current profits and projections of future profits. This may be a way to both target better and guard against potential NPAs, largely because poor profitability seems to be a good predictor of future default. It is clear however that choosing the right way to include profits in the lending decision will not be easy. On one side there is the danger that unprofitable companies default. On the other side, there is the danger of pushing a company into default by cutting its access to credit exactly when it needs it the most, i.e. right after a shock to demand or costs has pushed it into the red. Perhaps one way to balance these objectives would be to create three categories of firms: (1) Profitable to highly profitable firms. Within this category lending should respond to profitability, with more profitable firms getting a higher limit, even if they look similar on the other measures. (2) Marginally profitable to loss-making firms that used to be highly profitable in the recent past but have been hit by a temporary shock (e.g. an increase in the price of cotton because of crop failures, etc.). For these firms the existing rules for lending might work well. (3) Marginally profitable to loss-making firms that have been that way for a long time or have just been hit by a permanent shock (e.g., the removal of tariffs protecting firms producing in an industry in which the Chinese have a huge cost advantage). For these firms, there should be an attempt to discontinue lending, based on some clearly worked out exit strategy (it is important that the borrowers be offered enough of the pie that they feel that they will be better off by exiting without defaulting on the loans).
Of course it is not always going to be easy to distinguish permanent shocks from the tem- porary. In particular, what should we make of the firm that claims that it has put in place strategies that help it survive the shock of Chinese competition, but that they will only work in a couple of years? The best rule may be to use the information in profits and costs over several years, and the experience of the industry as a whole.
One constraint on moving to a rule of this type is that it puts more weight on the judgment of the loan officer. The loan officer would now have to also judge whether the profitability of a company (or the lack of it) is permanent or temporary. This increased discretion will obviously increase both the scope for corruption and the risk of being falsely accused of corruption. As we saw above, the data is consistent with the view that the loan officers worry about the possibility of being falsely accused of corruption and that this pushes them in the direction of avoiding taking any decisions if they can help it. It is clear that it would be difficult to achieve better