Table 7.

Effect of risk preferences on the share of financial wealth in stocks

			Risk tolerance proxy
Control for log risk tolerance	None	Categorical survey response	Ignoring response error	Modeling response error	Modeling response error	Including application covariates
Estimator	OLS	OLS	OLS	OLS	GMM	OLS
Category 3		−.026 (.010)
Category 4		.022 (.012)
Category 5–6		.025 (.011)
Proxy			.008 (.003)	.146 (.054)	.162 (.060)	.152 (.056)
Male	.024 (.007)	.023 (.007)	.023 (.007)	.023 (.007)	.014 (.008)	.018 (.008)
Education
>16 years	.034 (.012)	.032 (.012)	.032 (.012)	.031 (.012)	.012 (.013)	.019 (.014)
13–16 years	.036 (.009)	.035 (.009)	.035 (.009)	.035 (.009)	.024 (.009)	.029 (.009)
<12 years	−.023 (.009)	−.024 (.009)	−.023 (.009)	−.023 (.009)	−.026 (.009)	−.024 (.009)
Black	−.029 (.009)	−.029 (.009)	−.029 (.009)	−.028 (.009)	−.024 (.009)	−.027 (.009)
Hispanic	−.035 (.012)	−.036 (.012)	−.035 (.012)	−.035 (.012)	−.034 (.013)	−.038 (.012)
Age/10	−.002 (.008)	−.001 (.008)	−.001 (.008)	−.001 (.008)	.006 (.008)	.006 (.008)
Log income	.002 (.005)	.003 (.005)	.003 (.005)	.002 (.005)	.004 (.005)	.003 (.005)
Log wealth	.046 (.002)	.047 (.002)	.046 (.002)	.046 (.002)	.047 (.002)	.046 (.002)
R2	.170	.172	.171	.171	.178	.177

NOTE: Regressions include 5,818 households with positive financial wealth and total income in 1992. Individual attributes are from the household's financial respondent. Share of wealth in stocks has a mean of .158 and a standard deviation of .286. Asymptotic standard errors are in parentheses. In the second to last column, the GMM estimates are based on the formula in (25), and the R² in the last two columns is based on the formula in (28). For the application subsample, the true-to-proxy variance ratio λ is 6.40. In the last column, the proxy is constructed from a model of log risk tolerance that conditions on the application covariates as well as the gamble responses.