Tuesday, June 27, 2006

US Demography: A Tale of Two Pyramids?

by Edward Hugh


I've already made a number of attempts here to try and get a better understanding of the dynamics behind US demography. This was an early attempt, where I try and link the rebound in the TFR that can be seen from the mid 70s to the start of modern-era large-scale immigration together with the slowing down in the birth postponement process in the US 'majority' population and the consequent 'recovery' of missing births (this process has also been noted in a number of societies in North Western Europe). The rebound can be clearly seen in this graph:



I also gave some suggestions as to why the current near replacement fertility reading may be less stable than some suggest due to the continuing decline in teenage pregnancies, and the steady displacement upwards in mean age at first birth. Both these indicators would suggest that a second 'tempo effect' movement downwards in aggregate US TFRs is to be anticipated.

More evidence that the US fertility profile may be less stable than many assume comes from looking at the fertility behaviour of the recent migrant population. Now as this article explains, the U.S. Hispanic population, which accounts for only 14 percent of the US population, was responsible for a staggering 49.2 percent of the population increase four-year population increase in the US over the last four years.

Even more to the point are the so-called 'vital statistics' — the ratio of births to deaths — of the two main population groups. The vital index for white non-Hispanics is 1.2 (approximately one birth for every death), while among Hispanics the figure is 8.2 — approximately eight births for every Latino death. This disparity is the result of the considerable differences in age structure and fertility between the groups.

A quick comparison of two population pyramids (the white non-hispanic one and the hispanic one may make this clearer). Here is the while non-hispanic pyramid:




It looks, as the prb article notes, pretty much like many of the European pyramids. Now here is the hispanic one:



So with a current TFR of 2.8, and a legalisation and stabilisation process now about to happen, and roughly the same wish as anyone else to get healthy, wealthy and wise, my guess is that the global hispanic TFR is now set to fall, and set to fall rapidly. So while in relatively discrete (and statistically not especially important) settings like Idaho and Utah, white non-hispanic fertility may be notably higher than average for religious or other reasons, it is far from clear that ideational forces will maintain the US population long outside the socio-economically driven global trends.

15 comments:

Will Baird said...

I am not going to attack what you've been writing, Edward. You have obviously been studying this much(!!!) more than I have.

I have to wonder though about your conclusions: the US is a delayed European country in demography terms. Now I am not saying the US is a gilded special case, but there are a couple things that just make me wonder about your conclusions.

First there was this article I readhere about the differences in fertility between people from their home countries and their emigrants to the States. James called it the 'weird ferility effect' or some such. If this article is accurate, then something...interesting is going on here. Even Europeans that come here have more kids.

Second is purely ancedotal and almost assuredly worthless. Most people my age and younger that I have met are espousing have more than two kids. Three minimum. A number of them are feeling like there was something missing coming from a two or one child family and want more. Now, right now my friends are skewed to those who have at least one child at a similar age to my own. That inclines me to worry about sampling problems. However, I heard via the rumor mill that there is a baby boomlet under way...*shrugs* ancedotal and useless.

What data do you have post 2001? That might help test your hypothesis.

Alright, ready for mocking here. :D

Edward said...

"Alright, ready for mocking here"

No, no mocking at all Will, thanks for taking the trouble to make some thoughful comments.

Firstly since we are talking about the future evolution of something here, nobody actually *knows* what is going to happen, so we all need to be relatively modest and conjectural in our claims.

The thing is, this is like an options 'put'. What will be the TFR of the US in 2015? Quite a lot, of course, depends on this.

So we need to like at all the arguments, and the evidence, and give a probability weighting.

Now the 'dominant hypothesis' in the US seems to be along the lines of the kinds of things you say. Since there is a lot of evidence of things being different elsewhere, I am simply saying it might be rash to put all your eggs in one basket, and at the very least you need to follow this situation carefully as it evolves.

There are three main possibilities: there may be lessons for the US for the rest of the world, or there may be lessons for the rest of the world from the US. Or some combination of these two.

This latter possibility is the most plausible. For example ease of entry and exit for young women into the labour market (the US case) is clearly of some benefit, as is a generally supportive gender environment.

OTOH there is the view which comes from parental investment theory that there is a certain logic to the quantity/quality trade-off, and to a postponement towards later first birth ages based on a move to higher value economic activities which prioritise higher human capital values (skill-bias).

The outcome is really a product of the mix of these two processes.

Now your point about the relatively higher fertility of new immigrants in the US is a well founded one.

There were a couple of interesting papers about this presented at PAA2006:

http://paa2006.princeton.edu/download.aspx?submissionId=61697

http://paa2006.princeton.edu/download.aspx?submissionId=61352

The real issue is what mechanisms do we have to explain this pattern, and are those mechanisms sustainable? One of the key factors in deciding how many children a woman has is the age at which she first gives birth. The upward movement in these ages gives rise to what is known as the postponement process. A collection of scholarly articles on this can be found here:

http://www.oeaw.ac.at/vid/meeting_postponement_prog.shtml

"Most people my age and younger that I have met are espousing have more than two kids. Three minimum. A number of them are feeling like there was something missing coming from a two or one child family and want more."

As far as I can see this is globally pretty typical, all the surveys say that people want more children than they actually have. The question is whether the desires are fulfilled, what are the constraints, and what factors might influence achieveability.

Take housing in the US. How would achieved fertility be influenced by a downturn (not a crash) in the housing market and higher interest rate driven service costs? What about responsibilities to parents. You belong to what is called the 'sandwich generation' that might see resources going out in two directions, as the parents of this generation live longer, but with imparied function in some form or another, and with fiscal constraints meaning family share more of the burden. There is also the question of the rising cost of having each child as more and more educational input is required to stay at the cutting edge globally. There are really lots of things to think about.

"However, I heard via the rumor mill that there is a baby boomlet under way.."

This is quite possible. There is one now in Spain. These are normally associated with a slowing down in the pace of postponement among the group in question, and the arrival of the 'missing births'. Such phenomena should not be read as changes in long term trends.

Relatively weak labour markets in some sectors can also encourage short-term fertility, as people decide not to actively seek work and have the child now. Of course they then need a job even more urgently to pay for it.

Bottom line: it really would be quite interesting to understand why migrant women are having more children, and whether this pattern will continue. I tend to doubt it, for the reasons I have suggested in the article, ultimately they will want a better life for themselves and their children, and this will mean postponement and less children, but only time will tell.

"What data do you have post 2001?"


The graphs in the other posts I link to comes from US final births 2003 (published in September 2005 by the CDC) This is the latest data I have access to. Till next September I suppose.

http://www.cdc.gov/nchs/data/nvsr/nvsr54/nvsr54_02.pdf

"You have obviously been studying this much(!!!) more than I have."

Yep, well don't be surprised. I consider all of this to be the biggest outstanding unresolved issue in the social sciences, and especially in macroeconomics.

Edward said...

Incidentally, I think this extract from the Swicegood paper on non-US born fertility is important:

"For many national origin groups, a substantial amount of recent childbearing has occurred outside of marriage, and the factors that differentiate recent fertility for unmarried women do not always operate the same way as in the case of marital fertility."

Extra marital childrbirth is extremely low in many of the sending countries (as it is in eg Southern Europe, S Korea and Japan etc). The freeing of these constraints and the relative ease of setting up some sort of independent home or other in the US must be a factor. Of course this does run a bit across the 'religious component' argument, but then maybe there are different influences in different groups.

Robert said...

Attitudes towards education might be the silver lining in the U.S.'s debt-driven consumer culture. More than once I've heard the expression, "If an education is worth buying, it is worth buying with borrowed money."

(Now, this helps in no way those who pursue educations that are not worth buying.)

S.M. Stirling said...

US Hispanic age distribution is heavily skewed by migration. Generally speaking, any migrant population is going to be disproportionately young.

'twould be useful to look at the comparative profiles of immigrants and the native-born.

S.M. Stirling said...

Also, "Hispanic" is a uselessly broad term.

Demographically Cuban-Americans and Puerto Ricans are as different from Mexican-Americans as Mexican-Americans are from Anglos.

In fact, Cuban-Americans and Puerto Ricans both have _lower_ fertility than non-Hispanics -- down around 1.6/1.7 as opposed to 1.9-2.0.

Mexican-Americans are also socially unrepresentative -- they're disproportionately rural/small-town and lower class in their origins, as opposed to the source population back in Mexico.

Heading for the US is sort of an alternative to moving to Mexico City or Monterrey (or a subsequent step after moving there).

S.M. Stirling said...

The US population contains a number of distinct fertility patterns. Some closely track world trends; others are very exceptional.

Noel Maurer said...

Hello, Edward. If I may, there is a logical problem with your prognostication that the recent falls in the teenage birth rate presages a future fall in the TFR.

The teenage birthrate has fallen in the last few years, while the overall TFR has remained steady. The birthrate for older age groups, therefore, must have risen over the same period.

These older mothers, of course, are more likely to have been teenage mothers than the current group of teenagers.

Your prediction, therefore, rests on two assumptions:

(1) Teenage birthrates will continue to fall;

(2) When the current crop of childless teenagers reaches their twenties and thirties, they will be less likely to give birth than women currently in their twenties and thirties ... even though the current crop of adult women were more likely to have already given birth as teenagers.

The first assumption seems unfounded. That is not to say that teenage birthrates won't continue to fall; it is to say that predicting that they will fall seems to be a atheoretical extrapolation of short-term trends. If you have another explanation I would be very interested in hearing it.

The second assumption seems based on some rather counterintuitive behavioral assumptions. Would you mind justifying them?

Edward said...

Hello Noel,

"If I may, there is a logical problem with your prognostication that the recent falls in the teenage birth rate presages a future fall in the TFR."

Well I'm not sure it's a logical problem, although I concede there may be an empirical one, in the sense that it's a prognosis that can be falsified by data rather than refuted.

Part of the problem with the US teenage pregancy issue is it's hard to know what is going to happen when we don't really understand why it is there in the first place. It is such a special US problem (although the UK has some of it too). Part of the problem may be lack of social mobility. You may simply have a lot of families who are just stuck.

Normally though, you would expect that as generations (and the economy generally) move up the value scale, people's social aspirations change, and their children start postponing, to get more education, experience, resources to start a home etc.

I suppose I'm being optimistic and assuming that this process (which seems to have started to some extent) will continue.

The US anthropologist Hillard Kaplan provides some arguments as to why it may not happen though, in the sense that the single-mum phenomenon means that a lot of girls don't wait to set up home at all, and stay with their mum (who may also be single). This is a self-reproducing 'trap', it is also matriarchal, which is, well odd.

On other issues you are right that those who have children as teenagers have much more probability of having more children later.

The issue arises with postponement, as there is a lot of evidence that this leads to unfulfilled desires to have children when people wait to their thirtees, and then come up against biological and other issues which make having children a touch more difficult.

Normally you would expect to see a lot of postponement taking place in the US over time, as people strive to maintain the economic level in a more human capital intensive world.

You may find this presentation useful on postponement:

http://www.oeaw.ac.at/vid/download/pce/dec03/pm/How_long_can_postponement_continue--GoldsteinLutz.pdf

The data they cite on doctors, lawyers and PhDs in the US gives a figure of 1.5 final completed cohort fertility for this group. So it really depends on how many well-qualified professionals you imagine the US will be producing in the future and how many cleaners and care workers. This is what will, at the end of the day, decide.

Also, it is only a detail, but obesity is a fertility negative, and this seems to be becoming a growing problem. So even in the lower income groups people may have more problems actually having children.

"The second assumption seems based on some rather counterintuitive behavioral assumptions."

I'm not sure which behavioural assumptions you are refering to here. Historically moving from quantity to quality and postponing to improve educational levels has started in the higher income groups and has then been copied by the lower ones. Rather like the idea that today's features on a Mercedes will be incorporated in tomorrow's Fiat Punto. These are my behavioural assumptions, and I don't find them especially counterintuitive, but then I don't know what your intuitions are.

Noel Maurer said...

"I'm not sure which behavioural assumptions you are refering to here."

I'll clarify. You've assumed that a woman who gives birth as a teenage is more likely to give birth in her twenties than a woman who has not.

That seems counterintuitive. Not wrong, just counterintuitive. What do the data show>

Edward said...

"You've assumed that a woman who gives birth as a teenage is more likely to give birth in her twenties than a woman who has not."

Well, let's be just a little more general here, lets not say in her twenties, but across her reproductive lifespan, and not that she is more likey to give birth at any point than a woman who hasn't, but that she is more likely to have more children than a woman who starts later in life.

"What do the data show."

Exactly that. The earlier the first birth the more children. That is the correlate, and it is pretty general. Of course the course of the trajectory can come down, across time, for eg in the third world, with 'stopping behaviour', tying tubes, access to modern contraception etc etc.

The woman who has the first child at 34/35 has a very reduced probability of being multi-parous. Not *no* probability, just a reduced one when compared with a woman who starts out at 18.

Noel Maurer said...

Where can I find the analysis? I'm surprised that a longitudinal analysis would show that a mother who first gives birth at 18 has significant more children across her lifespan than one who starts at 25, adjusting for all other factors, but if that's what the regressions show, then that's what they show. (Your point about 35-year-olds is a canard, of course.) I'd be interested in seeing the rigorous analysis; can you give me a link or published reference?

Edward said...

Hi again Noel.

"Where can I find the analysis?"

Well here would be as good a place to start as any.



or if you want to go into more depth in the problem, you could read Tomas Sobotka's entire PhD thesis which is online.

Basically what you need to think about are age-specific parity probabilities, these link the age at first birth to final number of children. In this sense they are a bit like supply and demand curve schedules in economics.

Of course, you can argue that some women exhibit 'non-normal' behaviour (rather like how consumption of potatoes varies non-normally with price in some socities) but such behaviour is precisely that 'non-normal'.

Really your initial intuition goes against the whole weight of demographic evidence.

"I'm surprised that a longitudinal analysis would show that a mother who first gives birth at 18 has significant more children across her lifespan than one who starts at 25"

Well, surprising or not, this is exactly what lies behind the 18th century explosion in UK population, and without it the industrial revolution might never have happened, or not have happened in the way it did.

"but if that's what the regressions show"

You don't need anything as sophistocated as statistical regression here, just addition and division. The birth data is there, as is the age of the mother, in a European context across many, many years.

"Your point about 35-year-olds is a canard"

Well, you can think that, but you would be mistaken, since it is the central issue. These doctors, lawyers and PhDs are simply the pioneers. There is considerable evidence of behavioural 'copying' across social classes in this area. Basically, that's how the industrial revolution thing spread.

"can you give me a link or published reference?"

Yep, sorry, I tried to put a link to this in my previous comment, but it didn't show properly. Here it is again. This isn't the original research, but Lutz knows what he is talking about, as he is one of the world's leading demographers, so we can take it that there is a study behind it, probably work by the American demographer Ronald Rinfuss.

Now.....

The US situation is an extremely complex one, and for precisely that reason it is interesting. You might enjoy reading this study by Steve Martin, precisely becuase in some way it contradicts the lawyers, doctors, phds finding. Not that he finds that people who postpone don't have less children, but because he finds that women who are still childless at 30 and who *don't* have a college degree (remember the behavioural copying effect)have subsequently less children than those who have postgraduate education. In neither case do people have a lot, since there is, of course, the biological window effect.

But Martin's finding is interesting, and bears thinking about, especially in the US/Europe comparison context. It would seem that more highly educated people in the US have a better possibility of realising their lifelong childbearing ambitions than do people of the same (very highly educated) level in Europe. Which raises the question, why? Labour market and gender issues obviously play a part, as do, I suspect, the presence of a large number of 'illegals' available for substitute domestic work at a very cheap rate. I am not asserting this, it is just conjecture. I think the situation is fluid, and we need more time and more studies.

Will Baird said...

But Martin's finding is interesting, and bears thinking about, especially in the US/Europe comparison context. It would seem that more highly educated people in the US have a better possibility of realising their lifelong childbearing ambitions than do people of the same (very highly educated) level in Europe.

This is interesting. It seems to fit my ancedotal experience pretty well. Also see below.

Which raises the question, why? [...]
the presence of a large number of 'illegals' available for substitute domestic work at a very cheap rate.


This does not.

Stupid, stupid question. This may apply tot he US only. Does personal mobility have anything to do with this? Mobility makes a pretty big difference here about kids. Without a car, its damned hard to get places to take care of kids: schools, hospitals, sports, etc. This is doubly so on the off hours or very weird hours. The better educated you are the better off you tend to be. The more wealthy you are, the more vehicles you can have. The easier it is to get around to do what needs to be done.

I have to wonder if there's a correlation. That's all. Cars => kids?

Doc Merlin said...

I think ultimately what is most important is how many children a couple tries to have, and ultimately that is tied to memes.

One big difference is how people expect to have to care for their children.
This is threefold:
1) Opportunity costs:
In areas where the cultural expectation is for women to stay at home with their child (Germany) and where driving is difficult (so its hard for the child's caretaker to travel for errands and such (New York City) you will naturally find a much lower birth rate

2) Cultural expectation on family size:
People don't just put off having children because of economic costs, also people have cultural expectations of when they should have children. This ties in with the lengthened adolescence that modern people have.
This also shows a key difference between Mexican immigrants and most whites in the US. Mexican
immigrants expect a large family and thus end up with much more children than someone who cuts themselves off at a lower number.
-----

Also, does anyone have a histogram of number of children per woman, to see the distribution? I would imagine that most of the loss in the developed world has been as a result of the higher child bearing brackets going from normally having 4 or 5 children now having 2 or 3.