Let's get dense

I've been reading Ryan Avent's new book The Gated City, and it provides some good arguments in favour of the point I was making about prosperous cities needing to grow up as well as out.

As I said, in the debate over the National Planning Policy Framework I think there has been an undue focus on the virtues of making land on urban fringes available to develop at low prices, because that will lower costs for businesses and households. Avent's argument, as I understand it, is that we should be most concerned with the costs facing high-productivity or highly growth-promoting firms (and the people they employ), and they are much more likely to want to locate in dense urban areas because of the positive impact of density on productivity, innovation and entrepreneurship. So we should be at least as worried about restrictions on densifying in city centres as we are about restrictions on expansion of suburban areas.

You could argue that this is a false opposition and that cheaper land on the fringe will reduce pressure on the centre. The problem is I'm not sure the two locations are good substitutes. To generalise a bit crudely, those that like fringe locations - sprawl developers, lovers of low density housing and low productivity firms- will benefit from cheaper fringe land, while those who don't - higher density developers, lovers of higher density areas and higher productivity firms - will not. Of course, compact, well-planned urban expansion could potentially give us the best of both worlds, but we don't seem to be very good at that in this country and the current draft of the NPPF doesn't really encourage it.

The bottom line is that real estate prices are generally highest in urban centres. Insofar as that represents the market signalling something, it is a strong signal to build more in urban centres, and only a weak signal to build more on suburban fringes. 'Land supply' isn't everything, at least as it's generally conceived: what we are allowed to do with already-developed land is just as important.