Have you heard? There’s a new asset class around. It’s called foreign exchange. Maybe you have heard, or maybe you’ve heard someone mention it and discounted it. But right now, it’s unlikely that you haven’t come across the idea. Which is a very different situation from, say, a decade ago. At that time, FX was either a nuisance, if you were an asset manager who didn’t want it messing-up returns, or a well kept secret among those hedge funds that used it to deliver profits.
But now, the cat is out of the bag. Most large investors will have at least a proportion of their assets in an FX product. And the variety and sophistication of these products is increasing all the time. Some of them are even trickling down to the retail level. So why has it suddenly been discovered, if it was there all along? And what was the problem with investing in this strategy before?
A brief history of FX
Part of the problem is that FX is a young market, as asset classes go. It is a relatively recent innovation that FX rates should be a free floating variable – up until the 1970s there were attempts to fix and regulate FX on a global scale. At that point the Bretton-Woods agreement collapsed, and it was finally admitted that free-floating exchange rates served the global economies better than the more rigid regime of fixed rates adjusted at intervals. The newly created market rate was very different from more familiar investments. Unlike interest rates, it had no natural range. Unlike equities, it had no expectation of growth. But like them both, it could move both fast and far – and it was a deep and liquid market.
But is it an asset class? It was not thought to be, certainly at first. Part of the problem has been that FX is a rate of exchange, not an index ultimately linked to ownership of some underlying item of value. To buy dollars, one must sell pounds, or euros, or some other currency. How could one ‘buy’ or ‘invest’ in FX? And, how was it expected to move? Initially FX rates were expected to evolve in response to the costs of similar goods in different countries, and to change according to relative interest rates in the different countries. Thus, if a basket of goods in the UK cost one pound, and a similar basket in the US cost two dollars, the FX rate would be equal to US$2.0 per £1. The interest rate effects were calculated as follows: if in the UK and USA annual interest rates were 10% and 5% respectively, then at the end of the year, there would be 10% more money in the UK and 5% more money in the USA, thus the exchange rate would move to (2×1.05)/(1×1.10) = 1.91. These simple relationships were supposed to define FX, apart from minor variations. But as the market evolved, it became clear that the rates declined to obey these expectations.
As the market evolved in the 1980s, exposure to international assets grew. FX was often omitted from consideration in both the investor and the corporate sectors, which was frequently to the detriment of both. Equity and bond returns, for example, can be wiped out or reversed by FX moves, and many companies learned to their cost that FX was an important consideration in overseas expansion. But currency risk remained relatively limited, and the larger investments tended to be in near neighbour countries where there were strong links between the currencies. Some realisation began to emerge that predicting FX movements was not as simple as looking at relative prices and interest rates, and momentum and macro-based forecasts became popular. It became clear that if one could correctly judge where an FX rate was headed, there was money to be made, regardless of the question of whether it was an ‘asset’ or not.
In the 1990s the market began to take off. Spreads narrowed, liquidity spread to emerging currencies, and FX developed beyond just market makers and takers. Corporations, previously the major users, became less important as investors became aware of FX risks and opportunities, and began to actively hedge and take positions (see Fig. 1). Capital flows increased enormously. Proprietary traders, CTAs and hedge funds began to develop as a community, and their interest in trading FX grew along with them. Currency crises like the ERM, Mexican and Russian events began to be seen as opportunities. The highly predictable era preceding the introduction of the euro strengthened investors’ belief that currency movements could be reliably forecast, and the depth and liquidity of the new European currency tempted bigger players to the market. Currency overlay and active management became more accepted as investment strategies. Take-up was strongest initially in the US, which was traditionally wary of currency risk and keen to outsource, but quickly spread to Europe.
Fig. 1 illustrates the changing composition and growth of the FX market. The labels of the constituents of the bar chart are those given by the Bank of International Settlements (BIS) and benefit from some explanation:
• “Other financial institutions” covers financial institutions such as smaller commercial banks, investment banks and securities houses, mutual funds, pension funds, hedge funds, currency funds, money market funds, building societies, leasing companies, insurance companies, etc. This is the profit-focussed part of the market.
• “Non-financial customers” refers to mainly non-financial end-users, such as corporates and governments.
• “Other reporting dealers” includes general inter-dealer flow which trades only for liquidity purposes.
For more detail about the BIS data and definitions see BIS triennial report 20072.
In the later 1990s, statistical analysis and rule-based trading began to become popular, as the difficulties of analysing and replicating judgement-based returns became apparent. Hedge funds began to launch with heavy currency components, and their assets under management grew rapidly. Prolonged equity drawdowns and sideways periods made investors appreciate the relatively modest impact which unlevered currency management could deliver, while hedge funds made full use of leverage to deliver a previously untapped source of alpha.
It is worth looking at the last bar on Fig.1: in 2007, more than US$3 trillion per day was traded in the FX markets. This is now the deepest and most liquid market in the world. Even in the darkest moments of 2008, FX trades could still be done fairly smoothly, albeit at higher bid-offer spreads.
How does it work?
So, assuming that FX rates are in some sense predictable, how does one make an investible product out of them? First of all, one needs to understand how FX is traded. It is essentially a margin or zero upfront capital trade, which immediately leads one to realise that leverage is simple to apply. In general, trades are month-end forwards, which are settled at expiry. As even a volatile month will only contain FX moves of a few percentage points (FX rates have similar, though lower, volatilities to equities) then it is clear that the amount of capital which can be considered ‘at risk’ is not the entire notional amount of the trade. There are various methods of calculating this capital amount, but it is usually conservatively considered to be something like 30% of the notional amount – implying a leverage of about three. This leverage yields an asset class with a similar volatility to that of equities. The capital itself is held and invested, and used to fund any losing trades.
This is simple in concept but not in execution. Apart from all other considerations, is that of the physical deal processing. Most diversified strategies demand the capacity to physically book and process multiple FX deals, potentially every day. Not all the investors who would like to take advantage of currency returns have this ability. Thus a new product type has evolved – the active FX return index, whereby investors may buy (via a fund, or swap, or note) exposure to an active FX strategy, which is managed and administered by a third party. The interest in these products has ballooned in recent years as more traditional investments delivered devastating underperformance.
Interestingly enough, the motivation behind the earliest indices and index-like return series – in both the equity and FX spaces – was not necessarily to enable them to be traded. The earliest series of “market returns” were created for benchmarking purposes. Initially, active currency strategies had enjoyed a generous zero benchmark, with overlay providers arguing that whatever additional returns they provided should be regarded as pure added alpha. However, this argument began to wear a little thin as it became apparent that a mechanistic and blindingly simple implementation of the popular carry trade strategy would have delivered an IR of about 1.0 since 19901. Plan sponsors began to question the added value offered by managers who fell short of this target. We have discussed the appropriate benchmark extensively1, but it will suffice to say that the market is beginning to come to a consensus that a simple mixture of carry, momentum and valuation models is an appropriate way to benchmark active currency returns. Managers in future will have to deliver “true alpha” – some return above and beyond beta – to satisfy investors.
However, just as in the equity world, where indexation became popular 25 years ago, once an index is available many investors want to buy and sell it directly. This simultaneously bypasses all the problems traditionally associated with currency as an asset class, and permits exposure to hedge fund like returns without paying hedge fund levels of fees. Moreover, a level of flexibility becomes available which was previously unprecedented. An investor who feels the carry trade is under pressure can take out a short contract on the index for a month. The need for separate indices, representing different styles of trading (like carry, trend or volatility) becomes apparent, once more mirroring the developments in equity and bond space where style and geographic indices are commonplace. In response to this market interest, Citi launched the CitiFX Benchmark (Beta) and Alpha Indices early in 2007. The Beta is designed to represent the “market” return in the world of active FX, to be simple, and easily replicated. The Alpha suite of indices, in contrast, are style specific, and designed to outperform. Citigroup was not the only FX provider to launch indices at this time. Indeed, a plethora of similar series was created in 20073.
FX versus other asset classes
So how does FX measure up as an asset class? It is a little difficult to say over the long term, as there are no traditional, well-loved indices of the likes of the S&P and FTSE to refer to. Indeed, earlier than 1975 or so, there were very few floating FX rates. So we have to create what we call a ‘backtest’ – a what-if scenario. Using historical FX rates, we ask the question, ‘If I had followed this strategy in the past, what would I have made?’ This way we can generate a hypothetical series of returns which may be used for analysis – for example, we could find the optimal mix of FX and fixed income for a portfolio. This is only possible to do for rules-based strategies, for obvious reasons. And, a hypothetical series of returns is never as good or convincing as actual profits. Nevertheless, if we dig a little deeper, we find a number of interesting similarities between FX and a traditional investment like the S&P index.
What is the S&P500 Index? It is a weighted combination of share prices. How is the weighting determined? By looking at the number of common shares outstanding, and normalising everything relative to a selected base period (1941 – 1943 in the case of the S&P 500). How are the companies selected? Using a complex set of criteria designed to ensure that the companies in the Index are the leading companies in the leading industries. These criteria can include liquidity/turnover ratios, market value, financial statements, and many more. This is starting to sound like a rules-based trading strategy.
Moreover, though the index is often graphed as going back to the 1930s, it did not in fact appear in its current form until 1957. The results prior to this date are obtained form modifying the returns of a different index – ie, they are backtested. This is often not appreciated. Suddenly FX and equity indices are looking more and more alike.
Let us try to put the best comparison that we can together, for the longest reasonable time period, of FX and other asset classes. The results here were published4; we reproduce some of the graphs here, updated to the current time. In Fig. 2 we show the results of the carry trade since 1975. This is a very simple and unoptimised version of the trade; in each of all the 45 major g10 crosses, a forward trade is put on once per quarter and rolled at expiry. The four lines on the graph are:
• The total carry return, with no leverage. This is the actual return which would have been earned by the trade.
• The pure forward point return. Had the FX spot rates never moved, this is the return which could have been made – the ‘perfect carry trade.’ In fact the actual carry return is not very far off this theoretical line.
• The spot component. The return of the carry trade without the interest rate differentials. Though overall it has made only a small difference, the 95 – 06 period was interesting in that spot moved contributed in a positive way to the carry trade profit, meaning that forward rates were actually counter-indicators of spot direction.
• The returns if spot had always moved to the forward, and uncovered interest rate parity (UIP) had been preserved. This was the way that the FX rate market was assumed to evolve when floating FX rates first became traded.
The returns are not very high – a few per cent a year – but recall that FX trades may be levered by about a factor of three to approach the risk levels of an equity index. Suddenly this looks interesting. Now, we compare this trade with some other asset classes. We have data for this since 1985 rather than 1975 but it is still more than 20 years of comparison. In Fig. 3 we show the results over the period for equities, FX carry, commodities and government bonds, and this time we use the appropriate leverage of three for the FX carry.
Looking at these graphs, it is perhaps an understatement to say that the carry trade is a credible asset class. It delivers the second best information ratio overall, and is in general uncorrelated to the other instruments, though this is possibly changing. But its returns – taking drawdown, risk, and value into account – outstrip those of the equity market.
Why do investors buy equities? They do indeed have a longer traded track record than FX, but when one dissects the nature of the actual index construction, the differences become less significant. Perhaps it is simply that we are used to them. With 2008-2009 eroding almost 20 years of equity returns, it is starting to become harder to justify the dominance of this asset class in our portfolios. THFJ
This article has been prepared by the named authors in their personal capacity and not as employees of Citibank NA or any of its affiliates (collectively “Citi”). The views expressed herein are those of the authors and may change without notice. They may also differ from the views held by Citi.
1. The CitiFX Trading Strategy Benchmark: Beta 1, Jessica James, Philip Gladwin; Investor Currency Monthly – February 2007.
2. BIS triennial report 2007
www.bis.org/publ/rpfx05t.pdf, page 36.
3. Systematic FX model structures: A new generation of investment products’, Jessica James, Investor Currency Monthly – September 2007
4. Uncovered Interest Rate Parity and the FX Carry Trade, Jessica James, Aysu Secmen and Kristjan Kasikov – Quantitative Finance 9 No 2 (March 2009) 123-127
ABOUT THE AUTHORS
Dr Jessica James is Global Head of the Foreign Exchange Quantitative Investor Solutions team. She began her career as lecturer in physics at Trinity College, Oxford.
Aysu Secmen is head of Quantitative Investor Solutions in New York. Dr Secmen holds a BS in Mathematics from Bogazici University and a Ph.D. in Mathematics from Texas A&M University.
Kristjan Kasikov is a Director in the Foreign Exchange Quantitative Investor Solutions team. He specialises in quantitative FX alpha generation and optimal hedging strategies.