Risk is caused by: positive or negative outcomes of events, or positive or negative volatility of a random variable. When the events or the random variables affect the firm, they are called risk factors. There is plenty of knowledge on how to measure the probability of occurrence of an event, or the volatility of a random variable. In my opinion, however, this measurement, based on past data is not only of little utility, but also can be somehow misleading creating a false sensation of certainty. In certain types of risks, learning from past data is easier than in others. The really important thing we want to learn from the past is the reasons why a certain variable has a certain behavior, or why an event occurs. In other words, we need to learn the determinants of a risk factor. The problem is that these determinants are not always the same; they tend to change over time and from one situation to another. Learning about risk determinants will help us understanding the behavior of the risk factors that affect our firm.
A risk determinant is the reason why a certain risk factor has a given behavior, is what causes such a behavior. We can then state that a risk factor is a function of its determinants; RF: f (risk determinants). Any given risk factor has several determinants, for example, currency devaluation can be caused by several different political reasons and or by several diverse macroeconomic causes. These causes can change over time and across events. Take for example the case of oil prices; everyone will probably agree that supply and demand issues are the main causes for their volatility. On the supply side, we can probably count (i) political issues on the oil producing countries, (ii) technological advances on the production capacities, (iii) interest rates, etc. On the demand side, we need to take into account issues like; (i) climate, (ii) growth of car market, (iii) use of alternatives sources of energy, etc. Each of these variables are likely to cause volatility in oil prices, however, the specific influence of each variable is likely to change over time. For example, political issues in producing countries can be very important at some point in time (they were crucial back in the ‘70s), and have no effect at another one. This is exactly the problem when we measure data from the past: if during the time span of our data set, there is no effect of a particular variable, our model will predict no effect of that particular for the future. So, there are three potential problems here, either (i) the effects of the variable were not captured by our data set because it did not affect the risk factor during the study time, or (ii) it affected the risk factor but we did not know it so we did not have the data to measure it, or (iii) it did not affect the data in the past, and this is the first time it happens. In either case, our backward looking models will fail to correctly measure for the future.
A good example of what I have just stated is what happened with gas prices in the US. As we can see in Chart 1, by mid-2008, the prices dropped from 13 to 3 USD/mmbtu in a few months. This incredible, and unexpected, price drop occurred at about the same time in which gas producers found the way of extracting shale gas, thus dramatically increasing the expected gas reserves.
Chart 1 – Natural Gas Prices
Before 2007, shale gas was expensive and difficult to extract, therefore gas reserves outlook were expected to follow a slight decrease over time, as can be seen in Graph 2, pushing prices upward. When technological innovations allowed gas producers to extract shale gas at a competitive price, the expected gas reserves grew starkly, and prices dropped accordingly (the projections in Chart 2 show that by 2040 around 50% of the total gas production will be shale gas).
Chart 2 – US Natural Gas Production (2012-2040 estimated)
The relevant question here is; how many energy companies were expecting this to happen? This price drop caught most of the industry unprepared. People following gas price dynamics have seen this in advance? Past information on prices did not show any sign regarding the importance of the new extraction technologies and its potential in the determination of gas reserves… thus for somebody looking only at the data, this was invisible, however, for somebody understanding the determinants of the volatility of oil prices, this was probably something easy to see. It is just a matter of be looking in the right direction… The only way of having been able to see this in advance would have been to follow the evolution of the extraction technologies and its importance for the gas reserves. Whoever identified this might have anticipated the fall in prices.
Managers, unfortunately, tend to underestimate the importance of identifying these risk determinants. Short-term issues are usually assigned higher importance and therefore the relevance of risk determinants tends to be neglected.
Sometimes, risk determinants can be a bit detached from normal business operations. They are often discussed in Universities’ research labs, and not close to the firms’ operations. Just to give another example. The evolution of solar energy is still slow with respect to its potential. Solar cells are still expensive, but what if there were an innovation that allows them to improve its cost efficiency? For example higher efficiency in the form of smaller surfaces needed to generate more energy? This is likely to be caused by innovation on new materials: how many energy firms are closely following these innovations? There are several similar examples affecting other risk factors. The volatility of commodity prices and macroeconomic variables, the occurrence of certain political risks, the volatility of the market for talent, the ability to obtain and retain important knowledge in the firms, etc… All these risk factors have their key risk determinant. It is crucial that firms learn how to identify them and find the way of assigning them an owner that is in charge of following their evolution.
The problem is that, as always happens in basic research, lots of these paths take nowhere. It is also true that some of them also take to a large prize, but for most companies is impossible to follow them all. This is against the classic short-termism force that most firms face but is crucial for a good risk management. The owner of the risk determinants has to be aware of what is going on with that specific risk determinant. Control panels of key variables should be monitored to identify key indicators in advance. This would allow firms to have a much better grip of what is going on with the main risks they are facing, and be closer to profiting from the upside controlling the downside of each risk!
