This is a very long post, but skip on to the next one if you get bored. I got interested in immigration and am trying to work my mind around a chapter in George Borjas's 2014 book, Immigration Economics. Paragraph (0) is about law and the President's recent decree, but you'll find its ideas in many other places and liberals will probably find it unpalatable, even if they feel uncomfortable with expanding executive power. Paragraphs (1), (2) and (3) are more analytic. They about the economics of immigration, expanding on the idea of "the Borjas triangle" to do back-of-the-envelope estimates of the surplus gains and redistributional effects of increasing the labor force by 15% via immigration. They're simple enough that you can plug in your own estimates of the size of immigration and the inverse elasticity of labor demand and see what happens.
(0) I'm still trying to figure out the legality of Obama's immigration amnesty. Congressional leaders considered relaxing immigration laws, and there weren't the votes, so they didn't even bring a proposal to vote. Then the November elections were a clear repudiation of immigration relaxation and the President's party. So he unilaterally decreed a change in the law:
"So we’re going to offer the following deal: If you’ve been in America for more than five years; if you have children who are American citizens or legal residents; if you register, pass a criminal background check, and you’re willing to pay your fair share of taxes — you’ll be able to apply to stay in this country temporarily, without fear of deportation. You can come out of the shadows and get right with the law. . . ."
Prosecutorial discretion means that the President has broad powers. He could decide not to enforce the tax laws against Democrats, for example. He could even tell the entire Border Patrol to stop patrolling and take a holiday, or, indeed, give the entire army mass home leave. But here it seems he is going further. He isn't just refraining from prosecution; he's got an entire new statute, it seems, complete with detailed requirements and claims that it removes an illegal alien's illegality. Except Congress never voted on it and both House and Senate would would vote against it if it was brought before them. At the same time, what judicial remedy is there? Impeachment is appropriate in cases like this (or the non-taxation of people in the President's party), but the Democrats would block it. So is there any present or future legal structure that could be used to prevent a President from nullifying a statute?
November 24 Update: The Christian Science Monitor has a good q-and-a article on the law involved. It is the only article I've seen that explains the work permit aspect; apparently the statutes do allow the President to grant green cards to illegal alients whose deportation is deferred. The statute didn't envision his doing that with 4 million of them, of course, but if the newspaper is correct, that says Obama's within his authority— at least if Congress has the consitutional authority to delegate the immigration laws to the executive completely, which is dubious.
(1) I was just reading George Borjas's new book, Immigration Economics, after a discussion of immigration at lunch at the econ department yesterday. The Borjas immigration-gain triangle on p. 151 (apparently first shown in a 1995 article of his) is good. It's like the 1954 Harberger triangle for monopoly. The gain from immigration can be seen in a supply and demand for labor diagram. Make labor supply inelastic. The social gain is one half the reduction in the wage times the increase in the quantity of labor. The increase in labor is 15%, Borjas estimates, and the inverse elasticity of labor is .3 (the elasticity of demand for labor is about 3), meaning that wages fell about .3(.15), which is 4.5%. Thus, the increase in social surplus in the labor market is .5(.045)(.15)= 0.34% (not 34%). Since labor income is 70% of national income, the increase in social surplus from a 15% increase in the labor force is 0.24% of national income. It's small because it's a triangle. The loss to American workers, though, is the 4.5% reduction in wage times the initial amount of labor— a 4.5% loss in income. As a fraction of GDP, that's 3.2%. (Borjas has 2.7% because he's using post-immigration GDP; I'm using pre-immigration GDP, which is smaller).
American capital gains what labor loses, plus the triangle, so it gains 3.4% of GDP.
If no-immigration GDP would have been $13 trillion (post-immigration it actually is about $15 trillion), then the gain in American social surplus from immigration is (13)(.0024) = 31 billion dollars.
The loss to American labor is .034(15) = 442 billion dollars. American capital benefits by 473 billion dollars. Capital was receiving 30% of national income before immigration, by our earlier assumption of a 70-30 labor-capital split, and it has gone up to 33.4%. That's an 11% increase in the total return to capital. So we can see why American employers are so strongly urging increases in immigration but why their pleas have gotten nowhere in Congress despite the support of the Republican and Democratic leaders (who are more responsive to contributions than the backbenchers).
Borjas notes that in the long run more capital will enter the U.S.— from domestic sources and from abroad— so the long run gain will be zero, since the return to capital must be equalized between home and abroad. By the same reasoning, the long run loss to labor and gain to capital will be zero.
(2) Something I wondered about was how the gain to consumers figured into this. It is implicitly included, because this is a model about real returns to factors of production, not nominal returns. But I think it's useful to break it out. Let's use a simple quantity theory of money where the quantity is 1, so the price of the single output good is P = 1/Q. We also need a link between the extra 15% labor from immigrants and Q. The increase in output will be less than 15%, since capital is fixed. Suppose output rises by 10%. Then the price drops by about 9% too, since P_1 = 1/(1.1*Q_0). This allows us to think about how nominal wages and returns to capital have changed.
Recall that we have labor losing 3.2% of GDP in real terms (from a 4.5% wage decline) and capital gaining 3.4%. Since prices are falling by 9%, this means that nominal wages must fall by 13.5%. In the same way, the gain of 11% in capital's earnings will arise from a mere 2% increase in the nominal return. Am I right in concluding that if split the distributional analysis into three groups rather than two, we would say that labor suffers a 13.5% loss, capital gets a 2% gain, and consumers get a 9% gain? I think this is legitimate. The typical worker wears a separate hat as consumers, just as he wears a separate hat as capitalist— since the typical worker has a pension plan and a house, even if his capital earnings are much lower than his labor earnings.
Update, November 24. I now think this discussion is misleading. What it says about shifts in nominal income and prices is fine, but it doesn't capture the effect of immigration on consumer welfare. That can't be done in this simple model, because demand for goods is inelastic (which I suppose is because supply of labor is inelastic). What happens in the real world is that immigration lowers wages, as analyzed here, but in the product market, the extra output from the immigrants reduces the price of output, and creates real consumer surplus that way (while having a further effect of reducing demand for labor). To buy the extra goods, consumers would increase the amount of labor they supply. It would be nice to put together a model on this order of complexity that could separate out the consumer effects.
I wonder what the pattern is of labor and capital earnings for Americans at various levels of income. It's complicated by changes over the life cycle. Almost everybody is a 100% capitalist when they are old, and a 100% worker in his teenage years, putting aside the additional complication of bequest wealth.
Immigration thus has had an important effect on the distribution of income. If I remember rightly, the biggest change has not been an increase in the share of income of the average member of the top 10% of the population, though, or even the top 1%, but of the top 0.1%. Is that because even the median member even of the top 1% derives most of his income from labor, not capital? I don't know— I would have thought that the top 1% included a lot of small business owners and people in their 60's who had piled up substantial savings. Still, even people as rich as business school professors expect to spend less in retirement than in their last decades of working years, and that implies, I think, that their capital income in those last decades is less than their labor income. So perhaps it is plausible that an increase in the return to capital and a decrease in labor shifts income to the top .1% while leaving the share of the median member of the top 10% unchanged and reducing the share of the bottom 90%. (Don't trust my numbers here; I'm just wandering— though I'd appreciate corrections in the comments.) Of course, it's important to remember that the bottom 99.9% is still doing better in absolute terms, despite a lower share, since (a) technical change keeps income per capital growing, and (b) the rich pay far more than their share of taxes.
We have been assuming there is only one kind of labor, however, or, equivalently, that immigrants have the same distribution of skill as natives. Now, let's relax that assumption.
(3) It's also useful to think of this thinking of the particular type of labor provided by immigrants. Gardeners, professors, and engineers suffer from immigrant competition; lawyers, schoolteachers, and bank clerks do not. Suppose we guess that 25% of the labor force is in jobs with significant immigrant competition. We will call this group "unskilled labor" for convenience, without snark towards engineers and professors (that's paralepsis, of course). Then, the unskilled labor force has increased by 60%, not 15%. We can redo the calculations.
Now the wage response is .3(60) = 18%, which is the loss to the American unskilled workers. They were getting .7*.25 =$17.5% of the old national income so they have lost an amount equivalent to 17.5*.18, about 3.2% of the old national income. The triangle gain, in terms of old GDP, is (since unskilled labor was .7*.25 of national income), .7*.25*.5*.15*.6 =
0.8% of GDP, much larger than in the original example, though still small relative to GDP. It is no accident that both the redistribution effect and the increase in social surplus are much bigger under these assumptions. When social surplus increases, it's because there's been a big wage response, and that's what makes the redistribution effect bigger too.
The distributional effects are more complicated, because now we have other kinds of labor as well as capital. I've assumed that in each job, only one kind of labor is combined with capital; the increase in immigrant labor has no effect on the real wages of any labor except the unskilled. Here, it is crucial to think of wages in each sector as being paid "in kind"; in the skilled sector, workers have the same marginal product as before, and so they get to keep exactly the same amount of the output they produce as before.
As we've seen, the unskilled wage falls 18% in real terms. Capital in the unskilled sector is gaining the new surplus triangle of .8% from immigrant production plus a rectangle of .7*.25*18, about 3.2% of old GDP, for a total of a 4% gain. It goes from 30% to 34% of the old GDP level, an increase of 13.3% of 30%.
Compared with when immigrants had the same skill pattern as natives, the loss to labor is exactly the same, whereas the gain to capital has grown. Now, however, the loss to labor is concentrated on just 25% of the labor force.
What about the price effect? The amount of unskilled labor has increased by 60%, but the increase in output will be less. Diminishing returns to unskilled labor because capital is constant will be much bigger with a 60% increase than with a 15% increase, so our discount should be more than the 1/3 used earlier. Suppose it's 1/2: output rises by 30% in the unskilled sector. Unskilled labor used to produce 1/4 of the economy's output, but now it's 1.3*.25 = .325, so total output has risen by .075 to 1.075. Prices therefore fall to 1/1.075= .93, a 7% decline.
Now we can look at nominal changes. Unskilled labor's nominal wage falls by 25%, skilled labor's wage falls by 7%, and capital's return rises by 6.3%. Consumers are better off by 7% due to the fall in prices. It's interesting that capital's nominal gain triples from the 2% from matched-skills immigration. That's due to a combination of prices falling less and the real return to capital rising more. The real return to capital— that is, including the capitalist's role as consumer— rises by 13.3% compared to 11%. But see the November 24 update above– I now am dubious about the value of my output price analysis.