In a new article in the Journal of Alternative Investments, Scott Mixon and Esen Onur quantify the over the counter index dividend swap market. Along the way, they provide a good example of the scientific method: positing a relationship, testing it against the data, and then abandoning it when the data indicates they should.
Scott Mixon, since 2012 a staff economist for the Commodity Futures Trading Commission, came to the CFTC after long career in private-sector finance, beginning with a stint as a quantitative and derivatives strategist for UBS.
Esen Onur is also a CFTC staff economist, since 2011. Before that he held academic posts, including a professorship at California State University.
To review the sort of trade at issue: a dividend swap consists of a fixed and a floating leg. The cash flow for the floating leg, as these authors write, is “computed based on the dividends paid out on an index in a given time period, usually a year.” The dividend paid by any given stock within the index has the same weight within this calculation as the same stock has within the index.
A New Database
In studying the swaps, these authors are working with a newly available database. Until quite recently there were no publicly available repositories of OTC transaction data for such swaps. At present, though, large participants in the market have to register with the CFTC and report their transactions for a swap data repository (SDR).
The SDR goes beyond time and sales. It provides a position-level look at the market, though information about individual participants remains confidential.
In their overview of this data, Mixon and Onur make the following points:
- Size of the market. Two and a half billion dollars in notional market facing OTC dividend swaps are outstanding between dealers and end-users, another $4.4 billion amongst dealers. The S&P 500 accounts for the majority of dealer-dealer swaps.
- Velocity of trading activity. The Mixon/Onur analysis indicates some “reasonable rule[s] of thumb,” such as, one buy-side trade per week for EURO STOXX 50, and one buy-side trade every two weeks for the S&P 500 dividend market. Trading in other indexes is less active.
- The role of dealers. Mixon and Onur conclude that their data is consistent with the proposition that dealers are net short dividends, with leveraged funds being on the long side as end users. On the EURO STOXX 50, dealers are net short over half a billion US dollars notional.
Mixon and Onur consider the break down between dealers to end user trades on the one hand and dealer to dealer trades on the other to be “revealing.” Although approximately 80% of the S&P 500 trading is between dealers, that is true of only 20% for the EURO STOXX.
The FTSE 100 market is less active than either of them and “virtually all of the reported notional trades” there are between dealers.
A Conjecture Tested and Found Wanting
Here, though, is the paradigmatic example of statistics in the service of science and of Popperian falsification. Specifically with reference to the EURO STOXX market, Mixon and Onur conjecture that “dealers accumulate floating dividend risk through the issuance of structured products for retail investors and occasionally demand liquidity from buy-side firms in order to reduce that risk.”
This is a reasonable conjecture but, alas, when our scholars checked their data, they found little evidence for this hypothesized triangular relationship among dealers, retail investors, and buy-side firms. The buy side is typically the demander of liquidity and pays a small premium over futures screen prices to get the long floating dividends. Likewise, the buy side accepts a small discount to futures screen prices when short floating dividends.
Dividends have a central role in asset prices, so one would intuitively expect that (as these authors write), “indicate dividend swap prices might be a powerful source of information for researchers.” They caution, though, that “the OTC nature of the market might cloud some of the conclusions.”