No Pain, No Gain

On June 8, the WSJ editorialized about the gains from immigration:

The President's Council of Economic Advisers recently added up all these benefits, updating the procedures used by the National Academy of Sciences, and concluded that the value of immigrants to the overall economy is a net positive $30 billion a year.

This statistic inspired David Frum to whip out his calculator:

Well let's do some sophisticated economic research. Let's open Google, type "US" and "GDP" and see what we get. Ah, here it is: the very first entry ...

United States — GDP (Official Exchange Rate): $13.22 Trillion (2006 Est.)

Now let's use another sophisticated tool - that little calculator on my computer desktop.

30 billion over 13.22 trillion  ...  Why that's ...  0.22% percent! An extra one-fifth of a penny!

Frum rediscovered a wheel that economists have been aware of for some time: as a fraction of GDP, the gains from immigration are remarkably tiny.

The CEA leaked the $30 billion number to the WSJ hoping to influence the first round of the Senate vote. Just in time for the second round, the CEA puts out a "white paper" detailing their calculations. As the NYT puts it after reading the leaked report: "White House Report Lauds Immigrants' Positive Effects."

One of the beauties of economics is that it provides a rigorous way for thinking about how one should calculate these benefits. And the rigor pays off--it teaches us a brand new insight: the net gains from immigration are closely related to the wage losses suffered by natives. The greater the wage loss, the greater the net gain!

And therein lies an irony. The first method used by the CEA to calculate the $30 billion number updates calculations from the 1997 National Academy study (pages 151-152), which is itself an updating of this 1995 study.

(Full disclosure: I was a member of the National Academy panel that did the updating, and I am the author of the 1995 study that first estimated the "immigration surplus").

My original paper (p. 7) has the formula that can be used to calculate the net benefits from immigration in the simplest textbook supply-demand framework. For reasons that will become obvious very soon (even for those who have an allergic reaction to simple arithmetic), it's worth writing down the formula and working through the exercise:

Net gain from immigration as a fraction of GDP =  - .5
                          times "labor's share of income"
                          times "wage elasticity"
                          times "fraction of workforce that is foreign-born" squared

"Labor's share of income" (that is, the fraction of GDP that goes to workers) is around 0.7 in the United States. And, according to the CEA, the fraction of the workforce that is foreign-born is now around 0.15.

So the only number left to stick in the formula is the value of the "wage elasticity." This is a very contentious number, for it gives the percent change in the wage resulting from an immigration-induced one percent increase in  the size of the workforce. The CEA assumed that the wage elasticity is -0.3, so a 10 percent immigration-induced increase in the size of the workforce reduces the native wage by 3 percent. (This is the estimate used by the NAS study, and it is at the lower end of the -0.3 to -0.4 range reported in my 2003 study of the labor market impact of immigration).

Let's stick in the numbers:

   Gain from immigration = -.5 times 0.7 times -0.3 times .15 times .15

which equals .0024. In other words, immigration increases the income accruing to natives by 0.24% of GDP, or a gain of just over $30 billion.

But, as they say, that is not the end of the story. The same model that generates the $30 billion net gain implies that workers suffered a substantial wage loss. In fact, the total wage loss suffered by native workers is given by this other formula (p. 8 of my 1995 paper):

Wage loss as fraction of GDP = - "labor's share of income"
                          times "wage elasticity"
                          times "fraction of workforce that is foreign-born"                               times "fraction of workforce that is native-born"

Let's stick in numbers:

        Wage loss = -0.7 times -0.3 times 0.15 times 0.85

which equals 0.0268, or 2.7% of GDP. Since GDP is around $13 trilliion, the implied wage loss is around $350 billion. We are not talking about change anymore!

Finally, if you add the wage loss of $350 billion to the net gain of $30 billion, you get how much employers benefit--almost $400 billion. Of course, some of the gains made by employers eventually trickle down to consumers. But surely employers must keep an important portion: why else would they spend so much trying to lobby Congress for a never-ending supply of cheap labor?

This is all based on a particular economic model--and obviously the answers are sensitive to the underlying assumptions. The CEA, in fact, reports an alternative set of estimates (in the $30 to $80 billion range) based on calculations of the wage impact made by economists Gianmarco Ottaviano and Giovanni Peri (here). But that model also faces various difficulties:

1. The higher benefits result entirely from accounting for the possibility that immigrants increase the productivity of natives who have the same education and work experience (although the importance of this point is often glossed over). In other words, the immigration of high school dropouts who are 30 years old makes natives who are high school dropouts and 30 years old more productive! Put bluntly, young low-skill immigrants made young low-skill African-American construction workers more productive. Maybe--but just think about it for a second. Wouldn't most people be skeptical about this?

2. Suppose we buy into it nevertheless. The Ottaviano-Peri simulation then suggests that immigration raised the wage of natives by 1% to 2%, but simultaneously led to a huge reduction (of around 20%) in the wage of the pre-existing immigrant population. If correct, there must have been a substantial widening of the wage gap between immigrants and natives in the 1990s. In fact, no such widening took place. If anything, the immigrant-native wage gap narrowed slightly for the most educated workers.

I think the most credible estimate of the net benefit from immigration is probably around $30 billion--with four crucial asides:

1. If we are going to accept the $30 billon estimate as "the" estimate of the gains, then we should also accept the same model's predictions of the wage losses. And those losses are staggering.

2. The $30 billion is the net gain resulting from immigration in the short run. Over time, the economy adjusts to the presence of immigrants. Existing firms expand and new firms open up to profit from the lower wage. As the wage rises, however, both the gains and losses from immigration dissipate.

3. The $30 billion net gain does not account for the impact of immigration on the cost of providing social services. Any fiscal losses need to be subtracted from the $30 billion gain.

4. The same factors that determine the size of the net gain for the United States also determine the size of the wage loss for American workers. The larger the losses, the more beneficial immigration is for the United States.

An ironic lesson for supporters of higher immigration levels: Lift up your blindfolds and start arguing that the wage losses are staggering. You have nothing to lose but the chains that now prevent the free flow of labor into the United States.

Comments

[Sami B]

Prof. Borjas,
Thanks very much for your engagement of the Ottavanio/Peri work. I'm especially interested in seeing further discussion, particularly of Peri's most recent study on California, which takes a slightly different estimation approach.
If the "wage losses" are indeed as huge and problematic as the evidence suggests, then this leaves me wondering whether or not the going wages at time t prior to the immigrant influx at time t+1 were not too high. There may be a more complex labor market/GE model that I'm missing, but the mere fact that foreigners at the same skill levels are willing to work for lower wages suggests to me that the "native" labor at time t was merely accruing rents in the form of income beyond the marginal productivity of their labor.
Is it completely off base to view anti-immigration arguments based on protecting "native" wages as yet another form of rent-seeking just as we would similar anti-trade liberalization positions. I truly appreciate Prof. Rodrik's work on trade and his position that there is certainly an element of procedural justice that may be missed in advocating trade liberalization outright and I can appreciate similar arguments being made on immigration. However, I still wonder whether or not from an economic perspective the concern for "native" wage losses takes an unduly static approach to assessing the potential gains from enhanced global labor mobility.
I believe there's much more research that needs to be done on all of these issues, and the "right" answers have certainly not yet unambiguously emerged, but economic theory certainly provides us with strong tools for taking a more nuanced approach to the more fundamental issue of borders. Alesina et al appeal to economists to more strongly consider the endogeneity of borders, and it seems like there might be a number of unexplored implications for the study of migration in this regard.
Thanks again for your post. I look forward to further discussion of Peri's work.

[Luke Lea]

Employers to American and Mexican workers: 'Let's you and them share, and we'll take the gains.' That's awfully big of them!

[TK]

"The $30 billion net gain does not account for the impact of immigration on the cost of providing social services. Any fiscal losses need to be subtracted from the $30 billion gain."

Why, after the news articles about the expense of uninsureds closing hopsitals, the education costs, welfare costs, subsidized housing costs, criminal justice costs ad nauseum would they not include those costs? Perhaps because they wanted a certain result?

[jimbino]

I'd like one of you economists who has learned to use a calculator answer me this:

Given the choice between inviting a potty-trained Mexican worker to Austin or allow an Amerikan breeder to add just one more ignorant Texas kid to our schools at my great expense, which is ultimately better for the economy, and by how much?

I can't believe the answer would be the Texas brat!

Arronax : Those of us who read this blog and its postings expect rigorous scholarly discussions from different points of view. Next time you post, please engage the arguments being advanced, if you are intellectually equipped, and refrain from ad hominem attacks. Cheap shots have no place in a scholarly forum.

[bob]

Thanks for this excellent piece.

On your point 3 you talk about subtracting the fiscal loss from the $30 billion. Isn’t it possible that immigrants improve the fiscal position? Or would including the tax they pay be double counting (sorry my economics has gone rusty). Moreover isn’t there a long-term benefit in terms of improving the demographic position of the population with regards to social security?

[Peter Schaeffer]

All,

Professor Borjas writes

"Any fiscal losses need to be subtracted from the $30 billion gain"

Robert Rector of the Heritage Foundation has already published extensive work on this subject. Using Robert Rector's numbers, it would appear that the net cost of social services was in the range of $89.1 billion per year in 2004. Of course, it would be higher now. This is based on an estimate of 4.5 million unskilled immigrant households and a net cost of $19,588 per household, per year. See "The Fiscal Cost of Low-Skill Immigrants to State and Local Taxpayers" (http://www.heritage.org/Research/Immigration/tst052107a.cfm).

Subtracting $89.1 billion from the supposed $30 billion gain, gives a net loss of $59.1 billion per year. Of course, with Instant Amnesty these numbers will rise substantially.

This means that unskilled immigration is making the United States $60 billion poorer each year. Clearly, publishingly the truth about the dismal economics of unskilled immigration is yet another job Americans (at least at the WSJ) won't do.

[Paul]

Regarding Bob's query about Social Security: Today's immigrants are dominantly low-skilled and, consequently, poorly paid. (This is even more the case for most illegal aliens -- and, of course, many of them aren't having taxes withheld, hence aren't contributing to Social Security in the first place.) So they don't pay much into Social Security. And their returns from Social Security are relatively much higher than the returns to better-paid workers. So these unskilled immigrants turn out to be a heavy burden for Social Security, not a means of rescuing it.

[Memomachine]

Hmmm.

@ Bob

I think you're assuming that all of these illegal aliens are young and thus would have many decades to input money into the SS system. This is not necessarily true.

Consider that the current amnesty bill does not require age limits or barriers so 60 year old illegal aliens would be allowed amnesty just as a 18 year old. Also this current amnesty bill allows Z visa holders to chain immigrate their *parents*. Thus each actual illegal alien worker could chain immigrate *two* older or elderly parents who would be eligible for SS benefits.

Not a net positive.

[Bobby]

I simply do not believe that you can come up with such a relatively small number estimate on what the gain from illegal immigration will be. I know some people will say 30billion small?, Yes, it is when you are talking about possibley 20-30 million people here illegally. Which brings me to another point. Even the Professor does'nt know what the real number is. All of NASA"S budget is less than 30billion.

[CTL]

Bobby, Dr. Borjas' point wasn't that the $30 billion was a precise and completely reliable figure, but simply that IF you accept the method by which it was derived, you've also got to accept that method's implication for wages.

[Kann]

""But surely employers must keep an important portion: why else would they spend so much trying to lobby Congress for a never-ending supply of cheap labor?""

Bingo.

Dr. Borjas, love your work. Today there was a paradoxical story about Senator Trent Lott. In opposing border fencing he gave the analogy of the fence at his farm. He stated however high he built the fence his goats could climb out.

So, assuming Trent Lott is an economically rational actor, why does he have a fence at all?

[bob]

Thanks for responses.

Was just trying to make points that

Immigrants are generally young and have higher fertility rates. Furthermore immigrants generally have higher labour force participation rates. This certainly wont save social security but would be a benefit?.

If immigrants are low skilled initially then over time they can acquire skills, just as other immigrant groups in the US have done in the past.

Lastly, making positive steps to sort out the illegal immigration problem would mean more tax revenue and less rent for unscrupulous employers.

[D Flinchum]

RE: Social Security. GWB has already signed a totalization agreement with Mexico whereby citizens of Mexico who work in the US will be eligible for SS benefits. It doesn't need to pass both houses of Congress to take effect. Once he sends it to Congress, it will become law in 60 days unless one house votes against it. In other words, millions of workers from Mexico will be able to claim SS benefits without even being US citizens. The US has totalization agreements with a number of other countries but Mexico is by far the country that will have the largest number of workers making claims under totalization. I have assumed that GWB is simply waiting for the immigration bill to pass before he sends this to Congress.

Totalization agreements enable a citizen of a participating country who is working in another participating country to pay "SS" (or whatever it's called) in only one of the countries. It would make sense for Mexican citizens to pay into US SS because not only it is more generous than the Mexican program, but it also allows the worker to qualify years sooner.

The overwhelming majority of Mexican immigrants - especially illegal ones - tend to be low skilled with scant education. They will tend to contribute less to SS than they will receive later relatively speaking; and what they receive later will go much farther in Mexico than in the US. The US could see billions of dollars leaving the US in the form of SS payments, just as remitances do now. Good for Mexico, not so good for the US since the money won't be spent here. The relatively small amount that these workers pay into SS could end up costing the US - and future retirees - dearly.

Also, Mexico allows citizens to have dual citizenship. We may well see Mexican citizens who have US citizenship as well for the benefits (Medicare, etc) that are in the US but not Mexico, but who live in Mexico.

One wonders how GWB squares all of this with his assertion that SS is in deep trouble now.

[Ames Tiedeman]

This is the doom and gloom I see...or it is an opportunity if I am correct. In any event, I am deeply worried.

Gold will go way up, maybe to $1,500 an ounce or higher because the dollar will fall for years. The dollar will keep falling and here is why:
The U.S. cannot sustain 800 bilion a year trade deficits. We cannot export our way out of this mess. The only answer is a sharply lower dollar to drive manufactruing home and to lower the trade deficit. The dollar has much farther to fall. What you are seeing is a long term effort (it will take 20 years) to get the trade deficit back under 1% of GDP. We are currently running a trade imbalance of nearly 6% of GDP. No nation can do this. The IMF would be stepping in to help any nation if its trade imbalance went to 6% of GDP becuase its currency would collapse! The U.S. is different, but still, we cannot sustain a trade deficit of this magnitude. People must understand that when we buy an item from say China, we pay in dollars. The Chinese company we just bought from them goes to an Exchange Bank in China and converts those dollars to Yuan. The Chinese banking system (Chinese Government) is now sitting on those dollars. They can either 1, buy oil, 2, buy Treasuries, 3. buy U.S goods, 4. buy U.S. Corporations, 5. other. Over time if we (the U.S. ) continue to run a trade deficit we could simply be completely bought and controlled by foreigners. Warren Buffet has explained the situation as being like a rich Texas farmer who loses a small piece of his land year after year and never notices for a while. When he then notices, tragedy sets in because he no longer controls his land. So in sum, we need to get the trade deficit way down. This is why the Fed has abandoned the dollar. It wil be going down for the next 20 years. That is how long it is going to take to correct this imbalance mess. Bottom line: A lower dollar will equal higher inflation and higher GOLD prices.