Social Networks and Housing Markets

Abstract

We document that the recent house price experiences within an individual's social network affect her perceptions of the attractiveness of property investments, and through this channel have large effects on her housing market activity. Our data combine anonymized social network information from Facebook with housing transaction data and a survey. We first show that in the survey, individuals whose geographically-distant friends experienced larger recent house price increases consider local property a more attractive investment, with bigger effects for individuals who regularly discuss such investments with their friends. Based on these findings, we introduce a new methodology to document large effects of housing market expectations on individual housing investment decisions and aggregate housing market outcomes. Our approach exploits plausibly-exogenous variation in the recent house price experiences of individuals' geographically-distant friends as shifters of those individuals' local housing market expectations. Individuals whose friends experienced a 5 percentage points larger house price increase over the previous 24 months (i) are 3.1 percentage points more likely to transition from renting to owning over a two-year period, (ii) buy a 1.7 percent larger house, (iii) pay 3.3 percent more for a given house, and (iv) make a 7% larger downpayment. Similarly, when homeowners' friends experience less positive house price changes, these homeowners are more likely to become renters, and more likely to sell their property at a lower price. We also find that when individuals observe a higher dispersion of house price experiences across their friends, this has a negative effect on their housing investments. Finally, we show that these individual-level responses aggregate up to affect county-level house prices and trading volume. Our findings suggest that the house price experiences of geographically-distant friends might provide a valid instrument for local house price growth.

Informative Social Interactions

“…Exploiting then di¤erences between these responses and responses about population wide stock market behaviour or information, enables our novel triple circle approach to separate pure information exchange and mindful imitation from mindless imitation, while controlling for unobserved factors in ‡uencing how the respondent perceives others and the economy in general. 5 We …nd that respondents' perceptions about the shares of their …nancial circles that are informed about the stock market or actively participating in it are systematically and positively related to the accuracy of respondents'perceptions of past stock market returns and of expectations of future returns; to the probability of stock market participation; and to the risky portfolio share conditional on participation. Respondents who perceive their …nancial circles to be more informed or more widely participating in the stock market have perceptions of returns that are closer to the truth, controlling for their perceptions of the other circles, a wide array of respondent characteristics, as well as perceptions of the respondent's social standing in terms of professional accomplishment, education, or wealth.…”

“…The assumption that the network is weighted by signal precision captures the fact that investors put more importance on good quality information they receive from the social circle. Given the information network, investors'information sets are de…ned by I i = fx i ; pg; 8i = 1; :::; n (5) because also asset prices are allowed to transmit information in equilibrium, and investors rationally anticipate it. We also assume that the random variables X; Z and i are all jointly independent.…”

“…R t B i R t : Answers to question C42 in our survey enable probabilistic elicitation of respondents'perceptions about the most recent realized cumulative stock market return over a three-year period. 24 We focus on the mean of each respondent's subjective probability distribution over the size of the realized three-year stock market return, which we denote Table 2a under columns (3)- (5) show that the respondents'perception gap about facts, as measured by their mean perception error, is a strongly signi…cant predictor of their forecast error, regardless of whether peer variables are included in the regression (columns 4 and 5) or not (column 3). This extends the results of the in ‡ation literature to the stock returns literature.…”

“…Corresponding …rst-stage regression results are reported under columns (3) and (4), respectively, in the online appendixTable A2.3 7 SeeTable A1in the online appendix, under columns (5) and (6), respectively. Corresponding …rst-stage regression results are reported under columns(5) and(6), respectively, in the online appendixTable A2.3 8 A good recent example of this approach is Bailey et al (2019), which uses Facebook data to assess peer e¤ects in product adoption.…”

Redistribution of Local Labor Market Shocks through Firms' Internal Networks

“…Prior research has focused on production (i.e., input-output) networks (Acemoglu, Akcigit, and Kerr (2015), Acemoglu et. al (2016), Barrot and Sauvagnat (2016)), social networks (Bailey et al (2016)), and banking networks Rosengren (1997, 2000), Schnabl (2012)). By contrast, our paper studies how shocks propagate across different regional labor markets within the same (non-tradable) industry sector through firms' internal networks of establishments.…”