Nonfundamental representations of the relation between technology shocks and hours worked

Estimating the response of hours worked to technology shocks is often considered as a crucial step for evaluating the applicability of macroeconomic models to reality. In particular, Galí [1999] has considered the conditional correlation between employment and productivity as a key tool for building an empirical evaluation of Real Business Cycle theories and New-Keynesian models. Impulse-response functions are often identified by means of Structural Vector AutoRegressive models. However, a structural Moving Average model of the economy cannot be estimated by VAR techniques whenever the agents' information space is larger than the econometrician's one, that is when we face a problem of nonfundamentalness. We consider how factor models can be seen as an alternative to VAR for assessing the validity of an economic model without having to deal with the problem of nonfundamentalness. We apply this method to the well known business cycle model by Galí [1999], which originally was estimated using a VAR, and retrieve alternative nonfundamental representations of the relation between technology shocks and hours worked. Such representations always yield a positive correlation between productivity and hours worked when conditioning on a technology shock. This result is more robust than the results by Christiano et al. [2004], because it is independent of the transformation used for hours worked and moreover is perfectly consistent with the unconditional correlation observed between the common components of the variables considered.