کاهش ارزش و مناقصه تعادلی در عملیات بهره بانک مرکزی

Among the most puzzling observations on the euro money market are the discount in the weekly refinancing operations, the more aggressive bidding under uncertainty, the temporary flatness of bid schedules, and the development of interest rate spreads. To explain these observations, we consider a standard divisible-good auction with either uniform or discriminatory pricing, and place it in the context of a secondary market for interbank credit. The analysis links the empirical evidence to the endogenous choice of collateral in credit transactions. We also discuss the Eurosystem's preference for the discriminatory auction, the remuneration of reserves, and the impact of the recent market turmoil.

The euro money market, defined here as the market for euro-denominated short-term credit between counterparties of the Eurosystem, has been challenging economists by exhibiting a variety of puzzling features right from the market's inception in January 1999.1 One of these features has been that credit seems to be obtainable at more attractive conditions in the primary market, i.e., in central bank operations, than in the secondary market, i.e., in the interbank market. This is counterintuitive because the regular central bank operations in the euro area, the so-called main refinancing operations, are effectively highest-price auctions involving several hundred bidders. How is a discount possible under such tight competition?2 The first account and explanation of the discount has been given in a seminal contribution by Ayuso and Repullo (2003) who assume that the central bank has an asymmetric objective function and penalizes downwards deviations of the market rate more heavily than upwards deviations. As a consequence, the central bank follows a tight allotment policy, driving the market rate above the policy rate, which explains excessive bidding for central bank reserves. However, the explanation depends on the central bank's use of the so-called fixed-rate tender, through which funds are offered to market participants below market conditions. The model, therefore, cannot account for the spread between primary and secondary market conditions under the variable-tender regime which has been in place following June 2000. Another potential explanation of the spread between primary and secondary market conditions might be intermediation within the banking sector. 3 Indeed, Freixas and Holthausen (2004) have pointed out the role of money centers to distribute unexpected liquidity shocks within the euro area. Intermediation is useful in this context because it reduces the informational frictions of unsecured lending, especially in a cross-border context. However, in contrast to the case of unsecured interbank lending considered by Freixas and Holthausen, all loans vis-à-vis the Eurosystem have to be fully collateralized. This suggests that informational frictions are not a central feature of lending allotted through Eurosystem tenders. 4 To the contrary, such intermediation is likely to cause unnecessary costs, for instance in terms of regulatory capital usage, 5 counterparty risk, or collateral handling. In sum, this suggests that we should find actually only limited intermediation, at least outside of banking groups, of refinancing that had been received in the Eurosystem tenders. 6 We conclude that intermediation is not likely to fully explain the spread in money market conditions. 7 To better understand this and other pieces of evidence, the present paper offers a theoretical framework that integrates two institutional features of the Eurosystem's variable-rate tenders. The first element, adopted from Klemperer and Meyer's (1989) analysis of oligopolistic competition, is an aggregate uncertainty (potentially small, but not negligible) about the quantity that is eventually allotted in the auction. 8 The second element is an endogenous choice of collateral pledged to secure the individual funding transaction. The framework is then used to study equilibrium bidding of commercial banks in Eurosystem auctions in the context of a competitive secondary market, where we allow both the uniform and the discriminatory pricing rule. 9 The characterization of the equilibrium is shown to have a number of testable implications, which are compared to the empirical evidence. First, we look at the discount between primary and secondary market conditions, as discussed above. The model predicts here that conditions offered through the discriminatory auction are typically strictly below marginal valuations even if there are many bidders. This suggests an explanation for the obscure underpricing. Second, empirical research tells us that with more uncertainty in the market, bids are on average placed at higher interest rate levels. Intuitively, these findings reflect the bidder's concern of ending with insufficient liquidity, i.e., the so-called loser's nightmare (cf. Simon, 1994). Here again, the predictions of our theoretical framework are largely supportive of the evidence. We identify, however, also a new effect that might shed additional light on incentives for bid shading in discriminatory auctions. Third, during relatively calm periods, bidders in ECB repo auctions have tended to submit excessively flat bid schedules at the level of the expected stop-out rate. We show here that the slope of bid schedules is “nearly” vanishing when bidders face little uncertainty and the liquidity of collateral assets is high. Again, this matches the evidence. Finally, we discuss another observation, which is the unexpected increase in the so-called Eonia spread with the introduction of an adapted implementation framework in early 2004. Also here, the predictions are in line with the evidence. Specifically, we show that a ceteris paribus increase in the size of the auction will lead to a wider spread between conditions in the primary and secondary money markets. We go on to study the central bank's decision on the pricing rule. Empirically, the ECB seems to have a clear preference for using discriminatory pricing in its main refinancing operations. Indeed, the uniform-pricing rule has been employed for the main operations only in early 1999. To explore this issue, we determine the revenue impact of the pricing rule, and find that in the identified equilibria, expected revenue is strictly higher in the discriminatory auction than in uniform-price auction. This is a somewhat unexpected result because the stronger bid shading in the discriminatory auction had sometimes been understood to actually reduce the auctioneer's expected profits. We also show that the difference may be even more pronounced when required reserves are, so the established terminology, remunerated at the marginal rate, which is the case in the euro area. Finally, we mention another advantage of the discriminatory format, which is related to the signaling role of tender rates. A remarkable piece of evidence was revealed when market turmoil triggered by the U.S. subprime crisis hit the euro money market. Specifically, it was found that following August 2007, counterparties of the Eurosystem would be willing to pay a premium above benchmark rates for participating in the auction. We apply our formal framework to comment also on these developments. The rest of the paper is structured as follows. Section 2 reviews the empirical evidence. Section 3 outlines the auction model. In 4 and 5, we characterize bidding equilibria of the uniform-price and discriminatory auctions, respectively, covering the cases of few and many bidders. Section 6 relates our predictions for many bidders to empirical observations for the euro area, while Section 7 discusses the Eurosystem's potential motivation for using the discriminatory format. Section 8 reviews some related theoretical literature. In Section 9, we derive predictions for bidding under market distress. Section 10 concludes. Appendix A, Appendix B, Appendix C and Appendix D provide, respectively, background information on the euro money market, anecdotal evidence on bid schedules in the Eurosystem auctions, a mathematical description of the allotment rule, and formal proofs of the propositions.

Refinancing operations are large-scale financial auctions that are used by central banks such as the ECB to provide an appropriate amount of reserves to the banking sector. The present paper has offered a model of refinancing operations in the context of a reduced-form market for interbank credit. In the model, the central bank invites a finite number of counterparties to submit schedules of bids, and only the highest bids will be successful. Counterparties that end up with insufficient funding from the auction must turn to the secondary market where credit can be obtained only against the most expensive sorts of collateral. We have used the model to investigate the weekly main refinancing operations conducted by the Eurosystem following June 2000. Four stylized facts have been analyzed. First, concerning the spread between primary and secondary market conditions, we found that bid shading is persistent in large discriminatory auctions, provided that aggregate uncertainty about the allotment does not vanish in the limit. This suggests a new explanation of the interest rate spread. We also analyzed in some depth the intricate issue of how uncertainty influences bidding behavior in repo auctions. Here our model predicts that both the absolute level of bids and their dispersion increase in the volatility of the expected market rate. Also this finding is largely consistent with the evidence, i.e., with the hypothesis of the loser's nightmare. A third regularity that turned out to be reflected in the equilibrium prediction is the flatness of bid schedules during calm periods. Finally, we looked at the “mysterious” increase of the Eonia spread following March 2004, and can offer also here a theoretical prediction that is consistent with the evidence. We then discussed the question why the ECB has been relying mostly on the discriminatory format in its main refinancing operations. We could show that with sufficiently many bidders, expected revenues are strictly higher under the discriminatory format than under the uniform-price format, despite the more pronounced bid shading. This revenue dominance may even become stronger when the central bank, as in the case of the Eurosystem, has chosen to reimburse interest paid on required holdings of reserves. However, we have also shown that tender rates in the discriminatory format are on average closer to the main policy rate (and less noisy), which might even be more relevant from a policy perspective. Finally, motivated by the credit crunch following August 2007, we derived the implications of a deterioration of collateral quality on equilibrium bid schedules and the bidder's marginal willingness to pay for credit after the auction. Here we found the prediction that illiquidity of collateral not only increases the average level of both tender rates and marginal valuations, but also bid dispersion and the volatility of marginal valuations. These general conclusions suggest that the model can also be used to study bidding behavior during a liquidity crisis.