FX Fearless

I live in Europe and the majority of this blog readers are European as well. The most recurrent complaint I receive is that I focus on USD investment without acknowledging the related FX risks for European investors.

The Model Portfolio is built with USD assets simply because it was easier for me to backtest, from a system and horizon length point of view. If my goal is to show the properties of a portfolio of uncorrelated assets, I risk losing the audience at hello because their circumstances are different.

So let me introduce you to today’s bogeyman, the EUR/USD FX rate:

From inception, the rate swung between a min of 0.825 and a max of 1.60, averaging around 1.20. The rate crossed and stayed above/below the average many times, with high volatility.

Here is a quick recap of gains/losses for a EUR investor who bought USD assets over different time horizons:

data from 1985 until 2020, taken from here

[A note to the reader: in this post I will refer to Dedalo Invest post for several reasons: I trust the source, I do not want to re-calculate all the numbers by myself (especially now that I do not have Bloomberg), many Italian readers asked me about that post so I can provide a comprehensive reply once and for all. That said, Dedalo has been pretty cheeky about the data he choose and someone might wonder if his “FX volatility is the devil” stance did not play a part. This is what EUR/USD did in his sample:

If you are a EUR investor buying USD assets, not exactly the best possible start eheheheheh. Especially considering that this happened a few years back:

Why he picked 1985 to start, if not to include the lowest point in the sample AND “tax” the EUR investor, is beyond me. Officially the EUR started to trade in 1999. Considering its relatively short history, many “add” additional past data by either looking at the Deutsche Mark or doing a weighted average of the mark (GER), franc (FRA), lira (ITA) and peso (SPA). Needless to say, a trader in the ’80 was most likely not looking at the lira or the peso as they would have done after the infamous Mario The Dragon Draghi “whatever it takes”.]

This graph represents the number of instances in which the FX rate moved by a certain % over a 30-year period. This is an important piece of the puzzle to combine with the table I posted above: the EUR investor is not only interested in how many times they would lose money but also in how much. Again, how much that right skew is influenced by the sample starting date?

Focusing solely on the FX component, it seems to introduce only instability. But if I limit myself to buying only assets denominated in EUR, I lose many investment AND diversification opportunities (even more if I live in Switzerland, the UK, Sweden and so on). If I hedge my investments, I have to consider additional costs while still dealing with a restricted investment universe.

How much FX diversification risk should you allow in your portfolio?

This is a tricky question to reply to. That’s why I would go with the unofficial Corey Hoffstein principle: “first assume markets are efficient“. Buy the market portfolio and let the market decide your FX exposure. Unfortunately, this works when dealing with broad stock indexes. The proposition is already trickier when dealing with bonds: “market-cap” indexes do really reflect the market portfolio in this case?

With trend following, you have the “manager alpha” sitting on top of a (generally) big collateral exposure in USD T-Bills. In the long run, the yield of those T-Bills compensates for some of the FX movement: if the margin was invested in another currency, you would have earned a different risk-free rate. But in the short to medium term, FX volatility drives the biggest impacts.

And so on: commodities are priced in USD and other risk premia are usually ‘isolated’ on top of a USD exposure.

Without an ISDA and some currency hedging, there is no straight applicable solution. Maybe use some 3x Bund ETF in the bond sleeve? Maybe WisdomTree would finally launch a global stocks and bonds NTSX-like?

If a butterfly flaps its wings in the Amazonian rainforest…

Let’s take my favourite example, the Golden Butterfly portfolio. Dedalo Invest has a few interesting graphs, that must be read along with relevant caveats. For example, he realises that you cannot build a proper EUR-only GB because there is no SmallCap Value ETF available (there is $ZPRX but it owns more than 30% of non-EUR stocks). Also, the equity lines and related rolling returns shown below are not based on the standard portfolio weights (20% each in this case) but he averages 11 ‘optimisation models’ ¯\_(ツ)_/¯

The blue line is the GB we all know (almost, due to those models he uses). The red line is the USD GB from the EUR investor’s point of view: its value is simply translated into EUR every day based on that day’s exchange rate. The green line is the EUR(poor) GB.

The red line is clearly more volatile than the green line (unfortunately the graph is not logarithmic so it is hard to visually understand what really happened in the first half of the sample); drawdowns are deeper as well (even if there is a clear distinction between the pre-EUR launch period and the real-EUR portion). For EUR investors, it is pointless to look at the USD-GB as a “low volatility” investment option because it is not.

But this is not 100% correct. Well, it would be if our hypothetical investor bought on the 21st of Jan 1986 and then held their investment (save for the rebalancing) until they sold 100% of the portfolio at a certain future date. Do you know anyone that invests this way?

Most likely, our EUR-based investor would buy every month a piece of the blue line at the exchange rate of that day. In other words, the FX rate component of the price paid is the average of the rates printed in the period: as the FX rate swings up and down, the investor slowly but surely averages that component. Sure this process is not perfect, the investor would have salary increases during their life, and their buying pattern would change but still, it would never be a one-and-done type of deal.

The total portfolio value would move day by day according to Dedalo’s picture. But again, this is relevant ONLY if the investor has to sell EVERYTHING on a random day. If we assume that the investor would use the portfolio for their retirement, they would sell a small part every month, mirroring what happened in the accumulation phase.

At this point, volatility is volatility: it does not matter if it comes from the FX rate, stocks or bonds. The investor would sell sometimes when the portfolio is “undervalued” (bad) and sometimes when it is “overvalued” (good). We will dive into this point in a sec.

A key element of the above graph is that the red line, in technical terms, generates more fucking money than the green line. A lot more. You can see it way better here:

The red line spends more time above the green line than below. The area drawn by the red line above the green line is bigger than the area related to the opposite: the investor in the red portfolio has a materially higher terminal value than the one choosing the green portfolio.

Let’s look at it from a retiree’s point of view. The red-portfolio pensioner and the green-portfolio pensioner decide to use the same SWR. To the red retiree, all the volatility that happens while the red line is above the green line does not matter. They are still winning. The volatility that matters is when the red line is below the green line. Visually, we can say that the danger zone is limited but can we quantify it?

PortfolioCharts to the rescue

We can repeat the simulation done by Dedalo in PortfolioCharts. The USD portfolio is the classic GB, the USD -> EUR portfolio is the same portfolio with Home Country = Italy, and the EUR portfolio is this one:

Dedalo uses a currency-hedged gold ETF in his simulations while PC doesn’t allow us to do so. As before, no SCV in the EUR portfolio because there is no ETF to do it.

USD

USD -> EUR

EUR

As expected, the USD -> EUR portfolio has the highest volatility. Unexpectedly, the USD -> EUR portfolio has a higher real average return compared to the USD one. Considering that the nominal returns should be higher for the USD version (unless Dedalo is completely off, remember those silly 11 optimisation models), this result is only possible if inflation in Italy was lower than in the US. Not really:

Anyhow, we are here to compare the USD -> EUR and the EUR portfolios so, does not matter. The USD -> EUR CAGR is 5.75% while the EUR CAGR is 5.20%: this means the EURpoor investor is compensated by the higher volatility of the USD ensemble by the higher returns he gets.

In the accumulation phase it is better to choose volatility than not.

Here are three other graphs that show that the USD -> EUR portfolio indeed grows faster than the other two options:

USD

USD -> EUR

EUR

What about the decumulation phase? How much the additional volatility is hurting the USD -> EUR portfolio?

USD

USD -> EUR

EUR

No Surprise here, the ranking in terms of SWR is USD, EUR and USD -> EUR. But:

• if you remember, the USD -> EUR portfolio grows faster than the EUR one. After 30 years of (assumed) savings, the USD -> EUR investor can withdraw a smaller portion out of a larger pot.
• the SWR looks at the WORST POSSIBLE SCENARIO. In the median scenario, investing in the higher-yielding strategy means leaving (again, technically) a boatload more money to your kids and grandkids.

Another possible alternative is just to withdraw as much as you would do in USD, i.e. at the beginning of the year you calculate 6% of the portfolio, adjusted for US inflation, and then convert that amount into EUR at the current exchange rate. Before you think I am crazy, all the retired Brits living in Spain basically do that: they have a fixed GBP annuity that is increased by UK inflation.

Fact is, part of the FX exchange movement year-over-year can be explained by the relative movement in inflation levels. There is a bit of a wash for the retiree between the fact that they can withdraw more if US inflation goes up and less cause the EUR get stronger ag the USD. The retiree’s yearly budget would fluctuate but less than you would expect looking at the FX rate only. If you consider that every retirement expert is already pointing to the solution that retirees should adjust their budget based on their circumstances…nothing new under the sun.

Going from a SWR of 5.6% to 6% means increasing the yearly budget by 7%. If you are not comfortable seeing the budget swing too much, you can use part of that 7% as a reserve: buy some EUR ST bonds when the USD is strong and sell them to fill the budget gap when the USD is weak. If you look at the EUR/USD graph, the pair didn’t spend many months far from its average.

Can I have my cake and eat it too?

The solution is pretty simple: use the market portfolio!

This equates to a CAGR of 5.66%: not the 5.75% of the USD -> EUR portfolio but closer to it than the 5.2% of the EUR one. (Before you scream that now I am using too many instruments, messier to rebalance, transaction costs, blah blah blah, in real life you can find one ETF for each sleeve, thank you Avantis!!).

Same SWR as the EUR portfolio.

In conclusion, I do not think there is merit in reducing diversification and renouncing investing in some of the best opportunities just because they are listed in a different currency.

A last thought on the USD. It is not a coincidence that most investment strategies are priced in the global reserve currency. If one day there will be a new reserve currency, or a mix of currencies, then the market portfolio will shift accordingly. The point here is not that you have to be invested in USD, rather that you should given the current circumstances.

What I am reading now:

Follow me on Twitter @nprotasoni