Paving the Road to Export: Assessing the Trade Impact of Road Quality

Abstract

Assessing the trade impacts of domestic transport costs is data demanding and analyses that examine the effects of road quality, a critical aspect in regional and public policy, practically do not exist in the international trade literature. The few studies available rely mostly on distance-based measures as proxies of transport costs which impede analyzing the trade effects of transport-infrastructure improvements. In this paper, we combine highly disaggregated records of export flows with detailed geo-referenced information of the Colombian transport network, including its road quality, as well as real measures of transport costs of shipping goods within the country to measure the trade impacts of improving road quality. We find that the trade effects of improvements in road quality are relatively small on average; however, there is considerable heterogeneity in the magnitude of the effects. We show that longer routes have larger shares of their roads in poor conditions; accordingly, the trade impacts of shipments originated in remote regions are found to be quite substantial.

Keywords:

• Transport costs
• road quality
• regional exports

Jel Classifications:

• F10
• R10
• O2

Notes

¹Port here refers to ports, airports or borders.

²For instance, there might be differential use of imported inputs across sectors generating heterogeneous impacts via the transportation of those inputs. To the extent that the use of imported inputs is sector related, the commodity-year fixed effect captures this differential impact. For robustness, we also include a commodity-region-year fixed effect in order to account for the possibility that the differential impacts of imported inputs are not only sector-related but potentially sector and regionally related.

³We use the same surveys in this paper to obtain the operating costs of the truck industry in the country.

⁴While we could potentially use these transport costs in our analysis, they are limited to the cities analyzed by the Ministry. Using these figures would not only restrict the geographical extension of the analysis (for example, not all the departments are included in these data) but would also induce some potential biases because not all the cities with customs are included.

⁵Specifically we exclude the industry HS-2dig: 27.

⁶For details see: http://www.wto.org/english/res_e/statis_e/technotes_e.htm

⁷Detailed results available from the author upon request.

⁸It is worth mentioning that the reduction in transport costs generated by this simulated improvement in road quality is likely to be a lower bound for two reasons: first, we do not have information on the road quality on the secondary or tertiary road networks of the country even though they are used – together with the primary road network – to select the least-cost routes. For all the roads on the secondary and tertiary networks we have made the conservative assumption that they are all in good condition. Second, the primary road network with information about road quality is limited to the roads under the jurisdiction of INVIAS, the public entity in charge of the roadways. This excludes the roads under concession contracts with the private sector. Roads under concession contracts in the primary network represent 17% of all the roads. Once again, we have made the conservative assumption that all these roads are in good condition.

This is the C-2 configuration: a 2-axis truck with a capacity to carry 9 tons. In 2005, trucks with this configuration accounted for 82% of the cargo fleet in the country and carried around 50% of all the land cargo (in tons) according to the Colombian Ministry of Transport.

Obtained from estimates by the Ministry of Transportation.

The value of n is low for roads where the dominating roughness amplitudes have short wavelengths, such as on a modern designed highway with a deteriorated surface with plenty of potholes. The value of n is high for roads where the dominating roughness amplitudes have long wavelengths, such as on an ancient designed rural low volume road (Ahlin and Granlund, 2002).

While there is consensus in the literature that maintenance, tire, repair and depreciation costs are affected by roadway conditions, the effect on fuel consumption is less clear. Many argue, for example, that there is no measurable difference in fuel consumption on paved roads of different roughness.

Note that this is a conservative increase because it is likely that the increase in transport costs when the road has an IRI in the 8–14 range (bad) will be higher than when the road has an IRI in the 3–8 range (regular). However, we apply the same 25% increase for both regular and bad roads because we do not have information on how much higher these costs should be in bad roads as the analysis in Barnes and Langworthy (2003) do not differentiate between road conditions for IRI's higher than 2.7.