As we noted previously, for the first time ever, primary dealers' corporate bond inventories have turned unprecedentedly negative. While in the short-term Goldman believes this inventory drawdown is probably a by-product of strong customer demand, they are far more cautious longer-term, warning that the "usual suspects" are not sufficient to account for the striking magnitude of inventory declines... and are increasingly of the view that "the tide is going out" on corporate bond market liquidity implying wider spreads and thus higher costs of funding to compensate for the reduction is risk-taking capacity.
As Goldman Sachs' Charlie Himmelberg notes,
Following several weeks of sharp declines, dealer inventories of long-term IG bonds and HY bonds have turned negative for the first time since the NY Fed started reporting corporates separate from non-agency mortgage MBS. While inventory levels are often negative for Treasuries, this is the first time in the available data that corporate bond inventories have ever turned negative (detailed data on corporates go back only a few years, but broader aggregates of corporates plus non-agency MBS have never gotten remotely close to zero).
To some degree, the inventory drawdown is probably a by-product of strong customer demand. The rally in credit spreads since Oct. 2 was helped in part by a resumption of mutual fund inflows (especially in HY) at a time when dealer inventories were already substantially depleted. In addition, HY new issue volumes – which drive dealer inventory holdings because they drive secondary trading volume -- have cooled somewhat over the past month as wider spread levels helped to deter issuers.
But these “usual suspects” are not sufficient to account for the striking magnitude of inventory declines. Exhibit 1 below shows how weekly dealer inventories for investment grade bonds have evolved over the past 2.5 years, and compares this to the changes implied by a simple model that estimates the effects of new issue volumes and mutual fund inflows (both ETFs and open-end funds) estimated on data through June 1, 2015. The plot shows that model-implied inventory levels have, in fact, been declining since early August. But, in fact, as the chart also shows, the model gives a very poor fit over the full sample period (R2=14%), and an analogous model for HY bonds is even worse (R2=9%). Whatever is driving the decline in corporate bond inventories, it is not visible in the (observable) changes in market conditions used here.
Implications for credit market liquidity and credit risk appetite
Negative inventory levels mark an impressive milestone for the trending declines in dealer inventory holdings of corporate bonds over the post-crisis period. Despite our failure to model them above, we feel fairly confident that the recent declines are the result of transitory market factors. But we are far less confident that the long-run trend declines of inventory levels and corporate bond liquidity are temporary. On the contrary, the more we study the problem, the more we are convinced that low market liquidity is the “new normal” for corporate bond markets.
In our view, the trend declines in dealer inventories reflect the rising cost of hedging and holding corporate bond positions. For example, when single-name CDS were more liquid – and they were far more liquid up until just the last year or two – dealers could bid aggressively for bonds and use liquid CDS markets to quickly hedge that risk until buyers on the other side of the market could be found (a search that for roughly half of corporate bonds can easily take days or weeks).
Such hedges have become progressively harder in the post-crisis period as the liquidity available in single-name CDS has steadily declined (the cause for which requires separate treatment). And while substitute hedging instruments like equity puts have remained more liquid, they risk falling into the greyer areas of the Volker Rule. Finally, almost any such hedge now carries considerably higher capital charges due to Basel 3, stress testing, etc.
The result has been a substitution away from increasingly costly “principal” trades toward “riskless principal” trades (i.e., trades for which buyers and sellers have already been located on both sides, and therefore require much less risk taking and less inventory by dealers, but also offer much less immediacy for buyers and sellers in the market, and therefore command much less bid-ask spread. Under this logic, declines in dealer inventories less causal than symptomatic.
There is still a large gap between this understanding of how the market is evolving and the narrative proposed by non-market participants. For example, a recent blog series on the website of the Federal Reserve Bank of New York notes the same reduction of corporate bond inventories, but concludes that liquidity is fine since bid-ask spreads have narrowed. For the reasons explained above, we find this argument unconvincing.
Instead, we are increasingly of the view that “the tide is going out” on corporate bond market liquidity. The cost of intermediating secondary markets will only continue to rise as trading activity in single-name CDS goes lower. The cost of principal trades will also rise further as new systems for managing the regulatory capital required for “stress tests” and new capital charges are phased in. As these higher costs are passed on to end users, we expect they will continue to sacrifice immediacy in favor of trades conducted on “riskless principal” basis.
This loss of immediacy implies a loss of bond market liquidity by definition. And since market liquidity allows for risk management, lower liquidity implies lower risk-bearing capacity of the investor base. We would argue this is already evident in the spread widening of the past year, and that spreads would have widened less had they not been amplified by liquidity concerns.