A while ago I wrote about a Bank of England paper investigating the effect of immigration on wages in Britain which found that “the immigrant to native ratio has a small negative impact on average British wages”. Then, some time after, I came across a paper by Julie L. Hotchkiss in the Southern Economic Journal which finds that “illegal immigrants actually raise wages for documented/native workers”. How to explain this contradiction in the economic evidence?
The first thing to note is that the Bank’s researchers, Stephen Nickell and Jumana Saleheen, are looking at slightly different question to Hotchkiss. Nickell and Saleheen are looking at immigration in general, Hotchkiss only at that part of it which is illegal.
Second, they are also looking at different data. Nickell and Saleheen are looking at data from Britain, Hotchkiss from the US state of Georgia.
But should this matter? Should not an economic theory which predicts that, ceteris paribus, an increase in immigration will raise/lower wages in Britain also predict the same for Georgia? In his excellent History of Economic Thought, Lewis H. Haney called the belief that the answer to this question is Yes ‘cosmopolitanism’ and the belief that the answer will be Yes at any point in time ‘perpetualism’.
That seems to depend on which theory you mean. As I wrote of Nickell and Saleheen’s paper, “In the short run at least, this is what economic theory would lead you to expect. If you increase the supply of anything relative to the demand for it, ceteris paribus, the price of that thing, whether it is shoes or labour, will fall.” But Hotchkiss applies a different bit of economic theory, that of comparative advantage, and finds the opposite. So which theory to apply?
The theory of comparative advantage is one of the old workhorses of economics. It states that Country A should specialise in the production of the good or service A that it produces most efficiently, even if it also produces good or service B more efficiently than Country B. This will lead to more efficient production and, via trade, make both countries better off. The more countries there are, the more they can specialise, the more productive they become, and the better off everyone is, so the theory goes.
Hotchkiss applies this to illegal immigration. Illegal immigrants “with limited English skills” (I’m quoting Art Carden’s write up here) coming into Georgia and finding jobs “frees up low-skill American workers who can then specialize in tasks that require better English”.
That makes theoretical sense, but it isn’t happening in Britain, according to Nickell and Saleheen. Perhaps this is because the premium on speaking English particularly well isn’t all that high lower down the value chain in the UK. Consider Mike Ashley’s Sports Direct warehouse where the signs are posted in English and Polish. There would seem to be little advantage accruing to a native English speaking employee over his or her Polish immigrant colleague from their greater language skills.
It is interesting to speculate on the microeconomic differences between the British and Georgian labour market that allow for so much more specialisation in linguistic ability lower down the value chain in Georgia than there would seem to be in Britain. After all, these differences would seem to determine whether we apply standard supply and demand analysis to the question of immigration’s impact on wages, or the theory of comparative advantage.
But how much of this is fitting theory to the facts? When I read Nickell and Saleheen’s paper I thought “Well, that’s what standard supply and demand analysis would predict”. Did Hotchkiss, I wonder, see her results and think “Well, that’s what the theory of comparative advantage would predict”? Imagine if the British results had also shown a positive impact on wages from immigration. Would I then have thought to myself “Well, that’s what the theory of comparative advantage would predict”? If so, whither economic theory?
Prior to conducting the analysis on British and Georgian data, we would not have known which theory to apply and, hence, could have made no predictions. If, for example, you assumed that comparative advantage applied to the British labour market you would have predicted that immigration would cause wages to rise, as in Georgia. When confronted with the result, that immigration actually causes wages to fall in Britain, you would conclude that your prediction was wrong and that your theory had been falsified. But if you had gone through the same process in Georgia, your prediction would have been borne out and your theory would stand.
These are fundamental questions of economic methodology. Economists generate theories which make predictions such as ‘if X then Y‘.Governments spend large sums of money according to these predictions based on these theories. Is it the case, however, that the appropriate economic theory is entirely contingent on the particular circumstance in which it is applied?
I generally consider myself a cosmopolitan perpetualist in economic methodology. This can lead to situations where, when data contradicts theory, I assume that the data is faulty, that it must be failing to take some relevant factor or magnitude into account. But if there is even a question over which theory to apply in the first place, whither economic methodology?