John Shea

Instrument Relevance in Multivariate Linear Models: A Simple Measure

The correlation between instruments and explanatory variables is a key determinant of the performance of the instrumental variables estimator. The R2 from regressing the explanatory variable on the instrument vector is a useful measure of relevance in univariate models, but can be misleading when there are multiple endogenous variables. This note proposes a computationally simple partial R2 measure of instrument relevance for multivariate models.

What Do Technology Shocks Do

The real-business-cycle literature has largely ignored the empirical question of what role technology shocks actually play in business cycles. The observed procyclicality of total factor productivity (TFP) does not prove that technology shocks are important to business cycles, since demand shocks could generate procyclical TFP due to increasing returns or other reasons. I address the role of technology by investigating the dynamic interactions of inputs, TFP and two observable indicators of technology shocks: R&D spending and patent applications. Using annual panel data on 19 U.S. manufacturing industries from 1959 to 1991, I find that favorable R&D or patent shocks tend to increase inputs, especially labor, in the short run, but to decrease inputs in the long run, while tilting the mix of inputs towards capital and nonproduction labor. Favorable technology shocks do not significantly increase measured TFP at any horizon, except for a subset of industries dominated by process innovations, suggesting that available price data do not capture productivity improvements due to product innovations. Technology shocks explain only a small fraction of input and TFP volatility at business-cycle horizons.

Do Supply Curves Slope Up

This paper examines the short-run responses of price and quantity to exogenous demand shocks for disaggregated U. S. manufacturing industries, using prior information on input-output linkages to identify industries whose fluctuations are likely to function as approximately exogenous demand shocks for other industries. I find that demand shocks induce positive covariation between price and quantity for 16 out of 26 sample industries, controlling for observable cost shift variables. When sample industries are pooled, I estimate that a demand shock which initially raises industry output by 1 percent generates a price increase of 0.182 percent within one year. I find that input-output instruments detect upward sloping supply curves more readily than least squares or other commonly used demand-shift instruments.

Complementarities and Comovements

Short-run interindustry comovement may be due either to common shocks or to complementarities that progagate shocks across sectors. This paper assesses the importance of input-output linkages, aggregate activity spillovers, and local activity spillovers to comovement in postwar U.S. manufacturing. I find that input-output linkages and local activity spillovers are important to comovement, while aggregate activity spillovers are not important. I find that complementarities are important to aggregate volatility, even after I remove observable aggregate shocks from the data. Local spillovers are particularly important, explaining between 15 and 36 percent of manufacturing employment volatility. The importance of local spillovers stems from the fact that industries that cluster together in space also comove through time, contrary to the Marshallian idea that industries with negatively correlated fluctuations should cluster together in order to stabilize local demand.

The Input-Output Approach to Instrument Selection

This article proposes a new method of selecting demand-shift instruments for disaggregated industries. I use prior information from input-output tables to identify industries whose output fluctuations are likely to function as approximately exogenous demand shocks for other industries. After motivating this idea theoretically, I implement the input-output approach using data from the 1977 detailed input-output study. I conduct a systematic instrument search for over 450 U.S. manufacturing industries and find over 200 industries possessing plausible instruments. I conclude with a brief application, showing how input-output instruments can be used to estimate the short-run supply curve of the cement industry.

Comovement in Cities

Recent research has shown that industries that locate together in space also move together over the business cycle, and that this correspondence between spatial and temporal comovement is important to aggregate volatility. This paper asks whether this correspondence is due to local common shocks or to local spillovers. I examine interindustry comovements within seven large US cities, and find strong evidence for local spillovers. I estimate that local spillovers explain roughly one-third of manufacturing employment volatility at the city level. Local spillovers do not appear to result from transport costs and locally traded goods.

Should we test the life cycle--permanent income hypothesis with food consumption data?

Abstract Most household consumption studies use food consumption data. I show that in aggregate data one cannot reject the life cycle hypothesis using food, but can reject it using other goods. Existing micro tests of the life cycle hypothesis thus may have low power.