Thursday, January 10, 2019

The Economics of a Border Wall

A recent study by economists Treb Allen, Melanie Morten and Caue Dobbin (2018) is not yet published but has already has received much publicity. The authors study the economic impacts of the border fences that already exist on the US-Mexico border. The authors write:

“…the Secure Fence Act signed by President George Bush on October 26, 2006…authorized the construction of reinforced fencing on locations of the border in California, Arizona, New Mexico, and Texas. Between 2007 and 2010, 548 miles of wall were constructed along the 1954-mile U.S.-Mexico border, bringing the total wall to 658 miles.”

The authors find that, “…the border wall expansion harmed Mexican workers and high-skill U.S. workers, but benefited U.S. low-skill workers...”  This is a conclusion from the structural econometric model developed by the research team. The model is highly stylized (workers have only two skill levels and produce only one good, for example) and it is difficult for someone without a Ph.D. in Economics who hasn't carefully read the entire 100-page report to appreciate that these conclusions are rather indirectly drawn from the data.

This hasn't prevented the media from interpreting it as a comprehensive social cost-benefit analysis. For example The Economist recently summarized the findings of the Border Wall study and concluded “it did more harm than good”. They write

So what effect did the first 550 miles have? Not much, suggests an analysis by economists at Dartmouth and Stanford Universities. …the economists …calculate that it reduced the number of Mexican citizens living in America by only 0.6%.

But perhaps it shored up American wages? Sadly, no. College-educated Americans came out worse, losing about $4.35 per person in annual income. Low-skilled workers without a college education came off only slightly better—gaining 36 cents of income—because of reduced competition in the labour market. This is a pittance next to the $7 per person paid for the wall’s construction (to say nothing of the big maintenance and personnel bills to come). Because of lost workers, American GDP probably shrank by $2.5bn (0.02%) overall.

Taken at face value (as opposed to interpreted within the framework of the study's theoretical model)*, none of these facts may matter in an economic appraisal (i.e. a cost-benefit analysis, or CBA) of the 2006 Secure Fence Act. Why? Because consider the fact that wages fell for skilled workers is actually good for the firms that employ them and possibly for consumers who pay lower prices for the products the skilled workers produce. In other words these wage effects can be seen as “transfers” from employees to firms (or employees to consumers). More generally, wage changes by themselves are not measures economists use to appraise social investments.  

[* I hesitate to get into too many details of the theoretical model. Suffice it to say that the economists who wrote this report realize that wage changes by themselves are not measures economists use to appraise social investments, and in fact the 36 cent income gain reported for low-skill workers is actually a conversion of changes in welfare to "equivalent variation changes in per capita real income". Let me just say much of this nuance is "lost in translation" when news media reports on results from stylistic structural econometric models.]

Second, the fact that GDP fell is also not relevant in a CBA. GDP goes up for many reasons that aren’t necessarily good for society—if California cuts down the largest Giant Redwood Tree and sold it for $1,000, GDP would go up by $1,000, but then all the future generations would be deprived of seeing the tree and society as a whole would be worse off, even though GDP went up. The economist Paul Samuelson once famously suggested, if a man married his housekeeper, then, ceteris paribus GDP wouldfall. The upshot is GDP is also not a measure for appraising social investments.

Allen, Morten and Dobbin cannot necessarily be faulted if the news media inappropriately interprets their results and none of this is to criticize the authors' methods; they do an impressive job of what they set out to do, which is to develop and estimate a structural econometric model.  Nor is my point to argue for or against the Secure Fences Act or future border fences. My first point is to highlight that, given even economically sophisticated members of the news media have a hard time distinguishing between various types of economic analysis, economists should be a little more careful in interpreting their findings.  Allen, Morten and Dobbin should be commended for moving the ball forward but more remains to be done, and in particular we as a profession need to more quickly produce comprehensive cost-benefit analysis for evaluating public investments like these.

As it is, I haven't seen a comprehensive economic analysis--which incorporates impacts like crime, drugs, etc.--that really addresses the question of whether or not an additional mile or border fencing tomorrow would be worth it in terms of social value and opportunity cost.

1 comment:

  1. I think your premise can be stronger - are you pointing out fallacies, or supporting them? There are so many variables that were conveniently not covered, like reduction in benefit costs (medicare, social programs - (YUGE as DJT says), increased jobs availabilities, increase tax revenues (fewer under the table), reduced black markets, etc. Most arguments against conveniently show vary superficial 'studies', like $5bn cost, but only '.01% GDP...etc. Your conclusion was good and good summary.