Thursday, March 16, 2006

More On US Demographics

by Edward Hugh


"Is American society really so resilient that nothing can shake its foundation?" asked Claus in the last post. Well the answer obviously is no, since no society is that resilient. But is the US pretty well insulated from the current round of demographic shocks? This might be a more interesting question. I think the answer would be probably "yes", at least up to 2020 (the same probably also goes for the UK and France which are going to encounter relatively benign ageing in the short term). As ageing expert Axel Borsch Supan never tires of saying, Germany is now where the US will only be 20 years from now.

One part of the reason for this important difference is obvious, and the explanation comes in one single, simple word: immigration. These three societies (UK, France and the US) have relatively higher fertility rates due to their previous significant immigration. Let's look at the US case.

In the first place why is US fertility so different from many of the other OECD societies? Well some explanation can be found here.

And part of the explanation for the variation in regimes across ethnic groups comes from the fact that one of the groups (the hispanic one) is composed of a large number of relatively recent migrants.

Now in order to try and understand this process my first exhibit will be a graph of US fertility across the years. As can be seen, fertility was declining steadily until the late 80s when, surprise surprise, it starts to pick up again. What a coincidence that this revival comes at exactly the time when large scale irregular migration from Latin America begins to take off!














There is a clear break in the mid-seventies. Part of the reason for this is the start of large scale inward migration, and another part is probably a steadying up in the birth postponement process which had probably been in operation for some time, and with this steadying up comes the arrival of the missing births which had been displaced, and with these an upward turn in the TFR.

My second exhibit is a graph of US population by age structure. From this it is clear that something interesting happened between 25 and 30 years ago, since there is a distinct kink in the graphs.















And the third exhibit is a graph of foreign born population who entered the US between 2000 and 2003 by age structure. From this you can clearly see the concentration of recent immigrant arrivals in the childbearing ages.














12 comments:

Anonymous said...

My second exhibit is a graph of US population by age structure. From this it is clear that something interesting happened between 25 and 30 years ago, since there is a distinct kink in the graphs.

I wouldn't look at the kink so much as the peaks for those born in the late 1950s, and then again in the late 1980s. And then, I would look at the age structure graphs for "old" and "new" Europe in the report that captvk linked to on March 9:

http://europa.eu.int/comm/economy_finance/publications/economic_papers/2005/ecp236en.pdf

And then, I might observe the following: (1) The American boom generation that peaked in the late 1950s did succeed in replacing itself, more or less, eventually. (2) In contrast, the western European post-war baby boom did not peak until the mid 1960s, and this boom generation did not replace itself. (3) Interestingly, the EU-10 countries had their baby boom in the 1950s, and were replacing themselves quite nicely until the social system collapsed around 1990, and ever since then have been among the least fertile countries in the world.

Now here is a paradox: the U.S., which suffered significantly less than either western or eastern Europe during World War II, and has among the least developed social safety nets in the industrial world; and the Eastern bloc, which suffered immensely both during and shortly after World War II, and spent a fraction of its GDP on social programmes that would be unthinkable today in even the most thoroughly center-left of developed nations, had qualitatively similar demographics right up until Communism imploded, while western Europe boomed later than either, and had no echo boom.

I think these graphs also show that it's still too early to say whether the U.S. will be in 20 years where Germany is today. It will be another 10-15 years before we know whether today's American teenagers will reproduce to more-or-less-replacement or not.

Edward Hugh said...

"And then, I might observe the following:"

Nice try Robert, but I basically don't buy this story. First off there are three distinct fertility regimes in the US.

1/. The Afro-Americans

2/. The Latino migrants

3/. The rest (more or less)

It is possible to follow these patterns in the US in a way which it isn't elsewhere (unfortunately) becuse of the way you maintain ethnic id data.

"The American boom generation that peaked in the late 1950s did succeed in replacing itself"

No, I don't buy this. This is a story that the US is telling itself because of the whole way people are reading this *inside* the US - you have a discourse problem - but it is missing the big picture.

Fertility reductions are education specific, and the postponement is driven by educational levels.

The population in group three (which is the core of the boom generation) was well below replacement by the early seventies (so the boomers weren't replacing, but women were getting to be more educated and postponing, you were following the European pattern): see the Robert Drago & Amy Varner article linked to this post:

http://fistfulofeuros.net/archives/001805.php

Fertility then started to rise again in the US - but not from the boomers - as Latin migrants started to arrive in large numbers. Since this phenomenon is now being repeated to some extent in Spain the material mention by Marta Roig Vila and Teresa Castro Martín (see above post on Spain and migration) is relevant. Latinos generally tend to have more children outside wedlock of even stable partnership, and tend to have children much younger (we are already getting the first wave of the teenage pregnancies which so
tend to characterise the US picture here in Spain).

Again, you can see the fertility impact in the US education performance stats, which are so completely skewed by this massive inflow.

Clearly all this will change as the next generation of Latino women start to emancipate themselves and start to get more education.

Edward Hugh said...

"Interestingly, the EU-10 countries had their baby boom in the 1950s, and were replacing themselves quite nicely until the social system collapsed around 1990"

I don't buy this connection, which again Robert is I think a reflection of the way this is all being debated in the US. It is very hard to tie this down to specific social security arrangements, the whole movement is of much greater significance than this, and to some extent the US is 'living in denial', even on occasions laughing at others discomfort.

Funnily enough, France, which has had more immigration than most, is more or less in the US position:


Qu'il s'agisse de sa population totale ou de sa population active, la France ne connaîtra pas, à l'horizon 2015, à la différence d'autres pays européens, de problème démographique global qui justifierait un recours massif à l'immigration.

This is both true and not true, but that's for another day.

Anonymous said...


"The American boom generation that peaked in the late 1950s did succeed in replacing itself"

No, I don't buy this. This is a story that the US is telling itself because of the whole way people are reading this *inside* the US - you have a discourse problem - but it is missing the big picture.


You don't believe that it is true, or you believe that it is true, but somehow irrelevent? While delayed childbirth, even when compensated for, has effects of its own, the numbers I have for the completed cohort fertility rate in the U.S. are 2.01 for the 1950-1 birth cohort, 2.02 for the 1960-1 birth cohort. Comparison to various European countries can be found here:

http://www.demogr.mpg.de/Papers/Working/wp-2001-009.pdf

Are there migrant women in these numbers? Of course, especially in the 1960-1 cohort. But a highly convenient feature of the age structure of the migrant population in the U.S. is, not only do migrants tend to be more fertile than the native-born population, but their age structure peaks in the "baby bust" of the 1970s.

Now, a question that might be worth asking, but is probably difficult to answer, is, do migrants have *more* children than they would have had if they had stayed in their country of birth? That is, since migrants migrate in the expectation that they will be better off in a new place, if these expectations are realized, do they manifest themselves in increased childrearing due to perceived prosperity?

Edward Hugh said...

"You don't believe that it is true, or you believe that it is true, but somehow irrelevent?"

Neither of these. I believe it is true, and I believe that it is not irrelevant. I simply believe that the situation is more complex. I believe we are everywhere using aggregate data, and that aggregate data is often very misleading. That is why I want to break it down.

Before I go further, you are right to point out that I overstated my case when I contradicted outright your statemnent that the boomers reproduced. I should have been more subtle, some of them did (the poorer and less educated groups) and some of them didn't.

Part of the growing polarisation and inequality you have in the US is associated with this. Obviosly, says he (fresh back from the Tommy Lee Jones Melquiades Estrada film), if the people who live in trailer parks are producing more rapidly, more people may well live in trailer parks.

But this is a side issue.

The other issue you mention in this context - completed cohort fertility - is obviously more important, and raises the whole adequacy of TFR derivitives as an indicator. The graph I have put up for fertility is for TFRs not completed cohorts, and this does make a difference.

If you compare completed cohorts with European equivalents like UK and France then you will probably also find that the boomers came near to reproduction, but in all cases there was a significant tempo effect, as more educated women had children later.

This tempo effect is not without importance as it affects the age structure distribution of the population, and this can be seen in the original kink we were talking about.

Edward Hugh said...

When I talk about the US discourse here, I am referring to the work of two well known theorists who I think have had a lot of impact on how the problem is seen inside the US. Richard Easterlin:

http://www-rcf.usc.edu/~easterl/

and Diane Macunovich

http://bulldog2.redlands.edu/dept/EconomicDept/macunovich/index.html

The whole emphasis of this tradition is on cohort factors (like boom generations), and these are certainly interesting, I think they miss the central point.

This paper from Macunovich is reasonably representative.

Edward Hugh said...

"Now, a question that might be worth asking, but is probably difficult to answer, is, do migrants have *more* children than they would have had if they had stayed in their country of birth?"

This is, of course, a very interesting question, and again gives an illustration of the difficulty of composite aggregate data.

Maybe there are two effects, and they work in opposite directions (so some of this gets lost in the aggregate). Maybe there is an initial 'prosperity' boost (as there is with a Regan-style tax rebate on consumption), but them maybe the migrant begins to change her behaviour to fit in with the group she most identifies with in the new country and starts to decide to subsequently have less children. Almost certainly her daughter does. Not insignificantly many male migrant children may return to country of origin in search of a bride, but few female migrants willingly choose this road. Most family tension rises when the *father* attempts to impose this on the daughter.

Of course there are those children of migrants who find not prosperity but the poverty trap, and they then get assimilated into your growing trailer park and prison population. So again the issue runs in two directions at once, something which I think will be difficult to pick up without qualitative in-depth research. Again Roig Vila and Castro Martín talk about this issue a little.

Edward Hugh said...

I'm sorry if I'm going on a bit here Robert, but I do think that all of this is really quite important. All this differentiation between completed cohort fertility and period TFRs is very remioniscent of an earlier debate in Europe, when Europe was still in denial. This distinction is now virtually impossible to defend here in Europe as the implications of sustaining TFRs below 2.1 for any length of time become only too apparent, among other reasons for the structural damage they do to the age pyramid which Lutz has drawn our attention to.

The thing is this structural damage is being masked in the US by the presence of massive immigration, and that gives you a false sense of security.

But you should remember that large scale immigration won't be avaialble for ever. Once China and India themselves need immigrants the thing will be to all effect and purpose done, and we may only be some 20 years away from that. So if in the interim the US proves unable to listen and to learn, you will eventually find yourselves in a right pickle.

The US has a unique opportunity to learn from others, and apply state of the art best practice from a very early point. But in order to do that you need to appreciate that there is a problem, and to understand why what is happeniong to fertility is happening.

Anonymous said...

Only about 30% of the increase in US TFR's since the 1970's is due to immigration.

The rest is due to a sustained, if very gradual, increase in the fertility of the native-born.

Of course, in a country of 300,000,000 people and 3.5 million square miles, any generalization is dangerous.

But to generalize anyway, what's happened since the TFR bottomed out at around 1.6-1.7 in the late 70's or early 80's is that non-Hispanic whites have increased their TFR slightly but consistently, while all other groups have seen declines in their TFRs.

Hispanics, having started from a higher level, are still the most fertile, but they've also seen the steepest drops. Eg., Puerto Rico now has a TFR of around 1.7 and Cuban-Americans are around 1.6, about the same as in Cuba itself.

Higher natality for Hispanics in fact means higher natality for Mexican-Americans and migrants from Central America.

And of course "non-Hispanic white" is itself an imprecise term, covering 225,000,000 people.

Some parts of the US have TFRs rather similar to Europe's, though at the high end of the scale -- 1.6 for Vermont, for example.

Others like Utah are around 2.5-2.7.

The most important single divide seems to be religious. Those who attend services once a week (roughly 50% of the population) have TFRs of around 3 or a bit less. Those who never attend are at about 1. The rest are in-between.

Anonymous said...

Also, fertility is not as strongly correlated with education or income levels in the US as in Europe. Religious and regional differentials are more important.

Oddly enough, you can also trace a definite difference in fertility between backers of the two political parties!

At a given income/educational level, Republicans will have about .4 of a child more, with the difference increasing as you go up the income scale.

Anonymous said...

Differences in fertility between immigrants and their source countries are an interesting topic.

For example, Mexico's TFR is now about 2.3 or 2.4, but Mexican-Americans are about .3 or .4 higher.

At a guess, I'd say it's a reflection of social and geographical differences between the migrants and the source population. If the migrants were disproportionately poor and/or rural, for example.

Anonymous said...

Fertility could be going down because people are waiting to have children or they can't afford them. Immigration seems to be a way to combat declining population.