PrefBlog

The FINRA page titled “Independent TRACE Studies” leads me to some studies that have come out of mandatory TRACE reporting for Structured Credit. The data collected are not made public in raw form, but aggregate data is made available on a daily and monthly basis.

Structured Credit is a little far from my bailiwick, but some of the statistics cited are so entertaining I just had to highlight them here.

A 2013 paper by Hendrik Bessembinder, William F. Maxwell and Kumar Venkataraman is titled Trading Activity and Transaction Costs in Structured Credit Products:

After conducting the first study of secondary trading in structured credit products, the authors report that the majority of products did not trade even once during the 21-month sample. Execution costs averaged 24 bps when trades occurred and were considerably higher for products with a greater proportion of retail-size trades. The authors estimate that the introduction of public trade reporting would decrease trading costs in retail-oriented products by 5-7 bps.

An acknowledgement in this paper, by the way, introduced me to the Montreal Institute of Structured Finance and Derivatives, which appears to be funded by agencies of the Province of Quebec and provides modest funding to genuine research – only $300,000 p.a., but that’s not bad when compared to their 10-year total budget of $15-million (meanwhile, here in Ontari-ari-ari-owe, we fund ex-regulators and lawyers at FAIR Canada. Gag.) You learn something new every day!

Anyway, back to Bessembinder et al. and the amazing statistics:

Notably, less than twenty percent of the SCP universe trades at all during the twenty-one month sample period. One-way trade execution costs for SCPs average about 24 basis points. However, trade execution costs vary substantially across SCP categories, from 92 basis points for CBOs to just 1 basis point for TBAs. We show that trading costs depend in particular on what we term the product’s “customer profile,” which depends on issue size and the proportion of Retail- versus Institutional-size trades. Sub-products with an Institutional Profile tend to have lower costs. The highest average trading costs are observed for Agency CMOs (74 basis points) and CBOs (92 basis points), each of which has a low (22% or less) proportion of large trades. The lowest average trading cost estimates are observed for TBA securities (1 basis point), CMBS (12 basis points), and ABS secured by auto loans and equipment (7 basis points), each of which has a large (54% or greater) percentage of large trades.
…
Structured Products (SCPs), including asset-backed (ABS) and mortgage-backed (MBS) securities, comprise one of the largest but least-studied segments of the financial services industry. As of the end of 2012, there was $8.6 trillion outstanding in mortgage-backed securities and $1.7 trillion outstanding in asset-backed securities, implying that the SCPs markets are comparable in size to the $11 trillion U.S. Treasury Security market
…
Increased transparency has the potential to reduce the dealer mark-ups or bid-ask spreads, provide more information on the fair price of the security, and improve regulators and customers’ ability to control and evaluate trade execution costs. These ideas have been emphasized by Rick Ketchum, Chairman and CEO of the Finance Industry Regulatory Authority (FINRA):

“From the standpoint of investor protection, which is and always will be FINRA’s top priority, we simply must shed more light on the darker areas of the fixed income market.

That last quote from Rick Ketchum illustrates the big problem with securities regulation nowadays. They have swung so far over to the ‘consumer protection’ objective that they have – at least to some degree – lost sight of why capital markets exist in the first place: to get money from savers to those who want to invest in their businesses. I will certainly agree that investor protection is a worthy objective; and I will agree that it is related reasonably closely to the objective of having a well functioning capital market; but for it to be called the “top priority” shows very strange priorities.

So the first amazing statistic is:

The MBS database provided to us by FINRA contains almost 1.1 million distinct securities. The large number of securities reflects that a basic pool of assets may have more than a hundred tranches, each with a unique payoff structure, and assets can be re-securitized (By comparison, less than 5,000 companies were listed on the U.S. equity exchanges at the end of 2012). However, as Panel A of Table 1 shows, many of these issues are very small, as the 25th percentile issue size is less than $2MM. The median issue size is less than $5MM. However, the distribution is positively skewed, as the mean issue size is $22.8MM. The MBS securities are of long average maturity, as shown on Panel B, with a mean maturity close to 19 years.

Holy smokes! I knew there were lots, but I would have guessed ‘under a million’. Hundred-tranche structured products sound pretty amazing, too.

Table 2 reports on trading activity for MBS. Notably, only 17.8% of the issues traded at all during the twenty one month period studied. The mean dollar volume traded across the full sample of MBS securities is $106MM, with an average of only 4.1 trades in each security. Fannie’s issues average six trades during the sample, and the average trading volume for Fannie issues is almost three times as large as for the next most frequently traded issue (Ginnie). Freddie’s issues are traded significantly less than either of the other agencies. Non-Agency issues trade an average of only 1.8 times each, but surprisingly have the largest proportion of issues (23%) that trade at all. Non-Agency issues have an average of 3.5 dealers at issuance, compared to slightly over four dealers for each Agency issue
…
Table 3 contains information regarding the ABS data, which contains slightly over 300,000 issues (compared to 1.1 million issues in the MBS universe).5 The ABS issues are larger than MBS issues, with the mean ($114MM) and median ($29MM) issue size each close to five times larger than for ABS. Still, some ABS issues are very small; the 5th percentile of the issue size distribution is only $100,000 for ABS, compared to over $1MM for MBS. ABS issues have an average maturity of 23.2 years, about 5 years longer than MBS products.

Panel B of Table 3 reports on trading activity in the ABS market. Like MBS, ABS trade infrequently, but the percentage of issues that trade at all is almost 30%, considerably higher than MBS at 18%. The average number of trades per security is 4.97, but the trades are on average smaller for ABS; the mean cumulative trading volume for ABS is $16.3MM, compared to the $106MM for MBS issues. The likelihood of trading and mean number of trades is surprisingly homogenous across issue size terciles. However, average trade size and cumulative dollar volume is larger for ABS of greater issue size.

And what are these trades?

Table 5 reports on the distribution of trade sizes in SCPs. We consider a trade to be small if it is for less than $100,000 and large if it is for more than $1MM.
…
For comparison purposes, we examine the distribution of trade sizes for corporate bonds during the six months before and after the introduction of public transaction dissemination. Our analysis includes 1.9 Million trades in 10,108 corporate bonds phased into TRACE dissemination between January 2003 and March 2011.8 We find that 72% of corporate bond trades are small (less than $100,000), both before and after trades were publicly disseminated. We conclude that, on average, the market for corporate bonds is more similar to the retail-oriented markets for SCPs, including CMOs and MBSs, and is more distinct from the institutionally-oriented markets for CMBS and TBA securities.

And the cost?

The resulting estimates of customer trade execution costs are reported on Table 6. For the full
sample, the estimated average one-way trade execution cost is 24 basis points. Consistent with results previously reported for corporate and municipal bonds, trade execution costs for SCPs decline with trade size, averaging 83 basis points for small trades, 24 basis points for medium-sized trades, and only five basis points for large trades. Trade execution costs also vary depending on trading frequencies. Average costs for the least-heavily-traded tercile of securities are 31 basis points, compared to 28 basis points for the second tercile and 24 basis points for the most frequently traded tercile. The finding that trade execution costs for SCPs decline with trade size mirrors the findings reported for corporate bonds by Edwards, Harris and Piwowar (2007) and Goldstein, Hotchkiss and Sirri (2007) and for municipal bonds by Harris and Piwowar (2006) and Green, Hollifield, and Schurhoff (2007). The overall level of estimated trading costs for SCP is in line with estimates for corporate bonds.

Bessembinder, Maxwell and Venkataraman (2006) study institutional trades in corporate bonds, and report average one-way trade execution costs (prior to transaction dissemination) that average 10 to 20 basis points. Schultz (2001) also studies institutional trades in corporate bonds and estimates that trading costs average 27 basis points. Edwards, Harris and Piwowar (2007) study a broader cross-section that includes retail trades, and estimate that one-way trade execution costs for corporate bonds range from 75 basis points for very small trades to 4 basis points for very large trades.

And the effect of TRACE?

We first implement expression (2) for the full set of corporate bonds that became TRACE-eligible in March of 2003, including in the analysis trades executed six months before to six months after the initiation of public trade dissemination. We find that trading costs for corporate bonds were reduced after the introduction of price dissemination by 9 basis points for small trades, 6 basis points for medium trades, and 3 basis points for large trades. These results are quite similar to those reported by Edwards, Harris, and Piwowar (2006), who study the same sample but rely on more complex estimation techniques.

I take issue with the authors when they claim:

These estimates of lower trading costs for SCPs have important implications for security issuers, investors in these products and broker-dealers who supply liquidity. Improved liquidity that is attributable to post-trade price transparency has the potential to affect the valuation of the bonds themselves and lower yield spreads (see Chen, Lesmond and Wei (2007) for evidence from corporate bonds) for SCP issues. Additionally, the cumulative dollar impact of these trading cost reductions is potentially large. In the case of the transparency experiment for corporate bonds, Bessembinder, Maxwell and Venkataraman (2006) estimate annual trading cost reductions of about $1 billion for the full corporate bond market. In addition, they document the existence of “liquidity externalities”, by which improved transparency for some products can lead to improved valuation and lower trade execution costs for related securities.

As I pointed out in an earlier post, a tighter spread between the dealer buy price and dealer sell price does not necessarily indicate “fairer” prices, since the dealer may well quote only stink bids on customer sales so that a profitable re-sale can be executed quickly. This mechanism, if correct, would actually mean that the liquidity-seeker in the chain of trades is paying more for liquidity under TRACE and that both the interim and ultimate liquidity providers are making excess profits (I refer to this as the Shitty Price Hypothesis). The authors do not examine how the execution prices in the secondary market compare with new-issue prices, which renders their conclusion regarding the “improvement” in liquidity dubious.

In addition to this, the putative benefits of TRACE, estimated as “annual trading cost reductions of about $1 billion for the full corporate bond market”, does not make any attempt to compare this with the cost of the programme. And I don’t mean direct costs, either. If the Shitty Price Hypothesis is correct – and it is consistent with the finding of lower trading levels in the Asquith, Covert and Pathak paper, then actual liquidity has decreased, which means issuers will have to pay more for funds, which means that some bricks-and-mortar projects will be abandoned (this link in the chain is the entire basis for central banking policy rates) … and how much does that cost? Huh?

Anyway, the authors told us to “see Chen, Lesmond and Wei (2007) for evidence from corporate bonds”, so let’s look at Chen, Lesmond and Wei (2007) and see what they have to say.

The paper by Long Chen, David A. Lesmond & Jason Wei is titled Corporate Yield Spreads and Bond Liquidity and it turns out that the last named author is from our very own Rotman School of Management at UofT:

We examine whether liquidity is priced in corporate yield spreads. Using a battery of liquidity measures covering over 4000 corporate bonds and spanning investment grade and speculative categories, we find that more illiquid bonds earn higher yield spreads; and that an improvement of liquidity causes a significant reduction in yield spreads. These results hold after controlling for common bond-specific, firm-specific, and macroeconomic variables, and are robust to issuers’ fixed effect and potential endogeneity bias. Our finding mitigates the concern in the default risk literature that neither the level nor the dynamic of yield spreads can be fully explained by default risk determinants, and suggests that liquidity plays an important role in corporate bond valuation.
…
The notion that investors demand a liquidity premium for illiquid securities dates back to Amihud and Mendelson (1986). Lo, Mamaysky, and Wang (2004) further argue that liquidity costs inhibit the frequency of trading. Because investors cannot continuously hedge their risk, they demand an ex-ante risk premium by lowering security prices. Therefore, for the same promised cash flows, less liquid bonds will be traded less frequently, have lower prices, and exhibit higher yield spreads. Thus, the theoretical prior is that liquidity is expected to be priced in yield spreads. We investigate bond-specific liquidity effects on the yield spread using three separate liquidity measures. These include the bid-ask spread, the liquidity proxy of zero returns, and a liquidity estimator based on a model variant of Lesmond, Ogden, and Trzcinka (1999). We find that liquidity is indeed priced in both levels and changes of the yield spread.

Contemporaneous studies by Longstaff et al. (2004) and Ericsson and Renault (2002) also relate corporate bond liquidity to yield spreads.
…
Historically, the lack of credible information on spread prices or bond quotes has been a major impediment in the analysis of liquidity (Goodhart and O’Hara, 1997) and liquidity’s impact on yield spreads. We employ Bloomberg and Datastream to provide our three liquidity estimates. Among them, the bid-ask spread is arguably the most demonstrable measure of liquidity costs, while the percentage of zero returns is increasingly used as a liquidity proxy in a host of empirical studies.2 Despite the clear intuition surrounding the zero return proxy, it is a noisy measure of liquidity, since it is the combination of a zero return and the simultaneous movement of bond price determinants that more properly estimates liquidity costs, not the lack of price changes per se.
…
We find a significant association between corporate bond liquidity and the yield spread with each of the three liquidity measures. Depending on the liquidity measure, liquidity alone can explain as much as 7% of the cross-sectional variation in bond yields for investment grade bonds, and 22% for speculative grade bonds. Using the bid-ask spread as the measure, we find that one basis point increase in bid-ask spread is related to 0.42 basis point increase in the yield spread for investment grade bonds, and 2.30 basis point increase for speculative grade bonds.

So I don’t find anything objectionable in the conclusion; I’ve argued in this blog for a long time that liquidity is a major factor in corporate bond yields, far outweighing credit quality considerations. I will, however, point out that their primary liquidity estimator is at least a little suspect:

Data on the quarterly bid-ask quotes are hand-collected from the Bloomberg Terminals. Most quotes are available only from 2000 to 2003. For each quarter, we calculate the proportional spread as the ask minus the bid divided by the average bid and ask price. The bond-year’s proportional bid-ask spread is then calculated as the average of the quarterly proportional spreads. To include as many bonds as possible, we compute the annual proportional spread as long as there is at least one quarterly quote for the year. The bid-ask quotes recorded are the Bloomberg Generic Quote which reflects the consensus quotes among market participants.

I have to point out that Bloomberg quotes are suspect according to the Jankowitsch, Nashikkar and Subrahmanyam paper referenced in an earlier post, with almost half of actual trades executed outside the quote. This doesn’t necessarily mean that the Bloomberg quotation spreads are useless as a liquidity estimator, but it does mean that somebody has to do some work to show that Bloomberg spreads do in fact have a solid relationship to real life (e.g., that if the bid on bond A is less than the bid on bond B, then you can in fact sell B at a higher price than A).

So what it comes down to is that I agree with Bessembinder, Maxwell and Venkataraman that if TRACE does improve liquidity, then this is a good thing, but I will claim that you cannot measure liquidity in a practical way by comparing dealer sell prices with dealer buy prices if the Shitty Price Hypothesis holds.

As it happens, there is a paper by Nils Friewald, Rainer Jankowitschy and Marti G. Subrahmanyamz which seeks to validate the round-trip trading cost as a measure of liquidity, titled Transparency and Liquidity in the Structured Product Market:

We use a unique data set from the Trade Reporting and Compliance Engine (TRACE) to study liquidity effects in the US structured product market. Our main contribution is the analysis of the relation between the accuracy in measuring liquidity and the potential degree of disclosure. We provide evidence that transaction cost measures that use dealer-specic information can be eciently proxied by measures that use less detailed information. In addition, we analyze liquidity, in general, and show that securities that are mainly institutionally traded, guaranteed by a federal authority, or have low credit risk, tend to be more liquid.
…
For example, measuring liquidity based on the round-trip cost uses the most detailed information, i.e., each transaction needs to be linked to a particular dealer, on each side of the trade. Other liquidity metrics, such as the effective bid-ask spread, do not need such detailed trade information for their computation; but, transactions need to be flagged as buy or sell trades. Many alternative liquidity measures rely on trading data as well: However, they use only information regarding the price and/or volume of each transaction. On the other hand, product characteristics or trading activity variables represent simpler proxies, using either static or aggregated data.
…
Exploring the various liquidity metrics and focusing on the predictive power of transaction data, we show that simple product characteristics and trading activity variables, by themselves, may not be sufficient statistics for measuring market liquidity. In particular, when regressing state-of-the-art liquidity measures on product characteristics and trading activity variables, we find that the various liquidity measures over significant idiosyncratic information. Thus, dissemination of detailed transaction data, necessary for the estimation of liquidity measures, is of importance in the fixed-income structured product market. However, there is evidence that liquidity measures based on price and volume information alone (e.g., the imputed round-trip cost measure) can explain most of the variation observed in the benchmark measure, which uses significantly more information and certainly runs the risk of compromising the confidentiality of trader identity. In a second set of regressions, we explain the observed yield spreads using various combinations of liquidity variables and nd similar results: Liquidity measures provide higher explanatory power than product characteristics and trading activity variables alone. However, this result is mostly driven by price and volume information. Thus, details regarding the identities of the specific dealers involved with a particular trade or the direction of the trade are not an absolute necessity in terms of their informational value to market participants: Reasonable estimates of liquidity can be calculated based on prices and volumes of individual trades, without divulging dealer-specific information. This is an important result for all market participants, as it provides valuable insights concerning the information content of reported transaction data.

They acknowledge the Bessembinder paper and discuss the differences:

However, our paper is different from Bessembinder et al. (2013) for at least five important reasons, relating to various aspects of liquidity effects in the structure product market: First, while their analysis is based only on one single estimate of liquidity, we, in contrast, rely on a much broader set of liquidity proxies, which allows us to discuss the information contained in measures employing reported data at different levels of detail. Second, while Bessembinder et al. (2013) use a regression based estimate of liquidity, our round-trip cost measure (which serves as our benchmark) reflects the cost of trading more accurately, since it is based on detailed dealer-specific transaction costs, which are straightforward to compute, and does not depend, in any way, on modeling assumptions. Third, in their analysis, they focus solely on customer-to-dealer trades which constitute only a rather small fraction of all trades in the structured product market, whereas our analysis is based on all customer-to-dealer and dealer-to-dealer transactions. Fourth, unlike their study, we analyze different sub-segments (e.g., tranche seniority, issuing authority, credit rating) of the overall market in much more detail. These sub-segments have either turned out to be important in other fixed income markets, or are unique to the structured product market. Finally, a novel contribution of our paper is that we also analyze which of the liquidity measures best serves to explain yield spreads in the securitized product market.

So more particularly:

Thus, we ask how much information should be disseminated to allow for the accurate measurement of liquidity, compared to our benchmark measure using the most detailed information, in particular trader identity and trade direction, which certainly runs the risk of compromising the identities of individual traders or their trading strategies. Therefore, we measure the efficacy of liquidity metrics that require different levels of detail in terms of the information used to compute them. We analyze two aspects of this question, using different sets of regressions: First, we explore to what extent product characteristics, trading activity variables and liquidity measures using less information can proxy for the benchmark measure which is based on all available information. Second, we study which liquidity measures can best explain the cross-sectional differences in yield spreads for our sample.
…
Product characteristics are rather crude proxies of liquidity that rely on the lowest level of informational detail of all the categories.13 Thus, product characteristics are typically used as liquidity metrics when there is a limitation on the level of detail in the transaction data. In particular, we use the amount issued of a security measured in millions of US dollars. We presume securities with a larger amount issued to be more liquid, in general. Another important product characteristic is the time-to-maturity, which corresponds to the time, in years, between the trading date and the maturity date of the security. We expect securities with longer maturities (over ten years) to be generally less liquid, since they are often bought by “buy-and-hold” investors, who trade infrequently. We also consider the instrument’s average coupon as a relevant proxy. Despite the ambiguity of the relationship between the coupon and both liquidity and credit risk, we expect that instruments with larger coupons are generally less liquid.

Trading activity variables such as the number of trades observed for a product on a given day represent the aggregate market activity.15 Other similar variables that we calculate on a daily basis, for each product, are the number of dealers involved in trading a specific product, and the trading volume measured in millions of US dollars. We expect these variables to be larger, the more liquid the product. On the contrary, the longer the trading interval, which refers to the time elapsed between two consecutive trades in a particular product (measured in days), the less liquid we would expect the product to be.

Note that the Shitty Price Hypothesis negates this last assumption: dealers will set prices so they can exit their positions quickly.

Liquidity measures are conceptually based, and hence, more direct proxies for measuring liquidity, and require transaction information for their computation. However, the level of detail concerning the required information set varies considerably across measures. The liquidity measure that uses the most detailed information and, thus, serves as our benchmark measure, is the round-trip cost measure, which can be computed only if the traded prices and volumes can be linked to the individual dealer; see, e.g., Goldstein et al. (2007). It is defined as the price difference, for a given dealer, between buying (selling) a certain amount of a security and selling (buying) the same amount of this security, within a particular time period, e.g., one day. Thus, it is assumed that in a “round-trip” trade, the price is not affected by changes in the fundamentals during this period. Following the literature, the round-trip trade may either consist of a single trade or a sequence of trades, which are of equal size in aggregate, on each side. The effective bid-ask spread, proposed by Hong and Warga (2000), can be computed when there is information about trade direction available. The effective bid-ask spread is then defined as the difference between the daily average sell and buy prices (relative to the mid-price).

Many other liquidity measures use only the price and/or volume of each transaction, without relying on dealer-specific or buy/sell-side information. A well-known metric proposed by Amihud (2002), and conceptually based on Kyle (1985), is the Amihud measure. It was originally designed for exchange-traded equity markets, but has also become popular for measuring liquidity in OTC markets. It measures the price impact of trades on a particular day, i.e., it is the ratio of the absolute
price change measured as a return, to the trade volume given in US dollars. A larger Amihud measure implies that trading a financial instrument causes its price to move more in response to a given volume of trading and, in turn, reflects lower liquidity. An alternative method for measuring the bid-ask spread is the imputed round-trip cost, introduced by Feldhutter (2012). The idea here is to identify round-trip trades, which are assumed to consist of two or three trades on a given day with exactly the same traded volume. This likely represents the sale and purchase of an asset via one or more dealers to others in smaller trades. Thus, the dealer identity is not employed in this matching procedure; rather, differences between the prices paid for small trades, and those paid for large trades, based on overall identical volumes, are used as the measure. The price dispersion measure is a new liquidity metric recently introduced for the OTC market by Jankowitsch et al. (2011). This measure is based on the dispersion of traded prices around the market-wide consensus valuation, and is derived from a market microstructure model with inventory and search costs. A low dispersion around this valuation indicates that the nancial instrument can be bought for a price close to its fair value and, therefore, represents low trading costs and high liquidity, whereas a high dispersion implies high transaction costs and hence low liquidity. The price dispersion measure is defined as the root mean squared difference between the traded prices and the average price, the latter being a proxy for the respective market valuation.

The Roll measure, developed by Roll (1984) and applied by Bao et al. (2011) and Friewald et al. (2012), for example, in the context of OTC markets, is a transaction cost measure that is simply based on observed prices. Under certain assumptions, adjacent price movements can be interpreted as a “bid-ask bounce”, resulting in transitory price movements that are serially negatively correlated. The strength of this covariation is a proxy for the round-trip transaction costs for a particular nancial instrument, and hence, a measure of its liquidity. This measure requires the lowest level of detail as only traded prices, and not trading volume or dealer-specific information, are used in the computation.

Whoosh! That’s a lot of liquidity measures! And I thought I was obsessive!

The descriptive statistics and correlations presented in Section 5.1 provide initial indications of the informational value of the various liquidity measures. When analyzing the liquidity of the different markets and their sub-segments, the liquidity measures offer additional insights compared to the product characteristics and trading activity variables. For example, when comparing the different market segments, higher trading activity is not always associated with lower transaction costs. The correlation analysis hints in the same direction: There is low correlation between the product characteristics and the liquidity measures (the highest correlation coefficient is 0.26 in absolute terms) and between trading activity variables and liquidity measures (less than 0.20 in absolute terms). Thus, it seems that liquidity measures that rely on more detailed transaction data can provide important additional information, based on this perspective.
…
Table 10 shows the results for this analysis, presenting the six specifications. In regressions
(1) to (5), we use each of the liquidity measures in turn, plus all trading activity variables and product characteristics, to explain the round-trip costs. When we add just one individual proxy to the regression analysis, we find that the imputed round-trip cost, the effective bid-ask spread and the price dispersion measure are the best proxies, with R2 values of around 50% to 60%, whereas the Amihud and Roll measures slightly increase the R2 to around 40% compared to regressions without liquidity measures. When adding all the liquidity measures to the regression equation, in regression (6), we obtain an R2 of 67%, i.e., the explanatory power increases considerably when we include all these proxies. We consider this level of explanatory power quite high, given the rather diverse instruments with potentially different liquidity characteristics and the low number of trades per security and day, in general. We get similar results (not reported here) when explaining the effective bid-ask spread with liquidity measures using less information. Thus, we find evidence that liquidity measures using more detailed data can be proxied reasonably well by similar measures using less data. We further discuss this issue in the next section and analyze the importance of the disclosure in the context of pricing.

And correlation with yields?

Analyzing the effect of the trading activity variables in the full model, we find economically significant results only for the trading interval: An increase in the trading interval by one standard deviation is associated with an increase in the yield spread of 15 bp. The information contained in the other trading activity variables, e.g., traded volume, seems to be adequately represented by the liquidity measures. However, more important are the results for the product characteristics. The most relevant variable in the full model turns out to be the coupon. A one-standard-deviation higher coupon results in an increase of 137 bp in the yield spread. Thus, the coupon rate has the highest explanatory power of all the variables, indicating that a higher coupon is also associated with higher credit risk for certain products, in particular when there is no credit rating available. The amount issued shows important effects as well, where a one-standard-deviation increase leads to an 19 bp decrease in the yield spread: Larger issues have lower yield spreads. The maturity of a structured product is related to the yield spread as well, indicating that longer maturities are associated with somewhat lower spreads. However, compared with the other product characteristics, the maturity is of minor importance. Overall, the full model has an R2 of 69.9% with significant incremental explanatory power shown by the liquidity measures. Thus, liquidity is an important driver of yield spreads in the structured product market; therefore, the dissemination of trading activity information is important, given the size and complexity of this market.

And they conclude:

Exploring the relation between the various liquidity proxies and the depth of disseminated information, we find that product characteristics or variables based on aggregated trading activity, by themselves, are not sucient proxies for market liquidity. The dissemination of the price and volume of each individual trade is important for the quantification of liquidity effects, particularly for explaining yield spreads. However, we also provide evidence that liquidity measures that use additional dealer-specific information (i.e., trader identity and sell/buy-side categorization) can be efficiently proxied by measures using less information. In our regression analysis, we find that liquidity effects cover around 10% of the explained variation in yield spreads. Thus, the dissemination of trading activity is essential, given the trade volume and complexity of this market. These results are important for all market participants in the context of OTC markets, as it allows establishing an understanding of the information content contained in the disclosure of trading data.

Paul Asquith, Thomas R. Covert and Parag Pathak have written a paper titled The Effects of Mandatory Transparency in Financial Market Design: Evidence from the Corporate Bond Market:

Many financial markets have recently become subject to new regulations requiring transparency. This paper studies how mandatory transparency affects trading in the corporate bond market. In July 2002, TRACE began requiring the public dissemination of post-trade price and volume information for corporate bonds. Dissemination took place in Phases, with actively traded, investment grade bonds becoming transparent before thinly traded, high-yield bonds. Using new data and a differences-in-differences research design, we find that transparency causes a significant decrease in price dispersion for all bonds and a significant decrease in trading activity for some categories of bonds. The largest decrease in daily price standard deviation, 24.7%, and the largest decrease in trading activity, 41.3%, occurs for bonds in the final Phase, which consisted primarily of high-yield bonds. These results indicate that mandated transparency may help some investors and dealers through a decline in price dispersion, while harming others through a reduction in trading activity.
…
Proponents of TRACE argue that transparency makes the corporate bond market accessible to retail clients, enhances market integrity and stability, and provides regulators greater ability to monitor the market. They reason that with the introduction of transparency, price discovery and the bargaining power of previously uninformed participants should improve (NASD 2005). This in turn should be reflected in a decrease in bond price dispersion and, if more stable prices attract additional participants, an increase in trading activity (Levitt 1999).

Opponents of TRACE object to mandatory transparency, saying that is unnecessary and potentially harmful. They argue that “transparency would add little or no value” to highly liquid and investment grade bonds since these issues often trade based on widely known US Treasury benchmarks (NASD 2006). They further argue that if additional information about trades was indeed valuable, then third‐party participants would already collect and provide it, a view that dates back to Stigler (1963). Opponents also forecast adverse consequences for investors since, if price transparency reduces dealer margins, dealers would be less willing to commit capital to hold certain securities in inventory making it more difficult to trade in these securities. The Bond Market Association argued that the adverse effects of transparency may be exacerbated for lower‐rated and less frequently traded bonds (Mullen 2004). Lastly, opponents saw TRACE as imposing heavy compliance costs, particularly for small firms who do not self‐clear (Jamieson 2006). Thus, opponents argue that market transparency reduces overall trading activity and the depth of the market. Not surprisingly, similar arguments for and against transparency have resurfaced in response to the recent introduction of the Dodd‐Frank’s post‐trade transparency requirements for swaps (Economist 2011).

With all respect to the various debaters, and while recognizing that the above is an extremely quick summary of their thoughts, I have to say that all the quoted arguments miss the mark. The fundamental question is: what is the corporate bond market for? I claim that the purpose of the corporate bond market is to allow issuers to access capital at as little cost as possible; therefore, all regulation related to the bond market should be first examined through the lens of ‘what will this do to new issue spreads?’. While this is not the only thing to be addressed, it is the most important thing and it is something I rarely see addressed.

It was addressed, however, in a 2012 comment letter to FINRA from SIFMA:

Issuers face the ultimate risk from decreases to market liquidity since the public dissemination of trade information, as a general matter, makes broker-dealers less willing to take risk on large size trades. A reduction in liquidity will cause institutional investors to demand greater yield from issuers (to compensate for the reduced liquidity), or to simply refuse to buy new issues in meaningful size. Therefore, a careful balance between transparency and the preservation of liquidity must be struck. Data shows that dealers have recently chosen to (or been forced to, in the case of rules like the Volcker Rule) put capital to work elsewhere. This means that institutional investors will face greater difficulty selling a larger sized amount of an issue. Pre-TRACE, and pre-financial crisis, dealers provided a much larger outlet where they would take the risk temporarily while they worked to uncover a buyer. This outlet has been much reduced in recent years, due to a combination of regulation and other market structure issues. The real liquidity differential for larger vs. smaller “on the run” amounts has been meaningfully amplified, and eliminating caps on disseminated volumes would exacerbate this problem. At a much more specific level, it is more difficult to issue securities in smaller sizes when participant’s transactions are immediately made public and expose exact amounts taken down by particular investors. An increase in the dissemination caps will increase the threshold where these securities issuances are somewhat more challenging, and disproportionately harm smaller issuers. In each case, the macro and the granular, the result is a higher cost of capital for issuers.

Letting that issue slide for a moment and returning to Asquith, Covert and Pathak:

FINRA implemented TRACE in Phases because of concerns about the possible negative impact of transparency on thinly traded, small issue and low‐credit rated bonds. Examining issue size across all Phases, we find that trading activity decreases more for large issue size bonds, but that the reduction in price dispersion is uncorrelated with issue size. Credit ratings, however, matter for both trading activity and price dispersion. High‐yield bonds experience a large and significant reduction in trading activity, while the results are mixed for investment grade bonds. High‐yield bonds also experience the largest decrease in price dispersion, but price dispersion significantly falls across all credit qualities. Therefore, the introduction of transparency in the corporate bond market has heterogeneous effects across sizes and rating classes.
…
Price dispersion also decreases due to TRACE. This decrease is significant across bonds that change dissemination in Phases 2, 3A, and 3B, but is largest, 24.7%, for Phase 3B bonds. This finding is also robust across different measures of price dispersion and alternative regression specifications. Moreover, event studies show that the fall in price dispersion occurs immediately after the start of dissemination. It is important to note, if the transparency introduced in Phase 1 affects bonds that become transparent in subsequent Phases, our estimates are probably lower bounds on TRACE’s overall impact.
…
There are several welfare implications of increased transparency in the corporate bond market. One consequence is that it may change the relative bargaining positions of investors and dealers, allowing investors to obtain fairer prices at the expense of dealers. The reduction in price dispersion should allow investors and dealers to base their capital allocation and inventory holding decisions on more stable prices. Therefore, the reduction of price dispersion likely benefits customers and possibly, but not necessarily, dealers.

The implications of a reduction in trading activity are not as clear. Whether a reduction in trading activity is desirable depends on why market participants trade. A decrease in trading activity may be beneficial if much of the trading in a bond is unnecessary “noise” trading. On the other hand, if most trading is information‐based, a decrease in trading activity may slow down how quickly prices reflect new information. In addition, if the decrease in trading activity is the result of dealers’ unwillingness to hold inventory, transparency will have caused a reduction in the range of investing opportunities. That is, even if a decline in price dispersion reflects a decrease in transaction costs, the concomitant decrease in trading activity could reflect an increased cost of transacting due to the inability to complete trades.

Our results on the corporate bond market have two major implications for the current and planned expansions of mandated market transparency. The implicit assumption underlying the proposed TRACE extensions and the use of TRACE as a template for regulations such as Dodd‐Frank is that transparency is universally beneficial. First, it is not clear that transparency for all instruments is necessarily beneficial. Overall, trading in the corporate bond market is large and active, although, as seen, not comparable across all types of bonds. Many over‐the‐counter securities are similar to the bonds FINRA placed in Phase 3B. That is, they are infrequently traded, subject to dealer inventory availability, and trading in these securities is motivated by idiosyncratic, firm‐specific information. Therefore, the expansion of TRACE‐inspired regulations, such as those for 144a bonds, asset‐ and mortgage‐backed securities, and the swap market, may have adverse consequences on trading activity and may not, on net, be beneficial.

Second, our results indicate that transparency affects different segments of the same market in different ways. As a consequence, our results provide empirical support for the view that not every segment of each security market should be subject to the same degree of mandated transparency. Both academic commentators (French et al., (2010), Acharya et al. (2009)) and leading industry associations (e.g., Financial Services Forum, et al., (2011)) have articulated this position. Despite these recommendations, the expansion of transparency by the Commodity Futures Trading Commission (CFTC) in various swap markets, i.e. interest rate, credit index, equity, foreign exchange and commodities, in December 2012 and February 2013 was immediate for all swaps in those markets. This stands in sharp contrast to FINRA’s cautious implementation of TRACE in Phases. The fact that the effect of transparency varies significantly across categories of bonds within the corporate bond market suggests that additional research will be required to evaluate the tradeoffs associated universal transparency in other over‐the‐counter securities.

There is one assertion in the above with which I take particular issue: One consequence is that it may change the relative bargaining positions of investors and dealers, allowing investors to obtain fairer prices at the expense of dealers. Long time Assiduous Readers will probably be snickering to themselves, having determined that I am probably going to complain about the use of the word “fairer”, since I don’t know what “fair” means, and they’re quite right.

By “fair”, I assume the authors mean “at a price closer to the dealers’ cost than otherwise”, but that is not necessarily “fair” when examined in a broader context.

Suppose, for instance, that you are a bond dealer – horns, pitchfork, cloven hooves and all – and somebody asks you to bid on something. OK, so you do – but why do you? The answer, of course, is to make a profit and as a rational economic actor you seek to maximize your profit. But you’re not seeking to maximize your profit on every possible transaction or even to maximize your gross profit; you’re seeking to maximize the annual profit of your desk expressed as a fraction of your capital. This has a number of implications; for instance, you might give regular customers who deal exclusively with you slightly better prices than the other ones, simply to ensure that these guys never have any reason to consider going anywhere else.

But the most important consideration for purposes of this discussion is the question of maximizing profitability as a fraction of capital. That’s what determines the firm’s capital allocation and that’s what determines your bonus. And for a single given transaction, we can write the following equation:

Desirability = (Sell – Buy) / (Capital * Days)

Where the gross profitability is the Sell price less the Buy price, Capital is the amount of capital used when financing the position and Days is the number of days you have to hold the thing in inventory until it’s sold (or bought, if the position was initiated with a short sale). In this equation I am ignoring the Carry (the difference between the yield of the bond and the cost of financing it); I’m also ignoring default risk and lots of other considerations, with the objective of keeping this simple.

Under the pre-TRACE regime, one way to maximize trade desirability was, obviously, to maximize the difference between your Sell and Buy prices, but TRACE makes that a lot more difficult; after all, that’s the whole point of TRACE and Asquith, Covert and Pathak have made a solid argument that it is not happening to the same extent under TRACE as it was in the good old days. So for practical purposes, when the dealer is putting a price on taking a position, he is doing so with the knowledge that gross profit is capped.

The “Capital” term in the simplified equation is set by regulation and the bond desk has no control over it. As far as they’re concerned, it’s a constant.

Therefore, in order to increase the Desirability of the trade, the only avenue left open to the dealer is Days, which is inversely related. If they can make their $0.50 per bond profit in one day, that’s a whole lot better than if it takes a month! Therefore, when taking a position, they will concentrate their energies on how they will flatten their position. This will, of course, be much easier if they offer their position to a potential buyer at an attractive price. Therefore, I claim, TRACE will lead to the initial seller getting a really lousy price for his bond, which is turned over in short order to the ultimate buyer who gets a really good price.

There is evidence for this in the secondary GIC market, which has to be one of the most ridiculously infinitesimal markets in the world, but which exists at the major dealers not so much as a money-maker, but as a service to clients. Some GICs are transferrable and the dealer will buy them from the owner at a really, really lousy price – I think the bid yield is about 150bp over the market rate, but I confess I’m not too sure of that. I have a major dealer’s offering sheet from 2012 on hand, which is headed by the statement: “ALL Secondary GICS offered at approximately +50bps over today’s Best GIC Rates (on the [Redacted] System”

This is a great deal for buyers, and I have often recommended to clients that they open accounts at a major dealer for the purpose of access to new issues and access to secondary GICs. And why I am I saying this? Because I think the buyer will get a “fair” price, just like teacher talked about in kindergarten? Hell, no! It’s because I think the buyer will get a really good price, courtesy of the really, really shitty price that was offered to the poor sucker who needed to cash his GIC early.

Now this example comes from a market that barely exists, but I claim that it shows in sharp relief the problem with TRACE – which is that it encourages prices that are not “fair”, but prices that really stick it to the liquidity seeker in order to reward both the interim and the ultimate liquidity supplier.

And I will claim that this cannot be considered a Good Thing. This is an increase in the cost of liquidity, which leads to a decline in liquidity, which leads to an increase in the liquidity premium demanded for holding a position, which leads to higher coupons required from the issuer at issue time. And I claim that this is a Bad Thing because the purpose of the corporate bond market is to allow issuers to source cheap capital.

Note that none of these assertions has been tested, but for now we’ll call it the Shitty Price Hypothesis. It has the advantage of actually providing a causal mechanism for the reduction of trading experienced under TRACE: say you’re a portfolio manager and there’s a wave of redemptions. You have to raise cash. In the old days, you could utilize the opportunity to rebalance and improve your portfolio slightly. Got too much junk in the portfolio? Fine, raise the cash by selling it. But if all you see is stink-bids, you’re almost forced to move up the credit quality ladder and sell something more liquid. Thus, TRACE has made it more difficult for you to do your job.

To be fair, the authors make what might be an indirect allusion to this at the end of their Section 6:

In addition, the bond market is a dealer market, so dealer inventory will affect trading levels and the potential impacts of TRACE. Dealers only hold inventory in those bonds with sufficient trading activity to cover their carry cost. Thinly traded bonds may require dealers to have higher spreads to cover their holding costs. Since TRACE reduces price dispersion significantly, the benefit of holding bonds in inventory decreases. TRACE reduces price dispersion the most for high‐yield bonds, so the incentive to reduce inventory is strongest for those bonds. Thus, lower trading activity in high‐yield bonds post‐TRACE may be the result of a supply‐side response of dealers.

Another paper I found while updating myself on academic commentary about TRACE was by Rainer Jankowitsch, Amrut J. Nashikkar and Marti G. Subrahmanyam, titled Price Dispersion in OTC Markets: A New Measure of Liquidity:

In this paper, we model price dispersion effects in over-the-counter (OTC) markets to show that in the presence of inventory risk for dealers and search costs for investors, traded prices may deviate from the expected market valuation of an asset. We interpret this deviation as a liquidity effect and develop a new liquidity measure quantifying the price dispersion in the context of the US corporate bond market. This market offers a unique opportunity to study liquidity effects since, from October 2004 onwards, all OTC transactions in this market have to be reported to a common database known as the Trade Reporting and Compliance Engine (TRACE). Furthermore, market-wide average price quotes are available from Markit Group Limited, a financial information provider. Thus, it is possible, for the first time, to directly observe deviations between transaction prices and the expected market valuation of securities. We quantify and analyze our new liquidity measure for this market and find significant price dispersion effects that cannot be simply captured by bid-ask spreads. We show that our new measure is indeed related to liquidity by regressing it on commonly-used liquidity proxies and find a strong relation between our proposed liquidity measure and bond characteristics, as well as trading activity variables. Furthermore, we evaluate the reliability of end-of-day marks that traders use to value their positions. Our evidence suggests that the price deviations are significantly larger and more volatile than previously assumed. Overall, the results presented here improve our understanding of the drivers of liquidity and are important for many applications in OTC markets, in general.
…
Using a volume-weighted hit-rate analysis, we find that only 51.12% of the TRACE prices and 58.59% of the Markit quotations lie within the bid and ask range quoted on Bloomberg. These numbers are far smaller than previously assumed. Since these marks are widely used in the financial services industry, our findings may be of interest to financial institutions and their regulators.

The evidence that so many actual prices are outside the pre-trade quote is supportive of the Shitty Price Hypothesis, but more detail is needed!

And now 144a (exempt) bonds are being TRACEd:

Corporate-bond brokers may face a squeeze on profits as regulators start publishing prices for almost $1 trillion of privately sold debt, if the past is any guide.

The Financial Industry Regulatory Authority, seeking to “foster more competitive pricing,” plans to start disseminating trading levels for securities issued under a rule known as 144a on its 11-year-old Trace system within the next year. That means the notes, sold only to institutional investors, will face the same price transparency as publicly registered corporate bonds for which buyers demand half a percentage point less in yield spreads. Brokers typically are paid larger fees from higher-yielding debt.

Firms from Knight Capital Group Inc. to Gleacher & Co. and Pierpont Securities LLC sold or shuttered credit units this year as corporate-bond trading volumes fell to the lowest proportion of the market on record and smaller price swings shrink potential profit margins.
…
Stamford, Connecticut-based Pierpont, one of the dealers started after the 2008 collapse of Lehman Brothers Holdings Inc. decided to exit the high-yield bond and loan business this month. New York-based Gleacher said in April that it was exiting fixed-income trading and sales. Knight in Jersey City, New Jersey, sold its credit-brokerage unit to Stifel Financial Corp., according to a July 1 statement.

Jefferies Group LLC, the investment bank owned by Leucadia National Corp., said profit plunged 83 percent in the three months ended Aug. 31 as trading revenue fell to the lowest since the depths of the financial crisis.