Sunday, 7 July 2013

Large Markets, Gains from Trade, and a Puzzle

The latest paper in the UK Government’s Scotland Analysis series was released last week by the Department for Business, Innovation and Skills. The BIS paper describes the benefits of a large market for business, workers and consumers. This detail is fine: economists typically do believe that trade is usually mutually beneficial and so barriers to trade, such as a national border, are damaging at the margin. Indeed, I have written a paper on the economics of independence for Catalonia from Spain (see Comerford, Myers & Rodriguez Mora (2013)) in which this is the very mechanism at work. Sub-national entities appear to trade much more with the other parts of the nation state of which they are a part than independent nations do with their neighbours. Sharing a state seems to be a trade enabling technology, and this trade is valuable, with positive effects on welfare. Preliminary estimates (using the methodology of Comerford, Myers & Rodriguez Mora (2013)) of the welfare cost to Scotland of creating a national border between Scotland and rUK are of the order of 5% of Scotland’s GDP.
Successful small independent nations engage in more international trade than their larger counterparts. So perhaps small nations develop institutions that are conducive to international trade, and their international borders cause smaller frictions than the international borders of larger nations (for whom international trade is less important)? However, model calibrations do not suggest that the extra trade that we see in the data is due to a greater ability to engage in such trade: the international borders of small nations seem to cause frictions that are equivalent to the frictions caused by the international borders of larger nations. Instead small nations seem to be compelled to engage in more international trade due to the small size of their domestic markets. Therefore we cannot expect the reallocation in Scotland’s trade from rUK to the rest of the world to produce welfare effects that mitigate the 5% estimated cost – these reallocations are already assumed in that calculation.
This is starting to sound conclusive: trade is good and eliminating international borders is a mechanism that generates more trade. However, if this were the case, then we might expect to see a systematic positive relationship between country size and wealth. We see no such relationship in cross country data (see Rose (2006)). A quick back of the envelope calculation of the elasticity of per capita GDP with respect to national population using 2005 data from the World Bank shows that any relationship that there is between population and wealth would seem to be negative[1].
Country Grouping
Elasticity of GDP per Capita w.r.t. Population Size
Whole World
Perhaps the importance of trade is overstated? This is possible but the calibrations used in Comerford, Myers & Rodriguez Mora (2013) are consistent with the latest estimates and advanced methodologies in the literature (see Arkolakis, Costinot & Rodriguez-Clare (2012) and Simonovska & Waugh (2013)). If the positive effects of trade are real, then conditional on the real wealth distribution of countries, this implies that smaller countries must compensate: they appear to have higher intrinsic productivity (i.e. a higher productivity before considering the impact of trade). Comparable results to the above table but now for the elasticity of this intrinsic productivity with respect to national population are:
Country Grouping
Elasticity of intrinsic productivity w.r.t. Population Size
Whole World
If we could interpret these figures as representing a causal relationship, then they suggest that Scotland would more than double its intrinsic productivity on achieving independence[2], an effect which dwarfs the 5% estimated cost of a border with rUK. However, we cannot interpret these numbers as causal and we can imagine clear endogenous mechanisms that may generate such observed relationships: perhaps productive regions, being already wealthy, choose independence whereas less productive regions, who see their wealthier neighbours, choose to join a larger state; this would give the appearance of small states being associated with high productivity. On the other hand, we may also imagine mechanisms that would enhance productivity in small independent states when compared with the peripheral regions of larger states: perhaps there are spillovers from having head office functions rather than branch offices, and head offices preferentially locate to be close to decision makers at nation state level rather than local authority level; perhaps simply economic governance works better in smaller states.
Whatever explains the discrepancy between the predictions of models with Gains From Trade, and the observed relationship between country size and wealth, I do not mean to suggest that net benefits for Scotland from independence are to be expected. I simply point out that the BIS paper’s focus on the costs from the loss of the UK single market is a partial analysis, and one which, in particular, cannot account for the success of small open economies.

[1] The results in the table should be interpreted as e.g. countries in the EU15 with twice the population of their peers have, on average GDP per capita that is 14% lower than in these smaller peers. *, ** & *** respectively represent significance at 10%, 5% & 1% levels in these naïve regressions. For these results to represent a true statistical relationship, it would need to be the case that log Population was causal for log GDP per capita, but that log GDP per capita was not causal for log Population. This is clearly not satisfied, so what we are looking at is simple correlations in observables rather than any structural relationship.
[2] Change in intrinsic productivity ~ exp(-0.3*ln(5million/60million)) > 2

No comments:

Post a Comment