My post on rent control set me wondering: is there really such a thing as a downward-sloping demand curve for housing?

I ask because of simple introspection. If house prices were to fall sharply, I would not buy more. Nor would I buy less if prices rose. This is simply because I'm content with the house I've got. My "demand curve" for houses is a vertical line at zero, and I guess the same is true for millions of owner-occupiers. Granted, a fall in prices would allow some renters to buy. But their demand would simply rise from zero to one, and stop there. That's not the smooth line or continuous function of the textbooks.

Sure, there might be some property speculators who would increase demand as prices fell. But these are a small minority.

How, then, can we speak of a conventional demand curve for houses? If such a thing exists at all - which I doubt* - it is likely to exist not at the individual level, but in aggregate; a small drop in price would allow a few people to become owners rather than renters, and a larger drop would allow more to do so. In this sense, an aggregate relationship isn't simply a typical individual's behaviour writ large.

It's not just housing of which this is true. In a study of the Ancona fish market Alan Kirman and colleagues showed that, for very many individual buyers, there was no negative relationship between price and quantity demanded, and in many cases there was a positive relationship. But there was a downward-sloping demand curve in aggregate. They say:

The aggregate characteristic, the negative relation between price and quantity, is not a reflection of individual behavior.

I say this for two reasons. First, to question "representative agent" theory. What's true at the aggregate level need not true for a typical individual. Instead, outcomes are the product of complex emergent interactions.

Secondly, to counteract a common bias in our thinking about the role that networks and interlinkages play in economics. Insofar as we think about these - and we don't do so enough - we consider them as a destabilizing force; for example, when links (pdf) between banks created systemic risk, or when information cascades cause stock markets to fall. This, though, is only part of the story. As Kirman has shown, aggregation can also generate conditions for stability - such as downward-sloping demand curves - which are not necessarily present at the individual level.

* In practice, aggregate price and quantity demanded are positively correlated, simply because both are driven by the same things; incomes, credit availability, interest rates etc. To get a downward-sloping curve thus requires some careful controls. And even then we might have the problem of extrapolative expectations: a fall in prices might lead to lower demand, to the extent that potential buyers expect prices to fall further.