Guest post by Gedeon Lim
This is the thirteenth in this year’s series of posts by PhD students on the job market
Much of modern development efforts are channelled through local government structures. In 2020, for example, 248,160 Gram Panchayats in India will receive transfers of $8.3 billion; 41,913 Barangays in the Philippines will receive $2.5 billion; and in Indonesia, 74,000 rural villages have received annual transfers of $5.1 billion since 2014. Yet, despite the amount of resources devolved to local government, chief executives – Sarpanch (Gram Panchayat heads), Barangay captains, or village chiefs – are rarely paid a living wage for their time as local leaders.
Given that local leaders play important roles, paying them a decent salary might be an important first step in bringing about better performance and development. In my job market paper, I study the following question: In the context of locally elected leaders, how do differences in the size and nature of pay to these local politicians affect long-run economic development? I answer this using an historical natural experiment in West Java, Indonesia. In the nineteenth century, this homogeneous region was split, and in one part (90 villages in Cirebon) chiefs were awarded cultivation rights over village rice land (tanah bengkok), but not in the other (103 villages in Priangan), see the map below. To investigate the effects of this differentiation on development, I spent a year in Indonesia conducting a retrospective survey of 965 respondents in these 193 villages.
I find that villages where chiefs received higher land rents are richer and more developed today. Why? I find evidence consistent with historical selection and contemporary alignment of chief incentives. Higher land rents attracted historically chiefs who built more village schools in the recent past. So much so, that chief candidates today in these villages are not more educated than the average villager. Furthermore, chiefs today in such areas have weak re-election incentives: on average, only 17% of chiefs rerun for office. And yet, chiefs today continue to perform better, but why? Given the lack of re-election incentives and selection on education, I propose a third theory: the within-village nature of bengkok land works like stock options. Cultivation rights over village rice land gives chiefs a stake in local village development, and aligns the incentives of chiefs with that of villagers.
Studying impact of variation in pay from cultivation rights over rice land, tanah bengkok
Exogenous variation in political pay, especially at the local level, is hard to come by. The traditional institution of tanah bengkok in Indonesia provides plausibly exogenous differences in land rents. Tanah bengkok are cultivation rights over rice land awarded to a village chief as compensation for his (almost always men) services. The chief, in turn, is obligated to sharecrop or lease out the land to villagers at fixed prices. Villagers harvest the rice for the chief and sales from the rice harvest serve as his rents.
Importantly, higher rents are unlikely to work without robust disciplining mechanisms. This is where Indonesia, again, presents a nice setting. Chiefs have been democratically elected since Dutch colonial times. Importantly, for interpretation of later results, all chiefs before 1979 were elected for life. After 1979, however, chiefs today are elected to 6-8 year terms. Chiefs also play an important role in local development. For instance, more educated chiefs can construct more public goods through greater lobbying efforts.
To study this, between January – May 2019 present and ex-village chiefs were interviewed in 193 villages. If a chief was no longer alive or unable to communicate, we interviewed village elders or officials who were alive during that chief’s rule and familiar with his rule in the village. During our interviews, not only do we ask respondents about chiefs, we also ask them about the education and occupation of all candidates that ever stood for elections. Thousands of hours of qualitative interviews were coded, resulting in a dataset on the electoral history of 931 chiefs and all 2,291 candidates that ever stood for office between 1979-2014.
From survey data, I find that chiefs in bengkok villages receive an annual compensation (34.4 million IDR or $2,293) roughly three times more than non-bengkok chiefs (12.2 million IDR or $813). The bulk of these differences is driven by returns from the bengkok rice harvest. In terms of outside options, annual compensation for bengkok chiefs are roughly equivalent to the wages of a mid-level civil servant (33.6 million IDR or $2,241) whereas non-bengkok have an annual compensation closer to that of a day labourer (18 million RP or $1180).
Having higher land rents for leaders made villages richer and more developed today
To estimate the causal impact of higher land rents, I use a fuzzy spatial regression discontinuity design. This design exploits the discontinuous jump in tanah bengkok across the historical Priangan-Cirebon border in West Java. This border did not exist before Dutch intervention, and the discontinuity in tanah bengkok arose from 1830 Dutch efforts to co-opt village chiefs based on perceptions of pre-existing practices. Hence, Dutch perceptions that tanah bengkok was a traditional practice in Cirebon (it existed in the central areas away from the Priangan-Cirebon border) but not Priangan led them to impose tanah bengkok on one side (Cirebon) but not the other (Priangan).
I find that higher bengkok land rents cause stronger nighttime light intensity (as a measure of economic development), lower child mortality, and higher years of education among the population. These villages have more upper-level government and villager funds (attained from informal and formal lobbying upper-level government agencies). An increase in 1 hectare of bengkok leads to a 0.22 standard deviation increase in village public goods such as access to roads.
Why do chiefs perform better? Historical selection
Better educated leaders can be part of the story. An increase in 1 hectare of bengkok leads to the candidate pool having 0.3 years more education. Candidates are also 3.6 percentage points more likely to have previous occupations as career bureaucrats. Controlling for average education of villagers, however, these effects disappear.
This suggests a role for historical political selection. I document that in bengkok villages, villager education from cohorts born in 1920 till today are consistently more educated than those in non-bengkok villages. This is a result of more historical village schools constructed through greater chief efforts. Chiefs elected to office before 1979 were elected for life and the absence of re-election incentives suggests that higher land rents attracted chiefs of higher quality. Positively selected chiefs were so effective at school construction that chiefs today are no longer more educated or elite than the average villager.
Given the lack of modern-day selection and re-election incentives across the border (only 17% of chiefs rerun for office and this is not different across borders), persistently positive contemporary development suggests a third mechanism: the within-village nature of bengkok might be important. Bengkok potentially functions like stock options for CEOs by incentivizing chiefs to invest in their village. Bengkok rice land is only as productive as, for example, village roads and infrastructure that support rice sales.
Greater roads construction is suggestive of this interpretation. I do not, however, have a direct measure of chief incentives towards villager interests. Instead, I test for orientation towards elite interests. Weaker orientation towards elite interests might signify stronger orientation towards villager interests, especially in the context of positive long-run development outcomes presented above. As a proxy for orientation towards elite interests, I test for the likelihood that a village voted for Golkar, President Suharto’s party, in the first post-democratization elections in 1999 (Suharto was the longest-standing president of Indonesia from 1965-1999). I find that bengkok villages are 18 percentage points less likely to vote for Golkar. Given rich literature documenting the role of chiefs in turning out the vote, these findings are consistent with bengkok rents attracting chiefs who are less dependent on supra-village interests. In turn, these chiefs consistently act in the best interests of the village.
Under the 2014 Village Law Fund (Undang-Undang Dana Desa 2014), villages receive annual direct transfers of US$70,000 to individual village bank accounts for development purposes, broadly construed. Chief compensation, however, has not kept pace with these large increases in funds (chiefs today, in some cases, get an irregular monthly salary of $200). Moreover, my findings suggest that optimal design of compensation for these leaders should take into account not only the level, but the nature of compensation. My research also highlights benefits for development if broader attempts to raise the compensation of local leaders in India and Africa succeed and provides conditions under which such measures might be more effective.