Many interesting trends are sweeping through the currency overlay industry as the sophistication of clients and currency managers improves with each passing year and with clients implementing innovative mandates. The most pronounced trend has been to consider currency as an alpha source rather than as a simple hedging strategy.
As a result, there has been pressure on currency overlay managers to conduct research to stay abreast of the innovation sought by clients. Some clients have tried to push the envelope by investigating alpha options (volatility) strategies to complement the mainstay of the industry – fundamental and technical strategies. However, most managers are still tied to the basic strategies as clients gain familiarity with plain vanilla programs.
In this note, we highlight a number of innovative ideas that an alpha-oriented client requested in a recent search to highlight the direction of the industry and to help other clients who are grappling with similar issues. The four critical areas of innovation in this mandate included: (a) budgeting risk rather specifying currency ranges; (b) restrictions, if any, on specific currencies can be done in the context of their role in overall risk and liquidity rather than on underlying allocations of equity managers; (c) expanding the scope of currencies to include emerging markets; and (d) insisting that risk be allocated to various trades based on an optimisation procedure.
Here, we evaluate each of these issues and demonstrate that the currency programme created for such a client greatly improves the performance for clients. The diversification benefits of one additional area of innovation, namely including alpha oriented currency options strategy is covered under separate cover (Muralidhar and Neelakandan, IPE March 2002).
Risk budgeting
While clients have known for some time that budgeting risk (eg, budget tracking error1) to a currency overlay manager for off-benchmark positions is the most appropriate way to establish risk controlled currency overlay programmes, there has been a tendency for clients to establish ranges around individual currency allocations. This has probably been driven by the fact that clients feel that it is easier to monitor physical limits on positions than risk levels. However, with greater availability of risk models and increased focus on risk budgeting, the trend is now towards allocating a tracking error budget to a currency manager rather than ranges around a specific benchmark currency allocation. In such a framework, the client gives the manager a target tracking error (eg, 200 bps/year) and a range within which the manager is expected to operate2.
At the simplest level, ranges suffer from one critical problem – they actually permit the manager to take varying degrees of risk depending on market volatility and may permit risk taking beyond a threshold desired by a client. Suppose a client permits a ±20% off benchmark range around the allocation to yen exposures because it translates into a 200 basis points of maximum tracking error. However, if the volatility of the exchange were to double during market crises, then the same positions permit the manager to effectively take 400bps of risk, double the level desired by the client. Tracking error is important to investors, especially institutional investors, because it provides a measure of the variability of a manager’s returns around the benchmark. Hence budgeting risk explicitly is clearly the way to go as it controls the variation in performance around a benchmark that a client can tolerate.
Liquidity tiering
As simple as this may sound, clients will obviously feel uncomfortable with mandates that have a broad tracking error budget, with ranges and with no further guidelines. Issues that concern clients are a potentially large allocation to illiquid currencies. Possible add-ons to the risk budget could be (a) liquidity tiering of currencies; and/or (b) restrictions on the marginal contribution to risk of particular currencies. Liquidity tiering essentially provides guidance to the manager as to how much of the total positions can be taken in specific currency pairs. The client creates groups of currencies that exhibit similar liquidity and then the overall allocation to these currencies is capped.
For example, euros, US dollars and Japanese yen are extremely liquid and require little or no cap on total positions. Other currencies such as New Zealand dollars and Mexican pesos may be capped as they are less liquid. This may be particularly important for clients in illiquid base currencies, where historically there has been a tendency to extract alpha from currency pairs where the base currency is one leg of the transaction (eg, Swiss franc mandates). This may be an inefficient use of risk and hence liquidity tiering can be used to overcome this situation. An alternative way is to restrict the marginal contribution of risk of the currency pairs that fall into lower liquidity tiers. The only difficulty with marginal contribution strategies is that they are more effective in monitoring risk than in budgeting risk.
Adding emerging market
currencies
Traditionally, currency mandates have focused only on developed market currencies as managers have not provided systematic products to manage emerging market currency portfolios. Emerging market currencies have declined dramatically and have been prone to crises, making standard currency models using moving average technical rules or simple fundamental rules useless. As a result, there has been a tendency to think of emerging market currencies as currencies that need to be removed from portfolios, but this is difficult to do on a systematic basis because of the high cost of hedging. The high cost of hedging emanates from the generally high interest rates in the emerging market countries.
However, instead of seeing this as a problem, high interest rates in emerging markets offer a unique opportunity to enhance returns, especially if the manager can also create exposures. When managers create exposures, the access the high interest rates (ie, are long the high interest rates). Furthermore, permitting cross hedging (eg, buying Mexican peso and selling Brazilian real) allows a manager to offset the high cost of hedging by financing the trade in another emerging market currency which is reasonably highly correlated and with attractive interest rates. Such a mandate puts the onus on currency managers as such trades are non-obvious and most managers are not in a position to examine all possible relationships among the approximately 24 tradable currencies globally.
Dynamic optimisation of
risk allocation
This takes us to the next innovation: requiring managers to optimise the risk to currency strategies. Typically, in what we refer to as the ‘classical’ approach, managers have built good currency models for specific currency pairs (eg, US$/yen, US$/E, E/yen). The typical techniques used under the classical approach such as moving average cross over rules or fundamental analysis make it easy to build such models. Thereafter, currency managers either statically optimise the weights to each of the currency models or naively equally weight them because optimisation techniques do not give stable results. A typical portfolio under this approach could have the following recommendations:
o US$/EUR: +5%
o US$/¥: –3%
o E/£: +3%
This then gives us a residual portfolio as follows:
o US$: +2% (+5%–3%)
o E: -2% (–5%+3%)
o ¥: +3%
o £: –3%
Two problems afflict the classical approach: (a) by assigning a static weight to a currency model, risk may be taken inefficiently in the portfolio as there are times when certain currency pairs move sideways and risk will be allocated to them; and (b) inconsistencies can creep up into currency views (eg, the US$/E can be bullish on the dollar, the E/Swiss franc model can be bullish on the euro, yet the US$/Swiss franc model may be bullish on the Swiss franc). As a result, rather than modeling individual currency pairs, the newer approach to currency management allows for a dynamic optimisation of return subject to the risk allocated to a portfolio of all possible currency pairs. In such an environment, it is possible to ensure that inconsistencies are precluded; moreover, the optimiser can select which currencies to be bullish on and simultaneously determine the funding currencies (not necessarily a one-to-one currency trade). A typical portfolio in this framework could look like, which can (and will) be markedly different from the classical approach:
o US$: +5% o E: –2%
o ¥: –10% o MXN: +9%
o BRL: –2%
A hypothetical mandate and
some simulations
We briefly describe the mandate and demonstrate some interesting results from the simulations for this mandate. This mandate was to be run against a 50% hedged benchmark and the client was going to maintain a passive benchmark separately. Hence, the active currency program was to be viewed as a purely alpha strategy.
The first issue was to preclude any leverage and hence a conservative approach was taken in the definition of leverage. Essentially, the notional values of both legs to the transaction were added for each currency position (eg, for a US$10 E/¥ position, this trade counted at US$20). Many currency managers and clients consider a more relaxed version of leverage, which converts both legs into base currency equivalent and only counts the larger leg.
The second major constraint was to manage the portfolio to a risk budget. For simplicity, we will assume that the client was willing to budget 4% annualised tracking error to this programme. In addition, rather than precluding any currency pairs from trading, the following liquidity tiers were adopted:
o US$, euros and Japanese yen
No more than 30% of the notional principal;
o Developed Europe and Commodity Currencies
No more than 15%;
o Liquid Emerging Markets (eg, Mexican peso, Czech koruna)
No more than 10%;
o Illiquid Emerging Markets (eg, Korean won, Indonesian rupiah) No more than 5%.
The simulation was performed from January 1995–November 2001 with live transactions cost data included. The model underlying the simulation was the model used to generate trades for our more aggressive global currency programme. The optimiser attempts to take advantage of fundamental and technical currency strategies to add value in these currencies, with certain risk management enhancements to protect against emerging market crises and fat tails in currency returns. In addition, attribution by currency pair is difficult under an optimized scenario, so we decomposed the risk into two independent sources: G-10 currencies and emerging markets. The results are presented in Table 1 and we provide the entire period data as well as data segmented over two sub-periods (1995–97 and 1998–2001). The key critical results are:
o Adding emerging markets raises information ratios dramatically from 0.77 to 1.31 over the full period and also over the sub-periods. These ratios are quite impressive considering those achieved by developed market only strategies3. Moreover, the risk allocated to G-10 currencies exceeds that allocated to emerging markets4.
o In the years in which the developed markets have offered good opportunities (1995–97), the optimiser chooses to take the risk in these currencies and is adequately rewarded for doing so, with over 70% of the returns emanating from G-10 currencies. In three key years (1998, 1999 and 2001), emerging markets provide a hedge against weak G-10 performance.
o Adding emerging currencies, reduces maximum drawdowns from 3.07% (G-10 only) to 2.62% (total product).
o The final product is still very highly correlated to the G-10 return stream indicating that developed markets largely drive the returns.
o The risk is maintained under the 4% cap over the entire horizon and over the sub-periods.
Conclusions
In this note, we set out to demonstrate that certain innovative approaches suggested by a client provide enormous value-added to the client. First, adding emerging market currencies to the pool of eligible currencies provides a good source of alpha and diversifies currency risk. Second, creating a dynamic allocation to risk framework allows for an efficient allocation of risk to best opportunities. This was demonstrated by the fact that the 1995–97 period was good for developed market currencies and later years are less favourable and the optimiser recognises the same. These may not be captured by the classical approach to currency management. Moreover, budgeting risk is easily accomplished in currency markets and clients can add liquidity tiers to concentrate risks in the most liquid areas. These constraints are more efficient in managing risks than ranges around a particular currency allocation.
As more clients consider currency overlay or as existing clients get more comfortable with currency overlay programs, they have to search for alternative sources of alpha or better ways to manage their existing portfolio. Previous research demonstrated the advantages of adding options to standard programmes. This paper elevates the discussion to a different level by incorporating innovative notions of risk budgeting, dynamic allocation of risk to currency opportunities and enhancing the performance through greater leeway in trading emerging markets to demonstrate how clients can implement innovative currency programmes.
Ryan O’Grady is director of investment research, Philip Simotas is president and Arun Muralidhar is managing director at FX Concepts, based in New York
1 Tracking error is defined as the ex-ante standard deviation of excess returns
2 See Litterman, R, J Longerstaey, J Rosengarten and K Winkelmann (2001), ‘The Green Zone... assessing the Quality of Returns’, The Journal of Performance Measurement, Spring, 5(3)
3 See Baldridge, J, B Meath and H Myers (2000), ‘Capturing Alpha through Active Currency Overlay’, Frank Russell Research Commentary, May 2000
4 Notice also that emerging market returns are uncorrelated with G-10 currencies

Topics