Investors have shown significant interest in shorting treasuries recently. Some investors are choosing to express this view by buying levered ETFs that promise twice the inverse return of an underlying government bond index. Investors in these ETFs are holding them for longer than 1 day and are therefore, in addition to betting on the direction of the underlying, also making a complex bet on its path and realized volatility. In this article we examine some evidence that investors are behaving this way, and in the next article we examine the complex nature of the bet they are making.
Investors in levered ETFs linked to treasuries are mostly short
The chart below shows Assets Under Management (AUMs) in levered ETFs whose underlying are a government bond index. As we can see, the overwhelming majority of assets are in the levered short ETFs compared to levered long ETFs. There is a total of $7.4 BN in AUM allocated to levered short ETFs while only $69MM in AUM allocated to levered long ETFs.
Investors are holding onto government bond linked levered ETFs for longer than 1 day
We can estimate the relative holding periods for ETFs by calculating the ratio of the number of shares outstanding to the daily volume traded for each ETF. This is an inexact measure of the average investor holding period for each ETF because it does not take into account that ETF market makers or index arbitrageurs may be responsible for a significant percentage of a day’s traded volume and may also be responsible for holding onto a portion of the shares outstanding for their own inventory. This makes it likely that the true holding periods for investors in these ETFs are higher than the numbers we calculate. However, we can use these ratios to compare relative holding periods across ETFs (this assumes that market makers/arbitrageurs are behind similar percentages of an ETF’s traded volume across all ETFs).
The chart below shows the average holding period (for ETFs with AUM > 100MM) for non-levered ETFs, levered ETFs and for levered short government bond ETFs. As we can see, the average holding period for government bond linked levered ETFs is half that for non-levered ETFs, and ~5 times higher than that for levered ETFs. Therefore, investors who are investing in government bond linked levered ETFs are holding onto their positions for much longer than they are for other levered ETFs and very likely longer than for 1 day.
Holding a levered ETF past 1 day is not the same as buying/shorting the underlying using leverage
The issue with holding a levered ETF past 1 day is that investors expose themselves to path dependency in the underlying. If you are not sure about what we mean, the video below starts from the very basics and explains how the payoff of a levered ETF is different from levering up and buying/shorting the underlying yourself. A key difference is that the payoff from levering up the underlying on your own is not path dependent and can be calculated based upon the prices of the underlying the day you enter and exit the position.
The reason for the difference in payouts comes from the fact that the manager of the levered ETF promises you a multiple of the daily returns of the underlying. To be able to promise you these daily returns, the ETF manager has to buy/sell some of the underlying every day to position himself to have a constant leverage ratio the next day. The short video below explains this process in detail for a 2x long ETF, but the same result holds for a 2x short ETF: the manager has to buy more of the underlying on a day when the underlying increases in value and sell more of the underlying when the underlying goes down in value .
This daily buying/selling causes the payoff of a levered ETF to be different from that of levering the underlying on your own (we are assuming that people levering up and buying/selling the underlying allow their leverage ratios to change across time) . The video walks you through why this is true:
In part 2 of this article, we’ll examine some factors that determine the difference in payoff between holding a levered long/short ETF and buying/shorting the underlying yourself with leverage.