The Role of Market Forecasting in Risk Management

The 1 August 2011 marked the 150th anniversary of public weather forecasts in the UK. These early forecasts, first published in The Times in 1861, were prepared by vice admiral Robert Fitzroy (who famously captained Charles Darwin’s HMS Beagle). Fitzroy, who began forecasting the weather using innovative methodologies such as synoptic charts (a technique involving the charting of weather observations taken simultaneously to predict future conditions, which is still used today), had at least one thing in common with many modern day financial market forecasters – his forecasts were often wrong, and he was frequently the subject of criticism and mockery, even from the editorial staff of his own publisher.

Financial market forecasts – including foreign exchange (FX), commodity price and interest rate forecasts – are often a key element of the corporate risk management process. When deciding upon an FX hedging strategy, for example, the future direction of FX rates can have a significant impact on the relative performance of various hedging strategies. As such, it seems only natural to look at market forecasts, either internally generated or from banks and independent analysts, before deciding upon a risk management strategy. However, just like Fitzroy’s early weather forecasts, financial market forecasts are also frequently inaccurate, and often dramatically so. Noted market strategist and blogger Barry Ritholtz once suggested (only half jokingly) that market pundits should be forced to include the following disclaimer after all forecasts: “The undersigned states that he has no idea what’s going to happen in the future, and hereby declares that this prediction is merely a wildly unsupported speculation.” So, if there is little evidence to support the ability of forecasters to consistently predict financial market movements, does this mean that forecasts should be ignored and their role in the risk management process eliminated?

The answer to this question will often depend on exactly how a forecast is used within the risk management process. While allowing unreliable market forecasts to determine hedging strategies can clearly be dangerous, this does not mean that forecasting cannot play a valuable role in the risk management process. The preparation of market forecasts encourages an analysis of the factors which may drive underlying market movements, and reviewing alternative forecasts requires the consideration of different possible scenarios and outcomes. Both of these activities are beneficial, ensuring market dynamics are better understood, and the effectiveness of potential hedging strategies can be reviewed under different future outcomes. However, forecasting can become less beneficial, and even detrimental, when ‘single-point’ forecasts are used and relied upon to guide hedging decisions. A common mistake is aggregating a number of different forecasts and taking an average forecast – this enforces a deterministic mentality, and reduces the value of the forecasts themselves, which individually might encourage the consideration of different scenarios. Crucially, using an average forecast will necessarily reduce the focus on potential extreme events, the awareness of which should really be the priority of the risk manager. Another danger of deterministic forecasting occurs when too much confidence is placed in the forecast, and alternative scenarios are ignored. This can lead to hedging activity, which is essentially speculative in nature, where risks are either over or under-hedged based upon an expectation of future market movements.

One of the most effective methods of avoiding the pitfalls of single-point, or deterministic, forecasts is to incorporate the concept of probability into the forecasting process. Generating probability ‘cones’, which provide various potential future paths for example FX rates, commodity prices etc. Along with the associated probabilities, this is a relatively straightforward method that can be used to ensure that a probabilistic approach to forecasting is used, and the dangers of single point forecasts are avoided. Probability cones can be generated quite easily using a Monte Carlo engine, and while they are far from perfect forecasting tools, they will often underestimate the likelihood of low probability outcomes; for example, they will at least ensure that the forecasting process considers a range of alternative scenarios, rather than just one. Using extreme forecasts, such as using the high and low forecasts of various forecast providers, is another useful way of ensuring that alternative outcomes are considered, and these forecasts can be used to stress test alternative risk management strategies. Even when using such extreme forecasts, however, it is important to avoid overconfidence; actual market volatility can exceed the expectations of even the most radical forecaster.

Given the limitations involved in forecasting financial markets, it is important to consider the appropriate role of market forecasts within the risk management process. The role of the forecaster should not be confused with that of a fortune-teller looking to predict the future. Forecasts are most valuable when they help treasurers to explore the range of factors that are influencing the markets, and the range of potential future outcomes, which may occur. Equally, forecasting can be dangerous when it encourages users to ignore alternative scenarios and emphasises a directional view. Incorporating a probabilistic, rather than a deterministic, approach to forecasting is an effective way of ensuring that forecasts support the risk management process by encouraging consideration of various alternative outcomes. If the weather forecast calls for a 40% chance of rain – in effect, a prediction of no rain – it might still be a good idea to bring your umbrella.


Related reading