Highway infrastructure and economic development: Measuring causal impacts of infrastructure investments using a three-step instrumental variable identification strategy

This paper provides an original third-step identification strategy using instrumental variables to evaluate the causal impact of highway investments on the local economy. First, we construct a novel national highway dataset at the municipal level in Brazil using the Growth Acceleration Program (PAC) (2007–2018) as a case study. Second, we rely on some of the main infrastructure project costs to propose several cost-related instruments to correct for measurement errors in the road variables. Third, we circumvent the omitted variable bias from the non-random placement of roads by building instruments based on global cost minimization methods, historical plans, and the propensity of a municipality to receive highway interventions. Our identification strategy allows us to identify relevant biases coming from both measurement error and omitted variables. Our preferred estimates point out a reliable road elasticity in the range of 0.011–0.017. From this, we calculate a non-biased return rate to highway infrastructure of 21.3% in Brazil, proving the high rentability of those investments in the developing world context.