British adopted zamindari or landlord-owned systems, as opposed to an owner-cultivator system, have worse developmental outcomes in the present day. They suggest (but don’t test) that the mechanism through which institutions persist is the inter-class conflict they generated and which continued to characterize the politics of these districts, long after landlordism was abolished. In landlord areas, the newly empowered masses likely continued to view the local elites as oppressors as depicted in the iconic 1953 film “Do Bhiga Zameen” (two bhigas of land) referenced above.
While some have faulted BI’s coding of what they classify as landlord districts, others have contended that the real action lay with the British decision to use direct or indirect rule across India. These papers are important and offer a nuanced picture of the British project in specific provinces. Largely, however, studies agree that BI were indeed on to something when they uncovered the link between the empowerment of landlords and long-run development. Which brings me back to questions I raised earlier: what are the exact mechanisms through which these institutional effects persisted, and in particular, how did landlordism shape future development?
Two recent studies by Alex Lee offer some useful answers and raise interesting new dilemmas. Lee (2019) shows that land-tenure systems led to very different levels of state capacity in the colonial era. In landlord districts, the British state invested a great deal less in revenue collection, and law-and-order institutions, relying instead on landlords to undertake these important administrative functions. In contrast, raiyatwari or individual cultivator-based districts experienced the full force of the British colonial project. From village-level revenue inspectors, to a local-level police constabulary, these districts witnessed ground-up capacity building. Using measures of per capita government officials employed by the British and the size of district-level tax collections, Lee first shows that non-landlord districts had higher levels of state capacity compared with landlord districts in the colonial era (Figure 1).
Next, when controlling for measures of state capacity in pre-independent India, the effects of land tenure systems on contemporary outcomes largely disappear. Lee takes care to note that land tenure systems may have mattered, but they mattered through how they mediated the development of state capacity rather than through the increased salience of class politics in the present day. What has persisted are the effects of weak state capacity not the original land ownership structures. Lee’s findings also resonate with those by Verghese, who uses process-tracing to compare districts in the Travancore princely state (indirect-rule) with the Malabar and Tinnevelly districts of the Madras Presidency (direct-rule) to show that where princely kingdoms empowered middle-caste farmers, they were already providing better public goods in the early 20thcentury than in districts where the British state sided with landlords as local intermediaries. Taken together, these findings suggest different reasons for, and solutions to the problems of development in contemporary India. The continued neglect of rural state capacity — in the form of village-level officials, police or local-level taxation — might be the reason for weaker development and not the collective action challenges presented by class conflict.
In a new working paper Lee raises a different question: to what extent did land tenure institutions matter to village-level outcomes in contemporary India? He shifts the unit of analysis from the district-level to the village-level because the village was the original “unit of treatment” for these institutions that set the terms of taxation between the British and Indians. He focuses on a single district, Agra, in the former United Provinces (Uttar Pradesh today). In this district, villages either had landlord (zamindari) or village-level (mahalwarii) responsibility for land taxation. He finds that in the 336 villages in his sample, land tenure institutions have no effect on public goods provisions in most of post-independence India.
What should we make of this null finding? It is possible that the real institutional innovation under the British might have been the adoption of individual cultivator-based rights (raiyatwari) and that both village-owned and landlord proprietary systems were equally likely to be poorly performing. Alternatively, it is possible that the level of analysis really matters. Focusing on the district might have overstated the case for land tenure systems in the first place, conflating district-level characteristics with institutional effects.
One further aspect that remains untapped in studies of Indian political and economic development is how much democratic institutions established in the late colonial era mattered to the persistent effects of early colonial economic institutions. In November 1920 Indians went to elections for the first time to choose provincial legislators who would have new powers to legislate and spend on public goods. In the 100 years since we know little about the characteristics of those elected, what their priorities were, and how much land tenure institutions shaped early party politics. It is possible that just like Lee finds that state capacity was mediated by land-tenure institutions, so was political capacity. The early development of party systems and the political agency of Indians rather than class inequality might have shaped post-independence politics and government performance. I will be back in my next post to talk about colonial era party politics and what we can learn from this period on why and how colonial institutions matter. See you next time!