Julieta Caunedo is an Assistant Professor of Economics at Cornell University. She also serves as an affiliated researcher at CEPR, Y-RISE, and ATAI, and a theme leader for STEG. Her research interests include macroeconomics and development, with emphasis on the impact of technology for productivity and the labor market.
Economic activity in developing countries is labor intensive. Most workers are employed in agriculture, and the majority of farming is family-run and low scale.[1] Indeed, family farming accounts for 80% of land-holdings in low and lower middle-income countries, as reported by the Food and Agriculture Organization of the United Nation’s World Census of Agriculture. The presence of moral hazard problems in labor markets attaches farmers to their land, devoting substantial managerial time to supervising workers, and passing on seemingly profitable opportunities in non-agriculture.
A long tradition in development economics argues that an essential condition for economic development is the increase in agricultural productivity, which reduces the demand for labor in agriculture and releases workers to other sectors of the economy. The mechanization of production has become a primary feature of modern agriculture and my own prior work shows that it is central to agricultural labor productivity growth.[2]
In response to the observed correlation between mechanized practices and agricultural productivity, governments in the developing world are increasingly intervening in the agricultural sector to subsidize mechanization. At the same time, private companies providing rental equipment services are expanding throughout Africa and Asia. But to date, little is known about the magnitude of the returns to mechanization, and the causal channels through which mechanization shifts labor supply for farming households and overall labor demand in agriculture.
In partnership with one of the biggest providers of rental agricultural equipment in India, we conducted a randomized control trial to increase access to rental markets for mechanization covering 7,100 farmers across nearly 200 villages in the state of Karnataka.[3] Farmers were given a chance through a lottery for subsidy vouchers that allowed them to access approximately a third of the average mechanization hours over the agricultural season. Vouchers were valid for redemption throughout the season, allowing farmers to optimally allocate the use of equipment across agricultural processes. Farmers could rent any equipment available at a custom hiring center, including tractors, rotavators, and cultivators. A subset of treatment farmers were given part of the value of the vouchers in the form of a cash transfer. These cash transfers help to disentangle income effects associated to the decline in the cost of capital and to measure liquidity frictions that may affect mechanization take-up.
We collected detailed data on inputs and output by production process, e.g. plant preparation, harvesting, etc., before and after the intervention. Importantly, we collected data on task engagement for different members of the household, as well as for hired labor. This information was combined with administrative data from our implementation partner, which contains the universe of transactions across all their custom hiring centers including hours of service, type of equipment, area serviced, prices, and repeated interactions.
We found that treatment farmers are 30 percentage points more likely than control farmers to rent agricultural equipment from the custom hiring centers. Treatment farmers increase mechanization of the their fields by an additional hour relative to control farmers, and cash transfers have no significant effect on these responses. This mechanization occurs disproportionally at the land-preparation stage. We find that mechanization lowers labor demand across all farming processes, and that the magnitude of the labor saved is larger in processes not being mechanized. Mechanization lowers family labor engaged in worker supervision, consistently with output standardization. The impact of mechanization on farmers’ managerial time is a novel channel for the transformative role of capital intensification on labor-intensive activities. Importantly, we show that family workers take on non-agricultural opportunities in response to the subsidy.
We use the behavioral responses from the experiment to discipline the structural model of endogenous task replacement. The model allows us to estimate key outcomes that are relevant for one to think about the role of subsidized mechanization services at scale. For example, the shadow value of family labor in the farm, which determines family-labor incentives to switch into non-agriculture jobs when worker supervision needs decline in the farm.
Finally, we use the model to assess farmers’ welfare changes from the intervention. Because the interventions shift incentives to work in non-agriculture and farming households’ optimal leisure allocation across all processes, income changes are not sufficient to assess welfare. We construct a measure of consumption-equivalent welfare for the average farmer and find that the intervention raised welfare by 1.5%. The main contributors to the welfare gains are the changes in total factor productivity mentioned above, followed by the improvement in the span-of-control in the farm.
Evidence on the path to mechanization for currently rich economies suggests that equipment rental markets were a stepping-stone to that process. Rental markets for equipment are growing rapidly in the developing world and they provide an important opportunity to release labor from the farms into non-agriculture. While our experimental evidence suggests important gains from mechanization (both in terms of welfare and income), the impact of these policies at scale remains to be assessed.
[1] Herrendorf, B., Rogerson, R., and Valentinyi, A. (2014). Growth and structural transformation. In Handbook of Economic Growth, volume 2, chapter 06, pages 855–941. Elsevier, 1st edition.
Adamopoulos, T. and Restuccia, D. (2014). The size distribution of farms and international productivity differences. American Economic Review, 104(6):1667–97.
[2] Caunedo, J. and Keller, E. (2021). Capital obsolescence and agricultural productivity. The Quarterly Journal of Economics, volume 136, no. 1, pages 505-561
[3] Caunedo, J. and Kala, N. (2021). Mechanizing agriculture. Mimeo.
