Minimizing Year-on-year Currency Impact on Earnings: a Layered Hedging Approach
Company treasurers and CFOs inevitably mean different things when they say: “We are fully hedged!” The degree and type of coverage referred to depend on a multitude of factors, all specific to the individual company and outlined in their respective hedging policy. Variables such as the type of exposures that are identified and subsequently hedged, the point in time exposures are initially identified, the goal of the risk management program, the average hedging tenor, the specific instruments the company employs and the technique used to implement these hedges all play a crucial part. Consequently, even if a company is fully hedged it does not necessarily mean that it will achieve the same risk reduction as another company making the same claim.
In fact, many analysts have been mystified by the sometimes violent swings in company earnings on the back of currency fluctuations, often despite claims from the corporate in question that it is fully hedged. The reason for this usually boils down to the fact that the hedging program employed by the company in question is simply not designed to manage this variable. Managing year-on-year earnings volatility is, by definition, a multi-year issue and it will therefore take a multi-year hedging program to come to terms with it.
Previous research1 has shown that using a layered hedging approach is quite successful in achieving this goal. The averaging effect gained by using multiple forward contracts, struck at different market levels through time, tends to smooth the year-on-year volatility experienced by the company quite significantly.
Most treasurers would agree that one of the most crucial parts of the hedging policy is the statement of the risk management objective. Some companies2 aim to reduce variability against a budget rate, while others focus on reducing transaction risk. Yet other companies, as indicated in the title of this article, aim to reduce the year-on-year currency impact on earnings. Regardless of the aim, it is crucial that the company constructs a hedging program that will assist in achieving this goal as best as it can. In our work with corporate clients we sometimes notice that this link is not always clear. The most common example is the corporate that manages risk towards an annual budget rate while under the impression that this program is also the optimal technique to reduce year-on-year earnings volatility. In our opinion, a static hedging program of this kind is not particularly well suited to manage earnings volatility. This is hardly surprising, as by definition each year is hedged in isolation from the next year, inevitably exposing the company to longer-term trends in the underlying currency market.
Having established the importance of linking the aim of the risk management effort with the layout of the hedging program we now turn our attention to how to best manage year-on-year volatility in earnings. Generally, the longer the hedging program, the more benefit it offers, as it provides the company with the time it needs to identify and react to longer-term shifts in the exchange rate. If so, why don’t all companies focusing on hedging earnings volatility use ultra-long hedging tenors then? There are multiple factors that in aggregate contribute to cap the maximum tenor. These include forecast error and opportunity costs. Even the accounting regulations issued by the International Accounting Standards Board (IASB) and Financial Accounting Standards Board (FASB) concerning the recognition of derivatives contribute to set a de facto maximum hedging tenor as the company needs to qualify any forecast exposure as highly likely to occur in order to achieve hedge accounting. Naturally, the further out the company forecasts, the more difficult it is to prove this high probability of occurrence. Bearing these factors in mind, the prudent company could start to implement longer-term currency hedges in order to manage earnings volatility.
In order to fully quantify and evaluate the merit of various longer-term hedging programs we conducted an analysis based on historical market data for a hypothetical company. While part of the analysis focuses on the difference between rolling and layered hedging programs, the main focus of the paper is to investigate whether a particular hedging tool is especially well suited for managing volatility in earnings. As an example we are using a euro-based company with long US dollar receivables. The objective of the analysis is to quantify the reduction in year-on-year volatility in earnings currency hedging would have provided in the period March 1989 – September 2005. Each hedging program contains one of four common hedging tools: forwards, at-the-money forward options, collars and participating forwards. The last two hedging instruments are structured at zero-premium cost in each period and designed to protect the 3% out-of-the-money level.
Furthermore, we make the assumption that the volatility in the all-in rate achieved by the various hedging programs examined, including any premium paid, is a valid proxy for volatility in earnings.
A hedging program can naturally be designed with any tenor in mind. Here we chose to examine programs with three-year tenors, as a longer-term example more clearly demonstrates the advantages of the layering hedging approach. It could also be argued that a good proportion of corporates could implement three-year hedges should the risk reduction offered be appealing enough. This is especially true if the companies in question adopt a layered hedging approach as only a fraction of the total coverage is initiated this far out. In fact, as the average hedge tenor of the layered hedging program per our definition is only half of that of the rolling program, this approach should alleviate most concerns raised above concerning longer-tem hedging. To elaborate, in the analysis we assume that the amount of hedge cover is a direct function of time. For a three-year hedging program this means that we increase the cover by 1/12 of the full amount in each of the 12 quarters.3 Thus, three years ahead of the underlying forecast maturing the hedge ratio is 1/12 of the full amount, rising to a 100% cover in the final quarter. Conversely, the rolling hedging programs examined hedge the entire exposure in the first permissible hedging quarter.
By using historical spot rates, interest rates and implied volatilities we are able to fully replicate the outlined hedging programs. All hedges are initiated at true market rates available on specific dates in history and subsequently evaluated as the hypothetical underlying exposures mature. This allows us to measure the performance of the programs over time. Figure 1 presents summary statistics based on this data. The upper number in each cell is the actual statistic, while the percentage numbers in brackets underneath each statistic show the improvement the various hedging programs offer compared to the unhedged case.
Of the metrics presented, the risk differential (Rd) may need a brief introduction. Rd is an in-house measure that provides additional insights into where the volatility, as measured by the standard deviation comes from. If that volatility is essentially due to positive returns, the Rd will be greater than one; otherwise it is smaller than one. For a perfectly symmetrical distribution, such as a normal distribution, the Rd is exactly equal to one. Obviously the larger the Rd the more desirable the hedging program.4
Naturally, a successful hedging program should exhibit a smooth pattern through time with no, or only a few leaps from one period to the next. As expected, the unhedged strategy resembles this the least. The standard deviation if no cover at all is initiated is 11% and the worst-case change from one year to the next is a colossal -22.2%. This large move occurred between Q1 2002 and Q1 2003 as the underlying EUR/USD spot exchange rate trended strongly, moving from 0.87 to 1.09 over the course of a year. Having no hedge cover when spot moves this violently is very detrimental. Had the company had the foresight to implement any of the examined hedging strategies they should have been able to reduce the negative impact of an occurrence such as this. Depending on strategy, the hedging programs were able to reduce the volatility and worst case outcome by as much as 50% compared to doing nothing. This is quite a significant improvement and goes a long way in demonstrating the value-added in having a hedging program.
There are, however, quite substantial differences in the amount of risk reduction that may be obtained by the various programs. As a group of strategies we find that the layered programs vastly outperform the rolling programs. This is especially noticeable when comparing the strategies comprising forward contracts. Despite hedging with three-year derivatives the company would hardly have succeeded in reducing earnings volatility at all, had they used the rolling approach. The resulting standard deviation is still 10.2%, a reduction of only 7% compared to the unhedged strategy. The worst-case change exhibits an equally disappointing improvement of a mere 6%. The layered forward, on the other hand, exhibit much greater risk reduction properties. With this program, the standard deviation is reduced to 6.2% corresponding to a 44% improvement compared to the unhedged case. What is more, the worst-case outcome is improved by 47%. In other words, simply extending the horizon of the hedging program does not in it self help companies reduce earnings volatility, as all the rolling hedging program accomplishes is to track the spot market with a lag equal to the hedge tenor. While insulated to negative market movements for a period of time, unfavourable rates will inevitably catch up with you.
Examining the other type of hedging strategies we see the same pattern. All layered strategies examined are able to cut the worst-case outcome in about half. In terms of volatility reduction, the collar and participating forward exhibits the largest improvement by reducing the standard deviation by approximately 50%. In this aspect the vanilla option based strategy lag in performance, as it is unable to reduce the standard deviation to the same extent. This is, however, a regularly occurring phenomenon in analyses of this kind. While a vanilla option is very effective in providing protection, it fully retains the potential to benefit from positive movements. Large up-side observations, due to favourable market movements, tend to drive up the volatility displayed by this instrument compared to other hedging strategies. In our opinion this is a misnomer as the deviations experienced in this case is desired. This is the reason the Rd statistic is also included in the analysis. Needless to say, both the rolling and layered options based strategies exhibit quite high Rd ratios.
The improvement on an instrument-by-instrument basis is clearly depicted in Figure 2, where the standard deviation and worst-case numbers from Figure 1 have been broken out to facilitate an easier comparison.
A further point to highlight is that while the layered hedging programs as a group exhibit quite favourable risk reduction properties, this increase in performance relative to the rolling programs is to some extent due to a lower ability to participate in beneficial movements. Returning to Figure 1 and comparing the respective group of programs we consistently find lower averages, lower best case outcomes and lower Rd numbers for the layered strategies. This means there seems to exist a trade-off between the degree of risk reduction a strategy can offer and the ability to part-take in positive market movements. However, it can be argued that from a traditional risk management perspective the layered hedging programs are still preferable, as the goal of a passive hedging program should not primarily be to generate a positive return, but to reduce risk.
Companies manage their currency risk against a variety of variables. Some tend to focus more on achieving certain budget rates, while others try to minimize the risk associated with individual transactions. Yet other companies take a larger view in focusing their risk management efforts towards reducing volatility in earnings. This article has demonstrated that simply extending the hedge horizon of a rolling hedging program is not enough to effectively reduce earnings volatility.
The company must also consider the way it implements the hedge instruments. More specifically, it was shown that a layered hedging program possesses much more desirable risk characteristics than a rolling program. Of the examined strategies, the participating forward exhibited the best improvement in terms of risk reduction, with the collar as a close second. The former strategy was able to reduce both the standard deviation and worst-case outcome by about 50%, while at the same time allowing for some participation in beneficial movements as indicated by rather favourable Rd and best-case statistics.
We strongly suggest that companies that want to reduce earnings volatility take note of these results. It would appear that a considerable amount of risk reduction could be obtained by initiating hedge coverage early and over an extended period of time. Also, compared to a simple rolling hedging program, a layered approach should reduce the negative impacts from issues such as forecast error, opportunity cost and accounting impacts usually associated with longer-term hedging.
To further confirm our findings identical analyses have also been conducted for a EUR functional company with sales in both Pounds sterling and Japanese yen. While not included in the article, these results point in the same general direction as what has been outlined here. The fact that we find similar results in different currencies adds validity to the overall message conveyed.
1‘The hedge tenor dilemma’, D. Hirigoyen, The Treasurer, May 2003.
2‘CitiFX First Annual Corporate Risk Management Survey’, M. Daniel and D. Hirigoyen, The CitiFX Currency Advisor, Corporate Edition, Issue 8, October 2005.
3Layered programs with a one or two year maximum tenor would add a further 1/4 or 1/8 worth of cover each quarter respectively. These results are not included in this article.
4‘Re-defining Risk: Implications for Portfolio and Risk Management’, A. Miyamoto, Currency Issues for Asset Managers, Issue 25, May 2005.