MARKETS: Spread of Spreads

The Data

A closer inspection of the data confirms that broad impression. The tables below summarise the data. The rise in the spreads’ average OI share has strong statistical support. Investors’ spreads share of OI, in most markets, from 2018 averaged around twice the pre-2018 average. And, the averages since 2018 are around the same as the maximums before 2018.

Some variation between markets is also evident. The major divergences are in the wheat trio. The investor spread share for SRW wheat (Chicago) and HRW wheat (Kansas) is larger than the other markets. And HRS wheat (Minneapolis) is smaller than the other markets. The HRS divergence likely signals something about the investors. Mandates constrain many spread investors to markets in major commodity indexes (e.g. S&P GSCI) that do not include Minneapolis futures.

One obvious question is whether more spread positions mean less outright positions. Are investors, in aggregate, taking spread positions instead of outright positions? If so, in the limit, OI will not grow at all if spread positions simply replace outright positions. The question cannot be answered directly with publicly available data from the CFTC. In general, visual observation reveals no obvious drop. No consistent correlation between investors’ long or short positions and their spread positions. Low correlation, of course, does not mean there is no outright to-spread displacement. But it does suggest the displacement is not large. We’d conjecture that, for many investors, outright and spread strategies co-exist. And investors will favour either strategy based on the opportunities they perceive in markets.

---

The Pattern

Looking forward, is more investor spread trading now a permanent feature for these markets? Or will the recent experience prove to be episodic? We conjecture that the outcome will likely be a bit of both. The permanent part is that the average proportion will be higher. The episodic part is that spread participation will range around that higher average.

The experience so far has been swings around that higher average. There are persistent reasons why we will likely see such swings. One reason is that investors’ participation, spread or otherwise, depends on allocations to commodities. When money leaves commodities for other asset classes – equities, bonds, etc. – then investor participation will decline. Another reason is that a material (at least) proportion of investors are discretionary. Thus many investors can, based on their return expectations, choose whether to hold spread positions.   

---

So Then What?

If investor spread activity is now a permanently larger factor in these markets, what are the implications? How do these spread positions, and the associated flow, impact prices and other traders? 

Partial

Spread positions will have little directional impact if transacted actually or nearly simultaneously. The flows do have the potential to impact spreads between contracts if they are largely one-way. If the investor flow is largely short-near/long-far, then that creates pressure for spreads to widen. That ‘pressure’ though is a partial, or first-step, assessment. The final impact will depend on how other traders respond. Or, in other words, the market’s reflexivity.

Counterparties

Physical traders (the CFTC’s “producer/merchant/processor/user” category) are likely the main source of the reflexivity. The predominant flow in these markets is between physical and investor categories. The CFTC does not publish separate spread positions for physical traders, only long and short position totals. The market though has long ‘understood’ that physical traders run significant spread positions. Merchants in particular. And those spread positions are designed to lock in the ‘carry’ in commodity prices. So, physical spread positions tend to be short-near/long-far.    

A 2019 CFTC study^1,2 supported that understanding. The study provided some evidence on the extent of physical traders’ spread positions. The study covered the 2015-18 period. And had access to unpublished data. So, the authors were able to estimate physical traders’ separate spread and outright positions. The resulting data showed that physical traders’ spread positions were around 10% of OI during 2015-18. That proportion meant that total spread positions were around 34% of OI. Much larger than the 22% accounted for by swap, investor and other traders. The study covered four markets: corn, soybeans, HRW wheat, and SRW wheat. For the first three, the physical share of spread OI varied between 20-40%, and spent much of the time at higher proportions. The share in SRW wheat was lower but that reflected other factors.

Complementarity

Bringing the investor and physical sides together, we suspect, produces a complementarity. The physical traders tend to buy-near/sell-far. And investors tend to sell-near/buy-far to, among other things, reduce the return ‘bleed’ from contango. So, the two generate offsetting flows in spreads. Obviously, both positions are contingent on the state of the market. Clearly, those typical positions do not work in backwardated markets. That contingency is another reason to expect the scale of spread positioning to vary over time. Thus, the complementarity will also vary over time.

Reflexivity

This complementarity has several possible implications. One is that investors allow larger physical spread positions than otherwise. The presumed mechanism is the interaction of trade size, market capacity and (price) spreads. Physical traders’ buy-near/sell-far positions are limited because they move the spread against them. And, in the limit, become unprofitable. Investors expanding their sell-near/buy-far positions would raise that limit. So, physical traders would now be able carry more production forward from periods of seasonally high supply. And, physical spread positions’ share of OI is now likely materially larger than the CFTC’s 2015-18 estimate of 10%. That mechanism also suggests another implication for spreads. When investors have complementary spread positions, spreads are less likely to linger at extreme levels.

---