Poverty and Life Outcomes: A Theory of the Changing Relative Importance of Resources and Personality

Why do the poor experience low-quality outcomes in areas such as health, education, and crime? Bryan Caplan has recently been arguing "that social scientists need to search for factors that cause both poverty and irresponsible behavior. Such as? Low IQ, low conscientiousness, low patience, and plain irrationality" (also here and here; critical comment here). The position he is disputing is what you might call the Trading Places interpretation: low-quality outcomes are the consequence of the social situation called poverty. According to this view, there's a big fat causal arrow running from poverty to outcomes such as those mentioned above. In Caplan's model, there are two big fat causal arrows from a summary variable you could call "personality" (which encompasses low IQ, low conscientiousness, etc.) to both poverty and other outcomes of interest (a textbook "third variable" explanation).

I'm ambivalent about Caplan's position. On the one hand, his actual line of argument is deficient bordering on the ridiculous - among other things, it suggests he's never heard of opportunity costs, which would be kinda surprising for a professor of economics. On the other, I agree with the conclusion that both social scientists and laypeople are much too quick to interpret a correlation between poverty and some problem outcome as a poverty effect, and that much more weight should be put on personality as an alternative explanation. Sociological work is perhaps particularly bad in this respect.

My own model is a bit more nuanced. Let's start by imagining we had found a way to partial out those bits of the poverty-problem correlation that are due to poverty as a consequence (e.g., you become unemployed, and hence poor, due to bad health). It seems to me that 100% of the remaining correlation can be explained by "personality" (Caplan's focus) plus the structural restrictions brought about by poverty (standard sociological focus). It follows logically that the portion of the variance explained by one of those two factors must go up when the proportion of the variance explained by the other goes down, and vice versa.

My model applies to societies that today are relatively affluent and meritocratic, such as the USA, Germany or Japan. Back in the day - let's say 150 years ago - the common explanation for the poverty-problems correlation used to be in terms of "personality", but actually much of the reason for its existence was structural. Then structural explanations became more popular, and with it efforts to tear down the structural impediments to better life outcomes. These efforts were effective to a large degree: opportunities actually became more equal. This is another way of saying that the portion of the correlation explained by structural restrictions went down, which, in turn, means that the effect of personality must have gone up.

If true, this implies a neat "perverse" effect (in the sense in which the word is used in sociology): the more one type of explanation for the poverty-problems correlation is believed, the less true it becomes. 

So far my argument may have brought to mind programmes that allow young people from low-income families access to universities (and hence better-paying jobs), for example. But there's more to it, namely intergenerational transmission and the shadow of meritocracy.

In order to demonstrate the logic of the argument, let's first point to the fact that children tend to resemble their parents (their personalities are positively correlated). Let's also note that societies can vary in the degree to which they are meritocratic. One could say - non-normatively - that some people "belong" at the top (the intelligent, industrious, etc.), while others "belong" at the bottom (the dull, lazy, etc.). "Belong" here means simply that we would expect these people at certain positions in society, and these expectations should be the stronger the more meritocratic a society is. But we cannot predict the degree of "fit" we observe simply by looking at how meritocratic a society is now. We also need to look at the past: If a society has been highly meritocratic for a long time, children will tend to already start out in the place of the social structure where they "belong". This is due to the positive correlation between children's and their parents' personalities.

This view, if true, has a number of implications.

(i) You cannot judge the degree to which a society is currently meritocratic by looking at the social mobility which it exhibits. Holding current meritocracy (and parent-child resemblance) constant, we would expect societies with a long tradition of meritocracy to exhibit less mobility. Hardly anybody who talks about this topic understands this.

(ii) We should expect a lower "fit" between personality and social position in those groups who can't have benefited much from past meritocracy, such as recent immigrant groups.

(iii) If you make a society more meritocratic, you should expect the biggest movements right away. For example, programmes that allow children of low-income parents access to univerity will be most effective shortly after their introduction. After a few generations, there won't be so many many highly talented kids left among the potential beneficiaries.

(iv) Caplan's model is more true today than it would have been 150 years ago, because it became unfashionable.

The model also explains why the left (as in "people who favour lots of social mobility") used to like IQ tests, now don't.

An illustration: the social gradient in health
In the 19th century, when these things were first measured, it became apparent that the poor experienced worse health than the rich. In the middle of the last century, it was expected that this poverty-health correlation would soon disappear in affluent western countries due to improvements such as clean drinking water for everybody and universal healthcare (doesn't apply to all nations). Dirty drinking water-type illnesses did indeed disappear, but the social gradient stuck around. These days the leading causes of morbidity and mortality are too much food, alcohol and nicotine. According to some (standard citation), the continuing significance of socioeconomic status as a predictor of health is due to high socioeconomic status allowing access to a multitude of varied resources, such as money, networks and knowledge; access to such resources will be beneficial no matter what the specific causal channels from status to health are. Others contend that individual behavioural choices are important, and according to Linda Gottfredson's theory, the actual fundamental cause is not socioeconomic status, but general intelligence, which correlates positively with the former. My view is that Gottfredson's focus on intelligence is too narrow; that you must also include factors such as high time discounting and preferences for risk taking and that these personality factors are more important now than they used to be. Note how nicely this is in line with the change in the leading causes of poor health: dirty water then, too much to drink now.

Odds and ends
(i) Very careful readers may have noted that my dynamic general model of meritocracy and intergenerational mobility made an unspoken assumption; namely, that the personality traits that were beneficial for parents will also be beneficial for children. The less true this is, the more intergenerational mobility we should expect, holding all else equal.

(ii) I made a clean distinction between "structural" and "personality" influences. But the former may influence the latter. For example, Daly and Wilson (scroll down to p. 1271) hypothesize that growing up in a neighbourhood where life expectancies are low causes people to discount future consequences of actions more (they also present some evidence, but it's low-quality). If this type of explanation is true, it is not crystal-clear whether to model time discounting simply as a personality variable or a consequence of structural disadvantage. Which solution you choose will depend on the exact question you're asking, but take this as a note of caution to give it some thought.

Where credit's due
With a bit of inspiration from Thilo Sarrazin and Steven Pinker not otherwise acknowledged.

No comments: