mast image

Special Report

Impact investing


Editor's Choice: Why defining 'currency beta' is harder than you might think

Related images

  • Editor's Choice: Why defining 'currency beta' is harder than you might think

Portfolio engineering with FX exposures would be much simpler if we could identify its beta. Martin Steward looks at the range of candidates

Currencies are brilliant, as any active currency manager will tell you. All those non-profit-seeking participants (central banks, hedgers, tourists) lend a candy-from-a-baby element to trading them. Profits arrive on top of, rather than instead of, those from other assets, so returns can remain uncorrelated. But managers also speak of returns uncorrelated with other currency managers, or even indices of currency managers. Many even claim that the different trading styles they run are uncorrelated. And the most amazing thing? Those claims are justified. FX is a giant alpha playground.

So why all the fuss about FX beta? Since Bilal Hafeez at Deutsche Bank manufactured an index based on systematic value, momentum and carry trading strategies and James Binny did the same (adding volatility) at ABN AMRO a few years ago, it has been a hotbed of discussion. “It’s a huge topic of debate,” says Thanos Papasavvas, head of currency management at Investec Asset Management. “We’ve been having this conversation with investors increasingly over the past six months,” confirms Dori Levanoni, co-head of global macro research at First Quadrant.

To understand why, just try to imagine a global equities manager making all those claims. Even managers with ‘unconstrained’ mandates are really constrained at the margins: the more money they gather, the more liquidity they need and the more their portfolio will look like the market cap-weighted index. They get paid for illiquidity, but only because it is a risk. Indeed, that is why most managers are given tracking error limits: alpha is great, but we invest in equities because we expect to be paid for taking systematic beta risks.

By the same token, if something attracts a systematic return for risk in FX, you would want your FX manager to exploit it. On the other hand, if that risk exhibits a particular profile (a fat tail, for example), you would want to know what your manager was doing to control it - especially if he was claiming to be trading something else. “FX beta is really about having tools that enable you to describe how a manager approaches the challenge,” says Diane Miller, a principal at Mercer. “It helps us understand the source of their returns through time.”

But perhaps, most importantly, identifying systematic (and therefore sustainable) returns from FX enables investors to integrate it fully into strategic asset allocation. Those returns, in other words, would define the FX asset class. It is easier said than done. No-one is in the dark about how the S&P500 works. By contrast, Hafeez’s Deutsche Bank Currency Returns index and the RBS Multi-Currency index originated by Binny happily churn out different numbers despite nominally trying to capture the same risk. “That illustrates a basic problem with identifying FX beta,” notes Miller. There are no agreed calculations for momentum or value.”

Indeed, despite co-authoring the research that underpins the AFX Currency Management index (which identifies a mix of moving averages as a good proxy for momentum strategies), Aviva Investors’ head of currency management, Pierre Lequeux, still draws attention to all the dimensions across which FX can be traded. “Multiply all those together to identify every single trading opportunity, and the sheer number makes it difficult to conceive of a generic benchmark or a beta,” he says. “The only valid benchmark would be some kind of peer-group benchmark.”

Let’s not give up yet. All of that suggests that FX beta is just difficult to pin down, rather than non-existent. But there is a further question related to the strategic asset allocation issue: what sort of beta should we be looking for? One kind of systematic return is simply serial correlation - momentum and mean reversion are examples - that is purely an effect of aggregate investor behaviour. Their existence across asset classes does not guarantee that the mean to which returns revert is not zero. “In FX markets this term ‘beta’ has been abused,” argues SSgA’s head of currency management, Collin Crownover. “Some would use it synonymously with systematic trading styles, which I think is completely inaccurate.”

Non-zero long-term returns come from betas associated with economic value production. Equities are a call on profits from making and selling things. In aggregate they are a call on economic growth. Similarly, high bond yields act as a sort of tax on economic growth, which means that high prices act as an engine of it. There is little debate that the main candidates for FX beta - momentum, value, carry - exhibit serial correlation. But, do any represent a risk transfer associated with economic value production? Or, as Schroders’ currency fund manager, Clive Dennis, puts it: “If you buy currencies, are you investing, or are you a ‘spiv’ [speculator]?”

Momentum seems the easiest to dismiss. Some muse that momentum generates an illiquidity premium (in the form of a trade-off between short and long-term liquidity providers), but these remain tentative suggestions.

Value is much more promising. If you believe that a currency has a fundamental fair value relative to other currencies (and goods), then you must also believe that some economic drivers underpin those valuations. Expressed as purchasing power parity (see our article on PPP), valuation appears tightly related to economic growth differentials and perceptions of macroeconomic risk.

The academic consensus is that currencies do not have a fair value - or at least that it is very difficult to model with macroeconomics. Nonetheless, observing that something, at least, makes FX prices mean revert over time, it is easy to imagine why a valuation methodology like PPP ought to work. This ties in with all those non-profit seekers who are trading the 25% of global GDP that flows across FX borders. That flow can respond, albeit slowly, to price changes and conversely, prices of goods and currencies can respond to trade flows. These dynamics are presumably behind the most obvious current systematic transfer of risk in FX - the return generated from long-emerging market/short-developed market positions.

“Eastern European currencies returned about 50% over the past 10 years as part of their being integrated into the euro-zone,” observes Henrik Pedersen, CIO at Pareto. “Look at GDP outperformance and the premium that has been paid in terms of interest rates relative to inflation, and you can see that it represents about 1%. That must be some kind of risk premium.”

Cyclicality premium
The risk, presumably, is exposure to economies experiencing high volatility as they quickly grow their footprint in global markets. Crownover at SSgA calls this a ‘cyclicality premium’. “Foreign investors in the markets of very pro-cyclical economies will see their investments losing value via FX just as their portfolio is down thanks to economic fundamentals,” he observes. “These lousy diversification properties of pro-cyclical currencies should result in a risk premium for investing in those currencies.” We
begin to build a story in which relative GDP outperformance leads to changes in investment flows and, in turn, changes to FX hedging.

It was this kind of exposure that Schroders wanted to capture as a benchmark for its ‘beta-plus’ currency fund. Unconvinced that existing indices delivered this, the firm asked JPMorgan to create one. It crosses 34 currencies with the USD, GDP-weighted. “Over the last 15 years this index would have returned 5% per annum in USD,” says Dennis. “That’s roughly the same as a government bond index but with lower volatility. It also stacks up well against equities, where you might also expect global growth to be reflected.”

Of course, you might believe that that volatility is an inherent characteristic of this economic FX beta, and that the price action of more liquid securities has a better link with that beta than slow-moving and backward-looking GDP. This concept would begin to make sense of MSCI’s Global Currency index series as a kind of FX beta, for example: these indices weight each currency according to respective country weights in MSCI’s equity indices. Although it did not form any theoretical basis for the indices, designed to help investors monitor the contribution of FX to their equity portfolios, they express the investment-flows thesis. “If investors believe a country’s economy will outperform, they are likely to invest more in that country,” as vice-president of index research Raman Subramanian puts it.

The Parker BlackTree Currency index (PCBI), from Parker Global Strategies and BlackTree Investment Partners, pursues this idea rigorously. PCBI is weighted 50% to a portfolio of 22 style-diversified active managers, and 50% to BlackTree’s Currency Investment Strategies programme (CIS). Half of CIS is devoted to ‘macro pressures’, a valuation strategy keying into the themes discussed above.

“In FX investors have to look to other asset classes to understand valuations,” says BlackTree CIO, Alexei Jiltsov. “Effectively we try to describe the macroeonomic picture from the underlying movements of a full range of asset classes, before translating that into FX positions. That will generate consistent returns simply by capturing some well-defined economic themes, particularly global risk allocation.”

Figure 1 shows correlations that may result from these economic risk transfers. Interest rates are perhaps the most intuitive: as the differential between US and Japanese two-year interest rate swaps predicts higher rates in the US, long USD/JPY is pushed higher. The second chart shows the predictive quality of equity factors: the Australian economy is geared to materials and the Canadian economy to energy, and sure enough, the relative performance of the S&P Materials and Energy sub-indices tracks the value of long AUD/CAD. In commodities, copper prices track short EUR/PLN because Poland is a large copper exporter to the euro-zone. Finally, the ratio of equity to bond performance correlates with the outperformance of high-growth economies: in this illustration, short EUR/BRL. “These, we believe, represent risk premia in FX,” says Jiltsov. However, BlackTree is loath to describe any of this as ‘beta’. “The trading rules evolve and adapt and have risk management characteristics,” as Jiltsov explains. “It’s in that murky world between beta and alpha.”

This returns us to that issue of agreeing on a calculation of beta. No-one can wait to see whether or not the exhaustive PPP measure created by the World Bank and the IMF every three years is a good proxy for FX valuations. That is why we see the kinds of short-cuts described above, and the risk management wrapped around them. Creating those models and risk management protocols involves skill - just as valuing any security is regarded as a great skill. So whereas in theory valuation might seem a good candidate for FX beta, in practice it turns out to be far too complex and contested.

All is not lost. Our final candidate - the carry trade - may be directly related to the economic fundamentals described above (particularly those in PBCI’s ‘Global Risk Premium’ indicators and Crownover’s ‘cyclicality premium’), and it has a simple, uncontested definition. “There’s no argument over what drives carry,” as JPMorgan Asset Management’s head of currency research Frank del Vecchio puts it. “It’s the interest rate differential, and you can model that and explain its relationship with other asset classes back to the 1970s.”

The carry trade exploits a systematic phenomenon known as the forward rate bias (FRB). If you buy spot JPY/AUD while simultaneously buying forwards, over the medium term you might expect your returns to be equal to those from simply holding spot JPY/AUD: the forward rate should predict spot, or at least its error should be randomly-distributed around a mean of zero, because JPY should appreciate or depreciate against AUD based on their relative interest rates. But this theoretical ‘interest rate parity’ has been in doubt for at least 30 years and there is now plenty of empirical evidence that the forward rate is systematically biased towards predicting too low a price for spot.
According to Record Currency Management, which specialises in carry, for the G10 between 1978 and 2009 this bias has held for 54% of monthly periods in EUR/GBP (the weakest predictor) and 61% in EUR/JPY (the strongest). Under interest rate parity assumptions, that should have been 50%.

This looks like a risk premium. Those who believe that it is a transfer of fundamental economic risk argue that FRB is driven by sustained real interest rate differentials, which are themselves a result of different inflation expectations and current account conditions between economies. Buyers of deficit countries’ assets take on the risk that their currencies will depreciate to bring the deficit down; and the inflation volatility risk associated with deficit economies.

“While there are other factors than carry at play in currency markets, we regard those as simple long-term inefficiencies associated with investor behaviour,” says Record’s CEO James Wood-Collins. “But inefficiencies associated with some kind of economic function we elevate to beta, and FRB is the one about which we are most confident and assertive.”

While Record is a true believer, it is far from alone. “Carry returns are basically the transfer of investment from less productive to more productive regions,” says Papasavvas at Investec. “Countries that want to attract capital have to offer a higher rate of return to do so,” agrees del Vecchio at JPMAM, “just as you have to offer higher yields on your bonds.”

So if carry is a beta, what are its characteristics? Carry defined by the RBS index exhibits correlation of about 0.67 with the MSCI World index - perhaps what one might expect from an economic risk premium. But the FTSE Currency FRB5 index, constructed by Record in collaboration with FTSE Group to replicate a carry strategies in USD, GBP, EUR, CHF and JPY, exhibits correlation of just 0.07 against global equities.

How, if FRB is expressive of an economic risk transfer, can it also express such low correlation with other assets that are calls on growth? We should consider other characteristics of its return distribution. As Figure 2 shows, FRB5 has a higher peak and slightly longer decay on the left than the normal distribution. Its kurtosis (tail risk) might seem low enough at 2.10, as might its negative skew at -0.81. But compare that to the distribution from MSCI’s equity-market weighted currency indices: its emerging markets index has an almost identical profile to FRB5’s; while its global ex-US, USD-based index has zero skew and kurtosis at just 0.30.

To explain these numbers, it is worth remembering that carry’s returns constitute what is left from the FRB once you have taken off any movement in the spot price. Investors can factor expected short-term spot volatility into their pricing of carry trades, but they cannot predict realised volatility. Get a bout of risk aversion and a flight to the low interest rate currency funding your carry could end up forcing you to sell spot rates at levels that cut deep into your FRB returns. This was what hit carry traders in 2007-8: high rate currencies sold-off massively in spot markets - not least by carry traders rushing to a suddenly crowded exit - wiping out FRB premiums from positions that had rolled for years.

“In six months, 15 years of carry-trade returns were wiped out,” notes Levanoni at First Quadrant. “The problem with carry is that the whole cycle of the trade destroys all the alpha. Fundamentally, you have to find someone who will buy it off you at the price that you’re not willing to pay yourself. If you don’t, you lose all your money. That’s why the theory of uncovered interest parity exists.”

In other words, against the theory that carry is about moving capital from global laggards to global growers, we have the theory that it is simply a return to momentum - bidding the spot price of high interest rate currencies upwards - that suffers massive occasional losses from mean-reversion. It might appear to work because it relies on the sovereignty of central banks over their short-term rates and their tendency to raise them in the face of higher expected inflation. But because they also have a remit to consider exchange rate stability, central banks are also the ultimate mean-reversion spot trader and are likely to start selling into that rising market - or at least make enough of a complaint to prompt other traders to sell.

As Levanoni implies, uncovered interest rate parity may exist, after all: “It’s honestly not clear to us that carry at the short end of the curve has a non-zero mean. It really might be just like selling a deep out of the money option, and its distribution certainly looks like a leveraged equity short-volatility bet.”

If this analysis is right, the mean-reversion effect in carry would be very different from that in other markets - where it rarely wipes out economic returns unless the forced selling of leveraged positions is involved. It is possible that the big losses to carry in 2007-08 were indeed, like those to equity, down to leveraged positions. “You can see how leverage-dependent investors in a context where credit became very free leads to herd-like behaviour,” reasons Wood-Collins. “If we succeed in our ambition to get FRB recognised as an asset class by institutional investors, it then starts to react to counter-cyclical behaviour like rebalancing, which might help reduce the influence of those pro-cyclical players.”

It is a compelling idea - but one that depends upon some faith that the profits to carry were real economic profits wiped out by leveraged sellers, rather than paper momentum profits wiped out by interest rate parity. Although most of the people IPE interviewed for this article assume the former, it seems difficult to rule-out the latter. That is a pretty good description of how the industry thinks of FX beta in general: it assumes economic fundamentals are relevant, but finds it difficult to demonstrate that they are - because transmission mechanisms are so complex.

As BlackTree’s Balraj Bassi observes, the first generation of beta, that led to equity indices, was simply about identifying non-diversifiable average risks, while the second generation identified consistent sources of return within those averages - value, growth, small-caps, momentum, and so on. “But the FX beta debate jumped to the second generation straightaway because it is so difficult to identify the simple average of returns to holders of FX positions,” he says. “So carry, for example, rather than the equivalent to the S&P500, is the equivalent of a long-high dividend versus short-low dividend stock position, which would be an exposure to global growth.”

But the fact that the S&P500 is also a call on economic growth helps us explain how we can systematically tilt towards that risk within the S&P500 universe by overweighting high-dividend stocks. You cannot have the second-generation beta without the first. How can we claim to find a call on economic growth in second-generation FX beta when we cannot define first-generation FX beta? Are we investors in currencies, or merely speculators? The question remains open.

Have your say

You must sign in to make a comment


Your first step in manager selection...

IPE Quest is a manager search facility that connects institutional investors and asset managers.

  • QN-2543

    Asset class: Search of an Asset manager / Advisor managing / Advising a risk-based equity derivatives overlay program.
    Asset region: Global Developed Markets Equities, Global Emerging Markets Equities, Swiss Equities.
    Size: CHF 700-2100 million.
    Closing date: 2019-06-17.

  • QN-2544

    Asset class: Transitional Real Estate Debt.
    Asset region: North America (USA/Canada).
    Size: $50-100mn.
    Closing date: 2019-06-17.

  • QN-2546

    Asset class: Real Estate Equity Fund (non listed).
    Asset region: Europe.
    Size: Total CHF 600m, approx. CHF 100-300m per fund investment.
    Closing date: 2019-06-28.

Begin Your Search Here