To sharpen this analysis further, we repeat our estimations using a matchedsample. We do this in two ways. We select only control firms that match thetreated firms by their size and industry. We also implement a propensity scorematching and select control firms that are a good match according to a largerset of observable firm characteristics (profitability, tangibility, size, growthopportunities, and industry). We find that the negative effect of an increase inemployment protection on leverage is robust to these changes.