Borrow to invest, invest to borrow

Most empirical studies of investment in liquidity constrained firms assume the firms' constrained status to be exogenous. However, the same studies report evidence of a significant link between the financial status of the firms (constrained vs. unconstrained) and their size (small vs. large). Motivated by such evidence, this dissertation analyzes a model of investment in which the financial status as well as the firms' size are endogenously determined.;The financial status is measured in terms of the collateralizable net worth, that is, the difference between collateral and borrowing. Collateral is specified as a linear increasing function of the stock of physical capital. For given debt, then, the financial status of a firm is a function of its capital stock. In turn, the financial status contributes to determine the firm's investment and borrowing capacity, as the marginal interest rate on debt is increasing in borrowing for given collateral.;Solution to the model delivers a restricted VAR in capital and net worth. The model restrictions identify shocks to net worth separately from technological shocks. The elasticity of investment to net worth could be interpreted as a measure of the magnitude of the credit channel of monetary policy. Estimation on a sample of publicly traded industrial US. companies yields a better fit than an unrestricted VAR(1). The optimal aggregate level of net worth is estimated to be insignificantly different from zero at the steady state, implying that the elasticity of investment to net worth is also insignificant. The aggregate elasticity becomes significant but remains negligible even if the firms are borrowing in excess of their collateral, which is the case in the sample. Moreover, the expected aggregate rate of investment is found to be independent of the firms size. One way to interpret these results is that while fully allowing for firms' heterogeneity and for a role of financing constraints on investment, the model yields aggregate predictions that are consistent with standard dynamic representative-agent models and perfect capital markets.