Production function estimation controlling for endogenous productivity disruptions

We propose a modification to the standard production function estimationframework that incorporates endogenous disruptions in the productivity processdue to lumpy firm investment. We investigate the differences between our novel(disruption) framework and the existing (baseline) approach on a largeproprietary panel of Greek Manufacturing firms. The two models lead tosignificantly different production function estimates, which in turn affect theresults of subsequent inference. They imply different average levels of productivityand different magnitude of endogenous disruptions. The baseline model cannotcapture the full dynamics of disruption costs, which extend across time.Decomposing Aggregate Productivity Growth into its components givessubstantially different results under the disruption model