Thursday, August 23, 2007

Quick Thoughts on A Farewell To Alms

The most striking thing about the book to me is how much is done with a lot of a little. By that I mean that Prof. Clark has amassed a huge variety of bits of data to make his various cases. While it's frustrating that the little bits aren't bigger bits, it's hard to complain considering how far back in history we're looking at.

After arguing against a number of the conventional theories of the industrial evolution [edit: I swear I meant to write "revolution"], he argues modern growth came about because the characteristics that generate economic success were propagated throughout society by the relatively high fertility rates of the wealthy. The main piece of evidence for this hypothesis is from English wills which show that men with more surviving children tend to have more assets to bequeath. He admits that the rich are more likely to have wills but argues that this isn't a large concern since many men with modest occupations are found in the data too. But that's not quite the main concern. Kids are expensive, right? Then men with lots of kids are probably less likely to have assets to bequeath in a will. But this is probably a larger issue for poor men than rich men. So it seems like the poor men with lots of kids are most likely to be missing from the data. Put them back in and the positive relationship between wealth and children might shrink or disappear.

8 comments:

Will said...

We know the population growth rate. From that we can get the average children per person.

Clark's data give us the number of children per rich dude. This number is much higher (almost double) than the average.

There needs to be an empirical version of "QED"...

Jason said...

What does that have to do with the population growth rate? One should make a more thorough argument before bandying about QEDs.

Jason said...

Besides, if Clark thought the accuracy of the "poor" folks data was irrelevant, why didn't he say so?

Will said...

If population is stagnant (or close to) for generations, it means the average folk had 1.05 children per person, no?

If the average rich dude has 3 or 4 kids then he's having more kids than the average (i.e. the less wealthy).

I suppose you can think of some strange distribution where really poor people have lots of kids, middling people have none and rich also have many, but this distribution wouldn't fit the rest of Clark's data (i.e. the accuracy of the poor folks, or the not-so-poor folks on your interpretation, data matters). BTW, I'm looking at table 6.2 in AFtoA.

I always forget where the burden of proof lies... Clark's theory, i.e. Malthusian logic applied to social classes, predicts --- actually it assumes --- rich people would have more kids. Clark's data doesn't reject this theory.

You have a different theory (weird distributions)... but the data say nothing about it.

Will said...

Also, chapter 6 has a pretty good discussion both defending the data (e.g. sex ratios are what you'd expect if girls weren't being omitted, poor folks disproportionately gave to non-genetically related heirs, evidence from other data sources) and backing up the theory (e.g. wealth predicts reproductive success better than social status).

Jason said...

I don't have strong priors on whether or not kids increase or decrease with income. (Especially with respect to preindustrial England!) For modern societies, the empirical literature (on the causal relationship) is a seriously mixed bag. If I had to bet, I'd bet on a positive relationship. That said, I'm still concerned that the extent of the relationship may be exagerated by the will data for the reason I mentioned.

And clearly if rich people are having more kids than average, then poor must be having fewer than average. But you have to admit that the structure of the argument that you are making is not ideal. We really want to compare accurate data on the number of children by rich men to the number by poor men. Clark must agree since this is exactly the argument he makes in chapter 4 using the will data. He doesn't make the argument that you are making here. I suspect that it's because it's a bit weird although not unreasonable.

Anyways, haven't gotten to chapter 6 yet. I didn't realize he was going to revisit fertility. I was looking at table 4.6.

Will said...

My priors on kids is that the number you have is a non-linear inverted u-shape (is there a name for those things?).

This comes from the macro demographic transition literature where societies average fertility increases with income and then declines after a certain level. I imagine that what happens in the time series must happen in the cross-section.

Will said...

Oh, yeah. I also believe the distribution of wealth was all on the left side of that inverted u in the Malthusian era.