The above quote comparing liquidity to leverage comes from Goldman Sachs’ head of Global Credit Strategy, Charles Himmelberg. Historically, leverage “is the tinder that turns a financial fire into an inferno,” as The Financial Times put it recently. However, since February’s flash crash, Himmelberg has again and again sounded the alarm that the algorithmic transformation of markets means liquidity, not leverage, should be the preeminent, catalytic concern as quantitative tightening progresses and volatility returns. “I routinely field questions from clients asking where the risks are building up, and this is the one I worry about,” he told The FT earlier this month. “Financial markets have changed pretty dramatically since the crisis.”

In these pages, we have sought to understand the implications of the algorithmic and passive revolution, one of the most profound changes to the global financial system in history. And we keep coming back to liquidity as the key threat—if a market shock causes algorithmic strategies to simultaneously unwind or the passive herd to flee, supply could rapidly overwhelm demand, causing widespread panic. Now, mounting evidence suggests the threat is intensifying. Speaking to The Wall Street Journal, Jeffrey Cleveland, chief economist at Payden & Rygel, put current market liquidity in stark terms: “It’s like going into a grocery store and there’s nothing on the shelves.”

Whether inflation, interest rates, margin pressure or a domestic or global political crisis, it is not yet clear when a market shock will come and what will be the catalyst. However, the more threats mount, the more essential it becomes to understand how an ever-illiquid market might react under stress. Based on all available evidence, pain will come far faster and with far greater severity than most anticipate.

Post-crisis bank regulation is at the heart of the liquidity problem. As The FT documented recently:

Since the financial crisis, regulatory changes have aimed to purge leverage from the system, primarily by severely curtailing the role that banks have traditionally played in providing liquidity…Instead of relying on balance sheets to lubricate trading, banks have become more like brokers and embraced the algorithmic techniques of high-frequency traders…On the whole this has made markets more efficient, with the difference between the price investors have to pay or will receive for a financial asset—the “bid-ask spread”—narrowing to record lows. By this measure even the bond market today looks more “liquid”. But some analysts and investors said markets that may seem superficially more efficient were more prone to “liquidity crises”, where waves of sellers suddenly overwhelmed the depleted capacity of market-makers such as banks to absorb and intermediate the buy and sell orders.

According to Mauldin Economics, Dodd-Frank requirements have reduced major bank market-making abilities by as much as 90%. In more anecdotal terms, Goldman Sachs’ president David Solomon told The FT that the bank used to have 500 people dedicated to making markets in equities and now has only three. A similar dynamic is apparent in the bond market. In a paper released in October, “Customer Liquidity Provision: Implications for Corporate Bond Transaction Costs", the Federal Reserve Board concluded: “Among trades where customers are demanding liquidity”where dealers aren't just matching customers, but actually providing immediacy themselves“we find that these customers pay 35% to 50% higher spreads than before the crisis.”

Post-crisis regulation has no doubt made “too big to fail” banks less vulnerable to violent market swings. According to a recent study of nine big European and U.S. investment banks, hedge fund Capstone concluded that aggregate “value at risk”—a rough measure of how much the banks can expect to lose on any given day in markets—had fallen from a peak of roughly $1.9 billion in 2008 to $377 million at the end of 2017. The problem is, regulation didn’t eliminate the risk, it just transfered it from big banks to remaining market participants. Meanwhile, big banks are not there to serve as the backstop in the event of a spooked market, providing liquidity by buying into dips.

Compounding the threat, algorithms have seized unprecedented control since the financial crisis (chart on left), which has made markets more prone to flash-crashes. In recent years, this has been apparent in even the biggest, most liquid markets (chart on right):

Source: Goldman Sachs

In a recent note, J.P. Morgan’s quant guru Marko Kolanovic—who has repeatedly warned of a “Great Liquidity Crisis”—encapsulated the dynamic he fears:

A host of systematic strategies have the capacity to exacerbate moves by reinforcing price trends. As volatility increases, selling from such strategies tends to result in one-way flows in index products such as futures and ETFs…Those declines in turn cause market liquidity to dry up…There are fewer traditional players willing and able to step in to slow one-way moves. Value investors, for instance, have been hurt by more than a decade of underperformance, and may have less capacity to mitigate a selloff.

To support his concern, Kolanovic points to the fact that futures-market depth plummeted more than 90% during February’s flash-crash. A recent report by the Bank of England provides further evidence. It dissects the 2015 shock caused by the Swiss National Bank scrapping a foreign exchange cap and the liquidity crisis that resulted. The report concludes:

The Swiss franc event is probably the most significant shock to FX markets since computerised algorithmic trading has been prominent…Studying the reaction to this shock, we find that algorithmic trading contributed to the decline of EUR/CHF and USD/CHF market quality on the event day and afterwards as they withdrew liquidity and generated uninformative volatility.

Since the financial crisis, QE mitigated the threat illiquid markets could pose, providing a funnel of indiscriminate capital, thus suppressing volatility. Capital provided by hedge funds and other non-bank buyers and lenders did not retreat from markets. And the mom-and-pop passive herd was not spooked into panic. Flash-crashes happened, but they remained largely contained. However, as quantitative tightening progresses and volatility returns to markets, the frequency, severity, and contagion-potential of flash-crashes may increase, with February a clear warning shot.

Adding to the threat, volatility is already deepening the liquidity problem. As The Daily Dirtnap’s Jared Dillian points out: “There is an axiom in markets: volatility and liquidity are inversely correlated.” Since February, liquidity decreases have been reported in everything from S&P 500 ETFs and small-cap stock trades to the world’s deepest bond market, U.S. government debt.

Of course, which market faces a liquidity crisis and when depends on the timing and nature of an external shock. We will continue to track and dissect threats as they materialize. For now, Himmelberg offered the key, top-line perspective in his most recent note, released this week:

[High-frequency trading] knows the price of everything and the value of nothing…Their informational advantage over human trades under normal conditions has allowed them to grow to become the dominant liquidity providers in all of the largest, most liquid markets. In our view, this at least raises the risk that as machines have replaced people, and speed has replaced capital, the inability of the market’s liquidity providers to process complex information may lead to surprisingly large drops in liquidity when the next crisis hits.