However, the blog provoked a very interesting debate in the comments section, addressing much more real, substantive and difficult issues than David's original articles. In particular, Andrew Green from Migration Watch noted that (almost in passing) I had said that GDP per capita was a better measure of the economic benefits of migration than GDP; this in turn led to a lengthy discussion with Martin Wolf. The full set of comments (and there are a number of interesting contributions from others) can be be read here, but I thought that it was worth pulling out my exchange with Martin, which we also continued on email. I have edited it for readability, but don't think I've changed the main points. I am very grateful to Martin for allowing me to reproduce his words in this format.
"Aggregate GDP is, as I frequently noted at the time, a ludicrous measure of the benefits of immigration. That the Treasury used it was scandalous. Indeed, if that were the goal, elimination of controls on immigration would have been the right policy, because that would have maximised GDP.
Leave that absurd position aside. The question, then, is what weight should be put on the welfare of future immigrants. If the weight is very much higher than zero, then one is getting ever closer to the cosmopolitan welfare position that David is describing. So what was (and now should) the weight be, Jonathan?
To put it in the context of your debate with David, let us agree that UK immigration policy is not concerned with the welfare of Burundi. But what is the value it places on the welfare of the Burundians who will arrive, as a result of a liberal policy?
I would argue that this weight should be zero. Under a zero weight on the welfare of future immigrants (surely the position of most British citizens), the policy question becomes not what is the impact on GDP per head, or productivity per worker, but the impact of immigrants (of different kinds) on GDP per head (and its distribution) for those already in the UK. That was never the criterion used by the previous government, so far as I know.
To defend my assumption that the weight on the welfare of potential immigrants should be (close to) zero, I would note that if one adds foreign aid, net contributions to the European Union and that part of the defence budget which might be of benefit to foreigners, the sum is certainly considerably less than 5 per cent of total public spending (with aid at about 1.5 per cent of spending). Thus, 95 per cent of spending goes on 60m British residents, while 5 per cent goes on the rest of the world’s roughly 7bn people. So, revealed preference suggests that the weight placed by the British on the welfare of foreigners is not much greater than zero per head."
"What should the weight be? I agree there is a case for zero when we're talking about immigration policy decisions. But there's an obvious philosophical time inconsistency problem there. When does the weight become positive, and when 1, as immigrants become settled members of UK society, and eventually citizens? Let's leave that for now, and stipulate, for the sake of argument, that it's zero.
Even if we make this assumption, I do think that there is a strong practical argument for GDP per capita being the primary economic objective of immigration policy, even if the welfare weight on new immigrants is zero. Given a significantly progressive tax and benefit system, then the purely fiscal effects imply that high productivity immigrants will raise the welfare of the existing population, because they'll pay in more than they take out. That is even before you take into account labour market impacts (high productivity immigrants will presumably exert downward pressure on wage inequality) and productivity spillovers, which you would expect to be positive from high productivity immigrants. So it is difficult to think of circumstances under which an immigration policy directed at increasing GDP per capita wouldn't also increase the welfare of existing residents. Do you disagree?
And as an illustration of this, it's obviously absurd to suggest the previous government didn't regard the distributional impact of immigration as a criterion alongside GDP and/or GDP per capita. While I was at DWP, I wrote papers, and commissioned external research, on the unemployment and wage impacts of immigration. Ministers and my superiors didn't allow me to do this because they thought abstract academic papers were a good use of civil service time and money, but because they were keenly interested in the distributional impacts of immigration policy, especially on the unemployed, young and low-skilled.
I would also note that if you don't agree with this logic, and think we should ignore the fiscal effects and other spillovers, then perversely you could actually end up at a more liberal position, because migration just becomes economically directly analogous to trade. In that case of course the natives benefit, for the usual reasons we all agree on.
Abstracting from this, I do find the GDP/GDP per capita argument slightly overdone. As you and I well know, the government's objective function has many arguments, of which aggregate economic outcomes (GDP, GDP per capita, productivity) are only one, and far from determinative. If all policies, immigration and otherwise, were directed at maximising either GDP, GDP per capita, or any other single variable, things would be very different indeed."
"On the economics, the impact of moving a large number of people into a country is very complex. But I am inclined to believe, on the evidence I know, that liberal immigration of highly skilled people is likely to benefit both the economy and nearly all the people inside it. The only important qualification comes via the impact on the scarcity of resources (land) and costs of housing.
The impact of immigration of relatively less skilled people is more complex, distributionally, at least. I think the white working class has almost certainly been harmed by immigration of relatively unskilled people, not least because it has reduced the incentive to train them.
I think your question about the time inconsistency of the weight to be placed on the welfare of potential immigrants is very interesting. This seems to be similar to the question of abortion - a foetus has no moral value, perhaps, but the baby it would become, if not aborted, does. Similarly, a foreigner has no weight, but a foreigner does have weight if allowed to enter. Such somewhat arbitrary lines are difficult to draw. My answer has been that countries are like clubs. They can decide who members are. Once you are a member, you matter to the club. If you are not a member, you don't.
Finally, I think the trade analogy doesn't work. Standard trade theory assumes fixed factors of production, the most important of which, from a welfare point of view, is labour. If you assume, instead, that labour and capital are freely mobile, we are in the world of regional economics. Assume that a given country is capital-abundant, before opening. Then free immigration will reinforce whatever tendency free trade gives towards factor price equalisation. Wages will then fall towards the global equilibrium level. The benefits accrue to local capitalists and landowners. While aggregate world economic output will rise, the effects on the welfare of nearly all residents of capital-abundant countries will be negative. In theory, they could be compensated. In practice, particularly with mobile capital, they won't be."
"I think we are in broad agreement on the impact of relatively skilled immigration.
On low skilled immigration, I am less convinced. I agree it is an empirical question. But I think the evidence of significant harm is thin to nonexistent:
On low skilled immigration, I am less convinced. I agree it is an empirical question. But I think the evidence of significant harm is thin to nonexistent:
- as you know, the overwhelming conclusion of the econometric work on labour markets is that impacts are zero/insignificant, and certainly dwarfed by other factors (skill biased technological change, the national minimum wage, and so on)
- on incentives to train, I understand the theoretical case, but am not aware of any empirical work in the UK. What I would say is that - as you and Alison [Martin's wife, author of an excellent report for govenrment on vocational education] know - British employers have historically not done well at training lower skilled young people, and this long preceded the large expansion in low skilled immigration
- nor does the incentive to acquire educational qualifications appear to operate on individuals in this way. As your colleague Chris Cook has so convincingly shown, attainment of working class white kids in London has improved remarkably over the last decade as they have become increasingly surrounded by the children of immigrants. Meanwhile, attainment of working class white kids who live in areas where they neither attend schools with immigrants or compete with them in local labour markets has fallen wellbehind. We do not know why this is the case, but it is hardly consistent with your thesis.
I'm not sure I follow your argument on trade. As you say the standard trade thought experiment is to assume fixed factors by country and see what happens when you liberalise goods trade; and the standard result is a tendency towards factor price equalisation and overall welfare gains for both countries. The same is surely true of liberalisation of labour movements.
Consider for example a tradeable, low skilled industry with not much capital required and small barriers to entry - call centres, for example. In this case, surely the impact of allowing Indian companies to bid for UK call centre contracts is formally identical to allowing Indians to come here to work in call centres? British call centre workers lose, displaced into other sectors and having to accept lower wages, while British consumers of call centre services gain. We can argue out whether the distributional consequences outweigh the efficiency gains and whether, in practice, the resulting overall surplus will be "fairly" distributed, but surely the theoretical model is mathematically identical.
I don't take from this that all migration is good in the same way that all trade is (more or less) good, because in fact the static welfare gains from trade are quite small, and - returning to the earlier point about the impacts - I do think that at some point low skilled immigration is likely to have seriously negative distributional consequences. But I do think that those who thought that it was right, on principle, to allow the UK coal industry (say) to be destroyed because the standard economics of trade argued that this was likely to benefit the UK economy overall should have to explain why similar arguments don't apply to immigration.
Finally, I doubt this leaves us very far apart in policy terms, and would be interested to what extent you agree:
- I am in favour of free movement of labour in the EU, notwithstanding the potential disadvantages of some unskilled immigration, since these seem to me to have been empirically very small so far compared to the broader economic and political advantages
- I favour a relatively liberal approach to skilled immigration coming here from outside the EU, for the reasons we discussed above;
- moreover, I think there should be a presumption that those who've been here a few years and have demonstrated an ability to make a contribution and some degree of integration should be allowed to stay permanently ("join the club") in your phrase
- I think the government's objective of reducing net migration, and in particular setting a numerical target, is subject to all the usual objections for this sort of target in the economic sphere. And it's likely to be particularly damaging in this policy area, since the "marginal" migrants - those who policy can keep out - are likely to be skilled workers and students. Even worse, the target also forces government to seek to expel students and workers who've come here on temporary visas but wish to stay, usually because they've attained some form of economic and/or social integration."
"Let me comment/clarify a few important points.
First, I agree that the conclusion of the econometric work is that the impact is insignificant. I think this is because the immigration itself has been quite small, in relation to the labour market. It is not difficult for me at least to imagine that the impact would be substantial if, for example, unskilled immigration were to be expanded dramatically (on which further below).
Second, your comment on training by UK employers is well-taken. I am interested in your point that the performance in school of working class children is better if they have to compete with children of immigrants. My question, though, is what happens when they go out into the labour market to compete with fresh immigrants with already high qualifications. It may be that this has no effect on the behaviour of employers. But I would need to be convinced of this.
Third and most important, I made my little analysis of trade vs migration too simple. I will try to give a fuller view of what is going on.
The tendency towards wage equalisation generated by liberal trade alone, though present, is not overwhelming. This is partly because countries are not working with the same know-how (though this point has become weaker, in recent decades). More important, the large amount of non-tradeable activity and the possibility of non-overlaps in production between rich and poor countries diminish the likelihood of factor price equalisation.
A high-income country ends up specialising in ultra-high-human-capital-intensive output, while relatively unskilled workers no longer able to work in tradeable activities find at least tolerable opportunities in non-tradeable activities. Wages of relatively unskilled people are then no longer determined by opportunities to trade. Meanwhile, opportunities to trade are very important to many developing economies.
But free movement of labour (the analogy with free trade) would, I fear, be a far more powerful force for wage equalisation, to the benefit of landowners and owners of capital (including high-level human capital), since immigrants would then penetrate even the non-tradeable sector. I think this could be very socially disruptive indeed.
To come to the policy conclusions:
- First, I would accept free movement of labour in the European Union, though there should be a significant time-period before immigrants could have a right to domestic welfare benefits. The – to my mind, reasonable – assumption would be that if people were unable to support themselves from work, they would return home, unless/until they had developed a substantial attachment to the UK.
- Second, I would favour a liberal approach to skilled immigration, giving a particular preference to people who have qualified in scientific and technical subjects in UK universities. I would also auction work permits for different periods to employers or individuals and let the price tell me whether the policy was too restrictive. Anybody who has worked in the UK long enough, on these terms, would become eligible for citizenship. (We agree on this.)
- Third, I would be restrictive on immigration of relatively unskilled people.
Note that I have deliberately avoided discussion of non-economic aspects."
"Since this is my blog, I promised Martin the last word. I will only note that by the end I think that, as I said in my last substantive passage, we are pretty much in agreement on most of the key points and policy implications."