The Role of Positional Concerns in Determining Herding: Evidence from US Residential Property Markets

Introduction: This study examines the determinants of herding in the U.S. housing market with the focus on the effect of psychological biases on herding. People are motivated to copy others due to a psychological need for social behaviour. If homebuyers see large local inequalities in homeownership, they will have a stronger desire to overconsume relative to areas with more equal housing distribution. This mimetic behaviour will therefore lead to higher levels of herding, with resultant poor economic outcomes. Therefore, we test if areas with more extreme distributions of housing, income and consumption (i.e. areas where people have greater degree of positional concern) will exhibit more extreme herding, while controlling for fundamental economic and housing market variables. The development of behavioural finance in the 1970s (Kahneman and Tversky 1979) and then its crossover into asset markets shortly after (Shiller 1982) has allowed behavioural factors in real estate to enjoy a growing research interest. It has been shown that an appreciation of sociological and psychological factors provides an understanding of market dynamics beyond the conventional efficient markets approach (West 1988, Lux 1995). Herding is defined when behaviour is correlated across individuals (Devenow and Welch 1996), especially where it leads to sub-optimal investment decisions and bubble formation. Herding can be rational if informational inefficiency and agency issues lead investors to copy others (Bikhchandani, Hirshleifer et al. 1998), or irrational when they are susceptible to psychological biases (Devenow and Welch 1996). One motivator for the presence of these psychological biases is the established economic presence of positional goods, whose value is derived from their social status and where the value is based on relative, rather than absolute, distribution. The overconsumption of positional goods creates negative externalities which have wider economic repercussions (Frank 2005). It has been shown that excessive social positional concerns lead to poor outcomes in physical and mental health (Frank 2008), and also sub-optimal investment decisions, such that people over-consume and over-leverage. Evidence for the presence of herding in institutional investors is mixed (Barber, Odean et al. 2009), and the research in direct residential markets is limited, however the latter has found some significant results (Hott 2012, Ngene, Sohn et al. 2017). The existing evidence suggests that higher levels of herding can lead to bubble formation, and the resulting welfare loss can be extensive (Deng, Hung et al. 2018), as most clearly seen in the fallout from the Global Financial Crisis. Considering the scale of residential property as an asset class (over $33 trillion in the USA alone), and the dual nature of housing as an investment asset and consumption good, then there is an obvious requirement for further research. Data and Methodology: Urban areas are based on the US Census Bureau definitions of metropolitan statistical areas (MSA) which are the 384 urban centres with a population greater than 50,000. In total they represent over 281 million inhabitants or 85% of the population. The actual house price data is taken from Zillow Research, and the data for social, economic, demographic and property variables is taken from the American Community Survey, administered by the Census Bureau. The time period tested is from 2005 until 2018, and so a panel approach has been used. We test the main hypothesis using the following model: CSAD=f(fundamental factors)+f(positional factors) + Where CSAD is the cross-sectional absolute deviation, a herding measure. The CSAD can be derived from the relationship between price changes in the overarching MSA and the underlying ZIP code locations. It is expected that areas with more extreme distributions of housing, income and consumption will also exhibit more extreme herding as a result, with these positional determinants being highly significant. Results and implications: The initial empirical results have highlighted the importance of MSA size in determining the drivers of herding behaviour, with smaller urban areas being more impacted by new housing supply and availability of housing finance, whilst larger cities are much more strongly impacted by the relative distribution of house prices. Areas with older housing stock on average exhibit lower levels of herding, which motivates an interest in the causal connection between development and herding behaviour. These results have motivated deeper investigation into the role of MSA size in determining herding behaviour, as well as local regulatory conditions and supply elasticities which, along with ownership structures, show evidence of interesting results. The inclusion of various market efficiency-based interaction variables as mechanisms for the propagation of herding behaviour will also add to the literature on this topic, which is of particular interest in property due to its unique market structure. The research has several important implications, firstly in aiding the identification of market signals which can assist with better informed market forecasting through a deeper understanding of market dynamics. In terms of portfolio construction and risk management, a better incorporation of social and economic drivers could result in superior risk-adjusted returns.