The pre-exit performance of German plants : how long is the "shadow of death"?

This paper deals with the question whether firms' employment performance in the periods preceding their exit can be regarded as casting a 'Shadow of Death' on their final leaving the market. This aspect is of high relevance for politicians and other decision-makers, because by knowing more about possible indicators of market exit and possible problems of firms, they might have instruments at hand to detain unwanted firm deaths. Although it can be seen as a stylized fact that exiting firms exhibit a worse economic performance in the time before exit, a systematic and detailed analysis of the pre-exit perfomance seems warranted. With specific regard to Germany, it is not clear how long the Shadow of Death is and which influence can be attributed to other factors specific to the firm, its employees or its environment. We study the pre-exit performance of German plants in terms of employment change and to this use a comprehensive data set that covers the whole of Germany and contains monthly plant data for the years from 1999 to 2012. Carrying out detailed sector-, region- and time-specific analyses, in a first step we are able to complement the existing literature on the Shadow of Death in detailed descriptive respects. Furthermore, it is not very clear how long the Shadow of Death is and which influence can be attributed to factors specific to the firm, its employees or its environment. Therefore, in a second step we use statistical matching approaches to identify the causal relations. We use a two-step matching procedure using covariate and nearest neighbour matching. In a final step, the conditional difference-in-difference estimator is applied. First descriptive results show that most of the exits take place at the end of a quarter in a year, especially at the end of the fourth quarter. Furthermore, the surviving plants start and operate with more employees. A Shadow of Death exists nearly as long as the plant exists, i.e. here eight or nine years long. On average and for larger plants, employment decline starts between 54 and 63 months before closure.