1 Introduction
In 1980, the median home in Scranton, PA, was worth more than half the median
home in New York City. By 2018, its value had decreased to one fifth of the New
York City home according to U.S. Census data. In the U.S. and internationally, there
has been a substantial increase in regional housing price differences since the 1980s
(Van Nieuwerburgh and Weill, 2010; Hilber and Mense, 2021). The spatial structure
of economic activity has changed considerably across countries in recent decades. A
prominent trend is increasing social and spatial polarization among different sub-
national housing markets. As housing is the most important asset for most households,
the increasing dispersion of housing prices and housing wealth have become the subject
of intense public debate.
1
From an economic point of view, rising price dispersion across segmented housing
markets could increase spatial misallocation of labor as productive workers are forced
to stay in places where housing is still affordable. For instance, Hsieh and Moretti (2019)
estimate that such misallocation slowed down the growth rate of U.S. GDP by one third
in past decades. An increase in local housing prices has also been shown to lead to
more misallocation of capital (Herkenhoff, Ohanian, and Prescott, 2018), to affect local
non-tradable employment (Mian and Sufi, 2014) and demand conditions (Mian and
Sufi, 2011; Mian, Rao, and Sufi, 2013; Guren et al., 2020), as well as consumer prices
(Stroebel and Vavra, 2019).
Why have housing prices risen more in some locations than in others? In the most
parsimonious framework, rental cash flows determine the value of housing assets: the
price of a house is equivalent to the discounted expected future rental cash flow it
generates (Poterba, 1984). An important implication – and the starting point for most
existing explanations of growing housing price dispersion – is that price and rent
dispersion should evolve in lockstep. Yet, as we will show, this approach is at odds
with an important stylized fact: rent dispersion has increased considerably less than
price dispersion in recent decades, both in the U.S. and internationally. Existing studies
that model housing price dispersion as a function of growing differences in local rents
(
e.g.
; Van Nieuwerburgh and Weill, 2010; Gyourko, Mayer, and Sinai, 2013) typically
overestimate changes in rent dispersion by a substantial margin.
We use a novel long-run data set of housing prices and rents for 27 major agglomera-
tions in 15 developed countries as well as long-run data covering the entire cross-section
of U.S. MSAs, and show that price–rent ratios in large agglomerations have increased
about twice as much as the national average since the 1980s. Moreover, new research
1
For instance, existing homeowners in high price urban areas have an incentive to restrict urban growth
to the detriment of new buyers (Ortalo-Magn
´
e and Prat, 2014). The increasing polarization of housing
wealth may have also contributed to political polarization at the national level (Adler and Ansell, 2019;
Ansell, 2019).
2