That is the topic of my latest Bloomberg column, and it is assuming no major increase in supply in the megacities themselves. Here is one bit:
We live in a special time where clustered activities are unusually important for economic growth. Some activities, such as dentistry and cement production, don’t cluster geographically very much, for obvious reasons. In contrast, finance (New York and London), information technology (the Bay Area), and entertainment (Hollywood and New York) are the most clustered. For whatever reasons, it makes sense to have many of the top decision-makers in one place.
Leading cities have become so expensive in large part because two of these clustering sectors — finance and information technology — have been ascendant. There is no particular reason to expect those trends to continue forever, and that will bind rents in affected cities.
Even tech will decentralize its gains over time:
If you think of a typical technology project, some of the gains go to the venture capitalists and the intellectual property holders, and some of the gains go to broader society, including consumers. Insofar as the gains are disproportionately reaped by the early project initiators, then yes real estate values in the Bay Area (and other tech clusters) will rise. But the most likely future for information technology is that it will spread its benefits more and more broadly into more and sectors of the economy. That scenario suggests a partial convergence of urban futures.
Another way to put the point is that intellectual property returns erode over time. In the early years of smartphones, a big part of the gain goes to Apple. As cheap imitators enter the market, prices fall and more of the gains go to consumers, or business users of the product, who are scattered across the country.
The article contains other points of interest.