Speech by SEC Staff:
This was prepared for the keynote address at the Second MTS Conference on Financial Markets: "The Organization and Performance of Fixed-Income Markets" in Vienna on December 16, 2004. I gratefully acknowledge stimulating conversations with Amy Edwards, Bruno Biais, Jerry Lumer, Mike Piwowar and especially, Jonathan Sokobin, as well as comments from conference participants. The Securities and Exchange Commission disclaims responsibility for any private publication or statement of any SEC employee or Commissioner. This presentation expresses the author's views and does not necessarily reflect those of the Commission, the Commissioners, or other members of the staff.
Historically, much of the focus in studies of the bond market has concerned the determination of the term structure of interest rates within the paradigm of frictionless markets. For example, my own work on bond market pricing has emphasized the valuation of mortgages and mortgage-backed securities,1 the influence of strategic behavior on bond payoffs and pricing,2 and the power of the principle of "no arbitrage" for bond pricing.3
Yet it has become increasingly clear in recent years that frictions are central to understanding important aspects of the structure of our bond marketplace and bond pricing.4 This has been an area that has started to attract attention from academic researchers. My remarks will focus attention on some questions that I feel are deserving of further study. The two broad issues which I would like to emphasize are frictions in the information market for evaluating the creditworthiness of bonds and frictions in the trading process.
The evaluation of individual bonds and the formation of portfolios are important in the bond markets. As in other contexts, information is a central commodity in these markets. Yet "information" is not the typical type of commodity; indeed it can be difficult for an information provider to price his services or even to identify a viable strategy for getting paid in the marketplace. At the root of this problem is the traditional perspective in modern economics that information is a public good in that, like police and fire protection, its benefits can be consumed by many individuals simultaneously, i.e., it is not cost effective to exclude individuals from its benefits.
As a result of the difficulty of excluding users from the benefits of the information and the increasing returns to scale nature of the production of information, the pricing of information for individual participants is delicate. The academic literature in finance has highlighted a number of different approaches to the problem including delegated monitoring to intermediaries (as illustrated by the classic paper by Diamond (1984), which offers an academic foundation for banking), the preparation of newsletters to a restricted set of subscribers, and providing the information through portfolio management, where the portfolio structure cannot be inferred or replicated by other investors).5,6
With that introduction to the economics of the pricing of information, I think it useful to address the underlying nature of credit ratings and how ratings influence the investment process. For example, Boot, Milbourn and Schmeits (2004) recently point out that there can be multiple equilibria in credit ratings because the credit rating of a firm influences the firm's investment policy, which in turn influences the probability of credit default and the credit rating. They also argue that credit-watch procedures can empower the credit-rating agency in its dealing with the firm and result in less risky investment choices and a higher effective credit rating.
These insights point to the meaning of credit ratings, especially when the rating agency is not a passive participant, but has considerable market power. I wonder whether this is compatible with the interpretation of the credit rating as reflecting the probability of default and its potential severity. The endogeneity of the credit rating seems important. An interesting example of this occurred at the final stages of Enron's demise in the fall of 2001. As Enron's negotiations with Dynergy evolved, Enron was not assigned junk-bond status until after the negotiations broke down and then within several days Enron filed for bankruptcy. While the credit-rating agencies were criticized for waiting to assign the junk rating so close to the bankruptcy, this criticism seems misplaced as it was inevitable given the structure of debt covenants and the triggers that such a downgrade would have imposed. In effect, the downgrade is a self-fulfilling prophecy. Therefore, it does not seem to be a fair criticism of the credit-rating agencies that the downgrade was within days of the ultimate collapse, at least in this case. If the analogous downgrade would have occurred during the negotiations with Dynergy, the failure of those talks would have been inevitable. Imagine how intensely in the aftermath of such a downgrade the credit-rating agencies would have been criticized for triggering the failure of the negotiations and the demise of Enron. In fact, the credit-rating agencies waited and did not downgrade Enron until the negotiations with Dynergy collapsed. Yet, what do credit ratings really mean when real decisions and the informational structure are endogenous to the ratings? In this vein I wonder about the action deferring the downgrade and trying to "pool" in the underlying equilibrium (between the states in which the merger occurs and does not occur) given a substantial likelihood that the deal would not succeed and the amount of Enron's off-balance sheet debt.
With that preliminary discussion of the meaning of credit ratings, I'd like to focus upon the pricing of ratings-not credit spreads--but who pays for credit ratings. If the ratings from a bond-rating service are publicly available, then it is not possible to exclude potential recipients-which suggests that a pricing strategy that focuses upon the users of the ratings may not be viable. Indeed, excluding some of the potential recipients from receiving the ratings, if viable, would diminish the economic value to the credit-rating firm of providing the ratings. As a result, the bond-rating services focus upon charging the rated companies, rather than the investors who use the ratings. There is considerable demand by listed companies to have their instruments rated. Consequently, the rating service is able to contract with and charge companies in return for rating their issues. One reason that underlies the ability to charge the issuer is that the bond-rating service can exclude companies that do not pay to have their companies rated. There are several rating agencies and many issuers do not feel they need ratings from all of them.7 Curiously, the bond-rating services evaluate companies who do not pay for a rating.
Of course, if the rating services evaluated non-paying companies that would appear to undermine the ability of the rating service to charge the rated companies, at least if the rating provided were equivalent for a company that pays the rating service and one that decides not to pay. This suggests that it is puzzling that the rating services evaluate companies that do not pay for ratings. To emphasize the point, note that this puzzle does not require that the unsolicited rating is costly for the rating agency to provide and in fact, would be reinforced by assuming such evaluations are costly. But, as in much of economics, the "everything else" assumption is crucial to the underlying argument. The most natural way to resolve the puzzle to an economic theorist would be if the unsolicited ratings were not as favorable to the rated company as the paid or solicited ratings. Indeed, unsolicited ratings do not appear to be as favorable empirically as solicited ratings.
Given the financial model of the rating agencies-namely charging the rated company-the practice of unsolicited ratings that tend to be lower than the corresponding solicited ratings, helps us resolve the puzzle of why the rating agencies would engage in unsolicited ratings. Indeed, in a recent article in the Washington Post, Klein (2004) presents allegations in which a company declined repeated requests to pay one of the major rating services for a solicited rating and then the rating service provided an unsolicited rating, which it diminished over time as the company's solicited ratings actually improved. While one may dismiss such an example as anecdotal (despite the control observation from the solicited rating in the example), the cross-sectional empirical evidence in Boyoun and Shin (2002) and Butler and Rodgers (2003) of lower unsolicited ratings than solicited ratings is quite interesting along this dimension. Butler and Rogers (2003) suggest that the empirical evidence of lower unsolicited rather than solicited ratings for the same firm may reflect the ability of the bond-rating agency to use interviews and other "soft" information in the case of solicited ratings. Absent selection effects which seem at least partially controlled for by pairing the solicited and unsolicited ratings for the same firm,8 this evidence suggests systematic downward bias in the unsolicited ratings. Such evidence, along with the mere use of unsolicited ratings, are consistent with the interpretation that the rating process is being used in the language of game theorists to "punish" firms that would otherwise not purchase ratings coverage from a particular credit-rating service (after all, for some small issuers a single rating may be sufficient due to the diminishing returns from employing additional ratings services). Of course, as game theorists recognize the punishment can be credible and affect the equilibrium incentives to purchase ratings, even if the punishment (i.e., a less favorable unsolicited rating) is not widely employed. Alternatively, a more benign interpretation for the lower unsolicited ratings is that the ratings reflect the probability of default and in some cases the assessment underlying an unsolicited rating relies upon considerable "pooling" because of the lack of access to "soft information." This still begs the question as to why there are unsolicited ratings. At the same time, I recognize that for the reasons outlined above, that the pricing of informational goods is delicate and that the lower ratings for unsolicited firms being rated helps support the overall generation of information by encouraging firms to pay for their own ratings.
One other aspect of unsolicited ratings that I should highlight is that Moody's does not distinguish between unsolicited and solicited ratings in its reports, while Standard and Poors does distinguish between these. The "punishment" meted out by the credit rater on a firm that does not purchase a rating is greater when the credit-rating firm itself does not distinguish between the solicited and unsolicited ratings in its public announcements as compared to when it is known that the rating is unsolicited. The market's response should reflect a smaller reaction to the less favorable rating when unsolicited ratings are clearly identified (a separating solution) than when the adverse rating is pooled with solicited ratings. However, when the unsolicited rating is not clearly identified, the lower quality of the information which is mixed with its solicited information may reduce the perception of the quality of the information being offered by the credit-rating agency to a greater degree and differentially affect the perceived value of its services.
This analysis raises a number of questions:9
At the root of many of these issues is the classic "public goods" nature of information. This is not an easy problem to remedy. How do we ensure in our marketplace that we are willing to pay for information? In my view this problem is at also at the root of some of the other issues in the financial markets in which conflicts of interest are central-e.g., the bundling of security distribution and the supply of information (how can analysts get paid?) and the bundling of transaction services with the demand for information ("soft dollars")-but that's a story for another day.
Now I would like to turn our attention to a different friction associated with bond investment, namely the nature of frictions within the trading process. Much of the attention by the Commission and by market commentators on frictions within the trading process has often focused upon our equity market. The recent debate about the National Market System for trading equity (Regulation NMS) is illustrative of the type of scrutiny often applied to the structure of our equity trading.
I'd like to begin by briefly reflecting upon some of the differences in the context of trading bonds from other instruments, such as individual equities or futures. For example, there are vast numbers of individual bonds (millions of municipal securities offered by over 50,000 state and local governments and authorities alone); while there are only thousands of individual equities and a smaller number of futures contracts. This limits the extent of natural trading liquidity in individual bond instruments. We know from basic market microstructure analyses that liquidity attracts liquidity. While liquidity can be self-sustaining, so can its absence. At the same time there are often extremely close substitutes that can potentially trade in a parallel manner. For example, while there are a huge number of individual names in municipal bonds (and these individual series can have a large number of different maturities) there are more modest numbers of ratings classes, maturity groupings, coupon levels, etc. In fact, in the municipal bond context it also is relevant to observe that many of the bonds are insured and therefore, rated AAA (see Nanda and Singh (2004)). Among such bonds the identity of the individual issuer does not seem especially significant, but the bonds are not traded interchangeably in the marketplace.
A somewhat parallel issue arises in the mortgage-backed security market in that the issuer of each underlying mortgage is distinct. However, here the market solution appears to be somewhat different. In the mortgage-backed security context liquidity is created by forming a pool and trading the risks in the resulting pool of loans making up the mortgage-backed security. Yet there is some heterogeneity across pools with respect to anticipated prepayments and valuations. For example, mortgages from a jurisdiction in which mobility is higher, e.g., like California, will tend to have faster prepayment rates. If the loans are trading at a discount due to rising interest rates, then the fast prepayment pools are especially valuable. In order to enhance the liquidity underlying these pools, the market trades these loans on a generic basis rather than on a pool-specific basis. However, the seller often has payoff relevant information on various pools and therefore will presumably deliver the least valuable pools. Another important facet of the mortgage-backed securities market is that there are real-time operational uncertainties about how much of an investor's ownership of a pool would still be available at the delivery time. Consequently, the trading process allows some variability in the quantity to be delivered (e.g., up to five percent more or less by face value). We would expect that the seller would then deliver the least valuable pools from the perspective of his own private information based upon their characteristics and the buyers would infer that the least valuable pools would be delivered (akin to the adverse selection characterized in Akerlof (1970)). The seller also would tend to over or under deliver on the quantity dimension depending upon whether the relevant pools were valued at a premium or a discount to the previously negotiated price. Notice that valuable pools will tend to not trade in a generic pricing structure (exactly the Akerlof (1970) adverse selection problem, i.e., valuable pools like valuable cars are not sold for a pooled price) and will instead trade "on special" or an identifiable pool basis. This illustrates the potential limits to liquidity in the mortgage-backed security market due to the intrinsic diversity of individual mortgages and even pools and the seller's private information. Frictions contribute greatly to both understanding the pricing of mortgage-backed securities and also the trading mechanism, such as the use of generic trading and the trading of mortgage-backed securities on special.
In municipal and in many corporate bonds there is only limited liquidity and limited volume of trade, as there would be in mortgage-backed securities if the individual pools traded separately. Traditionally, the bond markets (and especially the municipal and corporate bond markets) have many individual bonds so that there can only be limited liquidity available in individual bonds. However, many of these bonds are close substitutes and narrowly differentiated, which should be reflected in the organization of trading and the pricing of these instruments. While sellers are interested in selling specific securities, the interests of buyers are broader. Typically, buyers are not focused upon particular securities but instead are interested in purchasing a security with certain characteristics. Though the parallel to mortgage-backed securities is striking, we note the absence of a mechanism for trading "generic" corporate bonds and municipals. Of course, there is considerable heterogeneity in the interest in different instruments. Factors such as the own-state exemption from "state" income taxes imply relatively strong clienteles in municipal bond holdings (the market is relatively segmented-limiting the degree to which two bonds from different state tax environments would be close substitutes). The extent of product differentiation is an important influence upon the extent of liquidity and the effective trading spreads in the marketplace. Optimal security design should reflect the diverse underlying preferences of investors, including the resulting impact upon the structure of the secondary market and the resulting spreads and trading costs.
The actual effective spreads in the municipal and corporate markets are surprisingly wide (often about two percent of the face value) in markets without ex-post public dissemination of trade reports, as illustrated by several recent studies.14 These wide spreads might reflect adverse selection, trading costs or market power. Economic intuition suggests that the absence of trade reporting gives dealers considerable market power because it limits the investor's understanding of the size of the spread and the state of the market. In fact, in their paper being presented at this conference Green, Hollifield and Schurhoff (2004) use a structural model to econometrically identify the market power parameter and show that much of the spread in the municipal bond market is due to market power. The usual argument against transparency is that dealers need time to work large orders; if this argument were correct it would suggest that the spreads and price impact of large orders would be substantially greater than for small orders. The contention that the results reflect market power rather than the costs of working large orders seems greatly enhanced by empirical evidence in the various market contexts that the size of the spread in both the municipal and corporate bond markets is much greater for retail executions than for institutionally sized trades (see the papers cited in footnote 14 above). This is in sharp contrast to equity markets, where large orders have much larger price impact than smaller orders.
The nature of trade transparency also is different as illustrated in the equity case by the trade reporting of block (and other) transactions. Strikingly, in equity markets the context in which transparency is often evaluated is not whether there is ex-post dissemination of trade reports, but rather the extent to which order information is transparent on an ex-ante basis.
While we lack direct evidence about the optimal transparency regime, competition and transparency have proved important in other market microstructure contexts and I anticipate that this will prove to be the case for our bond markets. As pointed out in my survey with Bruno Biais and Larry Glosten, Naik, Neuberger and Viswanathan (1999) offer an interesting argument as to why transparency may be optimal by enhancing risk sharing. While the conventional argument is that trade disclosure can make it harder to supply liquidity to large traders because the market maker is in a difficult position in trying unwind his inventory after a large and difficult order, Naik, Neuberger and Viswanathan (1999) argue that in a transparent system the market maker would not need to scale back the size of the trade (the informational content of the trade would already be reflected in the marketplace), thereby enhancing risk sharing. A similar effect also arises in Vayanos (1999). The empirical evidence in Gemmill (1996) for the London Stock Exchange is consistent with the view that transparency does not reduce liquidity.
Some Concluding Observations on the Trading Process for Bonds
Of course, one of the fundamental frictions with respect to the bond markets and especially the municipal bond markets is the spreading of liquidity across many instruments. Frankly, I feel that a full commitment to trade reporting would greatly alleviate this problem, but there are still a variety of open questions with respect to transparency and optimal market design.
I also feel that the analogy to the mortgage markets is instructive. Trading instruments based upon their main characteristics (including the state), may be helpful and narrow the spreads. How can our marketplace design electronic platforms to exploit substitutes and to what degree should our market mechanism allow trading in these?
Finally, I should also note that the first line of regulation with regard to the degree of transparency is the relevant Self Regulatory Organization (in the case of corporate bonds, the National Association of Security Dealers (NASD) and in the case of municipal bonds, the Municipal Securities Rulemaking Board (MSRB)). An interesting facet of this SRO structure is the role of the dealers as members of the committee in defining the market handling of transparency. Dealers have considerable influence and interest in the governance of the bond markets through the SRO structure. How delegation should work in defining the regulatory mechanism does not seem completely clear, especially in light of conflicts of interest by market participants. In fact, the governance of SROs is the subject of a current SEC Concept Release (2004). Of course, the ultimate success of transparency will reflect market competition as well as regulation. The value perceived by customers helps determine whether the transparent prices are published in the financial press and whether brokers offer these prices to retail investors as a competitive service.
Two important dimensions in which there are significant frictions and costs in our bond markets concern the processes for credit evaluation and the trading of bonds. In my remarks today I have tried to focus attention upon issues that I think are worthy of attention from both academics and practitioners. Of course, I welcome your feedback and questions.
Admati, A. and P. Pfleiderer, 1990, "Direct and Indirect Sale of Information," Econometrica 58, 901-928.
Admati, A. and P. Pfleiderer, 1986, "A Monopolistic Market for Information," Journal of Economic Theory 39, 400-38.
Admati, A. and P. Pfleiderer, 1988, "Selling and Trading on Information in Financial Markets," American Economic Review 78, 1165-1183.
Akerlof, G., 1970, "The Market for Lemons: Qualitative Uncertainty and the Market Mechanism," Quarterly Journal of Economics 84, 488-500.
Anthony, V., M. Haines, and M. Aydogdu, July 2004, "Report on Transactions in Municipal Securities," Office of Economic Analysis and Office of Municipal Securities, Securities and Exchange Commission.
Beaver, W., C. Shakespeare, and M. Soliman, revised Sept. 2004, "Differential Properties in the Ratings of Certified vs. Non-Certified Bond Rating Agencies," unpublished manuscript.
Biais, B. and L. Germain, 2002, "Incentive-Compatible Contracts for the Sale of Information," Review of Financial Studies 15, 987-1003.
Biais, B., L. Glosten, and C. Spatt, 2005, "Market Microstructure: A Survey of Microfoundations, Empirical Results, and Policy Implications," Journal of Financial Markets, forthcoming.
Boot, A., T. Milbourn, and A. Schmeits, 2004, "Credit Ratings as Coordination Mechanisms," unpublished manuscript.
Butler, A. and K. Rodgers, June 2003, "Relationship Rating: How do Bond Rating Agencies Process Information?" unpublished manuscript.
Byoun, S and Y. Shin, October 2002, "Unsolicited Credit Ratings: Theory and Empirical Analysis," unpublished manuscript.
Cheung, Y., F. de Jong, and B. Rindl, October 2003, "Trading European Sovereign Bonds: The Microstructure of the MTS Trading Platforms," unpublished manuscript.
Dammon, R., K. Dunn, and C. Spatt, 1993, "The Relative Pricing of High-Yield Debt: The Case of RJR Nabisco Capital Corporation Holdings Corporation," Amercian Economic Review 83, 1090-1111.
Diamond, D., "Financial Intermediation and Delegated Monitoring," Review of Economic Studies 51, 393-414.
Dunn, K. and C. Spatt, 1999, "Call Options, Points and Dominance Restrictions on Debt Contracts," Journal of Finance 54, 2317-2337.
Dunn, K. and C. Spatt, 2005, "The Effects of Refinancing Costs and Market Imperfections on the Optimal Call Policy and the Pricing of Debt Contracts," Real Estate Economics, forthcoming.
Dunn, K. and C. Spatt, 1988, "Private Information and Incentives: Implications for Mortgage Contract Terms and Contracting," Journal of Real Estate Finance and Economics 1, 47-60.
Dunn, K. and C. Spatt, 1984, "A Strategic Analysis of Sinking Fund Bonds," Journal of Financial Economics 13, 399-423.
Dybvig, P. and C. Spatt, 1985, "Does It Pay to Maintain a Reputation? Quality Incentives in Financial Markets," unpublished manuscript.
Edwards, A., L. Harris, and M. Piwowar, Sept. 2004, "Corporate Bond Market: Transparency and Transaction Costs," unpublished manuscript.
Gemmill, G., 1996, "Transparency and Liquidity: A Study of Block Trades on the London Stock Exchange Under Different Publication Rules," Journal of Finance 51, 1765-1790.
Green, R., B. Hollifield, and N. Schurhoff, Feb. 2004, "Financial Intermediation and the Costs of Trading in an Opaque Market," unpublished manuscript.
Harris, L. and M. Piwowar, 2005, "Secondary Trading Costs in the Municipal Bond Market," Journal of Finance, forthcoming.
Jorion, P., Z. Liu, and C. Shi, 2004, "Informational Effects of Regulation FD: Evidence from Rating Agencies," Journal of Financial Economics, forthcoming.
Klein, A., November 24, 2004, "Credit Raters' Power Leads to Some Abuses, Some Borrowers Say," Washington Post, p. A01.
Klein, B. and K. Leffler, 1981, "The Role of Market Forces in Assuring Contractual Performance," Journal of Political Economy 89, 615-641.
McTague, J., January 10, 2005, "The SEC is Monopolists' Best Friend," Barron's.
Naik, N., A. Neuberger, and S. Viswanathan, 1999, "Trade Disclosure Regulation in Markets with Negotiated Trades," Review of Financial Studies 12, 873-900.
Nanda, V. and R. Singh, 2004, "Bond Insurance: What is Special about Munis?" Journal of Finance.
U.S. Securities and Exchange Commission, 2003, "Concept Release: Rating Agencies and the Use of Credit Ratings under the Federal Securities Laws," Washington, D.C.
U.S. Securities and Exchange Commission, 2004, "Concept Release Concerning Self-Regulation," Washington, D.C.
Vayanos, D., 1999, "Strategic Trading and Welfare in a Dynamic Model," Review of Economic Studies 66, 219-254.
|Home | Previous Page||