DP9894 Surprise, Surprise - Measuring Firm-level Investment Innovations
|Author(s):||Rüdiger Bachmann, Steffen Elstner, Atanas Hristov|
|Publication Date:||March 2014|
|Keyword(s):||expectation errors, firm data, higher moments, idiosyncratic shocks, investment, survey data|
|JEL(s):||E20, E22, E30, E32|
|Programme Areas:||International Macroeconomics|
|Link to this Page:||cepr.org/active/publications/discussion_papers/dp.php?dpno=9894|
Firms expect certain investment expenditures. Firms realize certain investment expenditures. The difference is an investment surprise. With the help of the IFO Investment Survey for the German manufacturing sector we measure firms? (quantitative) investment expectations and firms? (quantitative) investment realizations on a yearly basis and construct a panel of firm-level investment innovations. This paper documents its cross-sectional and time-series properties and thus provides direct, econometrics-free quantitative discipline on the idiosyncratic shock processes used in structural heterogeneous-firm models. We find: 1) there is excess kurtosis in investment innovations, but no significant skewness; 2) the cross-sectional average of investment innovations is procyclical; 3) the cross-sectional dispersion of investment innovations is countercyclical; 4) the cross-sectional skewness and kurtosis of investment innovations is largely acyclical; 5) the cross-sectional average of the firm-individual time series volatility of investment innovations is countercyclical and highly positively correlated with the cross-sectional dispersion of investment innovations; 6) measures of firm-idiosyncratic risk have sizeable fluctuations, in the range of aggregate investment fluctuations.