Analyzing Comovements in Housing Prices Using Vine Copulas

Abstract

Prior to the housing crisis, the Gaussian copula provided the basis for estimates of the degree of diversification of structured mortgage-based securities. The Gaussian copula's popularity stemmed not only from its link to the familiar normal distribution, but also from the fact that, unlike other copula-based models, it readily extends to higher dimensions. But the Gaussian copula has asymptotic independence, such that events, regardless of the strength of their correlation, become independent if one pushes far enough into the tails. Instead, this article forms multivariate models of housing price comovements using vine copulas. These more flexible models not only fit the data better, but they also uncover far stronger correlations between housing price movements, especially during extreme market swings. (JEL G21, C32, C51)

Crop price comovements during extreme market downturns

“…Financial economists have long recognised that univariate distributions of financial returns depart from normality (Blanchard and Watson 1982). Recent evidence finds similar results in multivariate distributions of stock prices (Lux and Sornette 2002), hedge fund returns (Li and Kazemi 2007;Boyson et al 2010) and housing prices (Zimmer 2015). Rodriguez (2007), using procedures similar to those employed in this paper, confirms departures from normality are especially prevalent in the midst of financial turmoil.…”

“…Of course, such instruments are not unique to crop markets. Structured mortgage-based securities are a common, albeit shrinking, presence in housing finance and are often cited as a root cause of the housing crisis (Zimmer 2015). And much like mortgage-based structured securities, assessing the riskiness of such assets requires estimates of the extent to which crop prices move in synchrony.…”

Time-Varying Correlation in Housing Prices

“…Existing methods to estimate correlations in housing price movements concentrate on complex distributional features of housing prices, but those methods treat the magnitude of correlation as time invariant (Li 2000;Zimmer 2014). Yet there are several reasons to suspect changing correlations.…”

“…Recent studies argue that, by relying on the assumption of multivariate normality, existing models of housing price comovements fail to adequately capture comovements during extreme tail events (Zimmer 2014). Thus, in order to allow for the possible presence of tail dependence, in both the CCC and DCC versions of the model, the disturbances υ t = υ 1,t , υ 2,t in Eq.…”

“…But the multivariate normal fails to account for correlations during extreme market swings. Those studies argue that such " tail" behavior could have been captured using other statistical distributions (Zimmer 2012(Zimmer , 2014Ho et al 2014).…”