In this article I am taking a brief break from writing about inflation. There have been lots of great stories and anecdotes recently about inflation. I loved the Wall Street Journal article about how “Inflation and Other Woes Are Eating Your Girl Scout Cookies”, and we have seen several contributions from former Treasury Secretary Larry Summers that are worth reading. One was an opinion piece in the Washington Post (“Opinion: The stock market liked the Fed’s plan to raise interest rates. It’s wrong.”) and one was a very good NBER working paper on “The Coming Rise in Residential Inflation,” in which he confirms and extends the normal way inflation people forecast rents and comes up with even higher numbers than I’ve been working with for a while. Incidentally, if you haven’t seen these stories before now, consider installing the Inflation Guy mobile app. I don’t curate every single inflation story; just the ones worth curating.
Moreover, the Fed increasingly sounds like they want to be aggressive with rates. That’s half the battle, though on the more important half (the balance sheet reduction) they don’t yet have a plan. I should note that saying hawkish things on half the plan isn’t really all that hawkish, especially when your notion of “pushing rates above neutral” means 3%: a level well below inflation. But it’s progress that these folks have finally realized that inflation is a real phenomenon and not just due to port congestion. They still don’t seem to see the role of money growth in causing that phenomenon, but it’s nice we’re making baby steps.
As I said, though, in this article I’m going to talk about market structure, and the deal with the devil we have made to seek ever-tighter-spreads at a cost of orderly markets.
Since the 1970s, the cost of trading equities has moved from a bid/offer spread of a half point or a quarter point ($0.50 or $0.25 per share), on round lots, plus large brokerage fees, to sub-penny spreads on any size trade, often with zero brokerage costs. The cost of bond execution has similarly declined, as has the cost of futures and swaps brokerage.
Volumes, across all markets, have responded to the decrease in costs. Some of this improvement in the median cost of trading has come from increased transparency, and a lot from increased competition.
Those improvements have not come without a cost, but at most times the cost is less apparent. The way the stock market used to be structured was around a number of market-making firms whose job it was to maintain orderly markets – including the distasteful task of being the buyer when everyone else is selling.
What this means for the profit of a market-maker is that they generally made steady, small profits (a quarter of a point on every share, day in and day out) and occasionally lost huge amounts in market panics. It’s a classic “short gamma” position of picking up nickels before the bulldozer, and well-understood by the market-makers to be so.
But that was the deal: you let the market-maker take his spread as an insurance premium, and collect on that premium when a calamity hits. Primary dealers in the government bond markets worked the same way: in exchange for the privilege of building an auction book (and being able to bid on the auction with that knowledge) and making spreads as a market-maker most of the time, it was understood that they were supposed to work to keep markets liquid in the bad times.
Then, we decided that we didn’t like paying all of these insurance premiums, which we called the “cost of trading” but could also be considered “the cost of providing continuous liquidity in bad times.” So stock prices were decimalized, which immediately started narrowing spreads. Electronic trading made the deal even worse because anyone could jump in front of the market-maker and be the bid or the offer, meaning that the market-maker wasn’t earning the spread. In many cases, there wasn’t any spread left to earn.
There is a parallel to something else I’ve written about recently, and that’s the trend over the years to lower and lower costs, and longer and longer supply chains, in manufacturing. Such a system is lower cost, but the price of that cost-savings is fragility. A long, international supply chain gets snarled much more easily and much worse than a short, domestic one. That cost/fragility tradeoff is the bargain that manufacturers made, although not thoughtfully.
Similarly, the price of the cost-savings from sub-penny equity spreads is fragility in the market-making system. It is difficult to find dealers who will accept the responsibilities of being the buyer or seller of last resort, and maintaining orderly markets, when that cost is not counterbalanced by an increase in profit opportunities during placid times.
As with international trade, we have begun to see the downside of this tradeoff when trading risks increase. Not that this is the first time, but it seems these days that liquidity conditions get sketchier more quickly now than they used to.
Of course, we saw this as recently as March 2020, when trading in credit got so bad that the Fed had to step in and backstop corporate bond ETFs by buying corporate bonds and ETFs under the Secondary Market Corporate Credit Facility.[1] Recently, the Nickel market basically broke when prices went vertical and the resulting margin calls would have put some LME brokers out of business (conveniently, the LME decided to just cancel the trades that they didn’t like, which means those brokers are still in business but probably won’t have a market to broker).
Prices went vertical partly because there are fewer highly-capitalized market-maker shops to stand in the middle and make orderly markets. Also recently, the European Federation of Energy Traders pleaded for “emergency funding mechanisms” so that they can continue to trade energy markets that have had greatly increased volatility recently.[2]
Now, the disturbing thing is that we are starting to see declines in liquidity even in fairly unremarkable periods. The last seven months’ worth of volatility in interest rate markets was higher than we’d seen in some years, but not exactly unprecedented. This month, 10-year Treasury yields are up 57bps. In 2002, 10-year yields fell 170 bps between May and October, in something close to a straight line driven by mortgage convexity. In about 6 weeks from May to June in 2003, yields dropped 81bps and then immediately reversed 129bps higher over the ensuing 6 weeks (same reason, different direction). I mention those two episodes because I was making markets in rates options and remember them not-very-fondly.
But these recent 57bps have been a lot more stressful on the market with fewer strong hands responsible for maintaining order. The chart below shows the BofA MOVE index, which measures normalized implied volatility on 1-Month Treasury options.
Recently, that index reached its second-highest level since the Global Financial Crisis. The highest prior level was in the March 2020 shutdown crash…understandable… and during the GFC banks were undercapitalized and in risk of failure. What’s the reason now?
We also see it in various market ecosystems. For example, there are roughly two dozen “Lead Market Makers” in the ETF ecosystem. In order to launch an ETF, you need to find someone to be the LMM.
The function of the LMM is to make markets in virtually all conditions. But it is exquisitely hard to get an LMM signed up nowadays because the math for them works out badly. If your fund is very small, they make a decent spread but on tiny volume so it’s not very lucrative.
As soon as your fund gets large, everyone else jumps in front of your markets, because they can and there’s money there to be made, so the LMM either makes no spread at all or makes a very small spread. Of course, those other Johnny-come-latelies will scatter the first time there is volatility, leaving the LMM there all alone to make orderly markets.
So the market-making itself is a bad deal for the LMM in almost all circumstances. Their models are only tenable if they are able to make money on the relationship with the ETF issuer in other ways – being a broker for fund rebalancing, etc. This means that fewer good ETFs come to market than otherwise would. I have lamented this elsewhere. And the root cause and ultimate result are the same: we’ve engineered a very low-cost, high fragility system for investors to deal in.
The bottom line is that as any insurance agent can tell you people really hate paying for insurance. But no one expects insurance companies to provide insurance without being paid at least a fair premium. What would happen if we did? Well, then we wouldn’t have any insurance. Financially speaking right now, we don’t have much insurance because it’s too costly to stand in the middle. That looks like a win, until something catches on fire.
[1] For the Fed to buy corporate bonds was long held to be impermissible, since the Federal Reserve Act listed the assets the Fed was authorized to buy and that list did not include corporates and equities. Clearly, this was meant to follow Bagehot’s dictum that a central bank, to avert panic, “should lend early and freely, to solvent firms, against good collateral, and at ‘high rates’”, but thanks to clever lawyers who note that the Act does not explicitly prohibit the Fed from buying these things the Fed has in recent years decided that since it wants to, what could go wrong?
[2] A sad aside is that the movement to remove “pricey, greedy market-makers” and replace them with bailouts provided by central bank or treasury is the opposite of what Dodd-Frank was supposedly trying to do in ensuring that systemically-important institutions were adequately capitalized. They’re adequately capitalized now, but they don’t provide the market-braking function they used to because that’s ‘speculative activity’ that penalizes capital severely.