Treasury Auction Manipulation Investigations and Litigation – Some Comments

There have been allegations of manipulation by major dealers of the auctions for U.S. Treasury securities. Apparently, at least twenty-five lawsuits have been filed, and the U.S. Justice Department and the New York Department of Financial Services are investigating. The back story to these lawsuits and investigations is not public, and the Treasury does not seem to be commenting.

In brief the allegation is that major dealers collude in keeping the yield up (or price down) in Treasury auctions. The complaints compare the auction results to trading in the when-issued market for the same security and contend that statistical analysis shows that there must be collusion.

It is certainly possible, but, after having read two of the complaints (State-Boston Retirement System v Bank of Nova Scotia et al, U.S. District Court, Southern District of New York, No. 15-05794 and Cleveland Bakers and Teamsters Pension Fund et al v. Bank of Nova Scotia, New York Agency et al), I do not think the issue is that clear. The statistical evidence in these complaints is badly presented, and the authors, while assuming the pose of experts on the government securities market, seem to have studied up on this market fairly recently. For example, their knowledge of bond math is limited.

One problem with their argument, as “Yves Smith” on her naked capitalism blog points out, bidders in Treasury auctions may effectively demand a concession in price to act effectively as underwriters for a sizeable chunk of securities.  She thinks, though, that the plaintiffs will avoid a summary judgment against them and will proceed to discovery. If there is evidence of collusion, such as emails or chat room discussions, then there will be a case to be made.

While this is interesting, what the commentary I have read misses is that prior to changes that happened in the government securities market, by government actions, market developments, and technology, the information advantage that primary dealers enjoyed was much more significant. For example, in the 1970s and 80s, primary dealers were the only ones allowed to trade at the major interdealer brokers, with the exception of Cantor Fitzgerald, which operated a government securities trading facility. Consequently, the primary dealers had access to an inside market which was not transparent to anyone outside the club. One firm, which was not a primary dealer, Lazard Frères, complained loudly about this state of affairs, but got no support from Treasury and made no headway in its complaint.

When it came to the auctions, at the time Treasury auctions were multiple price, that is, Treasury accepted bids at the highest prices (or lowest yields) until the amount offered was sold. Competitive bidders had to pay the price that they bid, even if that was higher than the average price. This type of auction poses the problem of the “winner’s curse” for participants; in other words, they run the risk of paying too high a price for the securities. Large investors, therefore, bid through the primary dealers, because they knew that the primary dealers were better able to know the real prices in the market. The primary dealers provided this service for free, because the amounts their customers were bidding for provided them useful information about the underlying demand for the securities. This system worked for the Treasury, but it undeniably gave the primary dealers an advantage and could be criticized as unfair.

Since then, the interdealer market has changed, and prices are much more freely available (though some firms, such as Bloomberg, charge a considerable amount for the use of their terminals). Also, the primary dealer advantage in the auction has been eroded, since Treasury now auctions its securities in single-price auctions. The best bids are still the ones accepted, but all successful bidders pay the lowest price accepted, thus doing away with the winner’s curse problem. (The argument from a cost perspective for Treasury is that the amount of money that Treasury “leaves on the table” is made up for or more than made up for by the higher prices bidders offer in this type of auction. In other word, bidders are not tempted to shade their bids but are more willing to bid based on their true demand curve in this type of auction, since they know that they will not overpay for the securities.)

In single-price auctions, there is less reason for investors to go through a primary dealer, and the amount of direct bids by investors has increased. This seems to be a fairer system for Treasury auctions. However, one can easily see that single-price auctions do leave room for collusion, and perhaps game theorists should study whether the opportunities and temptations to collude are greater in single-price or multiple-price auctions.

In any case, it will be interesting to see if the government investigations or the private class action lawsuits develop any hard evidence beyond the statistical analysis. One interesting point is that the statistical evidence for collusion apparently end after the Libor manipulation investigations become public.