If labor was perfectly mobile and suddenly there were massive dislocations requiring workers to relocate, you’d expect an uptick in the number of workers moving, right? A corollary is that if you do not see an up-tick in the number of workers moving, then either there was no massive dislocations or there is not perfectly mobility. Take the second case: with imperfect immobility workers would like to take jobs in other places, but they can’t. This looks like sticky wages. Sticky wages make aggregate demand effects important. These data suggesting labor mobility did not increase in the last recession, then, are consistent with both structural and cyclical problems in the labor market.
I was at a practice job talk at a nearby university on Friday. As preface to my work ((I’ll post on it as soon as I get my job applications sent)), I pointed out that workers change careers many times, 3 or 4 times, over their lifetimes where career change is defined as an occupation change that requires different tasks performed on the job, e.g. taxi driver to nurses assistant. BTW, workers change occupations about once every five years so only about a third of these are career changes. One of the audience members, an academic, objected saying that nobody he knows has ever changed occupations (not just careers, occupations).
On that note, I am sure everyone would find this MR linked paper useful: goofy titles get cited less often. At the link, its suggested that this is because “science is serious business”, but in my experience this result stems from the fact that most of these titles are “funny” but not funny. The problem is that the paper only looks at the effect of average “funny” titles. Since the average “funny” title isn’t funny, just “funny”, its picking up the effect of being “funny” not funny. I’m sure there are some really funny titles that get extra cites because of it.
Which paper titles are funny and not just “funny”?
Will Wilkinson has been schooling the internets about
division properly controlling for inflation when figuring real incomes, eliciting bored nods from those that have actually read Broda/Weinstein/Romalis (no links… too boring to write a post about) and confused indignation from those that want to talk about everything except real income inequality. Anyway, this prompted me to take another look at Broda’s (the intellectual ring master of this set of papers) list of working papers.
I found this interesting paper. Broda and Weinstein do the same sort of analysis for Japan that Broda/Romalis did for poor people in the US: he generates price indices (hah!) controlling for substitution and quality issues (but leaving out the fancy math to account for new products). Japan doesn’t make these corrections when calculating its CPI and so the authors just replicate the changes the BLS made in calculating the US’s CPI after the Boskin report but for Japan. They find that deflation was understated:
Ignoring the factoid that “deflation is bad”, this actually means growth in real consumption per person in Japan was higher than previously thought. Instead of growing at a rate 2% slower than the US, using the uncorrected Japanese official statistics, real consumption per capita has been growing at a rate only 0.7% slower than the US.
Here’s the per capita GDP for the G8 countries (relative to the USA):
You can take two lessons from this chart.
1) Japan hasn’t had bad monetary policy relative to, say, the continental European countries
2) Monetary policy just doesn’t matter in the long-run
Also, the more you stare at these real GDP per capita charts, the more you think the 1980s in Japan was an outlier (aka bubble).
Here’s all the Jolts data (minus job openings):
These are not seasonally adjusted so that’s why you get the inverted-U shape every year for the hiring and quits series and the inclined saw shape for firings. Still, you can see something happened the last two or three years. Firings (the pink line) spiked in early 2009 but several months before that quits (blue) and hirings (green) declined by about 20 to 30% each. Here’s a close-up of the most recent period:
A couple things that strike me about these graphs:
1. Firings had a one month spike in January 2009 (a month with a high number of firings to begin with) but the series has stayed about where it usually is (maybe 10% higher if you ignore the recent dip). The story of this recession isn’t of people losing their jobs.
2. Looking at the dramatic decrease in hirings, the story of the recession is that people aren’t finding new jobs.
3. However, because there’s much, much fewer people quitting their jobs the job market isn’t as tight as it could be.
One of the things I do in my dissertation is treat changing jobs as an investment decision. When a worker quits their old job and looks for a new one (maybe not in that order), they are forgoing human capital that made them productive at their old job and they’re investing in new human capital to make them productive in their new job. One possibility — given the JOLTS evidence and thinking of job switching as investment activity — is that like employers who are holding back on new invests (and thus holding back on hiring), workers are holding off on making new investments as well.
I’m not sure about the implications for aggregate demand, but I’m pretty sure this is bad for growth prospects.
A few weeks ago, Mike was asking me about the long term unemployed. I’m running some regressions using CPS data so I have the data he asked about just hanging out on my desktop. Public service:
The unemployed during this recession ((I looked at 2008 and 2009 data for folks that worked one week or more)) are different than those that don’t become unemployed. They’re younger, less educated, more likely to be male and less likely to be married.
Now, the long-term unemployed (greater than 25 weeks unemployed) are different from the short-term unemployed. They are even less educated, less likely to be white, less likely to be married and slightly older.
The never unemployed average 40.1 years old, the short-term unemployed 35.5 and the long-term unemployed are 36. Contrary to Tyler Cowen, unemployed workers above 40 with a college degree are not much more likely to be long-term unemployed (they’re 24% of the 40+ long-term sample compared to 21.9% of the 40+ short-term unemployed sample). Relative to all college educated workers, college educated workers older than 40 are just more likely to be unemployed.
Last time, I said that nobody really cares about the direct or partial effect of immigration on natives. This isn’t strictly true. If immigration policy were changed to only allow immigrants of a particular skill-type to come into the country, this would probably upset the people of that skill-type. Each individual immigrant hurts the employment opportunities of natives with the same skills as the immigrant. Immigrants, in total though, can have a positive impact on natives because immigrants have different types of skills. People, immigrant or not, that have different skills tend to compliment each other in the production process. They make each other more efficient.
Perhaps this is the source of the assumption of substitution I asked about before: we tend to imagine groups of people as being homogeneous masses, not individuals with a variety of skills. If immigrants were all the same, with the same skills, and natives all the same, with the same skills, and if those skills perfectly overlapped, then immigrants would compete directly with natives and hurt their employment opportunities. In reality, though, a random immigrant has different skills than a random native. Because they have different sets of skills, this random pair do not compete with each other in the labor market. In fact, on average, a random pairing of natives and immigrants would be more productive than the sum of its parts.
Even in the extreme case where all immigrants had exactly the same skills, this would only hurt the group of natives that had those skills. Everyone else would benefit from having co-workers with complimentary skill sets. Its a common belief that all recent immigrates have been very low educated. This, it is imagined, would hurt very low educated natives. The distribution of education among recent immigrants, however, has been U-shaped. There have been many very low educated immigrants and many very highly educated immigrants.
Even among natives with no high school degree, the pain of immigration has not been felt. There are two reasons for this. First, because it turns out that workers with no high school degree are strong competitors with those with a high school degree in the labor market, very low educated immigrants compete with a much bigger pool of natives, spreading out the pain of direct competition. Second, the pool of low educated natives benefit from the large group of highly educated immigrants as highly educated workers make the low educated much more productive. In other words, the negative effects of low educated immigrants on low educated natives are small and spread out, but the gains from high educated immigrants are acute. On net, the gains swamp the losses and even low educated natives benefit from immigration.
Capital accumulation also plays a key role. Like the Econ 101 growth theory would tell us, immigration, by increasing the labor supply, increases the return to capital. This will give investors an incentive to invest more and the capital stock will grow. Because capital makes labor more efficient, it would increase wages until they returned to their long-run value ((This means when capital fully adjusts, the average wage effect of immigration is zero. Different groups might be affected differently, but its a wash on net. It turns out all groups of natives benefit even if slightly from immigration. The only group that is hurt by new immigration is old immigrants.)). This undoes some of the negative direct effects of immigration, too.
All of these empirical claims were verified in a paper by Ottaviano and Peri just a couple of years ago ((A good discussion of that paper can be found here. The author reproduces many of Ottaviano and Peri’s findings using a slightly different data set.)). A random immigrant makes a random native more productive because the native is unlikely to have the same skills as the immigrant and because natives and immigrants, even if they have the same education and experience, are not perfect substitutes. Next time we’ll see that this is because immigrants specialize in different tasks than natives and because they induce innovations that complement their skills.
It has taken me too long to introduce the hero of this story. Borjas has had a series of papers, books and editorials since the mid-80’s that have each challenged the consensus economist’s view on immigration in support of the popular view. His work in the late 80’s uncovered the pattern of a secular decline in the quality of immigrants (in terms of educational attainment, work experience and unobserved skills) since WWII, a trend that is common sense today. He also introduced the idea of self-selection of immigrants and gave conditions for when we’d expect immigrants to be higher or lower quality. In the nineties, while setting the ground work for the work I’ll describe below, he started the conversation on the costs and benefits of immigration, sitting on a panel of the National Academy of Sciences evaluating the economic impact of immigration. He wrote papers encouraging researchers and policy makers to balance the fiscal costs of immigrants (welfare, schooling, etc) against any potential production gains.
In his 2003 paper called “The labor demand curve is downward sloping,” Borjas challenged the consensus that immigration has a zero effect on native employment outcomes. He worried that studies comparing regions did not properly account for spillover effects from trade, capital movements and labor mobility. The problem with goods, capital and labor movements is that they all tend to equalize wages across regions. This means that when you compare regions that have had more or less immigration, you won’t be able to find a difference in wages even as everyone’s wages have gone down.
He suggested, instead, to compare groups that do not admit mobility. Specifically, he looked at the effect of immigration across skill groups where skill is defined by a combination of education and years of work experience. Intranational trade in goods doesn’t operate in this context. Add to that the fact that international trade is probably not important here and the trade channel for spillovers is shut down. Also, workers can only increase work experience one year at a time (thus they can’t decrease it or move to a level of experience more than one year ahead of their own), so its extremely hard for them to change it to respond to incentives and, in the short run, workers can’t change their education. These things suggest the labor mobility channel is shut down, too. Finally, if you assume, as Borjas does, that capital is not skill specific (or at least to the degree it is, it has a small spillover effect) then there’s no role for capital movement to cause spillover effects.
Below I reproduce his suggestion. On the x-axis is the change in the proportion of immigrants from 1990 to 2000 and the y-axis is the change in native wages. Each dot is a education-experience group ((Actually, I cheated and used age instead of work experience which gives the same results but is much easier to calculate.)).
The slope here is negative (-0.35) and quarter standard deviation more negative than the mean found by Longhi, Nijkamp and Poot. This result, which controls for spillover effects, confirms what Borjas says: “immigration lowers the wage of competing workers”.
There are three ways to critique these findings. First, attack the premise. If spillover effects are creating the zero correlation at the local level, we should be able to detect them. We should see native workers fleeing areas of high immigration or we should see industries in high immigration areas reshuffle to take advantage of the skills of immigrants, as trade theory would predict. David Card took up both of these challenges. In 2001, he wrote a paper finding no native flight and in 2007 he wrote a paper finding that least one type of trade spillover doesn’t seem to be emperically important ((That said, there is an opportunity for us to look for other ways trade affects native workers. For example, perhaps immigrants produce new varieties of goods. Much attention in trade theory these days — someone got a Noble for his work on this — is on the trade effects of new varieties. This wouldn’t be a spillover channel, I don’t think there’s factor price equalization in New Trade models, as much as a possible explanation for native/immigrant complementarity, which I haven’t talked about yet. I’m no trade theorist but perhaps the production based models used by Ottaviano and Peri, e.g., are isomorphic to the love of variety trade models. So where those authors see complementarity in production, trade models would see love of new immigrant produced varieties. We’d need micro evidence on immigrant entrepreneurship to differentiate to see which of these models best explains the evidence.)). If spillovers are a problem, they don’t appear to work through the trade and labor mobility channels.
Second, attack the assumption that capital isn’t skill specific or that international trade doesn’t matter. To be honest, I haven’t found any work that does either of these things…
So moving right along. Third, point out that there are much more important indirect effects of immigration than spillover effects. This will be the focus of my next post.
For those of us that believe immigration has a net positive benefit on society, Borjas is the perfect enemy ((He’s also a model academic. Besides his volumes of academic papers, he’s written a well-used textbook, several popular books on immigration and dozens of editorials on the subject)). He has applied honest, fair and constant pressure on our sure beliefs about immigration. His challenges have inspired reams of rebuttals; reams that could not otherwise have been used in support of immigration.
Here’s the cross-state plot of the effect of immigrants on native wages:
This time you’ll notice that I plotted the changes in the proportion of immigrants against the change in wages (these are changes from the 1990 to 2000 censuses). In effect, I’m controlling for fixed features of states. This makes the effect size (a slope of about 0.25 and statistically insignificant) a little closer to the average effect size reported by Longhi, Nijkamp and Poot.