Chapter 17, by Ghosal and Miller, suggests that an attempt to solve liquidity crises by rollovers will promote moral hazard and, conversely, that the better one can deal with such moral hazard problems the more one ought to solve all liquidity crises by rollovers. The basic argument is set out in Sections 18.1 to 18.3 of these comments.
This volume discusses both liquidity crises and solvency crises. Chapter 17 focuses almost entirely on liquidity crises. It does not consider the effect of solvency difficulties. I will discuss solvency crises in Section 18.4. Section 18.5 considers implications of this analysis.
The setup is one which produces crises during the life of an investment project. These can be caused either by a negative external shock or by a failure of the borrower to make sufficient policy effort. When the negative effects of either of these strike, a liquidity problem arises, but the creditors cannot tell which. The question is whether (a) the lenders will continue to lend in such circumstances, and (b) whether the borrower has incentives to make policy effort in these circumstances. There are two periods (the first between t = 0 and t = 1, and the second between t = 1 and t = 2) and there are only two creditors. The project is long-run solvent in that the shocks only cause temporary liquidity crises.
There are two things which can go wrong at the beginning of the second period. First, there can have been "low effort" by the borrower at the beginning of period 1. Effort can be thought of as "financial regulation" or "capital controls". Without effort there will be an outflow of money abroad. The sovereign has the possibility of making effort which would prevent this. Second, the productivity of the project can temporarily fall, due to a negative shock at the end of period 1.
If either of these things happens, the project has a liquidity problem at the beginning of period 2, in that the required payment of interest at this time is more than the project has yielded by then. 1 The creditors face a co-ordination problem. They may or may not run. If they do not run, then the project can be completed. If they do run, then there is a "crisis" and the