The timing of investment episodes in the Netherlands

We investigate the timing of large investment episodes at the micro level. On examination of a large sample of Dutch firms we find considerable lumpiness in the accumulation of fixed capital. We therefore define investment spikes and estimate the probability of these spikes, conditional on time since the last spike. We also control for unobserved heterogeneity. Our first result is that the probability of a spike is very high the year following an investment spike. We also find some evidence the hazard rate does increase as time since the previous spike passes by. Therefore, our results indicate the presence of convex adjustment costs or time to build lags. In addition, a tentative conclusion is that fixed adjustment costs are relevant. © Springer 2006.