Does wall street affect main street? examining potential spillovers from investor stock market sentiment to personal consumption expenditures

Abstract

Investor sentiment rather than aggregate stock wealth predicts changes in non-durable consumer spending across different economic cycles above and beyond housing wealth, labor income, and other macro-economic factors. A statistically and economically significant positive contemporaneous relationship exists between investor sentiment and current nondurable spending, whereas sentiment negatively predicts later spending. Consumers adjust their spending patterns along with sentiment about the stock market, then reverse sentiment-based consumption smoothing when biased market expectations do not come to fruition. Investors’ biased beliefs about stock returns affect consumers purchasing decisions and, thus, the overall .

Appendices

Appendix A: Definitions of variables

Variables with an asterisk are per capita.

Main variables

*Personal Consumption Expenditures (PCE). :

Nondurable, durable, and service expenditures from the Bureau of Economic Analysis. Data is annualized, seasonally adjusted, and inflation adjusted.

Investor Stock Sentiment (sent, sent_o). :

From Baker and Wurgler (2006) and Baker and Wurgler (2007), sent is an index of stock market sentiment indicators, standardized to be approximately mean zero with unit standard deviation. Sent_o has been orthogonalized to macroeconomic business cycle variables. See Baker and Wurgler (2007) for details. In order to take the natural log of sentiment, sent and sent_o are transformed to a mean of 100 and an annual standard deviation of 20. Hence, sentiment is assumed to be approximately as volatile as the stock market.

*Disposable Income. :

Total after-tax personal income, from the Bureau of Economic Analysis. Data is annualized, seasonally adjusted, and inflation adjusted.

*Federal Housing Finance Agency Home Price Index (FHFA). :

A quarterly U.S. home price index based on actual sales transactions and appraisal transactions. Monthly data is constructed from quarterly data with cubic spline interpolation. Data is inflation adjusted.

*S&P 500 Stock Index (SP500). :

The Standard and Poor’s 500 U.S. equity index.

*Deposits :

Total customer deposits in U.S. chartered banks, from the Federal Reserve. Data is inflation adjusted.

Control and robustness test variables

Unemployment rate :

The unemployment rate as a percentage of the active labor force. Data is seasonally adjusted.

*Industrial production :

Index of industrial production from the Federal Reserve. Data is seasonally adjusted and inflation adjusted.

*Consumer credit :

Total outstanding debt from consumer credit, from the Federal Reserve. Consumer credit excludes loans secured by real estate. Data is seasonally adjusted and inflation adjusted

Income/Debt :

The ratio of disposable income to total consumer credit.

*GDP :

Chained GDP, seasonally adjusted and inflation adjusted with the chained method.

*Saved income :

Disposable income less total personal consumption expenditures. Data is seasonally adjusted and inflation adjusted

Consumer confidence (Conf) :

The Conference Board’s monthly index of consumer outlook.

UM Consumer sentiment (Cons sent) :

The University of Michigan index of consumer outlook.

CBOE Volatility Index. :

The Chicago Board of Options Exchange provides two indices to capture the implied volatility of index-options. The VIX since January 1990 is implied volatility on the S&P 500 index and the VXO since June 1986 is implied volatility on the S&P 100.

Interaction of sentiment and SP500 :

The product of sentiment and the SP500 level.

Interaction of sentiment and SP500 squared :

The square of the product of sentiment and the SP500 level.

High and low sentiment indicators :

The high sent_o indicator variable has a value of one if sentiment is more than one standard deviation above its mean, and zero otherwise. The low sent_o indicator variable has a value of one if sentiment is less than one standard deviation below its mean, and zero otherwise. interactions.

Appendix B: Index option implied volatility as a measure of risk expectations

In models with a stochastic discount factor (SDF), stock market risk can be defined as the negative of covariance between the SDF and stock returns:

$$ Risk=-Cov(m,r)=-\rho_{m,r}\sigma_{m}\sigma_{r}, $$

(B.1)

where m is the SDF and r is the stock return. ρ _m,r is the correlation between the SDF and returns; σ _m and σ _r is SDF volatility and return volatility respectively. C o v(m,r) is not realized risk, it is the expectation of what realized risk will be. So σ _r is not realized stock market volatility, it is expected volatility, which is reflected in stock index option prices as implied volatility.

If σ _r goes up, then all else equal, risk also goes up. SDF volatility or the correlation between returns and the SDF would have to also change in order for implied stock volatility and expected risk to move in opposite directions. A sufficient condition for implied volatility to be a same sign proxy for risk is:

$$ Cov(\rho_{m,r}\sigma_{m},\, \sigma_{r})\geq 0. $$

(B.2)