Does Consumer Sentiment Forecast Household Spending? If So, Why?
研究密歇根大学消费者情绪指数（ICS）是否对未来家庭消费支出有预测能力，发现ICS单独可解释约14%的消费增长变化，且在其他指标之外仍有一定增量信息。
In the three months following the Iraqi invasion of Kuwait, the University of Michigan's Index of Consumer Sentiment (ICS) fell an unprecedented 24.3 index points, to its lowest level since the 1981-1982 recession.' This collapse in household confidence became the focus of a great deal of economic commentary and, indeed, frequently was cited as an important-if not the leading-cause of the economic slowdown that ensued. Concern was fueled by the well-known contemporaneous correlation between the ICS and the growth of household spending. Figure 1 shows quarterly averages of the index, 1978-1993, together with the quarterly growth in real personal consumption expenditures as measured in the national income accounts (Bureau of Economic Analysis). The correlation is impressive. Of course, it is not surprising that sentiment and the growth of spending are positively correlated. This correlation may simply reflect that, when economic prospects are poor, households curtail their spending and also give gloomy responses to interviewers. Thus, the contemporaneous correlation between sentiment and spending does not refute traditional life-cycle or permanentincome models of consumption. Nor does it necessarily make the job of forecasting changes in consumption any easier. From the point of view of an economic forecaster, the questions of interest are first, whether an index of consumer sentiment has any predictive power on its own for future changes in consumption spending, and second, whether it contains information about future changes in consumer spending aside from the information contained in other available indicators. In Section I, we present evidence that the answer to the first question is a clear yes: we find that lagged values of the ICS, taken on their own, explain about 14 percent of the variation in the growth of total real personal consumption expenditures over the post1954 period. Further investigation shows that the answer to the second question is probably yes as well, though here the margin is narrower and the evidence more murky. The ICS contributes about 3 percent to the R2 of a simple reduced-form equation for total personal consumption expenditures in the longer of the two sample periods we examine, but nothing in the shorter sample period (though the latter result is heavily influenced by the observation for 1980:2). For the major subcategories of spending, the contribution generally ranges between 1 percent and 8 percent. Overall, we read the evidence as pointing toward at least some significant incremental explanatory power. Therefore, we take as given for the remainder of the paper that sentiment forecasts spending, and we turn to the issue of how that statistical relationship should be interpreted. One possible interpretation is that sentiment is an independent driving factor in the economy, and that changes in * Carroll: Division of Research and Statistics, Stop 80, Federal Reserve Board, Washington, DC 20551: Fuhrer: Research Department, Federal Reserve Bank of Boston, Boston, MA 02106: Wilcox: Division of Monetary Affairs, Stop 71, Federal Reserve Board, Washington, DC 20551. We have benefited from the research assistance of Stephen Helwig and Christopher Geczy and the comments of an anonymous referee. The views expressed in this paper are those of the authors and not of the Federal Reserve Board, the Federal Reserve Bank of Boston, or the other members of the staff of either institution. IThe Conference Board's Consumer Confidence Index also plunged at the same time.