The fairytale of market liquidity

Repeated need for support from central banks at times of crisis highlights vulnerability of financial systems

John Dizard 

  Market crises are becoming more frequent, even though their effects on asset liquidity dissipate more quickly © AP


Market liquidity is the fairy gold of investors, trading desks and financial regulators. It is desirable, apparently essential to realising dreams of security, yet crumbles when attempts are made to grasp or even measure it. 

During periods of market stress, the ability to buy or sell assets without affecting their price becomes even more the stuff of a folk tale.

Customers and voters would be outraged if they knew the fragility of their power to turn investments into cash. But the evanescence of market liquidity is not a secret. 

There is now a very deep literature on the subject in central bank reports and financial economic journals.

As one very good 2019 study from the Banco de España puts it: “Market liquidity is not easy to measure. In fact, it is an unobservable variable that embodies several heterogeneous characteristics.”

Is cash on demand is an “unobservable variable”? This is not what the public wants to hear.

Their midnight worries push regulators into action. Esma, the pan-European securities regulator, issued a 21-page document in mid-July, “Guidelines on Liquidity Stress Testing in Ucits and AIFs”, which has applied to investment products sold in the EU since September 30. This had been carefully negotiated with the better informed and capitalised investment managers so as to be firm, but realistic, unyielding, yet flexible.

The problem is that market crises are becoming more frequent in recent years, even though their effects on asset liquidity dissipate more quickly. Each crisis requires a higher level of support from state treasuries and central banks, which in turn requires more issuance of “risk-free” assets. 

Fortunately, crisis-spawned regulations require the purchase of that same risk-free paper that can be turned into cash through repurchase agreements, or used as collateral for trading riskier assets.

Not quite a perpetual motion machine, this process requires careful oiling and maintenance, as well as dedicated people and resources. Because even the minimum liquidity management and related compliance staff are expensive, greater systemic fragility is increasing the market share of large asset managers.

Since the major asset managers cover the range of liquid fixed-income markets, as well as equities and alternative assets, a natural question would be whether they can, in effect, become their own markets by providing high-quality liquid assets to their portfolio managers to repo for cash or to meet collateral calls and redemptions, even to take advantage of crisis-created mispricings.

Vincent Mortier, deputy chief investment officer of Amundi, the €1.6tn French asset manager, says: “That is something we have explored, but so far we have always been acting as an agency, not as a principal. We are thinking maybe to be a principal, or directly to take on some repos. So far, all we have is a model, but it is something we are thinking about. Some [American] asset managers are doing that. It would be a game changer.”

Lending collateral from one group of customers to another would be cheaper and less tedious than putting bids on the screen or calling around the market, but it probably creates more political vulnerability.

In any event, as Mortier relates: “We had some discussions with the ECB but their first priority is to monitor the banks and not really the markets. To be fair our [liquidity] model was not ready then. We only developed it in June and did back-testing in July. It turned out that on a few really extreme days in March it was much better than other sources.”

By “other sources” Mortier means liquidity algos such as those on Bloomberg or Reuters. “Those do not take into account voice orders which are usually used for bigger trades or in crises. Especially for credit.”

The political problem here is that some of the most innovative and useful models to guide those voice traders are based on machine learning. That is very cool, as long as customers and regulators do not blame your gadget for an asset crash.

Then you might have to explain how a series of bid/ask spreads have so much less information than a set of ranked eigenvalues. Teaching math to people who have lost retirement savings is, I believe, even harder than sales. 

0 comentarios:

Publicar un comentario