Is here:

Agglomeration can be caused by asymmetric information and a locational signaling effect: The location choice of workers signals their productivity to potential employers. The cost of a signal is the cost of housing at a location. When workers’ price elasticity of demand for housing is negatively correlated with their productivity, skill-biased technological change causes a core-periphery bifurcation where the agglomeration of high-skill workers eventually constitutes a unique stable equilibrium. When workers’ price elasticity of demand for housing and their productivity are positively correlated, skill-biased technological improvements will never result in a core periphery equilibrium. This paper claims that location can at best be an approximate rather than a precise sieve for high-skill workers.

To try and simplify this a little bit, there is a wage premium associated with big cities, which is difficult to explain using plain vanilla economics — there should be an incentive for firms to move elsewhere, rather than pay higher real wages. To get around this, you can invoke a number of different models. Many include agglomeration externalities, where there are positive spillovers to concentration — the whole is greater than the sum of the parts. What this paper posits is that the premium can be explained in part by worker signaling. This is similar to the filtering story I’ve been telling at this blog for some time. When housing costs are high, high productivity workers will opt in because they know that their skills will allow them to earn enough to justify the higher housing costs. And firms may want to locate in a particular area because high housing costs will sort the local labour market for them — they don’t have to do as much work to hire a talented worker.

Interesting bit here:

Under the framework of agglomeration externalities, an increase in the ratio of high-skill labor in one region causes more than a proportional increase in the average real wage (or an increase in labor’s marginal product). In a locational signaling model, an increase in the ratio of high-skill labor in a region yields a proportional increase in the average real wage.

Of course, I’ve argued before that workers might be signaling not only that they are productive, but they also happen to be workers whose productivity levels are most externality-dependent. For instance, it’s expensive to live in San Jose. Therefore, we might argue that only a very productive worker would move to San Jose, because only a very productive worker could count on earning enough to pay San Jose housing costs; hence, high San Jose housing costs sort for productive workers.

On the other hand, a highly productive worker whose productivity isn’t dependent on externalities might do just as well, net of housing costs, somewhere else. But a productive worker who will most benefit from the collaborative, innovative atmosphere of a San Jose R&D lab, whose productivity is dependent on being around other productive workers whose productivity levels are likewise co-worker dependent, will strongly prefer to locate in San Jose. The price elasticity of demand for an externality-rich city among such workers will be quite low. And as per the paper above, selection of workers with externality-dependent productivity levels should lead to a more than proportional increase in average real wages, and increases in housing costs sufficient to deter most of those workers who are simply looking to signal what you might call independent productivity.

Distinguishing between these two types of productivity-signaling workers then allows us to posit a range of city types featuring a range of talent concentrations and a range of wage premiums.

</super wonkishness>

To step back a bit, I think it’s also worth thinking about what else we signal when we choose a city. Cities are the great relationship markets; it’s said that those who are bearish on the future of cities have already met their spouse. When you move to a place like New York, you signal things about yourself to prospective partners, and you select for a pool of partners that is quite different from a corresponding pool in Denver or Des Moines.

And I think that to a certain extent, people make these choices based on conceptions they have about themselves and the people they’d like to be. If you see yourself as someone who is interested in art, you may move to New York, not just because there is a lot of great art there, but also because you’ll meet people there who are themselves interested in art and who will nudge you toward more involvement with art and artists. Or you might move to Denver, because you want to be an outdoorsy person. People you meet there will typically be outdoorsy, and they’ll make it easier for you to become this outdoorsy person that you hope to be. At a more general level, people may simply feel that they’re “destined for bigger things”, or ready for a “simpler life”, and they may choose cities based on these feelings. Not just because they’re going where they want to go, but because they’re committing themselves to a certain lifestyle, and placing themselves in a situation where the people they come to know will act as constraints on them, pushing them to behave in a certain way.  After all, you can love art in Denver and be outdoorsy in New York.

It seems to me that people want to be a lot of things that they can’t necessarily become on their own. A move can be a means to commit oneself to a certain course, and to make it harder to back away from a desired goal or style of life.