I see that Cash-flow at Risk (CFaR) has come up in the ACT’s Linkedin Group. CFaR, like Value at Risk (VaR), is a summary statistic of estimated probability distributions of future outcomes.

At risk measures are used more in financial services and investment management. Broadly non-financial companies (nfcs) do not find particularly helpful – or significantly more helpful than alternative tools. Of course this goes even more for VaR as nfcs normally manage cash flow risk rather than value at risk, other than in minor aspects of their business. CFaR models are also often used in financial services, in modelling derivatives, etc.

But CFaR models were developed in response to demand from industrial firms for something analogous to VaR that financial services seemed to be using successfully (though subsequently use of VaR in that industry and, in Basel II in its regulation, has often been cited as a cause of the 2007-date financial crisis)1 ,2.

CFaR models can, nevertheless, be useful to nfcs in a limited way. Examples might be:

• In illustrating the kind of variability that might be expected even when the world is not falling apart and how that might be limited especially where the company has no comparative advantage in managing the risk (such as, in corporate treasury, financial price risks)3
• In demonstrating that the treasurer is familiar with aspects of financial economics fashionable in the late 90s and the 2000s the over-use of which has astonishingly survived the financial crisis of 2007-date, probably due to the current lack of alternative theories and to academic inertia
• In comparisons with other businesses – for example when looking at diversification acquisitions or when seeking to show a new Non-executive Director that their assumptions carried over from other industries need adjusting in looking at their new firm4 and
• In dealing with industry regulators (of utilities, for examples) as regulators and those who build financial models in governments like reducing complex problems to single (often irrelevant in the long-run) numbers and
• In industries whose products are broadly undifferentiated commodities, such as energy supply, metals supply, etc. whose exposures to macroeconomic aspects dominate product differentiation, brand image, etc.
• In providing something else for consultants to sell to nfcs.

But statistically based models should not be taken too seriously for the reasons given by Nassim Taleb, in his interview with Joe Kolman in Derivatives Strategy at the turn of 1996/65 :

The problem we have with statistics is that although we know something about distributions, we know very little about processes. A process is a distribution that has time in it, and things change with time. People look at fat tails and say, We can simulate distributions with fat tails. But the reason distributions have fat tails may be because these distributions don’t have stable properties over time.

[VaR captures] “The risks of common events perhaps, those that do not matter, but not the risks of rare events.

We can add that “at risk” models take as their statistical base (usually recent) history or our limited models of the future: during the recent financial crisis, investment bankers were reported as saying they were seeing once in 100,000 year events several times a day.

The risks that matter turn out from time to time to be statistically unobservable: we cannot establish a frequency – they are uncertainties rather than computable risks in Knightian parlance6 .

We best deal with Knightian uncertainties (where VaR or CFaR are unavailable) by using scenario analyses. Companies need to be planning robustness rather than optimisation7

Industrial projects often run for a decade or several decades, way beyond the horizon of the financial derivative dealers where “at risk” figures were started.

At risk figures give us an unhealthy sense of security followed by un-planned for disaster when a negatively impacting uncertainty strikes.

Of course, scenario analysis too requires careful use: our imaginations are too limited. We often fail to recognise that really big disasters come when more than one of our assumptions fail at the same time.

www.treasurers.org/node/6911 that quotes Nassim Taleb on the subject.