​The significant effect of high leverage is present also in subsamples defined by industry, liquidity ratio, and geography. The effect is most clear in industries in which firms decreased their aggregate investment most during the downturn, and it cannot solely be attributed to a 'regression to the mean'-effect. The persistent effect across subsamples points to the existence of a separate leverage or 'balance sheet' channel in addition to the effects of other variables conventionally included in investment relations. Consequently, the degree of leverage among non-financial firms may have implications for macroeconomic volatility. Furthermore, results indicate that the development in investment during the crisis was not primarily a result of more difficult access to finance for highly leveraged firms.