This article is the first in our new “Digital Strategy series.”

    With the advent of technologies such as machine learning and data analytics, corporate treasurers around the world are poised to unlock new and efficient ways to tackle age-old optimisation problems in both the risk management and capital management space.

    However, these digital tools can be fully leveraged only when they are combined with equally strong “domain knowledge” in the people using them. Perhaps counterintuitively, digital tools cannot simply replace human intuition, rather they require greater clarity of it to function to its full potential.

    The digital revolution presents not only a great opportunity to transform processes, but an ideal time for corporate treasurers to revisit more fundamentally the nature and extent of the risks and opportunities that they are striving to optimise.

    In this brand new series, we will look at various aspects of risk management through a digital lens, exploring how this partnership between man and machine can define a new world of treasury best practices.

    Risk Management 101: Classical Finance

    The classical theory of risk management assumes that people are “risk averse.” Simply put, it is the idea that people will not pay more than USD50 for an equal-probability bet of either winning USD100 or nothing. (Behavioural science confirms this to be true, in fact an experiment by Daniel Kahneman and Amos Tversky have shown that most people would not even pay USD46 for the bet, as we saw in our first article in Rethinking Treasury Series.)

    The concept of “risk aversion” underpins all financial asset valuations. Rational investors demand extra returns over risk-free interest rates for holding risky assets, and riskier companies will therefore face higher costs of capital as a “penalty” from investors.

    Takeaway of Risk Management 101: “Risk Minimisation” is “Firm Value Maximisation”

    From a corporate’s perspective, de-risking market exposures represents a low-hanging fruit when it comes to increasing a company’s valuation.

    Unlike equities and bonds, market risks typically faced by corporates are not compensated by any positive expected returns. In essence, FX, rates and commodities markets are simply avenues for macro risk transfers.

    Having open, unhedged exposures in zero-sum markets such as FX, rates and commodities, effectively amounts to speculation and these unproductive risks present potential “dead-weight losses” to a firm’s valuation.

    The solution, viewed through this academic lens, is straightforward. To the extent that exposures are highly probable and measurable, the ideal hedge ratio would be 100 per cent. (Note 1)

    Classical theories assume that humans make decisions based on absolute knowledge and logic. In reality, we make decisions based on relative knowledge and emotions.

    Risk Management 201: Behavioural Finance

    The academic “prescription” that the ideal hedge ratio should always be 100 per cent, may be difficult to fully embrace and that is a perfectly normal position to take.

    Classical theories assume that humans make decisions based on absolute knowledge and logic. In reality, we make decisions based on relative knowledge and emotions.

    Absolute knowledge entails always evaluating the net position of a company, whilst relative knowledge entails only looking at a compartmentalised position of the company, such as a particular hedge.

    When we make and evaluate hedging decisions based on relative knowledge, people tend to face higher “regret risk” with hedging than not hedging, due to our tendency to evaluate decisions after the fact and judge their success on what would have been the best outcome. (See sixth article of Rethinking Treasury series on “hindsight bias.”)

    Takeaway for Risk Management 201: “Regret Minimisation” not “Volatility Minimisation”

    As we saw in the ninth article of Rethinking Treasury series, regret minimisation may be a better alternative than volatility minimisation in terms of its intuitive appeal when determining risk management decisions.

    This is also why many companies “feel” that the ideal hedge ratio should be middle-of-the-road, i.e. closer to 50 per cent, rather than 100 per cent. Indeed, it can be shown that a 50 per cent hedge ratio does minimise the potential for large regrets for linear strategies.

    (It can also be shown that some non-linear strategies may potentially achieve an even better “regret” profile than a 50 per cent linear strategy. See the ninth article of Rethinking Treasury series for more detail.)

    Risk Management 301: Data Analytics to the Rescue?

    As products of our evolutionary past, behavioural biases are persistent and perhaps impossible to eliminate. (Note 2)

    Therefore, we think that rather than giving in to classical finance completely by suppressing our natural instincts, there may be merit in considering alternative metrics in risk management, such as those that are grounded in behavioural science.

    For example, for large and one-off exposures, companies may deem that the “regret risk” is too high. In such cases, companies should have the flexibility to choose a “regret-based” metric rather than traditional ones.

    On the other hand, for small and repeat exposures, companies may be more interested in smoothing out volatility over time and achieving long-run optimality. In such cases, popular metrics such as VaR and analyses that look at long-run backtested outcomes can be considered.

    Upon defining the appropriate metric, we can leverage historical financial market data and also apply techniques such as Monte Carlo simulations to produce analyses that have less historical bias.

    In summary, with all this data on our hands, we can apply machine learning, a rising technology used for data analytics, to help solve various risk management problems. The pre-requisite would be to identify and agree on an appropriate metric to use for a particular context.

    In subsequent articles of this series, we will explore specific examples of how machine learning can be applied in risk management problems. Stay tuned.

    Download Simplified Chinese version

    Disclaimer

    More, collapsed
    Join the conversation?

    Join our Linkedin group to get an unparalleled view of macro and microeconomic events and trends from a bank that is a leader in both developed and emerging markets.