Discussion Paper

No. 2012-54 | October 18, 2012
Determinants of Transport Costs: Are They Uniform across Countries?

Abstract

The author shows with pooled OLS estimations based on transport margins from international social accounting data that investments in improved road infra¬structure have the potential to significantly reduce transport costs. However, this result can only be clearly confirmed for industrial countries and is of primary importance for production and transportation of agricultural goods. For developing and transition countries, in contrast, the author finds other determinants such as weather conditions to be more important in determining transport costs. A key variable, especially in these countries, is corruption. Very high corruption has the potential to prevent positive effects from road infrastructure on transport costs or to even reverse them. This paper contributes to the literature on infrastructure investment by introducing and applying an internationally comparable measure of transport costs which can be calculated for a large and growing number of countries. The author concludes that investments in transport infrastructure can have substantial positive effects especially on agricultural production and the efficient marketing of agricultural products but only if specific additional conditions are given.

JEL Classification

O18 O11 R42

Cite As

Hannah Schürenberg-Frosch (2012). Determinants of Transport Costs: Are They Uniform across Countries?. Economics Discussion Papers, No 2012-54, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2012-54

Assessment



Comments and Questions


Anonymous - Referee Report 1
October 26, 2012 - 09:40

It is an interesting idea to use data on cost shares of transport for various sectors to measure the impacts of road improvements.

Data limitations may well affect some of the main outcomes. In particular the transport density index does not take into account the spatial structure of ...[more]

... the country. Also the quality of roads is most probably not well represented in the data. Including the corruption index is a promising way to provide a proxy for the quality of infrastructure via the lack of maintenance. For the interpretation it is, however, important to note that corruption may well affect all costs of production, not only the transport costs. When transport costs and other costs would be affected to the same extent, cost shares would remain the same.

The paper says little on sectoral composition in the various countries. Countries with a high share in the economy of sectors that strongly depend on transport will yield high values of the cost shares. It is therefore recommended to apply shift share analysis in order to correct for the sectoral composition of the economy before doing the regression.

The finding that weather has an impact on transport cost shares is interesting, although the interpretation is not unambiguous. It is indeed possible that weather affects the functioning of the transport system and hence the transport cost share. But there is also another possibility, i.e., via the production process itself. Production in agriculture but also in other sectors will depend on weather in general and extreme events in particular and this may well be another route where weather affects cost shares.


Hannah Schürenberg-Frosch - limitations
October 26, 2012 - 13:25

Of course data limitations might have an influence on the outcome especially due to the very heterogeneous set of countries. I have tried to do an extensive robustness analysis by including other variables, excluding outliers, including interaction terms instead of splitting the sample. Still, the analysis requires a large data ...[more]

... set and is thus limited as far as data quality and availability is limited.
If the weather influences other aspects of the production process as well as transport costs, this would impact on both the numerator and denominator of the transport cost share (as total production costs enter the denominator). Hence I would suppose that the transport effect would rather be underestimated than overestimated in this case.
It is right that economies with a higher share of transport intensive industries would also have higher transport cost shares. However, I am not sure whether it would be the right procedure to correct for this. These economies would also have a higher requirement for road networks and probably a good road network would be of more effect in these than in other, non-transport-intensive economies.
I have been on the search for a good measure for spatial dispersion for a long time but have not found any good measure that would be available for a large number of states. If I could get good data on spatial dispersion I would very happily include such a measure in my estimations.


Johannes Tiemer - Details
November 09, 2012 - 10:55

Disclaimer: I'm a colleague of the author, so I am potentially biased. Still I would like to comment on the paper and will do my best to measure the author's paper to accepted standards.

The paper is all in all well written, but some points do still need work ...[more]

... in my opinion:
I agree with most points of the first comment and thus do not restate them. The author already explained to me the inherent data-problems especially with regard to African countries. The reasons for these shortcomings were pointed out to me in a fashion that made me tend to agree that improvements in this part of the paper might entail disproportionate effort.
My suggestions for the paper to become more valuable as a contribution to its field are to go into much more detail regarding its limitations. The fact that rich data on development countries is hard to come by is well known. However not all authors in the field do have extensive first hand experience of the local conditions and customs. Providing scholars with more detailed descriptions of the data problems, suggested remedies a.k.a. the solutions used in this paper, would render this paper not only a contribution to empirical research, but also make it a helpful reference for future data-driven research.
As an example for this I would like to point out the way climate conditions are accounted for in the paper. With a more detailed explanation the reasons for using the index used over other fluctuation measures like variance become more accessible and thus more helpful.


Hannah Schürenberg-Frosch - details on data
December 19, 2012 - 13:48

In fact data limitations are not only found for African countries, in general transport cost/transport spending/transport usage data is not very rich. Even most industrial countries have no data on intranational transport costs. Transport data is often limited to international transport or to specific ways of transport. Thus, of course ...[more]

... one could go into much more detail at this point and this paper provides only one possible alternative, there might be others of course.


Anonymous - Referee Report 2
December 18, 2012 - 08:53

see attached file


Hannah Schürenberg-Frosch - Limits to interpretation
December 19, 2012 - 14:26

This paper aims at providing a possible alternative measure for transport spending requirements in production in different countries, given the fact that the referred solution to use cost/ton or a comparable direct cost measure is not possible in an analysis that is intended to cover a broad range of different ...[more]

... countries. Transport cost or transport spending data, even data on the availability and cost of vehicles is very poor, once you leave the OECD countries. Even for many industrial or middle income countries exact cost measures for transport costs are not available. Of course there exists data that covers international transport and transaction costs, but this paper explicitly wants to include intranational transport as well. I do of course agree that a direct cost measure would be preferable if it was available for a significant number of countries.

As this paper explicitly searches for structural differences between country groups it was inevitable to search for an alternative measure. Of course this is neither the only measure one could think of, nor do I state that it is better to use this measure compared to a direct cost/ton measure. It is just the best measure I could think and which is based on freely available data.

I am aware that input-output-tables are very complicated datasets that undergo various procedures that might bias the data. I have tried to take as many tables as possible from the same data provider (OECD for the industrial countries and IFPRI for the developing countries) and only considered alternative data providers if I could not get a table from one of these databases. Thus I hope to minimize the bias in the result due to differences in data collection and input-output accounting as e.g. the OECD intends to provide tables which are comparable across the respective countries included in the database.

I include not only the transport spending reported as "demand for transport services" but also transport related costs and vehicle maintenance and related costs in the calculation of the margin with the intention to include as much of the intra-firm transport costs as possible in the measure.

Of course the measure is inappropriate in the case where no transport at all takes place in one period and a decrease in transport costs then leads to an increase in the margin in the next period due to a strong increase in transportation. However, as my analysis is performed for the national level, this is a rather unrealistic example even for very low income countries.

For those variables that have been used as explanatory variables in other studies I will try to add references, however I only found a very limited number of references that could possibly be cited here.

Corruption may increase transport costs e.g. if corrupt military or police officers frequently stop transport vehicles on their way and request to be bribed or even only control the vehicle but thus increase the transport time by far. These frequent controls are common in most low and middle income countries and lead to a lower efficiency of transportation. High corruption may also increase the costs for permits e.g. for driving a truck or operating a transportation company. This would lead to higher prices for transportation and thus higher transport costs. The indirect effect through higher ineffiency in road planning is one of many possible explanations one could think of for the interaction effect.

The specific comments would of course be dealt with in a revision of the paper.