New bridges generated a 19% internal rate of return by mitigating the loss of wage income due to floods and inducing spillovers on farm productivity
Infrastructure is a major concern for policymakers, with billions of dollars spent by governments and international organisations. Studying the effects of infrastructure can be difficult, as the high profile and cost of such projects tends to incentivise placement in selected areas to maximise benefits. Moreover, despite some intensive data collection efforts (Asher and Novosad 2018 is one particularly impressive example), researchers are usually forced to rely on pre-existing data to study impact. This limits the scope of outcomes one can study, which is a potentially important constraint given that infrastructure impacts a variety of economic margins.
In Brooks and Donovan (2018), we investigate a relatively understudied benefit of infrastructure: providing rural villages access to larger, higher-paying labour markets in nearby towns. Because rural households tend to have multiple, intertwined income streams (Foster and Rosenzweig 2007), we consider the potential spillover effects back in the village, including changes in village prices and choices made on farms.
Context: Northern Nicaragua
We study these issues in the context of rural villages in northern Nicaragua, which are subject to seasonal rains from May to November. These rains flood nearby rivers and prohibit villagers’ ability to access higher-pay day labourer jobs in nearby towns. While flooding is not continuous, it is frequent. A village in our study experiences at least one day of flooding in nearly half the weeks in the rainy season, with an average flood lasting for five days. When the riverbeds are not flooded, they are easily crossable on foot, vehicle, or horseback.
From a research perspective, floods provide an observable and external (exogenous) shock that we exploit to study how income varies with changes in market access. Using bi-weekly data collected from villagers over the course of two years, we find that floods have a large negative effect on earnings from wage work. Specifically, a flood induces 18% lower wage earnings, while it increases the likelihood of earning no wage income from 25% to 32% compared to a week without a flood. The top panel of Figure 1 plots the distribution of income in flood and non-flood weeks.
Figure 1 Distribution of wage earnings before bridge construction
The bottom panel of Figure 1 shows why we might expect important spillovers; despite the fact that these are farming communities, nearly all households have at least some non-farm wage income. Changing access to wage opportunities therefore potentially impacts a large number of farmers.
Our intervention: Building bridges
To study the cost of inaccessible urban markets, we partnered with the NGO Bridges to Prosperity (B2P) to build footbridges over these riverbeds. These bridges are technically sophisticated and resilient structures that can hold people, cattle, motorcycles, and carts without risking failure during flash floods. Because of our participation in the construction process, we were able to collect three years of data starting a year before the bridges were built, which included agricultural production, wage earnings, asset holdings, occupational composition, and other attributes generally difficult to collect in such studies.
We compare villages that received a bridge with other villages that had similar need for a bridge but where construction was infeasible due to the characteristics of their riverbed. This was driven by issues like uneven or sufficiently wide riverbeds such that B2P could not build a bridge. Crucially, we show that these small topographical features are unrelated to any characteristics in the village, thereby allowing us to interpret any changes between these villages as the causal effect of the bridge.
Direct effect: Eliminating market access risk
These bridges indeed mitigated the risk of accessing towns, as the introduction of a bridge eliminated the aforementioned 18% decline in wage earnings during floods. This is particularly important given that daily wages are approximately 30% higher for men than those in rural areas, who do the majority of this more urban labour work. One can see the difference in Figure 2 – after the construction of a bridge, the distributions of earnings are quite similar during flood and non-flood weeks (especially compared to Figure 1).
Figure 2 Distribution of wage earnings after bridge construction
Spillover effects in rural villages
While the direct effect on earnings is interesting in its own right, it is not the totality of the benefit. First, the movement of some villagers toward higher paying jobs outside the village put upward pressure on the local wage. In fact, the average wage gap between the village and nearby town was eliminated with the introduction of a bridge. Not only were some villagers able to access higher wage jobs, but those that stayed were able to reap the benefits of their neighbours leaving through the higher induced wages in the village.
The second spillover effect was on the farm, where fertilizer expenditures increased by 63%. At the same time, farm profit increased by 74%. It is worth highlighting this point – despite the fact that expenditures on the two main inputs increased (there is little capital use in these villages), profit still increased by 74%. This implies that there was some economically significant constraint holding back farmers that, once a bridge was built, allowed them to increase profitability while increasing costs. Put in slightly more technical language, these farmers had marginal revenue well above marginal cost before the bridges were built.
Mechanisms: Testing a theory of self-insurance
What are the underlying mechanisms generating the results? The most intuitive explanation is that bridges lowered trade costs, allowing farmers cheaper access to fertilizer or higher sale prices for crops. However, this is not the case in our context. Since the rainy season overlaps with the cropping season, fertilizer is purchased pre-rain and harvest sold post-rain, and therefore not subject to this risk.
We instead show that the increase in farm profit is closely linked to the ability to access new labour markets. In particular, the option to work in town allows villagers to generate income when times are bad, thereby freeing resources previously held as precautionary savings to be spent on productive farm activities. Indeed, we find that farmers decrease the fraction of their harvest stored for own consumption from 90% to 80%. Infrastructure, therefore, facilities self-insurance and unlocks more productive investments on the farm.
Cost-effectiveness and looking toward the future
How important are these effects? Our results imply that the average bridge generates a 19% internal rate of return. At a cost of US$40,000 per bridge, these bridges pay for themselves in about 5.5 years. Thus, this type of infrastructure is cost-effective and compares favourably to some of the more successful recent interventions measured by cost-effectiveness. The spillover effects highlighted here are critical to this calculation. A full third of this 19% internal rate of return is due to the increase in farm profit alone, and not capturing this margin would severely underestimate the impact of the bridges in our study.
We think there is a broader message to take from these results. Although our study focused on the interaction between labour market access and agricultural outcomes, there are still numerous facets of rural life that require more study. For example, women we surveyed consistently highlighted issues of maternal health as a key benefit from these bridges, though our sample size limited our ability to study such issues. Accounting for the full effect of infrastructure requires understanding these various channels, along with the effects induced on the markets to which these villages connect. These are big questions, and require developing interventions at a macro scale (see, for instance, the work by Y-RISE on the importance of studying such issues at scale). We hope our results demonstrate that these types of interventions can and should be studied, and not dismissed on cost alone. Infrastructure is effective, both in terms of absolute magnitudes and return on investment, once we start to properly account for the different margins impacted.
Asher, S and P Novosad (2018), “Rural roads and economic development”, Working Paper.
Brooks, W and K Donovan (2017), “Eliminating uncertainty in market access: The impact of new bridges in rural Nicaragua”, Working Paper.
Foster, A and M Rosenzweig (2007), “Economic development and the decline of agricultural employment”, in T P Schultz and J A Strauss (eds), Handbook of Development Economics, pp. 3051-3083.