Periods of high volatility in house prices increasingly occur synchronously in the housing markets of different countries. Such contagion, or volatility spillovers are often captured by ARCH type models and “GARCH models have been used extensively to analyze cross-border volatility spillovers in asset markets” (Beirne et al, 2009, p8). While markets in different countries may show periods of correlation in performance, i.e., house price movements, this by itself does not necessarily imply contagion. Contagion can be taken to refer to the unanticipated transmission of shocks and as such can be differentiated from correlations that may exist in more normal market circumstances. Kaminsky et al, (2009) “refer to contagion as an episode in which there are significant immediate effects in a number of countries following an event” (p55) They contrast this with situations where effects on a number of countries take time, labelling the latter as a ‘spillover’. Masson (1998, 1999) refers to three types of contagion caused by the simultaneous impact of common shocks, spillovers due to inter-country interdependencies and pure or shift contagion resulting from sudden movements, e.g., the withdrawal of liquidity following a crisis in one country that then impacts on other countries. Interdependencies between countries in, say, liquidity flows are the main channels through which crises are transferred (Forbes and Rigobon, 2000 & 2002). Brunnermeier and Pedersen (2009) discuss the role of liquidity and postulate a liquidity correlation channel that can generate contagion. 

In this paper we examine contagion in house prices in selected OECD countries following the sub-prime crisis, beginning in 2007, focusing on the increase in liquidity caused by quantitative easing applied by central banks including the European Central Bank. The aim is to capture the effect of liquidity shocks after 2007 and 2014 measuring house price contagion among three countries two of which do not belong to the Eurozone. Unlike models using financial data, our data frequency is lower and is quarterly rather than monthly or daily. This reflects data availability and the slower transactions process for complex (multiple attribute) housing assets compared to more liquid financial assets. Our analysis builds upon correlation and we use the dynamic conditional correlation model developed by Engle (2001, 2002), Engle and Sheppard (2001), and Tse and Tsui (2002) – DCC-GARCH.