Wednesday, September 03, 2014

Second generation human capital benefits of migration

Summary: One important element in estimating the benefits of liberalized migration is the effect on the human capital of migrants' children. Taking human capital benefits of childhood in a developed country into account increases estimates of overall benefits, and may favor permanent migration relative to temporary guest worker programs. I discuss human capital benefits in health, language competence, educational attainment, gender equality, and attachment to location. Benefits are largest and important for immigrants with initially low human capital. Some but not all of these benefits can be attained using retained earnings of temporary migrant workers to help their children at home.


In a previous post about upward and downward biases in estimates of the GWP impact of open borders I mentioned that:
The place premium estimates match workers by education levels, so if the education levels or other skills of workers systematically changed, it would throw off the impact estimates. In particular, levels of education, host country language skills, and health would be higher in later generations after a massive wave of low-skill migration. 
Education levels in many developing countries are far below those in developed countries, and migrants close most of the extreme gaps in a single generation (or leapfrog the native population, in some cases). 
In the longer term, taking this into account should boost GWP impacts, but with a number of decades of lag for migration, childbearing, and propagation through the work force.

Such changes could be relevant in several respects to thinking about more liberal migration policies.
  1. As mentioned, GWP increases would be greater in later generations.
  2. Higher 2nd generation migrant human capital increases the ratio of high-skill to low-skill workers in destination countries, helping to cushion falling wages for low skill migrants and induce more total migrations.
  3. Higher human capital reduces the fiscal costs to natives of redistribution towards poor migrants.
  4. If high human capital has positive externalities, e.g. on institutions, then convergence helps to capture these benefits.
  5. While guest worker programs for low-skill workers, such as the massive ones in the Gulf states and Singapore, provide valuable income increases to many poor migrants, they may deliver less in the way of human capital benefits if migrants' children are not raised in rich countries.
This post will discuss some of the major channels for 2nd generation migrants to enhance their human capital relative to their parents: physiological health, language, educational attainment, gender equality, and attachment to location. It will also examine to what extent these benefits can be replicated by temporary migrants using earnings from their labor abroad, and to what extent they depend on children being brought up in developed country destinations.

Physiological health: disease, nutrition, pollution, public health
Rich developed countries indisputably make their populations biologically healthier, compared to the conditions of dire poverty in much of the developing world. One particularly striking metric for this is human height.

While within a particular rich country most differences in height between people of the same height and sex are due to genetic factors, and there are differences between populations around the world in the distribution of alleles affecting human height (e.g. within Europe there has been selection on alleles that increase height, increasingly so as one moves north, and this matches the phenotypic data), across countries and generations height differences are mostly environmental. Consider this graph from Hatton and Bray (2010):

data from Hatton and Bray: Long Run Trends in the Heights of European Men, 19th-20th Centuries Econ Hum Biol. 2010 Dec;8(3):405-13. doi: 10.1016/j.ehb.2010.03.001. url: http://www.ncbi.nlm.nih.gov/pubmed/20399715 data extracted from pdf: http://privatewww.essex.ac.uk/~hatton/Tim_height_paper.pdf . Code here: https://github.com/cooplab/Height_over_time

The book How Much Have Global Problems Cost the World? has an extensive discussion of the literature on malnutrition (often measured by height) and individual and national income by Sue Horton and Richard H. Steckel. Both at the individual level and the national level, differences in height are more predictive of income in developing countries where serious malnutrition is an issue. This graph of continental heights (note the data after 2010 are projections) showcases the nutritional gap between the global rich and poor.




Wikipedia's table of human height shows how African-American height (reflecting mostly West African ancestry with some significant European ancestry) far exceeds typical heights in Africa below the Sahara

Country/RegionAverage male heightAverage female heightStature ratio
(male to female)
Sample population /
age range
Share of
pop. over 15
covered[52]
MethodologyYear
U.S.177.6 cm (5 ft 10 in)163.2 cm (5 ft 4 1⁄2 in)1.09All Americans, 20–2917.40%Measured2003–2006
U.S.176.3 cm (5 ft 9 1⁄2 in)162.2 cm (5 ft 4 in)1.09All Americans, 20+91.00%Measured2003–2006
U.S.178 cm (5 ft 10 in)163.2 cm (5 ft 4 1⁄2 in)1.09Black Americans, 20–39N/AMeasured2003–2006
U.S.178.9 cm (5 ft 10 1⁄2 in)164.8 cm (5 ft 5 in)1.09White Americans, 20–39N/AMeasured2003–2006
Cameroon170.6 cm (5 ft 7 in)161.3 cm (5 ft 3 1⁄2 in)1.06Urban, 15+N/AMeasured2003
Ivory Coast170.1 cm (5 ft 7 in)159.1 cm (5 ft 2 1⁄2 in)1.0725–29N/AMeasured1985–1987
Malawi166 cm (5 ft 5 1⁄2 in)155 cm (5 ft 1 in)1.07Urban, 16–60N/AMeasured2000
South Africa169 cm (5 ft 6 1⁄2 in)159 cm (5 ft 2 1⁄2 in)1.0625–34N/AMeasured1998
Nigeria163.8 cm (5 ft 4 1⁄2 in)157.8 cm (5 ft 2 in)1.0418–7486.70%Measured1994–1996
Mali – southern Mali171.3 cm (5 ft 7 1⁄2 in)160.4 cm (5 ft 3 in)1.07Rural adultsN/AMeasured1992

In addition to the harmful effects of general malnutrition, deficiencies of micronutrients also impair health and productivity. For example, iodine deficiency can cause substantial decreases in IQ, and salt iodization is one intervention GiveWell is considering to help the global poor. The ICCIDD, which recently received a 'top charity participation grant' from GiveWell, provides this map of coverage for iodized salt consumption (the US and Canada do have salt iodized), showing that at this time iodine deficiency and the global poor are closely linked, although other regions still suffer effects from before they implemented iodization at various times:





Developed countries have already eliminated malaria and many other infectious diseases locally, and migrant children growing up in developed countries can develop their minds and bodies unhindered by these diseases. Deworming helminths (eradicated in rich countries) can deliver significant increases in test scores in at least some places. For other diseases extremely high levels of vaccination in developed countries provide herd immunity. [Map from New Scientist]



While increased income for individual migrants can let them pay for better nutrition and health care for their families back home, micronutrient deficiency and disease prevention benefit from national-level policies: when all salt is iodized by default parents don't need health knowledge and scarce attention to acquire it. Mandatory and broadly available vaccination programs or disease eradication efforts can only take full effect when undertaken for whole communities: vaccinating one's own family will not provide the full benefits of herd immunity.

These gains of coordination would favor permanent migration with children, relative to guest worker programs, unless migrants had much lower vaccination rates. However, in reality rich country governments such as the United States require vaccination as a condition of permanent migration.

Learning economically valuable languages
Language fluency is one of the most malleable features of workers while they are young, but is stubbornly challenging for the old, and so is a very important source of potential gains for 2nd generation migrants.

One Journal of Economics & Finance paper:
The results indicate that male immigrants suffer increasing penalties with decreasing levels of English skills. Among the male immigrants, those who are proficient, intermediate and poor in English will earn 15.9%, 27.8% and 28.8% less than their counterparts with English fluency, respectively, all others being equal. However, female immigrants who speak intermediate English suffer the greatest earnings penalty. Among the female immigrants, all else being equal, those with intermediate level of English skills earn 22.6% less than those speaking fluent English, while those who are proficient and poor in English earn 14.5% and 18.6%, respectively, less than the fluent group...
The economic penalty for being deficient in English is generally greater for immigrants
at the upper tier of earnings distribution and for each level of English proficiency (proficient,
intermediate, and poor). Specifically, among immigrants who are proficient in English, those at the 0.75 of earnings distribution quantile earn 16.9% less than their fluent counterparts, while 16.6% less for those at the 0.5 quantile and 14.4% less for those at the 0.25 quantile. Among immigrants who are intermediate in English, those at the 0.75 and 0.50 of earnings distribution quantile suffer a similar loss of earnings of around 27.5% and such loss of earnings is only 23.5% for those at the 0.25 quantile. Among immigrants who are poor in English, those at the 0.75 of earnings distribution quantile earn 29.1% less than their fluent counterparts, while 27.4% less for those at the 0.5 quantile and 22.3% less for those at the 0.25 quantile... 
In other words, the reward for English proficiency is greater for immigrants at the upper earnings distribution. The possible reason might be that fluency in English is more needed for the higher-ranking (higher earnings in general) jobs; while lower-ranking occupations, such as the goods production and assembly lines, are unlikely to be as affected by English proficiency.
Children raised in developed countries quite reliably learn the local language. A Pew report on 2nd generation migrants examined language skills in 1st and 2nd generation migrants to the U.S. of Hispanic and Asian backgrounds. It finds that ~90% of 2nd generation migrants from these groups are English proficient, vs 78% of Asian and 48% of Hispanic 1st generation migrants.
Those surveys, along with our analysis of U.S. Census Bureau data, also find that Latinos and Asian Americans differ markedly both in their language skills and in their views on the importance of maintaining the language of their ancestral home. While fully three-fourths (78%) of Asian immigrants in the U.S. speak English either very well or pretty well, about half of Hispanic immigrants (48%) are English proficient. This may reflect the fact that a large share of Asian American immigrants are highly educated and often come to the United States to obtain even more education, while a large share of Hispanics who come to the U.S. have less education and come primarily to work.

Among the second generation, half of Latinos have the ability to speak Spanish, while less than half of Asian Americans are proficient in the language of their ancestral home. And as with Hispanics and Asian Americans overall, second-generation Hispanics are far more wedded to future generations retaining their ancestral language than are second-generation Asian Americans.

Language Usage: About nine-in-ten second-generation Hispanic and Asian-American immigrants are proficient English speakers, substantially more than the immigrant generations of these groups. When it comes to retaining one’s ancestral language, there are sizable differences by race and ethnicity. Eight-in-ten second-generation Hispanics say they can speak Spanish at least pretty well; just four-in-ten second-generation Asian Americans say the same about their parents’ native tongue.
More liberal migration policies for the global poor would likely involve substantially lower levels of language proficiency, and correspondingly greater gains for the 2nd generation.

However, we might ask whether the same benefits could be obtained for children of temporary migrants left at home? English and other rich country foreign language training is available in many developing countries for those who can pay (e.g. using migrant earnings), so wouldn't the opportunity to achieve higher earnings lead potential migrants to learn the language themselves?
I am skeptical that such schooling-based approaches would come close to the effects of being raised in a country where the language in question is common. It is already common in developing countries for proportional returns to English proficiency to be in the range of those in the U.S.

A British Council report finds that:
[A] salaried professional with good English language skill can earn on average 30% more than someone with no English language skills in Nigeria, and 25% more in Pakistan. The average salary gap between professionals in these countries is lower, as their industrial and technological bases are typically less well-developed.
Parakash, Chin and Azam present some similar evidence for India (the left columns are for men, the right columns are for women):


These returns make me think that the opportunity for similar proportional income gains is not sufficient to make schooling-based language learning work reliably, given existing levels of proficiency, especially compared to the striking proficiency of 2nd generation migrants.

Gender equality and female labor force participation
Female labor force participation rates, as reported by the World Bank, vary dramatically around the world. In guest worker situations, this can lead to disproportionately male workforces. When whole families come without pre-arranged work, as in refugee cases, migrant employment rates (which differ from unemployment rates, as they count those not looking for work, such as students, retirees, etc) can be lower than the general population mostly because of varying female labor force participation. For example, while critics of migration have complained about much lower employment rates of refugees than natives, especially in Europe (employment rates are much higher in the U.S., over 50%, likely in part because of more flexible labor markets, reduced welfare benefits, and perhaps in part because of different migration patterns) this is mostly driven by female labor force participation. E.g. in the UK (one of the worst cases) bringing female labor force participation up to male levels among Somali migrants would almost double employment rates:
Somali-born migrants have the lowest employment rate among all immigrants in the UK.[37] Figures published by the Office for National Statistics show high rates of economic inactivity and unemployment amongst Somali immigrants. In the three months to June 2008, 31.4 percent of Somali men and 84.2 percent of Somali women were economically inactive (the statistics include students, carers and the long-term sick, injured or disabled in this group).[88][89] Of those who were economically active, 41.4 percent of the men and 39.1 percent of the women were unemployed. Employment rates were 40.1 percent for men and 9.6 percent for women. The male employment rate has, however, risen from 21.5 percent in 1998.[88]
Female education, improved labor market opportunities, and assimilation to local culture would all tend to raise female labor force participation in the next generation (and have done so for many migrant populatoins in the past), enabling increased economic output.

A PPIC study shows a major increase in female labor force participation among second generation Mexican-Americans, from 40% for recent immigrants, to 53% for earlier immigrants, to 68% in the 2nd generation:




Increased educational attainment
Understanding the effect of migration on the educational attainment of migrants' children is complicated by the fact that existing legal migration channels strongly select for high educational attainment. Exceptional educational success by migrant parents will be in part due to factors that are not inherited, such as lucky educational experiences, non-additive genetic effects, and favorable developmental noise. This would tend to drive some regression to the mean in the next generation, just as the children of PhD-PhD couples tend to have lower educational attainment and test scores than their parents. At least, we would expect this before accounting for different educational distributions in the two countries.

Feliciano (2005) raises an issue of interpretation for educational attainment and examines whether the degree of immigrants' educational selectivity-that is , how immigrants differ educationally from non-migrants in the home country-influences educational outcomes among groups of immigrants' children. This study uses published international data and U.S. Census and Current Population Survey data on 32 immigrant groups to show that as immigrants' educational selectivity increases, the college attainment of the second generation also increases. Moreover, the more positive selection of Asian immigrants helps explain their second generations' higher college attendance rates as compared to Europeans, Afro-Caribbeans, and Latinos.
Such a measure helps to distinguish between educational opportunities in the home country and the ability to take advantage of them. Compare an extremely able individual who grow up in poverty in a place where almost no one achieves a college education but perseveres nonetheless, to one who simply takes advantage of free high-quality government education in a rich country. If the Horatio figure's offspring manage to achieve a similar percentile in the educational distribution of a rich country, they would have to earn advanced degrees.

Feliciano's Table 1 shows the striking differences in selectivity, calculated using a net difference index (ND), for which "an index of .35 indicates that an immigrant's educational attainment will exceed that of a nonmigrant from the same country 35% more often than a nonmigrant's education will exceed that of an immigrant from that country."

The leaders in selectivity are Hungary, Indian, and Iran, with NDs of 0.85 or greater. China has an ND of 0.671, European countries cover a sizable middle range, Caribbean migrants are highly selected, and major US mainland migrant sources Mexico and Puerto Rico have the lowest selectivity at 0.208 and -0.064.


Taking this into account, it's worth paying attention to the distinction between highly selected and relatively unselected migrants, since extremely large migration flows necessarily could not be highly selected. Hispanic migrants are relatively unselected, and are broken out in the Pew survey data:




From the 1st to 2nd generation, high school completion improves from 53% to 79%, and college completion from 11% to 21%, while median income (not fully adjusted for age, which penalizes the younger 2nd generation) rises from $34,600 to $48,400.

Another study that specifically breaks out Mexican-American immigrants (for whom selectivity is lower than other Hispanic migrants) to the United States finds even more striking gains. The percentage of high school graduates was about 33% for earlier migrants, but in the 2nd generation this increases to over 70%! College degrees (including graduate degrees) roughly triple from ~4% to 12%. For migrant populations beginning from even lower bases, such as most potential migrants from the world's poorest countries, 2nd generation gains should be still greater.


Educational gains for 2nd generation migrants self-identifying as Asian, non-Hispanic White, or Black are much smaller. For Asian-Americans high school and college completion rise from 88% to 93% and 50% to 55%, while income rises from $65,200 to $67,500 (although there is a large 3rd generation increase which may reflect changing selectivity and flows over time). For Black migrants high schools rises from 86% to 95%, college from 31% to 40%, and income falls from $46,500 to $43,500. For non-Hispanic White migrants high school goes from 89% to 93%, college falls from 44% to 39%, and income from $60,600 to $63,200.




















The results for the highly selected populations are affected by the biases discussed above, figures are affected by intermarriage with natives who may be better situated to assist their children's assimilation, age and household size adjustments complicate the picture, and migrant flows and selectivity within these broad categories have not been static over time. Nonetheless, it seems clear that relatively (educationally) unselected Hispanic migrants, make very large gains in educational attainment and income (which also reflects going from 48% to 90% English proficiency). The highly selected migrant groups show modest gains, reflecting the limited room for further progress, but these may be attenuated by regression to the mean, understating the causal benefits.

Could migrants with low human capital, who would seem to constitute the vast majority of potential migrants under an open borders policy and receive by far the greatest gains in the 2nd generation, realize such gains by paying for their children's education using guest worker earnings? Increased income does lead to increased spending on private education and reduced withdrawal of children from school, but schooling quality is worse in developing countries and migrants enjoy better achievement than their counterparts at home in the same grade level (developed country education also contributes more to earnings than developing country education for the same nominal qualifications, in line with the test scores; some research finds that controlling for home country test scores/individual achievement eliminates differences in return to education). Private educational spending in poor countries is also lower than what rich countries use taxation and public provision to mandate.

On the other hand, the cost of developed country education is extremely high, and the investment does not pay off fiscally for decades. Rich countries might balk at the enormous fiscal costs of providing free education at current standards to the entire world population, which would reduce gains.

All told, it seems that the educational attainment benefits of temporary migration for work would again be at least somewhat less than those of permanent migration.

Attachment to location
As I discussed previously, even in the face of open borders and a sizable wage gap, such as the case of the United States and Puerto Rico, migration is slow, and can take generations. Increased income offers diminishing marginal returns, while life outcomes such as connection to family and loved ones, familiarity with local language, marriage prospects, shared religious affiliation, local folkways, and other factors can strongly root people to their homes. Many migrants say they would not migrate if they were making the decision again, and some psychological studies find significant costs in well-being for many migrants.

Stillman (2012):
compare[s] successful and unsuccessful applicants to a migration lottery in order to experimentally estimate the impact of migration on objective and subjective well-being. The results show that international migration brings large improvements in objective well-being, in terms of incomes and expenditures. Impacts on subjective well-being are complex, with mental health improving but  happiness declining, self-rated welfare rising if viewed retrospectively but static if viewed experimentally, self-rated social respect rising retrospectively but falling experimentally and subjective income adequacy rising. 
However, 2nd generation migrants grow up with their attachments to the new community and so do not need to pay the same psychic costs (although if the migrant community faces low status or discrimination this may cause other stresses), and are unlikely to migrate to the land of their ancestors while it remains undeveloped. This allows for stronger cumulative effects of migration. If in each generation the 20% of people with the most wanderlust migrate temporarily as guest workers, gains will be limited by that quantity. However if in each generation 20% migrate to rich countries, but few return, then over several generations most of the population descended from the sending country will be found in the receiving countries, as is now the case for Puerto Ricans, who are mostly found in the mainland United States after over a century of open borders.

While a large and accumulating guest worker community (as in Dubai) can bring many familiar elements of home to a new land, the ability to settle with one's family is essential to capturing this benefit.

2 comments:

Unknown said...

Thanks for the post, lots of interesting things here.

On the cumulative and indirect effects of migration, do we know much emigration from a country affects population growth there? I assume that more space available will tend to lead to higher birth rates than without people leaving, but this might easily be overcome by the increasing wealth from remittances.

Carl said...

Owen, migrants tend to be young, i.e. in their fertile years, so their movement can have a disproportionate impact on population growth. In the new country they tend to assimilate towards local fertility rates, although this varies by community and takes time.

I haven't looked into empirical work on the effect of emigration on fertility among those left behind, save through the change in the demographics at home.