|
...counterfactual in the English auction can be bounded in an informative way. These results form a basis for nonparametric estimation of counterfactual revenue differences. I find that the revenue gain from using the English auction would be in the range of 11%-19% of the gross underwriting spread, and that most of it would already be captured by using the second-price auction. The recent explosive growth of Internet English auctions, administered by Grant Street Group, provides external support to the claim that auction design matters in this market.
1. Introduction
* This article considers the estimation of counterfactual revenues in municipal bond auctions. The traditional design adopted by this industry is the first-price, sealed-bid auction, in which bids are delivered to the issuer of the bond by hand, phone, or fax, without the involvement of any intermediaries. Auction theory, however, tells us that under certain conditions, ascending auction formats may be revenue superior to the first-price auction. Establishing useful notation, let [R.sub.F] be the expected revenue from the first-price auction, [R.sub.S] be the expected revenue from the second-price auction, and [R.sub.E] be that from the ascending (English) auction. (1) The result of Milgrom and Weber (1982) states that the revenues from these three formats are ranked: [R.sub.F] [less than or equal to] [R.sub.S] [less than or equal to] [R.sub.E].
Despite the fact that this revenue ranking result has received a great deal of attention in the theoretical literature, I am not aware of any previous empirical studies of the magnitudes of these effects. (2) Municipal bond auctions are a natural setting to explore this question. First, this is a large and important market, and understanding how alternative auction forms can improve revenues for the issuers is of obvious practical interest. Second, unlike many other financial markets, the secondary market for municipal bonds lacks liquidity (Downing and Zhang, 2004; Harris and Piwowar, 2004; Green, Holliheld, and Schurhoff, 2004). This means that private information about resale prices may be important. A likely channel through which this information is conveyed to the bidders is the presale of bonds before the auction. (3)
I estimate a model of bidding that allows for common as well as private components in bidders' valuations. Several other articles in the literature have proposed methods to estimate similar models--these include Paarsch (1992), Hong and Shum (2002), Hendricks, Pinkse, and Porter (2003) and Bajari and Hortacsu (2003). Most of these articles use a parametric approach, with the exception of Hendricks, Pinkse, and Porter (2003). (4) Parametric estimation can suffer from misspecification, so a nonparametric approach would have obvious advantages. Unfortunately, full nonparametric identification fails in a common-values environment (Laffont and Vuong, 1996). (5)
The identification result in this article shows that counterfactual revenue in the second-price auction is identified and can be estimated nonparametrically, regardless of the general failure of identification. The argument can be summarized as follows. We know from Guerre, Perrigne, and Vuong (2000, hereafter GPV) and Li, Perrigne, and Vuong (2002, hereafter LPV) (see also Perrigne and Vuong, 1999) that in a private-values environment, the econometrician can recover consistent estimates of bidders' valuations from the first-order conditions of the Bayesian-Nash equilibrium of the bidding game. These estimates have been named "pseudo-values" by GPV and LPV, and this notion was extended to common-value environments by Hendricks, Pinkse, and Porter (2003). (6) In a private-values environment, pseudo-values can be interpreted as the estimates of counterfactual bids that would be submitted in the second-price auction. I show that the same interpretation continues to apply even in the environment with common values, and this allows me to estimate the counterfactual revenue [R.sub.S] nonparametrically.
The revenue in the English auction, however, cannot be nonparametrically identified. (7) To make inferences about it, I use a bounding approach. I show that the expected value of the highest bid in the counterfactual second-price auction, which is identified, is an upper bound for [R.sub.E]. (8) Consistent estimation of [R.sub.S] and this bound on [R.sub.E] presents no difficulties, since it can be performed by following LPV.
Turning to the empirical implementation, my dataset consists of 386 municipal bond sealed-bid auctions held in the state of California in the period from May 1998 to July 2001. It includes the bids, bidder identities, and various bond-specific covariates. No data on bidders' profits from selling the bonds are available, so the estimation must rely on the information contained in the bids. Estimation results are presented and discussed in Section 6, where several specifications are considered. Performing counterfactual comparisons, I find that [R.sub.S] - [R.sub.F] are in the range of .10%-.16% of the par value of the bond and 9%-13% of the gross underwriting spread (the difference between the initial resale price of the bond and the winning bid). I find that [[bar.R].sub.E] - [R.sub.F] are in the range of .14%-.23% of the par value of the bond and 11%-19% of the gross spread. It follows that the second-price auction would capture a significant fraction of the potential gains, 66%-81%, depending on the specification.
2. Institutional details of municipal bond auctions
* The U.S. municipal bond market is one of the world's largest security markets. Municipal bonds are issued by more than 50,000 state and local governments and their agencies to build, repair, or improve schools, streets, and highways, as well as for many other kinds of public works. U.S. households own municipal bonds either through direct ownership of individual bonds or through investment in institutional portfolios, including mutual funds, unit investment trusts, and bank trust accounts. Commercial banks and insurance companies are the major institutional holders.
Municipal bonds are typically issued either by a negotiated sale or by a sealed-bid, first-price auction, also known as a competitive sale in the industry. Although the focus of my article is on auctions, it is useful to briefly discuss and compare these two most common selling mechanisms. (9) In a negotiated sale, the investment bank serves as both the originator and the distributor of the issue. (10) In a competitive sale, the issuer hires a financial advisor for origination services, and solicits bids by posting a notice of sale in the major industry publication, The Bond Buyer. Issuers typically put series of bonds up for auction simultaneously, and the winning bidder receives the entire series (bids for partial quantities are not allowed). The reserve prices are rarely binding, and auction cancellations are also very rare. The bidders are mainly investment and commercial banks; the prime motive for acquiring the bonds is resale to final investors. For a few days, the bonds are priced at a uniform price--at par or close to par--by all the dealers in the winning syndicate. They remain for sale at that price until the bonds "break syndicate" and are allowed to trade at what the market will bear.
The advantages of a competitive sale should be clear to an economist: we often expect that if a seller is able to attract more buyers, the price will be higher. (11) But this intuition is firmly rooted in the private-values assumption and may not hold if common values are present. (12) If common values are important, the bids in a competitive sale may include a discount to allow for the winner's curse, while there is no similar effect in negotiated sales. Auction theory predicts that the strength of the winner's curse effect may be related to market uncertainty. (13) Ederington (1976) provides some supporting evidence for this hypothesis. Without making an explicit reference to a strategic model of bidding, he finds that the underwriting spread is more sensitive to market uncertainty in competitive sales than in negotiated sales. Dyl and Joehnk (1976) find that the distribution of bond ratings is skewed toward riskier bonds in negotiated sales, a behavior by the issuers that is consistent with the winner's curse effect at auction. (14) These studies suggest that both banks and issuers appear to respond to their strategic incentives in a manner consistent with predictions of auction theory. (15)
Why is private information important in this market, and how is it obtained? Approximately $1.7 trillion worth of municipal bonds are currently in the hands of investors. (16) While some of these bonds are actively traded, others may not trade for months. The liquidity of the secondary market in municipal bonds is manifestly low: the bonds are often held by investors until maturity. The Municipal Securities Rulemaking Board (MSRB) has recently begun to disclose transaction data in the secondary market. Following this event, several very recent studies have explored liquidity and pricing using the MSRB data. Downing and Zhang (2004) find that only about a third of the bonds traded more than once over the entire period, and that most bonds traded only two or three times. Harris and Piwowar (2004) find that municipal bond trades are significantly more expensive than equivalently sized equity trades.
This low liquidity of the secondary market can make private information about resale values significant. Robinson (1960) describes the bidding process for municipal bonds in detail and concludes that prior to sale, bidders frequently presell the bonds to their clients. Since the presale price is likely to be correlated with the market value of the bond, it may be an important source of private information. If the entire issue of bonds is presold, then the bidders would know their valuations perfectly (because they would know their resale prices), the environment would be with private values, and the theory predicts that the second-price auction would yield more revenue than the first-price auction. (17) The theory also predicts no additional gain from using the English auction. On the other hand, to the extent that presale is only partial, the bidders are likely to be uncertain about the price they will receive in the secondary market, so that their valuations have both a known private and an unknown common component. In this case the theory predicts that the English auction would yield even more than the second-price...
NOTE: All illustrations and photos
have been removed from this article.
Looking for additional articles?
Search our database of over 3 million articles.
Looking for more in-depth information on this industry?
Search our complete database of Industry & Market reports by text, subject, publication
name or publication date.
About Goliath
Whether you're looking for sales prospects, competitive information, company
analysis or best practices in managing your organization,
Goliath can help you meet your business needs.
Our extensive business information databases empower business
professionals with both the breadth and depth of credible,
authoritative information they need to support their business
goals. Whether it be strategic planning, sales prospecting,
company research or defining management best practices -
Goliath is your leading source for accurate information.
|
