Net Migration and Economic Growth Around the World – 1958 to the Present
In my last post, I used World Bank data to look at the effect of net migration on economic growth. Net migration is defined by the World Bank as the number of immigrants (coming into a country) less the number of emigrants (leaving the country). I showed that net migration as a share of the population in 2012 (the last year with for which this data has been reported so far) is negatively correlated with growth of PPP GDP per capita from 2012 to 2015. In other words, countries where the share of immigrants as a percent of the population was larger grew more slowly than countries with a smaller proportion of immigrants.
The natural question is… does this relationship hold over a longer period of time? In this post, I will show that the answer is yes.
As to data… I will use three series compiled by the World Bank: net migration, population, and PPP GDP per capita. Net migration data is reported every fifth year beginning in 1962, and it covers five years of activity. In other words, the net migration figure for 1962 is the sum total net migration for the years 1958 through 1962. Similarly, the net migration figure for 1967 is the total for the years from 1963 through 1967. Population is available annually going back to 1960. PPP GDP per capita is available annually, but only begins in 1990. To maximize the use of the available data, and still avoid situations where growth could be leading immigration, I looked at total migration from 1958 to 1992 as a share of the population in 1992, and compared it to growth in PPP GDP per capita from 1993 to 2015.
In other words, I took a look at (roughly) the percentage of the population that had migrated over 34 years, and compared that to the growth rate from the following year to 2015, which is a period of 22 years.
There are 155 countries for which the full sample of data is available. Here’s what that graph looks like:
The correlation is -0.10. Not huge, but negative. In other words, on average, for this large sample of countries, the the more immigrants arrived over a 34 year period (relative to the country’s population), the slower that country grew over the next 22 years. This was a weak relationship, but it certainly doesn’t support the notion that increased immigration leads to more growth.
Nevertheless, the sample contains a lot of countries that won’t provide many useful insights to industrialized nations like the US. Much of the sample is made up of countries that are very small. As of 2015, 59 countries have fewer than a million people, 24 have fewer than 100,000 people, and one (Tuvalu) had less than 10,000 residents. Additionally, many of the countries have very, very different economic systems, and in a few sad cases, nothing resembling an economic system whatsoever. The next graph includes only countries that the World Bank defines as being High Income, and which had at least 1 million residents in the year 2015.
Notice that the sample size drops to 36, and the correlation becomes -0.67. But… you’d be forgiven if you decided this result must be driven by outliers. For example, the point sitting to the far right of the graph is the United Arab Emirates. According to the CIA World Factbook,
immigrants make up almost 85% of the total population, according to 2015 UN data (2016).
On the far left side of the graph, we see Trinidad & Tobago, whose net migration from 1958 to 1992 equaled -23.5% of its 1992 population. In other words, roughly a quarter of its people had left the country over a generation and a half. More colorfully, Trinidad & Tobago lost the equivalent of the population of Tobago several times over from 1958 to 1992.
Limiting our sample to countries with a more normal range of net migration / population (pruning outliers one at a time suggests the range should be from -15% to 15%) gives us this graph for wealthy countries with population in excess of 1 million:
Our sample drops to 29 countries, and the correlation between the two variables is -0.6. This tells us that the UAE and Trinidad (with or without Tobago) were not really outliers but rather the most extreme observations on the curve. And what the curve shows is that among large, high income countries, on average, nations with the most immigration as a share of their population from 1958 to 1992 also happened to be the countries with the slowest growth from 1993 to 2015. This is similar to the result of immigration on growth that we saw in the previous post when looking at the short term. Coincidence? I don’t think so.
I note that I played around with the data a bit, changing the dates, etc. Doing so moves the correlation up or down, but the negative correlation remains with any sample bearing a resemblance to the one selected. Of course, violently heroic cherry picking can provide the desired, politically correct results, but they resemble violently heroic cherry picking when you do it. I will happily send you my spreadsheet if you want to try it yourself. Or, you don’t have to trust me at all; nothing I did is difficult to replicate.
Moving along, what policy prescriptions does this suggest? Well, it is clear (not from this data, but it is obvious) that some immigration is very, very good for a country. Many an immigrant have contributed unbelievable benefits to their new homes, both for the US and for every other country in the sample. But in recent decades, the data seems to show that the positive impact of many immigrants seems to be more than outweighed by the negative impact of many other immigrants. The trick is to figure out what causes some immigrants to have negative outcomes and others to enjoy positive outcomes. Is it cultural values? The skills an immigrant brings? Family connections? Whatever it is, we should gear our immigration programs toward generating growth and benefiting the country. Right at the moment, it seems most wealthy, large countries, to the detriment of their citizens, are doing very much the opposite.
—
Notes:
As always, if you want my spreadsheet, drop me a line at mike dot my last name (only 1 m in my last name!!!) at gmail. I’ve made the spreadsheet interactive so you can modify dates, weed out countries by various factors, select what goes on what axis, etc. If you contact me within a month of the publication of this post, I will send you the spreadsheet. If you contact me several years later, I reserve the right to have a computer crash n the interim, etc.
Data sources:
1. Net migration, by country available here. The most recent data is from 2012. Net migration is defined as the “total number of immigrants less the annual number of emigrants, including both citizens and noncitizens. Data are five-year estimates.” As an example, the US reportedly had net migration of 5,007,887 (i.e., positive) in 2007 through 2012, while Bangladesh had a figure of -2,226,481 (i.e., negative) in the same years.
2. PPP GDP per capita. Data available here. The last year for which data is available is 2015.
R-squared values?
Put in the actual PPP GDP per capita, and run a mutlivariate analysis. I suspect you will find, as Picketty points out, that the more advanced a country, the slower its growth. But people want to move from poor countries to rich ones, even if the poor ones are growing more quickly.
Many factors affect per capita GDP growth, and many affect immigration patterns. Making one-to-one comparisons in such instances is, frankly, bollocks.
Warren,
1. No regression was run. No R-squared, adjusted or otherwise.
2. Your explanation is one I have heard before and it works in theory but not in practice. Here’s the problem. You would expect the larger absolute value correlation to occur in the first graph, which includes a mix of advanced countries, developing countries, and undeveloping countries. Conversely, the correlation should be lower in the second and third graphs which are more heavily weighted toward advanced countries and therefore should display less variation in growth, regardless of the variation in net migration / population.
I don’t want to derail this thread, but I will note that this developed-countries-slow-down-meme depends on assuming that developed countries haven’t adopted silly policies that slow the hell out of their growth. Advanced countries can afford to shoot themselves in the foot repeatedly.
The what is this “correlation” number? If it is not the slope of a linear regression, what is it? What algorithm did you use?
This is a pox on this blog.
I do not even bother tot read them anymore.
Eventually, someone will ignore the point of this post and figure out that the data is pure bs.
When that is mentioned, Mike’s wife will show up and say “he is at the work” and will correct it when he comes back, and then he never comes back.
I see no reason for this man to be posting here.
Warren,
As to correlations, its just the correlation coefficient returned when you enter “correl” in Excel. More here, including the formula Excel uses: https://support.office.com/en-us/article/CORREL-function-995DCEF7-0C0A-4BED-A3FB-239D7B68CA92
Now, the R squared is related (https://www.quora.com/Correlation-coefficient-vs-coefficient-of-determination-whats-the-difference-in-simple-terms) but you don’t need to run a regression to compute the correlation.
If you don’t like the measure of the relationship that I report, ignore it. Look at the points on the graphs. Do the graphs look like people like they are telling the “immigration –> faster economic growth” story? Or do they look like they are telling the opposite story? In these posts I am purposely trying to present the results in a graphical and intuitive manner rather than pumping the data through R and doing something more sophisticated. A two d scatter plot is harder to pull sleight of hand and anyone can replicate it.
EMichael,
I’ve been writing posts since 2006. A fair amount of what I write is original data work. And I try very hard to be open and clear about what I did, going so far as to offer my data to anyone who wants it. So yes, I made a mistake. Over an 11 year span it will happen. Someone noticed the error. Which is fine. I have more of an interest in understanding the world than do in advocating a position.
Now… this post is no different than any others I wrote. I am being just as open about sources and method, and am offering up my spreadsheet. I am presenting the data in a very simple way. So my question to you is this, what happens if nobody finds the error that you are so sure is there?
Keynes gave some great advice about what you should do if you find yourself in opposition to the data. If nobody finds the error that you so know is there, will you follow Keynes’ advice? Or are you here to advocate a position without any interest whatsoever in whether that position fits the facts?
EMichael is way out of his league here, Mike….he should just be directed to go back to Daily Kos, where they censor any material that might be found confusing by snowflakes like him…
RJS,
I am more than willing to admit that in terms of the statistics in these posts I am way out of my league.
But I can read.
And throughout this series of posts, almost all of them have been corrected by others with knowledge of these statistics. As an example:
ex-gf
December 27, 2016 1:45 pm
Sebastion,
Mike is at the work. He seems to have left his sources off the post for which he probably feels mortified.
I can tell you that the data for this post comes from Census historical table P-04. Data is for individuals.”
Those errors, like these posts, show a clear pattern.
Mike,
Any migration from poor countries to rich countries will produce the data you are charting. The larger the migration, the more the denominator increases and the more per capita growth decreases.
EMichael,
I have to imagine there is at least one person reading your comments who is certain you are a sock puppet I cooked up to play some sort of “how insane does this sound?” game. And honestly, if I didn’t know any better I would conclude the same thing myself.
Arne,
This is the point I have been making now over s very large number of posts: moving someone from country X to country Y, they do not instantaneously endow them with the characteristics of people in country Y. Be careful. That is enough to get you labeled a racist by folks like my sock puppet, er, EMichael.
Mike:
EM is real.
“This is the point I have been making”
I don’t think so. You are suggesting human reasons for the correlations you see. I am saying you are not (necessarily) showing anything other than arithmetic artifacts.
Arne,
In that case, you are wrong. If someone coming Somalia is, upon his/her arrival equal in productivity to the average American, GDP per capita in the US should not change when that Somali arrives. That is the politically acceptable story – people are exactly the same. It is the argument that people like Beverly and EMichael and Longtooth and others have been making in comments in my posts. And if you don’t believe that everyone is the same, by definition you are a racist.
Now, there are two options other than “the same.” If the Somali is more productive than the average American, GDP per capita should go up when the Somali arrives. On the other hand, if the Somali is less productive than the average American, GDP per capita should go down when the Somali arrives.
Multiply that one arrival by however many others, and you get the results in the graph in this post and the results presented in various other posts so far.
Arne,
A follow-up to my example of the recent Somali arrival – I have been arguing that people have cultural baggage that can persist for multiple generations. So immigrants with a skill deficit often pass on that skill deficitto their children, and immigrants with a skill surplus also often pass on that skill surplus to their offspring. Immigration today affects productivity for a long time. The post above shows that to be the case, with “a long time” lasting at least one generation. From casual observation I bet it lasts longer than that, but that’s the extent I have from the World Bank.
“a long time” lasting at least one generation.” I have been saying this.
Deficits are much dependent upon the environment people end up in and the US is adept at creating and maintaining those environments.
Run,
I disagree. IIRC one of the newspapers had a story a year or two ago about a guy with 8 kids with one wife and however many more with another wife who was still in Syria. He was a goatherder in Syria, and he and his wife in the US were functionally illiterate. He wanted his second wife to be admitted to the US as well, along with kids from that marriage. Meanwhile, in the US he was having trouble learning English and couldn’t hold down a job.
I am willing to bet he and his wife also weren’t helping their kids with their homework, etc., despite having plenty of free time on their hands. In other words… whatever deficit there was would last multiple generations… and not through any lack of effort or well-meaning intention from Americans.
Mike:
Look at what you are doing and saying:
“8 kids with one wife and however many more with another wife who was still in Syria and he and his wife in the US were functionally illiterate. He wanted his second wife to be admitted to the US as well, along with kids from that marriage. Meanwhile, in the US he was having trouble learning English and couldn’t hold down a job.”
versus Reagan’s account:
“”There’s a woman in Chicago,” Reagan said, according to an article in the now-defunct Washington Star. “She has 80 names, 30 addresses, 12 Social Security cards. … She’s got Medicaid, getting food stamps and she is collecting welfare under each of her names. Her tax-free cash income alone is over $150,000.”
Dog-whistle politics targeting a sub group. I am sure these people exist in one fashion or another; but to say, this is an example of them all? No, not true. you original comment; “a long time lasting at least one generation” is true and typically you can see progress with the children in one generation.
Trump is creating enough villains in the US to deflect upon what his real intentions are. We do not have to create more.
Run,
What do you do with people who live up (or down) to a stereotype? Here is a story out of Germany about a refugee with 4 wives and 22 kids. The government of Germany is not talking but a financial manager estimated the family gets 360K Euro a year. That is paid for by someone’s taxes. His income back home came from a “car sharing and car service business” so in Germany, assuming he gets employed, it will be as what? An Uber or Lyft driver? Nothing wrong with being a driver for Uber or Lyft, but there is no shortage of them in Germany, I imagine, and more of them will drive down wages.
The next question… how much time and effort do you think he will invest in his kids? If the answer is he “less than his neighbors” then you are also saying the odds are his children will underperform, ceteris paribus. So his bad habits get passed on.
The point of demonizing externality-inducing behavior is not to to tar people who do not engage in that behavior. It is to eliminate the incentive of engaging in that bad behavior. A guy who creates a situation for himself where he will not be able to provide for his children under any situation other than “I just won the Powerball by myself” is harming his children and the taxpayer. And as I keep pointing out, that harm lasts a very long time.
One more thing – wasn’t overpopulation something to be avoided some years ago? After all, think of the resources and pollution. How do you avoid overpopulation if you reward people who do it?
Mike, Mike, Mike:
More Pink Cadillacs? So random examples exist in some form or another. Somehow in the past, people did have rather large families and they managed; but then, a family could get by on one income too. Families today expend as much time as possible on families considering most are two-income families. We can spend hundreds of millions for defense without argument and argue about expenditures on domestic issues other than to ban and deport people. The vast majority of which are not in the category you place them in today.
“A Malthusian catastrophe (also known as Malthusian check) is a prediction of a forced return to subsistence-level conditions once population growth has outpaced agricultural production.” Think we are there yet? In the US the 300+ million live on 5% of the land mass. In Germany and pre-migration, the death rate exceeded the birth rate of Germans. Without an influx of new, they were destined to be a country without labor to replace their own. I have no worry about the Germans, they are more than capable of deciding their needs. Germany birth rate 8.2/1000 followed by Italy and France. Germany surpassed Japan for the first time ever (2015) with the lowest birth rate per thousand.
By 2040, the White majority will be a minority. Not worried about taxpayers as much as basing policy upon pink Cadillacs and random examples going forth and pissing away $billions on war/defense, tax breaks for a few at the tip of the pyramid, stoking the fears of the surrounding population about minorities, walling off more cities economically, etc. There is a need to work through things and change behaviors while maintaining a stable population.
Urgh… poolry phrased sentence above…
“How much of his time and effort will each of his children get, relative to the time and effort his neighbor’s children will receive from their parents?”
mike,
Oh I am real, though the last several months reading your posts sometimes makes me think I am living in a dream(well, nightmare actually) where my typing of angrybear.com sends me to breitbart.com
Your variable “Annualized Delta PPP GDP per capita (1993-2015)” could be affected by factors other than immigration. I believe that your variable was also affected by the free trade treaties that were put into effect after 1993. (Beginning with NAFTA in 1994.)
I believe that those treaties were responsible for the “output gap” which appears shortly after the beginning of the 2008 recession. It did not appear sooner because the problem was being masked by low interest rates and a huge housing bubble which enabled a huge debt bubble.
Our situation became SIMILAR to the one that Charles E Persons described in “Credit Expansion, 1920 to 1929, and its lessons” in 1930. New sources of credit came into being and allowed consumers to pull forward their spending, which allowed producers to expand rapidly. Once consumers had reached the limit of their ability to borrow more, then spending returned to that allowed by their income minus the payments on their debt. Producers did not recognize what was happening until they were stuck with excessive inventory.
Charles E Persons words: “The essential point here is that during the period of introduction of these new financial devices and while the newly opened reservoirs of credit are filling, we have a temporary increase in the nation’s purchasing power. A combination of circumstances has rendered this expansion of large dimensions in the decade just closed. While the new credit is expanding to its justified limits, more than the normal of commodities may be sold. The industries affected are stimulated to expand their productive capacity and to build up their labor force to the level set by this temporary bulge in consumption. In the working out of competitive forces their great prosperity leads to expansion even beyond this limit as over-eager producers seek an enlarged share of the market and profits. Once the newly developed credit resources reach maturity, new debt created is balanced by installments due on previously assumed obligations. The nation can buy only such volume of goods as is covered by its current income. The check to expansion is sharp and is intensified by the excesses inevitably associated with periods of over-rapid expansion. Such a course of events is clearly proven by the evidence as to credit expansion in the period 1920 to 1929.”
Our situation was a little DIFFERENT in that the economy did not grow rapidly because of the debt. It just did not decline as it would have without the debt.
Quoting Larry Summers: “One of the reasons for my concern was that before the financial crisis, when we had the mother of all bubbles in the housing market, that was enough to propel growth to perhaps adequate levels, but not enough to produce any kind of overheating as measured by wage and price inflation or as measured by unemployment relative to traditional low points. Imagine the economy between 2003 and 2007 without the consequences of housing bubbles and overly easy credit. Housing investment would have been two to three percent of GDP lower, and consumption expenditure would have been considerably lower, as well, resulting in very inadequate performance.”
See: https://www.washingtonpost.com/news/wonk/wp/2014/01/14/larry-summers-on-why-the-economy-is-broken-and-how-to-fix-it/
Summers uses 2003 as a starting point but I believe that the starting point should be 1996 or 1998 at the latest. (Using data from the Greenspan & Kennedy study of home equity withdrawal.)
See page 16 table 2 line 1: https://www.federalreserve.gov/PUBS/FEDS/2007/200720/200720pap.pdf
You are depending on a correlation statistic, but that statistic can not separate out all the causes for changes to your variable for GDP per capita. Our world is messy, assuming a single cause for anything is usually wrong. In this case, the problems caused by the trade treaties must be at least as large as the problems caused by immigration. I believe they are much larger.
I am sympathetic to your argument that culture matters. But above all else, I don’t believe that you can prove your point using so few data points at a time of massive changes in the US economy. (Poorly policed global free trade, poorly policed residential borrowing, interest rates so low that speculation in one area or another was widespread, and eventually outright fraud.)
JimH,
Thanks. I will try to take into account other variables, including credit related variables in future analyses.
As to your last paragraph… culture is a hard variable to proxy. But I can keep piling up coincidences for which one of the few explanations that works is culture. That’s what I have been doing, to EMichael’s annoyance.