عملیات بانک مرکزی اروپا: بررسی تجربی مناقصه با نرخ ثابت

In an experimental study the auction procedure of the fixed rate tender is investigated with which the main refinancing operation of the Eurosystem was carried out in the past. In June 2000 the fixed rate tender was replaced by the “American” variable rate tender. This change was occasioned by the suspicion that banks grossly exaggerate their liquidity needs in their bids. The experimental results clearly support this hypothesis and demonstrate that the fixed rate tender method can be a regular invitation to the banks to continually raise their bids from auction to auction.

At the center of Eurosystem monetary policy stand the main refinancing operations, which are offered on a weekly basis with two weeks maturity. This procedure represents the primary way in which the Eurosystem makes refinancing available to the financial sector. As a rule, main refinancing operations take the form of repo operations.1 These operations also play a key role in clarifying the Eurosystem monetary policy line by the setting of interest rates. For this reason, until recently the fixed rate tender method was employed almost exclusively. In the fixed rate tender, the European Central Bank (ECB) sets the interest rate in advance at which it is prepared to make transactions. The banks submit bids indicating how much refinancing they would like to receive at that rate. If the sum of bids exceeds the planned allotment, banks are rationed proportionally to their bids. The main refinancing operations are also supposed to aid in gaining information about the money market, which is important for the evaluation and planning of monetary policy. For this, however, it is necessary that the banks indicate their true demand for refinancing in their bids. Consequently, the procedure should be designed in such a manner that the banks are induced to reveal their true demand. During the time in which the fixed rate procedure was applied, a continual rise of the bids could be observed. Since the sum of bids had increased onto a multiple of the monetary base of the whole Eurosystem, there is not any doubt that the banks exaggerated their true demand with their bids immoderately. The game theoretical analysis of Nautz and Oechssler (1999) shows, for the case of the planned allotment being strictly smaller than the aggregate true demand, that banks have an incentive to exaggerate their true demand for refinancing and that the rules of the fixed rate tender specify a game without equilibrium even in mixed strategies. Ehrhart (2000a) shows that this can even occur if the probability for the planned allotment being higher than the aggregate true demand is positive. The lack of clear theoretical reference points makes the fixed rate tender an exceptionally good candidate for an experimental investigation. This paper reports on a series of computerized experiments in which fixed rate tender games are carried out on a repeated basis. At the center of this study stands the question of whether and under what conditions the fixed rate tender method is responsible for the observed increase in bids. It is found that the bids can continually increase in the course of the repetition — sometimes explosively. Here, the ratio of allotment to true demand is of central importance. The smaller this ratio is, the greater is the danger that the bids will skyrocket. The experimental investigation also shows that even in a game with a unique equilibrium the danger of continually increasing bids far above the equilibrium level exists. Learning direction theory helps to identify the source of the phenomenon of increasing bids as a myopic best reply character of bidding behavior, as proposed by Nautz and Oechssler (1999) in their stimulating paper. Thus, the bids are determined by strategy and do not reflect true demand. Under these conditions, the bids fail to yield useful information for monetary policy. The experimental study implies that the positive trend in bids, which was observed in reality, can be attributed to the fixed rate tender method and not to changes in conditions on the money market. The weaknesses of the fixed rate tender have apparently also been recognized by the ECB. Since June 28, 2000, the main refinancing operations are conducted as variable rate tenders, using the discriminatory “American” auction (ECB, 2000b). This paper is organized as follows. In Section 2 the development of bids of the fixed rate tenders performed by the ECB is shown. In Section 3 the fixed rate tender is analyzed from a game-theoretical point of view. In Section 4 the experiment with three different repeated fixed rate tender games is presented.

In this paper it has been shown with an experimental investigation that the fixed rate tender can encourage the banks to increasingly exaggerate their true demand for refinancing. The incentive to do so is particularly strong if monetary policy is non-accommodative. It was possible here to identify the source of this development as the myopic best reply character of bidding behavior. The results of treatment I and II support the hypothesis that a non-accommodative policy is responsible for the phenomenon of continually increasing bids which was observed in reality. Treatment III shows, however, that even accommodative policy (in combination with a unique equilibrium of the fixed rate tender stage game) cannot prevent increasing exaggeration in the bids if the bid level has already increased in the past. The so called sophisticated myopic best reply rule offers a plausible explanation for this observation. The experimental analysis can be extended by setting up situations with more bidders and bidders with differential demands. The identified behavioral rule of myopic best reply, however, leads us to suspect that the evolution of bids will principally depend on the ratio of aggregate true demand and allotment in these cases as well. The ECB held onto the fixed rate tender for a long time because it did not want to give up control over the interest rate. The weakness of this method is, however, that the rules specify a game in which the bidders may have an incentive to exaggerate their true demand. It is possible to show that the rules of the fixed rate tender can be changed such that the dominant strategy for the banks is to bid their true demand (Ehrhart, 2000a). This method is primarily of theoretical interest, though, since it discriminates against large banks in favor of smaller ones. In June 2000 the ECB decided to switch to the method of the “American” variable rate tender. This is an auction at which the allotment interest rate equals the interest rate offered in each individual bid. For the purpose of signaling the central bank's monetary policy stance, a minimum bid rate is preannounced. Bids at a rate below the minimum bid rate are discarded. However, game theoretical analysis shows that in the “American” auction bidders do not have an incentive to bid truthfully. Optimal bidding requires bid shading, meaning that bidders underrepresent their true valuation (for example, Nautz, 1995 and Heller and Lengwiler, 1998). This theoretical prediction finds initial confirmation in experiments (Ehrhart, 2000b).