The recently proposed immigration reforms in the U.S. that seek to raise the bar for skill sets but also aim to reduce the number of legal immigrants by 50% over the next 10 years could prove a big negative for the nation’s gross domestic product (GDP) and employment.
Between now and 2040, the proposed RAISE (Reforming American Immigration for a Strong Economy) Act could shave two percentage points off GDP growth and cause a loss of more than four million jobs. However, in the short-term, reforms would have little negative impact on jobs and GDP, while wages would actually rise, although they would flatten out over time. Those are the top findings of a detailed simulator in the Penn Wharton Budget Model (PWBM), which helps analyze the potential impact of proposed policy changes.
“Immigrants tend to work pretty hard, they tend to have a very high attachment rate to the labor force, they are less likely to be on unemployment insurance and things like that, and so they are really a net positive,” says Wharton professor of business economics and public policy Kent Smetters, who is faculty director of the PWBM. As younger members of the workforce, immigrants also help pay for Social Security and Medicare for the elderly. That is a crucial benefit as the U.S. — along with other countries — grapples with the so-called dependency ratio, he adds. The simulator is available online and is free for use by the general public.
Smetters recently spoke with Knowledge at Wharton about the implications of the immigration reforms that have been proposed by the U.S. Senate and endorsed by President Donald Trump. (Listen to the podcast using the player above.)
An edited transcript of the conversation follows.
Knowledge at Wharton: Let’s get right to the headline findings. You have noted that net legal immigration into the U.S. is about 800,000 people per year. That may sound large, but it is actually about a quarter of a percent of the U.S. population (325 million and counting). You note on the website that whatever effects these policy changes might have are not going to be that great because they cover such a small portion of the population. Let’s talk first about the effects on GDP.
Kent Smetters: There are about 800,000 legal immigrants who come into the United States every year. About 45% of them are college-educated, and the rest are typically very unskilled. The way to think about immigration in America is [to adopt] a barbell approach: We have a lot of people who are very unskilled, and a lot of people who are college-educated.
The RAISE Act that was reintroduced by [Republican] Senators Tom Cotton and David Perdue last week tries to do two things. The first is that it lowers the total amount of legal immigrants every year to roughly about 400,000, or about a 50% decrease. But it tries to change the skill mix such that instead of 45% of them being college-educated, we estimate it to be around 75%.
“Our three programs alone – Social Security, Medicare and Medicaid – are going to absorb the entire federal budget at some point. They’re on an exploding path.”
What we do with the model is to simulate the likely impact [of those proposals]. Overall, the impact will be negative for both GDP and jobs. If there were no change in the number of visas issued, but all you did was to increase the skill mix, that would be positive for GDP, which would go up by about a third of one percent over time. But you will have a negative effect on GDP by reducing the total number of visas. The long-run impact will be about a 2% reduction in GDP, and a loss of about four-and-a-half million jobs.
That is the long-run effect, by 2040. We report intermediate years as well, and so by 2027 or 10 years from now, GDP will go down, but not by as much. The effect in terms of jobs is pretty small in the first several years, but over time the job losses increase roughly to four million by 2040.
Knowledge at Wharton: So you are saying there will be four million or so fewer jobs in 2040 if this bill is passed as it is written.
Smetters: That’s right. [The loss of] four million jobs [will be] in the base of 214 million jobs that we project for 2040, [which will] go down to about 210 million jobs. As a percent of total jobs, we’re talking about a pretty small number, but still it’s definitely a negative.
Knowledge at Wharton: Is it too small to have any effect on salaries?
Smetters: We did look at wages, and you have a couple of things going on there. The first is as you remove people, you have less labor, and that increases the amount of capital per worker in the short run. So wages actually go up by a small amount in the short run. But then eventually what happens is as more and more immigrants are out of the economy, and their capital is no longer here, it reverses course. Eventually, wages are unchanged over the long-term.
Taming the Dependency Ratio
Knowledge at Wharton: What is the dependency ratio? It seems like many countries are struggling with it — in Europe, Japan and even China.
Smetters: Right. The dependency ratio is a very important statistic when we think about major programs in the U.S. like Social Security and Medicare. They are funded on what is called a pay-as-you-go basis. You need a lot of younger people to help pay for those programs.
Over time, as the [number] of older people relative to younger people goes way up, we have an aging population. That is a problem for us because that makes it harder to pay for these programs that are mainly focused on the elderly. Keep in mind that our three programs alone — Social Security, Medicare and Medicaid — are going to absorb the entire federal budget at some point. They’re on an exploding path.
Immigrants typically come in at a younger age, and when you get rid of them you are worsening your ability to pay for those old-age programs. It is true that in the old days, some gaming would happen — immigrants would come in at middle age, qualify for Social Security with just 10 years of work, and then be able to collect benefits (until death). A lot of that gaming has been eliminated. For the most part, immigrants are a positive now for helping us pay for those entitlement programs.
Knowledge at Wharton: How was the gaming eliminated?
Smetters: It was an act of Congress that made it harder for people to qualify for benefits. They had to pass bigger hurdles to qualify based on a few years of work.
“If we didn’t change the skill mix but we just increased the number of legal immigrants, it would have a very big positive impact on the economy.”
U.S. vs. Japan on Immigration
Knowledge at Wharton: A lot of times when you talk about immigrants, people think about illegal immigration, which of course is a separate topic. But in general they think that maybe immigrants are taking other people’s jobs. But one thing that is interesting is the fact that the U.S. is a country of immigrants.
Japan has an aging population, and its total population is scheduled to drop by very drastic numbers over the next couple of decades. But they don’t have the kind of history of being open to immigration that the U.S. has, and it’s very hard for them to accept younger immigrants who might help them with their dependency ratio. In that sense we have some advantages over other countries, don’t we? Our culture sees immigration generally as a positive. Now we’re negotiating about how much of that is the right level.
Smetters: Right. No question that relative to Japan, our history of allowing for immigration has been a positive. Of course that can reverse somewhat with this bill. The one big difference between the U.S. and Japan, though — and one reason Japan has been able to get away with having fewer immigrants — is that the average Japanese household saves a lot more than the average American household. That is almost by about 10 times in terms of the percentage of their salary. So the Japanese have been able to afford a higher level of government debt, and are able to avoid some of the other negative problems that are associated with not having much immigration. That’s not been true in the U.S., and so immigration plays a really key role here.
Advantage Skilled Workers
Knowledge at Wharton: What else that is important was uncovered in your simulations on this immigration issue?
Smetters: The main points are that if you simply tilt the balance towards more skilled workers, that’s a positive because skilled workers are going to be net producers for the economy, not just in terms of the taxes they pay but in terms of job creation. But if at the same time you reduce the total number of immigrants, [the effect] goes in the opposite direction.
If the Trump administration really wants to figure out how we do immigration reform in a way that will be a net positive for the economy, it’s [about] changing the rules [that determine] how immigrants come in. In particular, [it will be] less about family ties and so forth, and more about the actual skills that they bring in.
Knowledge at Wharton: You also just quickly looked at what would happen if immigration increased by 50% or 100%, and the effects that would have on GDP and jobs. Could you talk about that for a moment?
Smetters: Even if we didn’t change the skill mix but we just increased the number of legal immigrants, it would have a very big positive impact on the economy. That is both in terms of total GDP and the number of jobs. Now, someone might say that this is obvious if you have more people. But it turns out that it even increases the amount of GDP per person, or how much money is available across everybody including native-born workers.
The reason behind that is that immigrants tend to work pretty hard, and they tend to have a very high attachment rate to the labor force. They are less likely to be on unemployment insurance and things like that, and so they are really a net positive, even at a per person level, not just at an overall level.
A Data-rich Simulation Model
Knowledge at Wharton: Could you talk briefly about your budget model simulator? I guess is it fair to call it an econometric model?
“We’re able to not only very closely replicate the demographic patterns in the U.S., but also see how those households will make changes in the face of policy changes.”
Smetters: It’s more than just an econometric model. We use econometrics behind it to validate the model, but the model is very detailed in that at the very base level it is a demographics model that starts with census-level data, but covers more than 30 different attributes of families. We then age the people in the population, [and track] lots of different life decisions they make such as about how much to work, how much to save, whether they get married, how many kids they have, whether they get divorced, how much education they give their kids, and so forth.
So it’s a very rich model. We’re able to therefore not only very closely replicate the demographic patterns in the U.S., but also see how those households will make changes in the face of policy changes. The econometrics comes in as we take the model back in time to see how this fancy model would have worked back in time. How closely would it have replicated the U.S. economy, and especially the demographic?
We spent a lot of the time on just that validation, making sure that it would have done well historically before we use it as a basis for projecting forward. On top of that model, we have coded up lots and lots of different policy rules. We’ve coded up almost every single individual tax reform out there, so it’s like a TurboTax-like feature to the model. We have very detailed Social Security systems, and we’re building on top of it just much more detailed government policy systems.
Knowledge at Wharton: It is worth noting that anyone can go online and play around with this – correct?
Smetters: Yes … you can go right to the Penn Wharton Budget Model website and play with the different simulators by just literally moving different dial controls to see what would happen, for example, to Social Security. There are many different options there of how we could fix Social Security, and you could move those dial controls [accordingly]. There are more than 4,096 combinations with Social Security alone.
Because it’s a very deep model that does a lot of big data and a lot of complex theory and so forth, you don’t have to wait a half-hour, or an hour for your results. We use cloud computing to pre-compute every single combination ahead of time so that you get instant results. [It is] just like doing a Google search, where Google scrapes the web first and doesn’t literally go out and search it every single time you type in something. So we do it ahead of time so that you can get instant results, and do lots of what-if analysis.
