We estimate that it is in 3yly, 2y2y and 2y1ly that the selloff would be largest (75-85bp — see Table 5) were the OIS curve to align with the median dots up to end of 2019. Options are attractive to position for a selloff in those forwards for two main reasons: 1. Volatilities in 1y and 2y tails appear to trade cheap on our macro model. More specifically, accounting for the relationship between implied vols and the first three principal components of the swap curve, we find that 2yly implied vol should be trading 22bp normal higher (Chart 52), and 2y2y vol should be 19bp higher. A simpler historical regression of 2y1ly vol vs the 2y1ly forward rate also suggests normal vol should be around 21bp higher (Chart 53). 2. Payer skews could richen in 2yly and 2y2y. While they are looking rich across 6M+ expiries in ly and 2y tails (based on payer-ladder breakevens/ATM vol}, we note that this has been the case for some time now and it’s rather in gamma on 5y+ tails that skews now appear richest on a 6m z-score basis. Relative to what has been realized in the past month, the payer skew appears just fair in 2y1y, while it is rich across tails in longer expiries. Chart 52: Market vs fitted level of 2y1y vol, based on macro model (*) Chart 53: 2y1y implied vol is 21bp too low on a regression vs 2y1y fwd 120 140 A ; a hee cif 100 ane ar 120 pt eee f ps oA | wa R| Do peel poe . - 80 42 Mf ma 100 gi a FY fr wee mae - oe" co | aa | Wd 80 i peueeSen re f % oe a ok 60 i o? y = 35.017x + 28.016 40 ww = R? = 0.8667 90 ——2y1y market implied vol 401.4" == 2y'y fitted 0 20 SESAASt Se eee ste saeeseetreese as ‘ a 2 ed 88 = E ae = ES &8 = 5 88 s 3 &8 s 3 g = past by ¢ — since July m= last Linear (past 6y) Source: BofA Merrill Lynch Global Research. (*) Based on a regression of log(2y1y vol) on the first , three principal components of log of rates (derived with a PCA ran since Sep1 1). Rsquare = 0.97 Source: BofA Merrill Lynch Global Research Another way to look at this trade is through an