Sansom, Bazil (2020) Intrinsic cyclicality and spatial coupling in U.S. housing market instability, 1975-2020. (PhD thesis), Kingston University, .
Abstract
Where did the national U.S. house price boom-bust that triggered the Global Financial Crisis come from? In this thesis I show that approaching U.S. housing as a network of sub-national markets with their own non-trivial intrinsic dynamics helps us to understand the character and timing of housing market instability, both at the sub-national and national levels. A detailed empirical analysis of U.S. house prices across historical time (Jan 1975 - Jun 2020) and geographical space reveals striking spatio-temporal patterns in the data. I argue these empirical results are hard to account for within existing theoretical frameworks (idiosyncratic shocks or bubbles), but are consistent with the intrinsic cyclicality of local markets combined with local spatial dependence between cycles in neighbouring markets. I show that a simple model based implementation of this hypothesis is able to easily replicate my key empirical results. My first thesis paper shows, using wavelet spectra of state level price series, that state level markets exhibit evidence of a permanent c.10 year cycle component over the entire historical sample period. My next two papers introduce instantaneous phase based methods as a spectrally focussed dynamic framework within which to study the relative timing of these cycles. I show that permanent cycles at the state level, whilst asynchronous during earlier periods, synchronised dramatically from the mid 1990s (coinciding with important shifts in housing finance in the U.S.), and that this synchronisation of cycles contributed significantly to the historically unprecedented national house price boom-bust of the 2000s. Moreover my analysis shows a clear and stable “traveling-wave” pattern in the relative timing of local cycles across markets over the entire historical sample period. My fourth paper investigates the relationship between the timing of “bubbles” and of cycles: combining bubble date-stamping strategies employed in the empirical bubble literature with instantaneous phase information obtained via complex wavelet analysis, I show a systematic relationship between the phase of the permanent cycle component and the timing of explosive bubble and collapse episodes. This result suggests that the underlying price cycles influence the likelihood a bubble may emerge or burst. In my fifth paper I show that both the synchronisation of cycles over time and the specific spatial patterns documented by my empirical analysis are easily reproduced in a simple model of endogenous local speculative house price cycle dynamics extended to a spatial network setting – in which price expectations are partly influenced by neighbouring markets. I argue, nevertheless, that my empirical results suggest shifts in housing finance may have played an important role in the synchronisation of U.S. housing cycles that I document. Taken together these different empirical and theoretical contributions suggest a possible significant paradigm level re-interpretation of U.S. housing market instability, with distinctive implications for theory, methods, policy and finance.
Actions (Repository Editors)
 |
Item Control Page |