VoxEU Column International trade

New roads to export: Insights from the Inca roads

Expanding road infrastructure is often justified on the basis of its presumed effects on exports. Yet, available evidence on to what extent these effects really materialise is very limited due to difficulties faced in convincingly identifying true casual relationships. Historical road networks can help overcome this endogeneity challenge. This column provides evidence for Peru based on the Inca road network and suggests that improvements in road infrastructure have had a significant impact on firms’ exports and thereby on job creation.

In policy circles domestic transport infrastructure is seen as a key determinant of exports. More precisely, among policymakers, there is a well rooted idea according to which new roads can generate increased exports. Statements in official documents introducing public export plans of developed and developing countries alike are illustrative in this regard. The report presenting the US National Export Initiative 2011 is a clear example. It states that “American businesses cannot participate in the global economy if they cannot get their products out the door … Deficiencies throughout America’s transportation system … severely impact the ability of businesses to transport their goods to global markets …”. Similarly, in Peru, the Plan Estratégico Nacional Exportador (National Strategic Export Plan) 2003-2013 states “The exporting sector … is one of the most affected by the infrastructure deficit, generally in transportation, which generates competitiveness losses relative to competing countries …”.

The identification challenge: What can be done?

Assessing whether such presumption in export plans holds true requires establishing causality. This is challenging because endogeneity is likely to affect the relationship between internal infrastructure and exporting. Road improvements might foster exports from regions receiving infrastructure investments. On the other side, it is also equally possible that increasing exports results in investment in these regions to reduce transport costs.

In order to address this potential endogeneity one has to resort to an exogenous source of variation in infrastructure, or instrument, which is not directly correlated with the economic outcome of interest but only indirectly through actual infrastructure. Several strategies have been recently proposed in the literature. Thus, for instance, Baum-Snow (2007) and Michaels (2008) examine the effects of highways on suburbanisation and relative demand for skilled workers, respectively, using the variation associated with the plans for the US Interstate Highway System, which were designed to strengthen national defence and improve connectivity among major metropolitan areas and with Canada and Mexico. In addition to these instruments, Duranton and Turner (2012) and Duranton et al. (2013) take advantage of historical data on the railroad network and the exploration routes between 1528 and 1850 to analyse the effects of interstate highways on the growth of major US cities and weight and value of bilateral domestic trade between large cities in this country, respectively. As for the impact of internal road transport network on exports, Volpe Martincus and Blyde (2013) estimate the primary – short-run – effects of domestic transport infrastructure on firms’ exports exploiting the random and exogenous variation in this infrastructure ('disappearance of roads') associated with the earthquake that took place in Chile in February of 2010.

The Inca roads and the current Peruvian transportation network

Between 2003 and 2010, more than 5,000 kilometres of new routes were constructed in Peru. This roughly amounted to a 14% net expansion in the country’s main (national and departmental) road network, and was made it possible through a 610% increase in the public resources allocated to road infrastructure over the same period, which accounted for 1.3% of the country’s GDP in 2010 up from 0.5% in 2003 (Figure 1).

Figure 1. Evolution of the total length of the road network and public investment in road infrastructure

Source: Volpe Martincus et al. (2013). Length is measured in kilometres. Investment is reported in $ millions.

New roads were asymmetrically distributed across regions. As a consequence, depending on the route(s) from the plants to the ports, airports or borders (hereafter ports) used in exiting the country, available transport infrastructure increased and distance travelled and internal transport costs incurred diminished for some exports while those for others remained the same. As discussed above, infrastructure improvements and spatial allocation of new roads in particular could be endogenous to exporting. In order to address this potential endogeneity, we (Volpe Martincus et al. 2013) use the Inca road network as an exogenous source of variation in this transport infrastructure to identify its impact on firms’ exports. Besides geography, historical, political, military, cultural and religious factors accounted for the spatial pattern of the Inca roads, which were thus built up for reasons entirely disconnected with current foreign trade. Also important, these roads are a good predictor of current road infrastructure (Figure 2). 

Figure 2. The current road networks and the Inca road network

Source: Volpe Martincus et al. (2013). Additions to the road network between 2003 and 2010 are coloured in black. The Inca road network appears in red.

Further, including an appropriate set of control variables, extensive tests show that it can be safely considered that the Inca road network affects today’s exports only through its correlation with the spatial allocation of new routes. In particular, the change in road infrastructure between 2003 and 2010 is instrumented with both the distance from the geographical origin of the exports to the nearest road that was part of the aforementioned network and the distance between this origin and current customs that could have been travelled along roads in this network.

Do new roads actually result in increased exports?

In examining the effect of this new infrastructure on firms’ exports, the analysis exploits two main datasets, i.e. transaction-level export data for the entire universe of exporting firms in 2003 and 2010, which inform the firm’s tax ID, the geographical origin of the flow, the product code (6-digit HS), the port through which the good exits Peru, the destination country, the export value in dollars, and the quantity (weight) in kilograms- and data on employment, sector of activity, and starting date for all these firms from the Peru’s National Tax Agency, SUNAT.

Estimation results based on data at the firm-product-destination level indicate that, as maintained in export plans, domestic road infrastructure has had a strong positive impact on firms’ exports. In particular, according to the instrumental variables estimate, the rate of growth of those exports whose routes to the main port experienced a reduction in their length due to the construction of new roads has been 39.8% higher than that of their counterparts whose route length remained the same over the period 2003-2010 (Figure 3). A simple back-of-the-envelope calculation reveals that, in the absence of changes in domestic road infrastructure, total exports would have been roughly 20% smaller in 2010. This finding is robust to several checks, including restricting the sample to differentiated products, using alternative specifications of the estimating equation, accounting for endogeneous location of firms and import behaviour patterns, incorporating additional instruments, and performing of falsification tests, among many others.

Figure 3. The estimated effects of new roads on firm exports, 2003-2010

Source: Volpe Martincus et a;l. (2013). ** means significant at the 5% level. Estimates that are not significant at the 10% level are reported as equal to zero.

As for the channels through which this effect arises, estimation results reveal that expansion in transport infrastructure, by leading to decreased distances to ports and therewith to lower transport costs, has translated into increased exports primarily through a larger number of shipments and thereby larger quantities shipped, but does not seem to have influenced unit values or the size of the shipments (Figure 3).

Is it only about exports?

Export plans also argue that increased exports made possible through better infrastructure will lead to job creation. Thus, in the words of the Peruvian export plan 2003-2013, “ … In a context of increasing economic globalisation, the exporting sector plays a fundamental role as a growth engine, in generating employment, and in fostering the development of nations … ”. In order to assess whether and how exports associated with infrastructure improvements influence employment, the predicted values from the estimated export equation are then used as a an instrument for the actual change in exports in an equation in which changes in employment is explained by changes in exports after conditioning by relevant covariates. Instrumental variables estimates suggest that a 10% increase in (the rate of growth of) exports -driven by improved transport infrastructure- leads to a 4% increase in the (rate of growth of the) number of employees. This estimated elasticity is similar to that estimated in Park et al. (2010) on a sample of Chinese firms over the period 1995-2000 using exchange-rate shocks as instruments for changes in exports.


Public investments in transportation infrastructure are often justified by arguing that resulting new roads or improved roads would help firms expand their exports. Larger foreign sales would then foster job creation. Robust evidence from Peru based on exploiting the Inca roads as an exogenous source of variation in current road infrastructure confirms that new roads have actually made it possible for firms to increase their exports. Furthermore, this deepened firms’ penetration of foreign markets seems to have in fact been associated with higher firms’ employment levels.

Note: The views and interpretations in this note are strictly those of the authors and should not be attributed to the Inter-American Development Bank, the World Bank, their executive directors, their member countries, Ministerio de Transporte y Comunicaciones, PROMPERU, or SUNAT. Other usual disclaimers also apply.


Baum-Snow, N (2007), "Did highways cause suburbanization?", Quarterly Journal of Economics, 122, 2.

Duranton, G and M Turner (2012), "Urban growth and transportation", The Review of Economic Studies, 79, 4.

Duranton G, P Morrow and M Turner (2013), "Roads and trade: Evidence from the US", University of Toronto, mimeo.

Michaels, M (2008), "The effect of trade on the demand for skill: Evidence from the Interstate Highway System", Review of Economics and Statistics, 90, 4.

Park, A D Yang, X Shi, and Y Jiang (2010), "Exporting and firm performance: Chinese exporters and the Asian financial crisis", Review of Economics and Statistics, 92, 4.

Volpe Martincus, C and J Blyde (2013), "Shaky roads and trembling exports: Assessing the trade effects of domestic infrastructure using a natural experiment", Journal of International Economics 90, 1.

Volpe Martincus C, J Carballo and A Cusolito (2013), "Routes, exports and employment in developing countries: Following the trace of the Inca Roads", Inter-American Development Bank, mimeo.

2,310 Reads