Disentangling the Effect of Housing on Household Stock Holdings : Evidence from Japanese Micro Data

Using Japanese household micro survey data during 2000-2015, this paper examines the effects of housing on household stock holdings. To disentangle the effect of housing asset (land value) and mortgage debt on households' portfolio share of stocks out of liquid financial assets, we apply the instrumental variable approach by Chetty et al. (2017) that employs differences in average land price indices across housing markets in the year in which household portfolios are measured and those in the year in which the house was purchased. Our estimates suggest that an exogenous increase in land value (holding mortgage debt constant) increases the portfolio share of stocks, while an increase in mortgage debt (holding land value constant) reduces it. We also find that an increase in land value and mortgage debt (holding home equity constant) does not affect the portfolio share of stocks, but increases the repayment of mortgage debt