For many years, the EU has prioritised the funding of large-scale transport infrastructure projects. Between 2007 and 2013 alone, the Trans-European Transport Network programme funded 348 projects at a cost of €7billion (TEN-T Executive Agency 2013). The goal of this investment has been to increase the integration of European markets. There is an interesting analogy here to Britain during the Industrial Revolution – when the advent of a turnpike road network (1760s), canal network (1790s) and railway network (1840s) changed the face of British transport (Bogart 2013).
What is market integration?
In both modern and historical transport revolutions, a common set of questions arises. What exactly do we mean by market integration? How should we measure it? To what extent did transport have a significant impact on market integration? Answering these questions is not as straightforward as one might suppose.
An obvious implication of integrated markets is that prices will be the same (or at least ‘similar’) everywhere: this is the so-called Law Of One Price. So a Fiat car should cost the same in the north of Scotland as it does in southern Italy, after adjusting for taxes. In reality, price differences will never be totally eliminated owing to transport costs; but lower transport costs – resulting from improved infrastructure – may compress price dispersion across Europe, for a given product. If it cost less to move Fiat cars to Scotland then the Scottish price would fall closer to the Italian price. So, across Europe, overall price dispersion would fall. We can thus track changes in market integration by examining how the cross section of prices changes over time; for example, does the standard deviation of the cross section of prices fall?
A lesson from history: Britain between 1770-1820
Jacks (2011) studied price dispersion across English county grain markets between 1770 and 1820 and concluded that they became less integrated over time: he put this down to the disruptive effects of the Revolutionary and Napoleonic Wars (1793-1815). Using the same data set, we find (2013) that price dispersion was indeed constant between 1770 and 1820, showing worryingly little benefit from improved transport links. However, it turns out that this is far from being the whole story.
A more subtle implication of integrated markets is that prices will move synchronously. When a price shock hits one market -- where a shock could be the arrival of new information, such as a change in government policy, or a real shock, such as a crop failure – then it will be transmitted to neighbouring markets. If transmission is fast then we say that markets are integrated. Following the seminal work of Ravallion (1986), it has become the norm amongst economists to measure market integration using ‘cointegration analysis’ (i.e. estimating an Error Correction Model for pairs of cities). This makes use of the time series variation in the data, as well in the cross sectional variation.
We implement the Error Correction Model approach for pairs of contiguous counties in England, such as Bedfordshire and Buckinghamshire, during the Industrial Revolution. One output statistic of the Error Correction Model is the ‘half-life’ of a shock. A shock to Bedfordshire prices (perhaps due to a harvest failure) drives a price wedge between Bedfordshire and Buckinghamshire; how long does it then take for half of the shock to be arbitraged away? A shorter half-life implies better-integrated markets, and it does indeed turn out that half-lives were falling between 1770 and 1820. Thus markets became more integrated in that sense.
But a further twist is that we show that the main effect of improved transport was not on the half-life (speed of adjustment), the statistic on which most analysts focus their attention. Two further aspects of the shocks also merit attention: the relative magnitudes of the shocks (do shocks tend to be the same size in Bedfordshire and Buckinghamshire?); and the correlation of the shocks (do shocks tend to hit at the same time?).
Surprisingly, the main effect of improved transport was to make shocks more similar in size. What is the intuition for this? Take two polar cases: London and Bristol can trade costlessly, or London and Bristol cannot trade at all (transport cost is prohibitively high). Suppose that the price of grain drops in London because the temporary Peace of Amiens arrives in 1801 and ships can now bring cheap Prussian grain from the Baltic. This will cause a large downward price shock in London, but what will happen in Bristol? If trade costs are infinite then there is no downward price shock in Bristol; if trade is costless then the downward price shock in Bristol will be exactly the same as in London. So falling transport costs (from infinity to zero) generate price shocks of increasingly similar magnitudes in different markets. This benefits consumers because it is a form of co-insurance: consumers lose more utility when prices rise very high than they gain when prices fall very low. So consumers benefit when trade is sufficiently low cost that shocks are spread more evenly across markets.
By contrast, the increase in the correlation of shocks over time in England was due to faster information transmission, rather than improved transport. It was stimulated by the spread of newspapers. If information is transmitted slowly then news arrives in one market substantially after another; price changes will occur with a lag and correlation between the two markets will be low. For example, grain markets were not open every day in England during the Industrial Revolution; in London it was Mondays, Wednesdays and Fridays and, in some markets, may have been only once per week. So if news arrived in Bedfordshire on Monday and was transmitted on to Buckinghamshire before the market there opened on (say) Wednesday, then the shocks would be highly correlated (both markets would move in the same direction in the same week). If the lag were longer then Buckinghamshire could react a week later and correlation of the shocks – as detected in the weekly data – would be lower. This is bad for consumers. If grain prices will eventually rise in a town, due to a national or regional shortage, then it would be better for consumers to know sooner and adjust their consumption earlier, rather than find out later and have to adjust more suddenly. When information transmission improves, wheat markets in Buckinghamshire can reflect news from Bedfordshire even if there has not yet been any transportation of grain between the two counties. So risk-sharing between the two counties can take place sooner and more completely. The improvement in information networks is thus complementary to the reduction in transport costs.
What all this shows is that market integration was indeed increasing during the Industrial Revolution: shocks were becoming more highly correlated across markets; shock sizes were more similar; and adjustment speeds (half-lives) were shorter. This raised welfare by lowering the real cost of doing business (physically moving goods) and by improving risk-sharing for consumers. But these improvements were completely masked by the increased turbulence of the wartime economy between 1793 and 1815. There were so many more shocks arriving, and they were so large, that overall price dispersion did not fall – even though the economy was absorbing the shocks more efficiently in 1820 than it had in 1770.
Returning to the issue of EU market integration, it will be interesting to see how the level of market integration changed between 2007 and 2013. It seems likely that price dispersion across Europe will be no lower in 2013 than it was in 2007 – indeed, it may well be higher. Observers may then say that spending €7billion on improved transport infrastructure was a colossal waste of public money. But it seems highly likely that the positive benefits of improved transport will have been masked by the disruptive impact of increased market turbulence since the financial crisis. Thus we should be wary of jumping to conclusions.
Improved transport may well have increased the speed of adjustment between markets, made shocks of more similar size and increased the correlation of shocks. These are all benefits to firms and consumers. Thus it may pay us to look at the data a little more carefully before deciding whether there might be further benefit from another round of infrastructure improvement.
Bogart, D (2013), “The transportation revolution in industrializing Britain: a survey”, unpublished mimeo, University of California-Irvine.
Brunt, L and E S Cannon (2013), “Integration in the English wheat market, 1770-1820”, CEPR Discussion Paper 9504.
Jacks, DG (2011), “Foreign wars, domestic markets: England, 1793–1815”, European Review of Economic History 15, 277–311.
Ravallion, M (1986), “Testing market integration”, American Journal of Agricultural Economics, 68, 102-9.
TENT-T Executive Agency (2013), “TENT-T projects in figures”.