The Flaw of Averages, Muni Style: SIFMA/LIBOR and Rates
Here’s a quick quiz. If over the last 10 years 1M LIBOR reset weekly averaged 2.814%, and the average of SIFMA / 1M LIBOR was 82.0%, what was the SIFMA average over the same time period (all rates unadjusted for day counts, holidays etc.)?
A. 2.05% B. 2.31% (2.814% * 82.0%) C. 2.62%, or D. None of the above but it seems like a trick’s in here somewhere
The correct answer is in fact A, which is a testament to how strongly the Fed has been stepping on the money accelerator over the last decade. Monetary policy aside, if you answered B (simply multiplying the LIBOR average by the SIFMA/LIBOR ratio average) you would’ve made a very common mistake which falls into the category of the Flaw of Averages. Overreliance on simple averages, partly induced by overreliance on simple spreadsheets, can very easily lead to errors of calculation and ultimately judgment. In this case, the seemingly more intuitive answer B is over 25 basis points wrong!
How does this work? When rates are low, SIFMA/LIBOR has been high and vice versa i.e. the two rates have been negatively correlated. If you don’t capture this fact in your analysis, you’re missing a critical component of how the tax-exempt markets have worked. This ultimately leads to over-hedging, misunderstanding of balance sheet hedges, and other unintended consequences.
Luckily, there are readily accessible public finance analytics that capture these very easily.