1Q19 Expected Returns Update

This is a quick post to share an update of this running model of expected S&P 500 returns using Federal Reserve data. As of March 31, the model predicted an 8.12% annualized return over the next 10 years. This has likely come down a bit further since then as the market rallied. As of today, we might be somewhere in the 6-7% range.

Data Source: Federal Reserve

Given there’s so much wailing and gnashing of teeth over macro risks these days it’s worth emphasizing a couple points.

First, this model is useless as a short-term timing signal. Don’t try and use it that way. If you’re looking for short-term signals you need to be looking at trend following systems and such.

Where I think there’s some utility here is as a data point you can use to help set longer-term return expectations and guide strategic asset allocation decisions (particularly when used alongside other indicators like credit spreads). When the aggregate equity allocation is close to 40% or above, it signals lower expected returns and argues for taking down US equity risk. Between 30% and 40% it signals “meh.” Probably not worth making any adjustments in this range. At least not on the basis of this model. At or below 30%, however, the model argues for adding equity risk.

Also, what I like about this model is that unlike indicators such as the CAPE or market cap/GDP what you are really measuring here is the aggregate investor preference for fixed income versus equities. When investors are very comfortable owning equities they bid up prices and expected returns fall. When investors are not comfortable owning equities they sell, prices fall and expected returns rise.

That’s the ball game.

No macro forecasting is required.

You don’t have to make any judgment calls on valuations, either.

What I would love to do eventually is run this for countries outside the US. What I suspect is that the ex-US models would show similar efficacy but with different “preferred” bands of equity exposure based on the culture of equity ownership in each country and whether or not there’s a significant impact from “hot money” flows from foreign investors.

I’m not aware of a straightforward way to find the data needed to do this. But if anyone has suggestions, please drop me a line.

The Permanent Portfolio In Action

May afforded an interesting opportunity to test the leveraged permanent portfolio strategy out of sample. (For previous posts on the permanent portfolio, see here and here) Below is data showing the results for two different leveraged permanent portfolio implementations, compared to the Vanguard Balanced Index Fund (an investable proxy for a 60/40 portfolio) and SPY. You can do a deeper dive into the data here.

Source: Portfolio Visualizer
Source: Portfolio Visualizer

NTSX’s laddered Treasuries provided better downside protection than the StocksPLUS bond portfolio here. But the gold exposure was also a major help, with GLD returning +1.76%. Obviously this is just a single month of performance, but the results are consistent with what you might expect based on backtests of the strategy.

Notice that the performance pattern is similar during the 4Q18 drawdown. In each case, the drawdowns are less severe than even those experienced in the 60/40 portfolio due to the diversifying impact of the gold. Because again, where the leveraged permanent portfolio shines is downside protection. You aren’t capturing all the upside of a 100% SPY allocation, but you’re capturing only a fraction of the downside.

Since December 2004, the PSPAX/GLD portfolio has captured 60% of the upside of SPY but only 43% of the downside. The asymmetry means PSPAX/GLD slightly outperforms SPY over this time period, but with less volatility. More importantly, the max drawdown is only a little more than half as bad.

Still, in my view the biggest problem the leveraged permanent portfolio presents for investors is precisely that its outperformance comes in down markets. This isn’t a sexy way to make money. It’s not the kind of thing that impresses people at cocktail parties. The behavioral challenges this presents should not be underestimated.

But personally, I’ll take a 10.62% safe withdrawal rate over cocktail chatter any day.