Between 1950 and 1990, the aggregate population of central cities in the United States declined by 17 percent despite population growth of 72 percent in metropolitan areas as a whole. This paper assesses the extent to which the construction of new limited access highways has contributed to central city population decline. Using planned portions of the interstate highway system as a source of exogenous variation, empirical estimates indicate that one new highway passing through a central city reduces its population by about 18 percent. Estimates imply that aggregate central city population would have grown by about 8 percent had the interstate highway system not been built.
You can quibble with the model specification but I accept the paper's general conclusion. A few points:
1. I am reluctant to call this a subsidy without qualification. Visit an old medieval city like Esternach, Luxembourg ("a formidable fortress erected by Count Siegfried in 963"). The main buildings are grouped together on a hill. It's value-enhancing that later governments adopted policies, such as near-free trade and national defense, which eased those constraints and spread out the population. I wouldn't say that the resulting population distribution of say Paris is best thought of as resulting from a subsidy because they're not all on top of a hill somewhere, I would say it is the result of greater wealth and trade opportunities and law enforcement, with some element of subsidy. So if you favor the construction of the interstate highway system, as I think most commentators do (try driving for long on Rt.1), it is a rhetorically loaded decision to invoke the word "subsidy" as the major mode of explanation.
2. Greater wealth, transport, and trade naturally cause people to seek out larger homes and greater living space. In a world with no policy distortions this may well be the dominant effect in various long-run settings. The rise of the suburbs is not all subsidy-driven by any means.
3. I recall the D.C. area quite well before it had either the Orange Line on the Metro (serving the Virginia suburbs) or Rt.66 going into Washington. The construction of both enabled the suburbs to spread further westward. But which was more important in driving this process? Mass transit also can encourage suburbanization, especially the city has abominable public schools. A longer and better Metro system — most of all with better parking — would have meant even more people moving to the suburbs.
4. Do not forget "white flight." For American cities, this paper suggests that each black arrival lead to, on average, 2.7 white departures. When it comes to DC, "black flight" has been a significant phenomenon and it is a major reason why the city has been losing population.
5. Canada is much less suburbanized than is the United States and the greater "flight from blight" in the U.S. seems to be a major reason (see p.8 here). When I think of U.S. suburbanization, I think of the failures of our municipal public sector as very important in this process. In contrast, Fairfax County and Fairfax City governments are of reasonable quality.
6. For competitive Tiebout-related reasons, it is no accident that the public schools are so often better in the suburbs than in major cities. Countries with strong social norms, such as Sweden, have this problem less.
7. One simple model (which I am not endorsing straight up) is that most American people with kids have a near-lexical preference for living in the suburbs. Anything that enables them to do so can be called a cause of suburbanization and measured as such. But isolating and measuring these marginal impacts, in the econometric sense, distracts us from seeing how general and how strong the underlying infra-marginal forces are and those are very often preference-based.