Can cost increases increase competition? Asymmetric information and equilibrium prices.

Suppose that, as before, there is a continuum of borrowers with types [theta] distributed uniformly in [0, 1]. We explicitly assume the existence of multiple lenders, composed of lender 1 and of a fringe of competitive lenders. We assume that each borrower has a cost of switching away from lender 1 that is drawn from a uniform distribution in [s, S]. To incorporate asymmetric information we assume, also as before, that lender 1 knows the type of each borrower, whereas the fringe knows only the distribution. The switching cost for each borrower is unknown by either lender, but the distribution of switching costs is common knowledge. The timing of the model is modified slightly, in that we assume here that both lenders bid simultaneously for all borrowers. (12)

We can now state the following result, whose proof is relegated to the Appendix.

Proposition 5. Let [r.sub.F] be the equilibrium interest rate offer made by the competitive fringe. An increase in [[delta].sub.1] leads to a reduction in [r.sub.F]: [dr.sub.F]/d[[delta].sub.1] < 0.

The intuition for this result is similar to that described earlier. When [[delta].sub.1] increases, lender 1 is less able to take advantage of its private information, and increases the minimum quality threshold of borrower for which it is willing to compete. This increase in the threshold for lender 1 leads to an improvement in the distribution of borrowers for the fringe for any given interest rate. Because the fringe is competitive, the improvement in the average quality of borrowers it finances must be followed by a reduction in the interest rates they offer, as otherwise some of the fringe lenders would obtain positive profits but would be subject to undercutting by other fringe members. (13)

6. Applications

* The framework presented in this article can be usefully applied to the analysis of industries where asymmetric information is pervasive, such as banking and insurance. In what follows, we first examine, in light of our findings, two issues that have been the subject of much recent discussion: the effects of the "too big to fail" policy in bank regulation and the recent regulatory action against a number of insurance brokerage firms. We argue that this model can explain some features and predictions for these industries which are difficult to understand with standard, symmetric information models. Then, we discuss the implications of our analysis in other contexts where informational asymmetries are important, such as foreign bank entry, insider trading, or competition for repeat business.

* Too big to fail and bank regulation. A long-standing concern among bank regulators is that banks may gain an implicit government subsidy as they grow in size, as financial markets may not see as credible the threat to close down a large but failing financial institution. The most commonly discussed aspect of this "too big to fail" (TBTF) doctrine is that banks may take excessive risk as they take advantage of the public safety net. A more recently voiced concern is that TBTF, by allowing institutions to borrow uninsured funds more cheaply, may confer an unfair advantage to large banks and hinder competition in credit markets. One proposal for neutralizing the funding subsidy enjoyed by big banks is to impose a greater cost on them through, for instance, a higher capital requirement, or by levying a "systemic tax" on such institutions (see the discussion in Soussa, 2000). These arguments cannot be reconciled with standard models, where a systemic tax may benefit small banks but is unlikely to benefit consumers (i.e., depositors and borrowers), as it would neither lead to an increase in deposit rates nor to a reduction in lending rates. On the contrary, such a tax would be viewed as potentially lowering consumer welfare to the extent that large banks passed some of the increase in costs on to their clients. By contrast, because an increase in capital requirements or the imposition of a tax for large banks is equivalent to an increase in the marginal cost for these banks, our model predicts that such a policy could increase competition and actually reduce lending rates if these banks possess private information about a large share of the borrowers in the market.

* Insurance brokerage and litigation. Insurance brokers have lately come under attack for the practice of rigging bids and accepting contingent commissions to steer business toward preferred clients. As a consequence, a number of large brokers have settled charges by some state attorneys general of illegal activities tied to these contingent commissions, but many brokers still remain the subjects of subpoenas and inquiries from state attorneys general and regulators. Smaller brokers in general have not been affected as much by these scandals, and so have avoided either suffering the reputation damage or paying the monetary settlements faced by large brokers. One possible consequence of these scandals is discussed in a recent study by Standard and Poor's (2005), which argues that "smaller brokers will seek to take market share from the largest brokers, increasing competition." A similar concern is evident from brokers' quarterly reports for 2004-2005, which catalog an increasingly competitive landscape and lower rates even as they deal with increased costs stemming from the litigation (Marsh and McLennan quarterly reports, 2004 and 2005). (14) As discussed above, it is difficult to interpret these observations in the context of standard models of oligopolistic competition in which a shock of this kind would be predicted to reduce the ability of some firms to serve the market, possibly resulting in a smaller aggregate supply and higher prices. These predictions and stylized facts are, instead, consistent with the predictions of our model, where a cost shock to the dominant brokers may indeed lead to increased competition and lower prices. (15)

* Domestic subsidies and foreign entry. Consider the case of a liberalization that opens up a domestic financial system to the competition of more cost-efficient foreign financial institutions. Suppose that concerns over the ability of domestic institutions to remain competitive lead the government to consider providing a subsidy to domestic banks. This subsidy could take many forms, such as the provision to banks of a line of credit at lower-than-market rates, increasing the extent of deposit insurance and thus lowering the deposit rate paid by domestic banks, or directly subsidizing the domestic banks through, for instance, a tax rebate. A relevant question for the design of such policy is whether this subsidy is likely to be passed on to bank customers, and the extent to which borrowers may benefit from the subsidy.

The analysis of this article suggests that such a policy may have the unintended consequence of decreasing competition and increasing the interest rates paid by all borrowers. The subsidy in this case acts exactly like a decrease in the marginal cost of extending a loan (a reduction in [[delta].sub.1]) for the domestic banks, which presumably also have all the existing information about the domestic credit market. As discussed above, however, such a decrease in the informed lender's marginal cost may lead to an increase in interest rates because it allows the informed lender to better exploit its private information and deters competition from new entrants.

* Repeat purchases. Although we have presented the analysis in terms of credit markets, the results in the model generalize to settings where firm-client relationships allow firms to extract future rents from their customers, and where the extent of such rents differs from client to client and is known only to the firm. As an example, suppose that firms compete in a market characterized by repeat purchases and switching costs, but where each customer has a different probability of future purchases. Furthermore, assume that these customers' characteristics are privately known to the customers' previous supplier, but are not observed by any other competitor. In that setting, as in our model, the informational advantage of the incumbent allows it to price discriminate across clients, offering better terms to customers with a higher probability of purchasing again in the future. Firms trying to compete away customers from the incumbent face an adverse selection problem to the extent that they are able to attract only customers with a lower than average probability of repeat purchasing. For any price offered by competitors, a marginal cost increase will force the incumbent to let go of clients with a relatively higher probability of future purchases. This in turn reduces adverse selection and hence leads to lower equilibrium prices.

* Market manipulation. Our results are also in line with those of the literature on stock market manipulation by an informed trader. Much of this literature argues that, although some traders may possess private information about the value of some securities, they may also need to trade for liquidity reasons, thus allowing trade to take place (see, e.g., Allen and Gorton, 1992). Our analysis suggests that shocks to inventory costs can also constitute a source of liquidity, in that increases in carrying costs for the informed trader diminish its ability at the margin to profit from its private information, and allow uninformed traders to purchase securities with higher valuations.

7. Discussion