Demographic transition reconsidered

I’ve been sitting on “A Farewell to Alms” for a few days after seeing a footnoted claim that I knew I’d have to look into and most likely write a post on. The note is in chapter 14 where Clark writes “And now the rich in England have fewer children than the poor, so if children are to be counted a blessing and not a burden the advantage now lies with the poor (though in some other advanced economies there is no difference between rich and poor in this respect).¹⁹” From the accounts I had usually heard, the demographic transition was supposed to be a universal phenomenon and that would over time engulf even the currently developing world. So what is Clark citing? Nicola Dickmann, 2003. “Fertility and Family Income on the Move: An International Comparison over 20 Years.” Dickmann acknowledges that lower income families did have more children back in the 80s, but this gap seems to have narrowed in more recent years. I should note that her method was to look at data for married couples (completely ignoring single mothers) within a certain age range in the USA, Germany, Canada, U.K, Finland and Sweden. The correlation between income and family size (and statistical significance) was generally small. It was positive in the last two and negative in the rest. I wouldn’t say it completely upends conventional wisdom, but it’s definitely something to consider. Two relevant EconLog posts are The Decline of the Rabbit Strategy and Kids, Opera, and Local Status. A recent OB post somewhat relevant, Humans Are Evolving. A lot of people may not bother clicking through to read the pdf, so tomorrow I think I’ll copy some of the tables into html below. UPDATE: Transcribing all the data got to be a chore, so I’ll just put up some images. The rules for transforming the text I can copy is pretty simple so next weekend if I’m not busy I’ll try to learn python and then replace it with html. UPDATE2: Done. I couldn’t completely automate everything. I’ll put up the code in a later post.

Some unrelated notes. For anyone wondering when I’ll start twittering, don’t hold your breath. I never update my facebook status (or bother logging on most of the time) because nothing happens worth talking about. If I have something to say, I’ll blog it. Maybe if I read twitter my eyes would open up to the possibilities, but I don’t. Also, Against Politics once hosted Richard A. Garner’s “If Hobbes Is Right, Then He Is Wrong”. After the site went down, google only turned up other sites giving that bad link. After bugging Aschwin about it, he restored it. Enjoy.

Statistical interpretation
Average number of children per woman and income level	USA 1979	West Germany 1981	Canada 1981	UK 1979	Sweden 1981	Finland 1987
Less than 5.000 int. Dollar	2,56	-	2,00	2,11	1,88	2,00
5.000 up to 10.000	1,96	2,00	2,07	2,14	1,83	2,00
10.000 up to 15.000	2,18	0,78	2,12	2,01	1,74	2,00
15.000 up to 20.000	2,11	1,14	2,04	1,84	1,86	1,27
20.000 up to 25.000	1,91	1,48	1,99	1,98	1,98	0,28
25.000 up to 30.000	2,00	1,77	1,86	1,69	2,00	1,00
30.000 up to 35.000	1,78	1,74	1,87	2,17	2,33	1,21
35.000 up to 40.000	1,96	1,83	1,65	1,71	-	1,51
40.000 up to 45.000	1,73	1,59	1,65	1,50	-	1,02
45.000 up to 50.000	2,10	1,43	1,73	2,50	-	1,67
50.000 up to 55.000	2,31	1,69	1,56	1,33	-	1,40
55.000 up to 60.000	1,33	1,67	2,75	1,50	-	1,26
60.000 up to 65.000	3,00	1,81	2,14	2,00	-	1,12
65.000 up to 70.000	2,61	1,24	1,25	2,00	-	1,29
70.000 up to 75.000	0,00	1,68	0,75	0,00	-	1,16
More than 75.000	3,00	1,88	2,43	2,50	-	1,61
Variance	0,49	0,25	0,21	0,34	0,04	0,20
Standard deviation	0,70	0,50	0,46	0,59	0,19	0,44
Maximum difference	1,67	1,22	2,00	1,17	0,60	1,72
Mean	2,03	1,48	1,87	1,81	1,95	1,36
TFR of elicitation year	1,81	1,43	1,70	1,82	1,63	1,59
Correlation Income – Fertility
Significance	0,002**	0,764	0,000***	0,059	0,241	0,000***
Correlation coefficient	-0,064	0,012	-0,102	-0,051	0,024	0,077

Eric Johnson Says:

November 16, 2009 at 2:21 am

> her method was to look at data for married couples (completely ignoring single mothers) within a certain age range

Why doesnt that render it totally worthless

Thorfinn Says:

November 16, 2009 at 11:00 am

Generally, there is a quantity/quality tradeoff with children, so you get fewer children as a country gets richer.

However, this does tend to stabilize at very high levels of development (US for instance, relative to Italy). At very high levels of income, families can afford to increase the quantity of children without compromising on the level of care of existing children. I don’t know about comparing across rich and poor within countries as much, but I remember that household sizes are larger for rich folks (though surely this reflects divorce as well).

Timing is also very important. Even if everyone has the same number of children; having them at 20 or 30 can create large differences.

Dr. Horsemeat Says:

November 16, 2009 at 10:07 pm
I don’t think that a nation’s fertility stabilizes at or near replacement rate as a result of economic development. The trend actually seems to go in the opposite direction.

As compared to America, the Total Fetility Rate for other highly developed nations is far lower. Germany: 1.41, Switzerland: 1.44. France: 1.89. This isn’t attributable only to the population density of European countries. Canada’s TFR is 1.57. Australia’s is 1.78.

In fact, the only developed country I could find with a TFR similar to America’s is New Zealand, at 2.11.

In 2005, the Total Fertility Rate for Non-Hispamic Whites in America was 1.844. For Blacks, it was 2.019. For Hispanics, it was 2.877. Without Hispanic immigration, America would have a TFR much closer to that of the rest of the developed world, that is to say, a sub-replacement TFR.

1. Thorfinn Says:
  
  November 18, 2009 at 11:58 pm
  http://www.nature.com/nature/journal/v460/n7256/full/nature08230.html
  
  Is Hispanic fertility that high? That’s higher than Mexico’s fertility rate. So it’s not just the case that we’re importing pro-natal immigrants. Something about America–the density, welfare system, plenty of jobs–is conducive to fertility.
  
  Divorce is also a lot more common among Mexican-Americans than Mexicans. The family structure that worked for inner-city blacks and rural whites broke down in the 60s, and it looks like it has also broken down for Mexican immigrants. Anyone know why?
2. Dr. Horsemeat Says:
  
  November 22, 2009 at 2:42 am
  Well, America is less densely populated than Mexico, though in many cases (I guess) Mexican immigrants are coming from rural districts and located in American metropolitan areas, so maybe that’s sort of a wash.
  
  I do believe that the welfare system and associated social services do contribute to a higher level of Hispanic fertility in America as compared to Mexico.

teageegeepea Says:

November 16, 2009 at 8:49 pm

Eric Johnson:
Variation within the set of married couples is still interesting. The two examples in Idiocracy were both married.

Thorfinn, good points.

November 17, 2009 at 2:44 am

Of course the TFR will probably not stay the same, but, for USA whites:

(1.844 / 2)^x

x
—
3 = .7834
4 = .723
5 = .666

2.0 is not the replacement rate, of course, so I have an optimistic bias. But I already did the math!

For all Germans, using of course 1.41/2 …

3 = .350
4 = .247
5 = .174

Four generations is of course a little over 100 years, I guess, for non-poor people.

Douglas Knight Says:

November 19, 2009 at 6:42 pm

What does this have to do with the demographic transition? It seems to be about some unspecified theory about why the transition occurs, but I’m not even sure of that.

November 19, 2009 at 8:51 pm

Thorfinn, Sailer noted the higher birth rate of hispanics in America compared to Mexico here.

Douglas Knight: the demographic transition is associated with a switch from the rich outbreeding the poor to the poor outbreeding the rich. That’s an impression that may rely too much on a subset of countries that have gone through the transition.