44 Journal of Economic Perspectives is a better channeling of resources to their ultimate uses. Namely the persons from group B-who have relatively high values for rates of time preference and for marginal returns to investment-command a greater share of current output. In any event the outcomes are non-neutral, and in that sense non-Ricardian The important finding from the inclusion of imperfect loan markets is that the government's issue of public debt can amount to a useful form of financial intermedia tion. The government induces people with good access to credit markets (group a)to hold more than their share of the extra public debt. Those with poor access(group B) hold less than their share, and thereby effectively receive loans from the first group This process works because the government implicitly guarantees the repayment of loans through its tax collections and debt payments. Thus loans between a and b take place even though such loans were not viable(because of"transaction costs)on the imperfect private credit market This much of the argument may be valid although it credits the government with a lot of skill in the collection of taxes from people with poor collateral(which is the underlying source of the problem for private lenders). Even if the government possesses this skill, the conclusions do not resemble those from the standard analy As discussed before, budget deficits can amount to more financial intermediation, and are in that sense equivalent to a technological advance that improves the functioning of loan markets. From this perspective it is reasonable to find a reduced spread between various discount rates and an improvement in the allocation of resources. If the government really is better at the process of intermediating, more of this activit that is, more public debt --raises perceived wealth because it actually improves the orkings of the economy In the preceding analysis, the imperfection of credit markets reflected costs of enforcing the collection of loans. a different approach, followed by Toshiki Yotsuzuka 987) in his extension of the models of Mervyn King H (1987), allows for adverse selection among borrowers with different risk characteristics Individuals know their probabilities of default, but the lenders'only possibility for learning these probabilities comes from observing the chosen levels of borrowing at going interest rates. In this setting the government's borrowing amounts to a loan to a group that pools the various risk classes. Such borrowing matters if the private equilibrium does not involve similar pooling. However, by considering the incentives of lenders to exchange or not exchange information about their customers, Yotsuzuka argues that the private equilibrium typically involves a pooled loan of limited quantity at a relatively low interest rate. Then the high-risk types may borrow additional amounts at a high interest rate. (The assumption is that this additional borrowing is not observable by other lenders. )In this case the government's borrowing replaces the private pooled lending, and leads to no real effects. That is, Ricardian equivalence holds despite the imperfect private loan market where high-risk people face high marginal borrowing rates. The general lesson again is that Ricardian quivalence fails because of imperfect credit markets only if the government does things in the loan market that are different from, and perhaps better than, those carried out privatel