Inequality, chemistry and crime

My latest column in Bloomberg (published Sunday night in the US, I think) looks a little at the link between socioeconomic inequality and crime. That there is such a link has been established rather convincingly by statistical analyses across many nations spanning decades. This was all laid out convincingly in the 2009 book The Spirit Level by Richard Wilkinson and Kate Pickett. See also this short TED talk by Wilkinson which covers the main points very clearly.

But I also referred (rather vaguely) in my column to two other lines of research that deserve some discussion. One is an idea in criminology known as "routine activity theory" which has increasingly become a branch of applied computational social science. In effect, it views crime as a kind of natural social chemistry taking place between potential offenders and targets, and has a good track record in accounting for empirical patterns of crimes of all kinds by considering simple factors such as patterns of human movement on streets, of street lighting, of building architecture and so on. It more or less gives up on thinking about the deep psychology of crime and its motivation. In general, this perspective asserts, you tend to get more crime where you have more targets coming together more frequently with more potential offenders, just as you get more chemical reactions whenever potential reactant molecules come into contact more frequently. I've written about this a little before in New Scientist (a copy of that article is on the web here).

The empirical success of routine activity theory is sometimes taken as evidence that factors such as poverty and social inequality don't matter in crime, as the theory doesn't explicitly consider such factors. This is a serious misunderstanding. After all, the theory focuses on the factors that bring together potential or motivated offenders with potential targets, and doesn't attempt to explore the factors that might give a person the motivation to commit crime in the first place. Here the analogy to chemistry is again constructive, I think. In chemistry, reactants coming together isn't enough. They generally need sufficient energy (or a catalyst) to help them overcome an energy barrier (caused by repulsive forces between molecules) before they can react. It's entirely plausible to think of prevailing social norms as a kind of energy barrier to the commission of crimes, as these forces tend to persuade people to avoid crime. In general, we should expect high levels of trust and strong social norms to suppress and discourage crime, much as a lowering the temperature of a chemical solution makes reactions go more slowly. (Low temperature increases the effective size of energy barriers to reaction, which must be overcome by collisions between molecules; these are stronger at higher temperatures.)

Analogies aside, the point is that routine activity theory actually fits together quite well with the existence of larger social factors that influence overall probabilities of crime. The factors considered in the theory can determine detailed patterns of crime, even as the larger global factors influence overall rates. The work considered by Wilkinson and Pickett strongly suggests that social inequality is one of the most important global factors encouraging crime (and other measures of social dysfunction).

The second body of work I wanted to add a little detail on looks at what creates high levels of inequality in the first place. On the drivers of inequality, of course, there is a huge literature in social science and economics and I don't pretend to be an expert on it. However, there is an interesting perspective motivated by physics that I think should be much more widely known. This paper from several years ago by Jean-Philippe Bouchaud and Marc Mezard analyses a very simple model of an economy and shows that a high level of economic inequality is a more or less generic and unavoidable outcome. It is something to be expected on mathematical grounds alone. The basic idea is to model an economy as a system in which wealth flows around among people by two fundamentally different processes. First, it flows between individuals when they make exchanges or trades, through contracts, employment, sales and so on. Second, wealth also accrues to individuals (or departs from them) on account of investments in instruments yielding uncertain returns. Importantly, the second factor contributes an element to individual wealth that involves a random multiplicative factor -- random because investments are uncertain, and multiplicative because, quite sensibly, the wealthier invest more than the poorer, and typically more in direct proportion to their wealth. The poor partake very weakly in this multiplicative channel to wealth growth, while the wealthy participate progressively more strongly.

Bouchaud and Mezard showed that the inevitable outcome in such a basic system is that a large fraction of the total wealth in the economy ends up being held by a small faction of the population. This is the case even if every individual is considered to have the same inherent money making skills. (The authors, of course, do not believe this to be the case; but making this assumption makes it possible to explore how natural economic dynamics can drive large wealth disparities even in the absence of any differences in skill.) It's the multiplicative nature of the returns on investments that makes it happen. This pattern in the distribution of wealth holds in every nation and was known long ago, being first described by Italian economist Vilfredo Pareto. This paper gave to my knowledge the first really fundamental explanation of why this pattern holds everywhere. Since then a number of works have taken this approach much further.

However, the universal form of this distribution does not preclude the possibility of some societies being more unequal than others. You can have a society in which the top 5% hold 90% of the wealth, or in which that same top 5% holds 99.9% of the wealth. Such variations are driven by the relative strength of multiplicative returns on investment versus the flows in an economy that might act to reduce inequality. Obviously, taxation is one mechanism that leads to a redistribution of wealth from richer to poorer and one should expect, in a broad brush way, that lower taxes should tend to associate with higher levels of wealth inequality.

I strongly recommend this paper -- it's short and you can skip over some of the mathematical detail -- as it shows how some very important global economic realities may have quite simple underlying causes. The bigger problem, of course, is learning then how to craft real world policies so as to keep inequality within bounds, avoiding the many kinds of social dysfunction it clearly seems to give rise to.