The Problems of Assessing Risk with Active Agents

Risk analysis is not perfect.  We do out best with available data to make forecasts subject to uncertainty, to estimate how things can go wrong and how likely they are.  We can make recommendations and I like to think that this has value for clients.

In all of this, one of the biggest challenges is to incorporate one aspect of individual behaviour into a forecast.  As an illustration, compare investing is a company producing consumer electronics with investing in an infrastructure project.

An infrastructure project, say a road or a railway, is strongly local.  It serves only those people wishing to travel between a set of origins and a set of destinations.  If there is a recession, if people become unemployed or fewer goods are needed to be moved around then there is less demand for the infrastructure and it may make a loss.

If we look at the case of a company producing consumer electronics it can look similar at first glance.  A recession reduces demand – indeed possibly more so if the goods are seen as luxury goods.  A risk analysis looking only at the effects of external factors might mistakenly suggest that this was higher risk than it is.

The reason that consumer electronics companies can manage this particular risk better is the flexibility in their market.  If there is a recession they can export surplus that can be sold on the domestic market – this is not ideal as there will be transport, tariff and possibly translation costs.  A road cannot relocate to serve more affluent communities.  The electronics company can scale back production to cut costs to better weather a recession; you can’t simply rip up a lane on a 3 lane motorway to cut your costs.

There is a flip side to this however – some account may need to be made for behaviour of competitors   Roads or railways do have substitutes, specifically other modes of transport, but they do not tend to be close substitutes.  Infrastructure tends to be a natural monopoly and it is somewhat protected against competition risk.  Electronic goods manufacturers on the other hand can lose market share to competitors products.

There are a few lessons to learn from this; firstly risk analysis has to look at the purpose of the assessment and the assumptions.  Is this assessment the risks in a “do nothing” scenario?  Are we looking to identify strategies to reduce risk?  Are we looking to asses the risk given a particular existing strategy to manage risk?

The second lesson is that this can highlight the lack of a coherent measure of risk; what is low risk for high probability changes in value can become very high risk in the context of more rare circumstances.  The value of a road or railway may be very susceptible to a number of parameters and may generally appear higher risk (say at a Value at Risk type measurement).  In extreme circumstances, say a global recession, the consumer electronics manufacturer may find no export markets to export to.  Investors in a failing road project still have a large asset – the loss given default has a soft upper bound.  If a consumer electronics company goes into receivership and its assets are sold off they may not fetch much of a price and depreciation on technology tends to be very high.


This entry was posted in Risk, Short Post and tagged , , , , , . Bookmark the permalink.

Leave a reply