Revisiting Market Liquidity: The Case of U.S. Corporate Bonds

“…a small reduction in liquidity from regulatory changes—even if present, which is not obvious—may be a reasonable price to pay for greater safety.” FRB Vice Chair Stanley Fischer Speech, November 15, 2016

Prior to the financial crisis of 2007-2009, many people took market liquidity for granted. So, when the ability to convert assets into cash eroded, the issue became one of survival for some intermediaries. Today, both investors and regulators are focusing on “the ability to rapidly execute sizable securities transactions at a low cost and with a limited price impact” (see Fischer). And there has been an intense debate about whether post-crisis regulations themselves have diminished the supply of liquidity (see our earlier post).

That debate is far from resolved. Many market participants attribute a perceived loss of liquidity to Dodd-Frank, Basel III, or other regulatory changes. Yet, as we discuss below, studies looking at traditional price measures of liquidity (such as bid-ask spreads) find it difficult to document the loss of liquidity in U.S. corporate bond markets. It is even more difficult to link changes in liquidity directly to any of the numerous regulatory changes. Importantly, liquidity concerns have not prevented a record pace of issuance by nonfinancial firms in recent years (see chart).

However, new research by Anderson and Stulz (AS) finds that there has been a decline in corporate bond market liquidity during episodes of heightened systemic risk. That is precisely when market liquidity should matter most to financial regulators. Anderson and Stulz also document the importance of non-price indicators—specifically turnover—as evidence of a persistent loss of market liquidity.

The remainder of this post describes recent work on corporate bond liquidity with a focus on systemic risks.

Net corporate bond issuance by nonfinancial firms (Billions of dollars at an annual rate), 1990-2016

Note: Four-quarter moving average. Source: Federal Reserve Board Financial Accounts of the United States.

Note: Four-quarter moving average. Source: Federal Reserve Board Financial Accounts of the United States.

Let’s start by keeping in mind that, compared to other U.S. fixed-income markets, liquidity in the U.S. corporate bond market has always been low. According to SIFMA, since 2002 (when data became available), daily turnover in the markets for U.S. Treasury, mortgage-related and municipal debt has averaged 23.6, 8.7 and 1.3 times larger, respectively, than in corporate debt. The principal reason for this low turnover is the fragmentation of the corporate market. Some leading borrowers have hundreds of different bonds outstanding, many of which rarely trade after they are underwritten (see Exhibit 4 in BlackRock Viewpoint Addressing Market Liquidity). In addition, there is no centralized limit order book for corporate bonds: while electronic quotes are increasingly available (see Sokobin), corporate bonds are primarily traded over-the-counter; and there are many “trade throughs” ― transactions that are not executed at the best available electronic price (see Harris).

Has the chronic weakness of liquidity in the corporate bond market deteriorated further in recent years? Several studies suggest that the answer is “no”. For example, using corporate trades reported through FINRA’s TRACE system, Adrian et al show that: (1) bid-ask spreads have narrowed; (2) the price impact of trades has fallen, even though average trade size has not fully recovered following the crisis; and (3) trading volume has risen (see chart). While the smaller size of trades could be interpreted as a loss of market depth, the declining price impact of trades suggests the opposite. Trebbi and Xiao conclude that recent episodes of regulatory intervention are associated with upward, rather than downward, shifts in market liquidity. Similarly, Anderson and Stulz find that, compared to the pre-crisis period of 2004-2006, price-based measures of liquidity (such as the price impact of trades) improved as regulatory changes were being implemented in 2013-2014.

Average daily trading volume in U.S. corporate bonds (Billions of dollars), 2005-March 2017

Source: SIFMA based on FINRA TRACE data.

Source: SIFMA based on FINRA TRACE data.

Yet, price measures may not capture the loss of market depth that worries market participants. As Anderson and Stulz note, trades that do not take place due to regulatory obstacles won’t be reflected in bid-ask spreads or in measures of the price pressure that results from trades. Importantly, while trading volume has risen in recent years, it has not kept pace with issuance. As a result, corporate bond turnover―the fraction of outstanding bonds that trade each day―remains somewhat lower than it was prior to the crisis (see chart). From the perspective of a portfolio manager holding a fixed portion of the outstandings, reduced bond turnover signals the possibility that selling has become more difficult.

Average daily turnover of corporate bonds (Percent of outstandings), 2002-2016

Source:  SIFMA and authors’ calculations.

Source:  SIFMA and authors’ calculations.

At the aggregate level, this turnover pattern in corporate bonds does not appear closely linked with the timing of regulatory changes (Dodd-Frank and Basel III were both approved in 2010 and implemented gradually over succeeding years). However, in their sample, Anderson and Stulz find that turnover was lower in 2013-14 than in 2010-2012, which was down substantially from 2004-2006. Moreover, this decline in corporate bond turnover contrasts with an increase of equity turnover over the same interval.

In theory, a decline in turnover could be benign. To the extent that higher pre-crisis turnover was associated with a period of overabundant liquidity (possibly due to costly and inefficient portfolio churning), a decline reflects a normalization of conditions. Other explanations that are of little concern include structural changes like the increase in the role of buy-and-hold investors, the temporary impact of record low interest rates on issuance, and the growing reliance on index strategies (see, for example, BlackRock).

That said, we are concerned by the apparent increased sensitivity of market liquidity to systemic risk. Containing systemic risk is the paramount objective of post-crisis regulation. To the extent that liquidity is more likely to evaporate during periods of stress, the system would be more, not less, fragile. It is for this reason that we find the Anderson and Stulz results worrying. Using spikes in the VIX—a market gauge of implied volatility in options on the S&P500 index (see our earlier VIX primer)—as a proxy for systemic risk, they show that recent jumps in the VIX have been associated with substantially greater increases in price pressure from trading than comparable shifts a decade ago. Of course, there is no perfect proxy for systemic risk: conceivably, the crisis itself conditioned investors to respond to the VIX rather than to the risks that it is presumed to reflect. However, we do not know of a better candidate for this experiment.

Again, these results are difficult to link to any specific regulatory change, much less an evaluation of whether these costs exceed the benefits. If, as the initial quote from FRB Vice Chair Fischer hints, higher capital requirements—risk-weighted or not—have led large banks to scale back market making, that may be a price worth paying to achieve a safer financial system. It also may be possible to counter an adverse impact of higher capital requirements on market liquidity through a more effective set of rules (see, for example, Duffie, regarding the effect of the pure leverage requirement on the trading of safe assets). And, as we argued before, resilience of intermediaries and resilience of markets are mutually reinforcing: healthy, well-capitalized institutions are more capable and more willing to make markets in periods of generalized financial stress.

Other factors also could offset any negative impact of higher capital requirements on market liquidity. For example, technological changes—like the advent of “all-to-all” electronic platforms that allow institutional investors to bypass market makers—and the willingness of large asset managers to supply liquidity directly may reduce the demand for large banks to make markets in corporate bonds. And, as we noted earlier, a willingness of big issuers to reduce the sheer number of bonds outstanding would lessen market fragmentation, and could give a powerful boost to market liquidity.

Nevertheless, we hope that evidence linking systemic risk and a loss of market liquidity pushes authorities to review and repair those post-crisis regulations that impede trading but do little to counter systemic risk. High on our list would be the overwhelmingly complex Volcker Rule which, as Richardson and Tuckman recently argued, imposes large trade-by-trade compliance costs, yet treats economically identical risk-taking in completely different ways (say, by allowing a bank to make a loan to a firm, but not to hold on its account that same firm’s bond). As former Fed Governor Dan Tarullo suggested in his recent valedictory address, this seems ripe for change.

Acknowledgment: We thank our friend and Stern School colleague, Bruce Tuckman, for very helpful suggestions.