From Quartz - The ROI on border walls is terrible, the following claim was made::

Low-skill workers earn one cent in extra income for every 19 cents the government spends on walling off Mexico

It further claims that every other demographic ends up financially worse off with the wall being built.

Unfortunately the actual study is pay-walled so I can't analyze it in detail. Are these claims accurate, and does the study have sufficient evidence of the claim to be considered a trustworthy source for policy decisions?

  • Strange claim. So they determined somehow that a wall between Mexico helps "low-skill workers", albeit a small amount? Seems there's a lot of assumptions in the claim. – fredsbend Jan 4 at 21:18
  • 5
    They're apparently running this all through a bunch of mathematical models, and reporting the results. Models... reflect the biases of the people who construct them. I'm not saying that the information is incorrect, necessarily, but it is highly unlikely to be unbiased. – Ben Barden Jan 4 at 21:25
  • @fredsbend why the quotation marks? – Andrew Grimm Jan 4 at 21:33
  • 1
    @AndrewGrimm You mean "low skill workers"? That's their phrase, not mine. It's also one I find has varying definitions. – fredsbend Jan 4 at 21:36

The full paper says:

Consider, for example, low-skill U.S. workers. By reducing the number of low-skill Mexican workers in the United States, the Secure Fence Act increased the scarcity of low-skill workers (the "relative skill supply"), increasing the wages of U.S. low-skill workers by $2.69. However, this effect was almost wholly offset by the fact that there were now more relatively more [sic] U.S. low-skill workers than Mexican low-skill workers (the "relative nationality supply"), which decreased wages of U.S. low-skill workers by $2.39. All told, the benefit of the Secure Fence Act for U.S. low-skill workers was tiny: an equivalent variation increase in $0.36 per worker.


the direct costs of wall construction, estimated to be $7 for each person (not worker) in the United States (see Section 2.1).

So the paper is studying the effect of the Secure Fence Act of 2006 which, through 2017, involved spending $2.3 billion.

  • 4
    One thing I'm not understanding from the quoted claim of the parts of the study I can see is the timeframe of these wage increases/decreases. Is that annually, hourly, etc? It says, for example, "increasing the wages of U.S. low-skill workers by $2.69." Okay, per hour or what? – fredsbend Jan 4 at 21:38
  • 2
    I don’t understand the part about something causing an estimated $2.39 reduction. – Andrew Grimm Jan 4 at 21:38
  • 2
    @dsollen One thing I noticed was the dates they base this on. 2007 to 2010. One of the most trying economic times in the last century, the first half of the so called Great Recession. – fredsbend Jan 4 at 21:40
  • 2
    @fredsbend from what I can see it's per capita income, so yearly. Though the numbers seem awfully tiny for yearly income difference. – dsollen Jan 4 at 21:54
  • 3
    @DavePhD Thank you, I can now see the article. Though the statistical equations are beyond my mathematical literacy to follow. I would say even ignoring the math the raw numbers used had quite a bit of conjecture, and considering how minor a percentage change in immigration they estimated it seems hard to get a valid calculation while controlling for the large economic impact of the recession, seems like trying to find a needle in an economic haystack. I'm inclined to consider the numbers the study as interesting but hardly conclusive after a first scan. I'll go over it more later. – dsollen Jan 4 at 22:06

No. The actual number is completely unknown: the 1-to-19 ratio is one group's best guess.

First off, you have to take into account: nobody has an idea how much the wall will cost. Government projects tend to be like that; you only know the final price tag once you've finished paying for it. So if the final cost is, say, 20 billion, that 1-to-19 ratio suddenly balloons to 1-to-76. Aka, it's off by a factor of 300%. Heck, a budget overrun of only one billion would still make the 1-to-19 number off by 20%. To give an indication how little certainty there is in this '5 billion number', check out this article by the NYT, where the estimates for cost range anywhere from 21.6 billion to 70 billion. As an additional example, the 2006 Border Fence was estimated to cost 1.4 billion... but the final deployment price tag was 2.3 billion (64% overrun)... and the upkeep over 25 years was pegged at 50 billion.

Second, you have to understand that it's based off the affect the border fence had. Which means we have to somehow have accurate numbers of the effect it had. I giggled quite a bit when I saw this in the linked article:

The economists calculate it reduced the number of Mexican workers in the US by 82,647;

... yeah, I'm sure all those digits of that number are significant. They had the effect measured down to each individual attempted border cross. So you also have to have faith in that number - if you think that number is understated and the real impact was more like 300k, then the 19-to-1 ratio goes down to 5-to-1. Here looks to be the article that first spawned the 83k number, and you can see that they arrived at the number by estimating how many illegal immigrants they expected before all this started, counted the amount afterwards, and attributed the difference to the fence. I don't know if there are any better ways to get an estimate for the effect, but to pretend the end number is extremely accurate is a stretch.

Third, since it's based on the fence... you have to assume the think-tank's modifications to effectiveness are accurate. I mean, the first thing to consider is the 'fence' consists of 350 miles of actual fence, and 299 miles of vehicle barriers... on a border that's 1954 miles long (33% coverage) - so how much would 'complete' coverage change the effectiveness? You're assuming their estimate of that affect is correct. Second, the fence is presumably easier to circumvent than the wall would be. How much easier? Well, you have to assume they've got that number nailed down, too. Third, the wall will cost more per foot than the fence - what's the difference in cost/effectiveness between the two different types of barrier? Again, they have a mathematical model for you; you just have to assume it's accurate.

Basically, what you're left with is something like:

There are 200 alien civilizations in the Milky Way; researchers uncovered this by using the Drake Equation

... aka, a nice easy-to-digest number... but that's determined by a huge array of assumptions and best-guesses.

  • 1
    This is a comment, not an answer. – Andrew Grimm Jan 4 at 22:49
  • 2
    I think it shows an appropriate level of skepticism, although this level is not really allowed here without accompanying data. Notice the other answer just cites part of the paper in question, but has received no criticism for being a comment, not an answer. – fredsbend Jan 5 at 1:29
  • 2
    You'll have to find some related and reliable sources to cite, or this might be deleted. – fredsbend Jan 5 at 1:29
  • Okay, I edited... that said, I think requiring sources on "Government projects tend to go over budget" and such is a bit overboard. The point of the answer is: that '19-1' ratio is depending on a huge number of factors that can't be known to a good degree of precision. – Kevin Jan 5 at 18:20

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .