The Municipal Bond Market and the Cost of Dealer Intermediation Erik R. Sirri Babson College. February 12, 2015
Topics for today 1. Description of the municipal market, key players, and regulatory framework 2. Overview of the MSRB study 3. Discussion of some current regulatory topics related to municipal bond market
Municipal securities encompass a broad range of instruments n A bond, note, warrant, certificate of participation or other obligation issued by a state or local government or their agencies or authorities (such as counties or special districts). n Examples include: General obligation bonds and revenue bonds Private activity bonds (IDB, house redevelopment) Tobacco bonds Moral obligation bonds Taxable bonds (BABs) Special tax/special assessment bonds (dirt bonds) 529 plans (municipal fund security see 2(b) of ICA) 2
Oldest domestic municipal debt pre-dates the founding of the United States
A $3.7 trillion market with a broad range of issuers n States, cities, and towns Infrastructure (roads, water, sewer) General purpose n School districts n Special agencies (Mass. Educational Financing Authority) n Bond banks n Conduit borrowers (private, 501(c)(3), IDB) n Hospitals and not-for-profits
Individuals are the dominant investors in munis (~ 65-70% of the outstandings) Closed-end Life funds insurance Banks Other MMMF P&C Insurance Households Mutual funds
There are a number of other key actors in the municipal markets n Broker-dealers and banks n Interdealer brokers n Municipal advisors n Regulators: MSRB, SEC, and FINRA n Legislatures
The market for municipal securities has a unique combination of traits n It is an over-the-counter market: no exchanges, little or no pre-trade transparency/quoting n High retail participation in the market due the to tax advantage of municipal debt n Limited disclosure due to Tower Amendment [see below] n Large number of offerings: Over 1.5 million muni securities and 80,000 issuers 29,000 corporate bonds and 5,000 listed equities n Bonds tend to be highly varied and differentiated: security, call structure, insurance, tax status, sinking fund, etc. n Limitations on bond characteristics, issuance process, and use of proceeds due to government concerns about arbitrage (e.g., yield burning) 7
Municipal market traits (cont d) n Geographically segmented market due to tax considerations n Bond issuance and refunding results from a political as well as a financial process n Municipal bonds cannot effectively be shorted n Bankruptcy is handled through Chapter 9, and half the states essentially prohibit its use without legislation n Regulatory oversight of the market and its participants attenuated and dispersed relative to that for corporate issuers [discussed below]
SEC authority over municipal markets is limited n The 1975 amendments to the 34 Act (Tower amendment) prohibit the SEC or other regulators from requiring issuers to file offering documents before they sell bonds n Exempt from registration and prospectus requirements of 33 Act n Exempt from registration and reporting under 34 Act n SEC anti-fraud authority does apply to this market: Section 10 and 17(a)
A key SEC tool is Rule 15c2-12 n n n n The rule essentially does indirectly what the SEC cannot do directly it regulates disclosure through dealer conduct Rule requires dealer underwriters to review official statements, and to ensure that issuers enter into an agreement to provide information to the MSRB on an ongoing basis, including financial information, operating data and audited financial statements If a required filing is not made, the issuer must file a failure to file notice with the MSRB Rule requires material event notices for certain events including: Delinquencies and non-payment defaults Bonds calls and tenders Defeasances Ratings changes
MSRB is the municipal market s SRO n The Municipal Securities Rulemaking Board was created by the 1975 Amendments. It is composed of 11 public, 5 b/d, 2 bank, and 3 advisor members [post-dodd Frank] prevent fraudulent and manipulative acts promote just and equitable principles of trade foster cooperation and coordination with persons engaged in regulating, clearing, settling, processing information with respect to transactions in municipal securities remove impediments to, and perfect the mechanism of, a free and open market
MSRB rulemaking n Adopts rules related to the dealer conduct in the municipal markets n Cannot write rules governing the behavior of other participants in the markets or their agents (trustees, and until recently, municipal advisors) n The MSRB cannot enforce it s own rules, not can it conduct inspections or examinations. This is done by FINRA and the SEC n Key MSRB rules relate to fair dealing, suitability, fair pricing, pay-to-play. It also operates EMMA, an analog to the SEC s EDGAR.
The MSRB study addresses three questions 1. What are the intermediation costs to customers associated with trading of municipal bonds? 2. How does transparency affect these costs? 3. How, if at all, do dealers characteristics affect customers intermediation costs? 13
A partial list of papers in this area n Municipal bond puzzle: Chalmers (1998) n Advanced refundings: Ang, Green, and Xing (2013) n Bond insurance: Bergstresser, Cohen, Shenai (2011) n Bond trading costs: Schultz (2001), (2012) Edwards, Harris, and Piwowar (2006) Goldstein, Hotchkiss, and Sirri (2006) Harris and Piwowar (2006) Green, Hollifield, and Schurhoff (2007a, b) Hollifield, Neklyudov, Spatt (2014)
The questions are answered with three sets of data n Transactions data for the period 2003 to 2010 Trade price and yield, quantity, date/time Dealer variables: n Trade type: Customer/Interdealer n Capacity: Agency/Principal n Dealer name I have no information about the customer n Bond cross-sectional information n Credit rating (over time) 15
CUSIP date time price size type buy_d sell_d 000369BZ0 1/13/03 14:07:00 98.5 20000 DB AAAA 000369BZ0 1/14/03 9:12:00 103.5 20000 DS AAAA 000369BZ0 1/15/03 15:08:00 100.75 55000 DB BBBB 000369BZ0 1/21/03 13:58:00 104.9 55000 DS BBBB 000369BZ0 3/18/03 11:43:00 101.14 20000 DB FFFF 000369BZ0 3/20/03 8:04:00 102.61 20000 DS FFFF 000369BZ0 6/25/04 14:29:00 96.5 10000 DB CCCCC 000369BZ0 6/30/04 14:22:00 97 10000 DB CCCCC 000369BZ0 7/2/04 15:38:00 100 10000 DS CCCCC 000369BZ0 7/6/04 9:36:00 100.74 10000 DS CCCCC 000369BZ0 11/2/04 14:38:00 99.326 35000 DB DDDD 000369BZ0 11/2/04 14:38:00 101.33 35000 DS DDDD 000369BZ0 12/8/04 14:51:00 99.25 20000 DB CCCCC 000369BZ0 1/12/05 15:41:00 100 5000 DB CCCCC 000369BZ0 1/12/05 16:49:00 101.3 5000 DS CCCCC 000369BZ0 1/19/05 9:56:59 101.28 20000 DS CCCCC 000369BZ0 4/28/05 11:42:20 100 20000 DB EEEEE 000369BZ0 5/2/05 9:09:21 101.15 20000 DS EEEEE 16
Because I focused on secondary market trading, I analyzed a subset of the data n There were over 71.2 million transactions in the original data n I removed the following: Floating rate bonds Transactions that occurred with 30 days of the bond issuance (dated date) BABs and taxable bonds Bonds that always traded at par Various other data cleaning filters n Filters down to about 43.5 million trades 17
Trade characteristics 18
Distribution of trade sizes Percent 19
Time between a same-day trade Percent 20
Dealer Characteristics Bonds/dealer Dealers/Bond Fraction0tx0retail Dealer0states 100%0Max 379,383 461 1.00 56 99% 75,310 70 1.00 55 95% 10,499 36 1.00 54 90% 2,841 24 0.92 54 75%0Q3 324 12 0.67 45 50%0Median 30 5 0.41 22 25%0Q1 2 2 0.14 8 10% 1 1 0.03 3 5% 1 1 0.01 2 1% 1 1 0.00 1 0%0Min 1 1 0.00 0 %0par0traded0by0top0100dealers 41.30% 21
Trading costs are difficult to measure for munis n No accepted practitioner measurement method n Bonds are highly differentiated n Trading is very infrequent n Traders use ad hoc judgment and experience in answering a question like Was that a good trade? n No agreed upon benchmark 22
Measuring intermediation cost of the market 23
Dealer chains can be complicated 24
Use algorithm to create trade chains n Sort all trades by CUSIP and time order n Search for a dealer buy n Follow the activity forward in time as the initial dealer sells inventory to other participants, using a LIFO algorithm to create chains that end with a dealer sell to a customer n For this study, 30-day/10-dealer max chain length 25
Number of dealers in a chain 26
Length of a round trip 30" 25" 20" Percent 15" 10" Series1" 5" 0" 0" 1" 2" 3" 4" 5" 6" 7" 8" 9" 10"11"12"13"14"15"16"17"18"19"20"21"22"23"24"25"26"27"28"29"30" Days 27
Actual trade sequence may have been different than reported date time price size type sell_d buy_d seq 7/11/03 15:27:00 99.53 10000 DB AAAA 2 7/11/03 15:27:00 100.53 10000 ID AAAA XXXX 3 7/11/03 15:09:00 101.53 10000 ID XXXX YYYY 1 7/14/03 10:30:00 102.07 10000 DS YYYY 4 28
Trading cost associated with dealer-intermediated trade pairs 29
Distribution of customer-to-customer trading cost 30
Costs show pronounced scale effects in trade size Lower Upper Cost+(%) $0 $5,000 2.31 $5,000 $10,000 2.12 $10,000 $25,000 1.84 $25,000 $50,000 1.51 $50,000 $100,000 1.13 $100,000 $500,000 0.73 $500,000 $1,000,000 0.38 $1,000,000 $3,000,000 0.24 $3,000,000 $5,000,000 0.17 $5,000,000 max 0.12 31
Transparency regime changed over time n Before 1995, transparency in municipal bonds was very limited. January 1995 summary data for inter-dealer trades in frequently traded securities (i.e. 4 or more times in a day) published on T+1 August 1998 customer trades added to determination of frequently traded securities; summary data for customer and inter-dealer trades in frequently traded securities published on T +1 January 2000 details of each trade in frequently traded securities published on T+1 October 2000 details of all trades published on a one month delay basis
Transparency regime changed over time May 2002 definition of a frequently traded security changes to one that trades 3 or more times in a day August 2002 details of all trades published on a one week delay basis December 2002 definition of a frequently traded security changes to one that trades 2 or more times in a day June 2003 frequently traded threshold removed; details of all trades published on T+1 with par amounts over $1,000,000 displayed as $1MM+
Transparency regime changed over time n Before January 31, 2005 trades by dealers were reported to the MSRB but these data were generally not re-disseminated to the public in real-time. n Starting on January 31, 2005 there was real-time (within 15 minutes) reporting or price and quantity for all trades below $1 million. n Trades above $1million had reported prices but were marked at +$1 million. Actual quantities were reported on T+5.! How did the improved transparency affect dealer intermediation costs?
OLS cost regression using dealer characteristics 35
Discontinuity regression one year either side of January 2005, with fixed effects (full sample) 36
Trading cost associated with 15-minute reporting 37
38
39
SEC/FINRA active on several enforcement fronts n n n n State of Illinois: related to the undisclosed risk associated with the funding of the state pension system Victorville, CA: for purposefully inflating the value of certain property securing a bond offering. Harrisburg, PA: For misrepresentations on the city s website, and for failing to provide investors with information about multiple Moody s downgrades to GO debt Regulatory focus on impaired or b/r issuers: Detroit; San Bernardino, Stockton, Vallejo and the town of Mammoth Lakes, Calif.; Central Falls, R.I.; Jefferson County, Ala.; and Harrisburg, Pa.
Policy issues before the SEC and MSRB n Bring pre-trade transparency to the secondary market No dealer quoting to capture and disseminate Unlike equity markets, it would likely not be practical to require dealers to publish a quote just because they are about to trade a bond often times dealers are simply called and asked to bid on a position If private systems were to be created for dissemination of quotes to the systems members or customers, such a system may be required to disseminate if it reaches a given size Alternatively, existing bond ATSs may be required to make their quotations public, though few sizable transactions are executed on these systems
Policy issues (cont d) n MSRB s new best execution standard Old: shall make a reasonable effort to obtain a price for the customer that is fair and reasonable in relation to prevailing market conditions New: require dealers to use "reasonable diligence" to determine the best market for a security and then buy or sell the security in that market so the resulting price to the customer "is as favorable as possible under prevailing market conditions. A list of factors for dealers to consider Does not address the point that muni bonds are sold and not bought what bond is sold to the customer may be as important than the price at which it was sold
Riskless principal markup disclosure n n n n n Bond markets are net markets with no commission. Dealers are compensated via a markup or markdown, and prices are reported inclusive of this cost While Rule 10b-10 requires disclosure of commissions, there is no requirement for disclosure of a markup Markups can be difficult to calculate if the market moves between the dealer s purchase sale A riskless principal trade occurs when the dealer lines up the buyer and the seller at the same time, and only nominally takes possession of the bond Such a markup can be straightforward to calculate, but will be unpopular with the dealer community as it leads to more cost transparency
Other policy issues n n n Reconsideration of Rule 15c2-12: showing its age Lack of reporting requirements for bank debt, swaps, and direct-purchase debt Move to more 8-K like event disclosure Lack or reporting about the creation of new direct or offbalance sheet financial obligations New interpretive guidance needed Increased standardization of debt terms and more plainvanilla offerings Some innovation: State of MA is continuously offering bonds through an ATS, providing updated disclosure via an iphone app, a getting good prices for these ATS-offered bonds.