Most investment in developing countries is financed externally, by borrowing from banks. Banks retain this central role because they are the best at monitoring projects and enforcing contracts when public information is limited and the legal and financial infrastructure is immature. They often do this by establishing long-term relationships with firms.
Banks in developing countries, however, tend to lend either too little or too much to firms, with two important effects. First, chronically low lending is associated with low levels of investment and growth, and, moreover, constrained lending restricts growth. Second, periodic surges in lending end in systemic bank and economic crises. For convenience, we refer to the first pattern as 'under-borrowing' by firms, and to the second as 'over-lending' by banks.
The role of financial agency costs in firms and banks helps to understand these related, and seemingly conflicting, lending patterns. Financial agency costs refer to the impact on lending behavior of the stakes that firms and banks have, respectively, in the outcomes of investment projects. These stakes grow out of features of the balance sheets of firms and banks. A simple version of how this works is as follows.
On the one hand, when firms have low net worth, they cannot put up collateral for investment projects. Then banks do not lend to them, since the firms have little to lose from project failure, and are more likely to choose or accept poor projects. This yields under-borrowing. On the other hand, when banks have low net worth, or there is deposit insurance, they also have little to lose from failure, and are more willing to lend to poor projects. This yields over-lending. If deposit insurance is implicit and systemic, then banks may require collateral for individual projects, while ignoring aggregate credit risk associated with inflated asset and collateral values. This simultaneously yields under-borrowing in some sectors where collateral is not available, and over-lending in other sectors where collateral is available.
To assess the explanatory power of this approach, we undertake a selective review and interpretation of evidence on patterns of bank lending, and on how financial agency costs contribute to them. The evidence on lending patterns is