 Chris O'Brien Director of Centre for Risk and Insurance Studies at Nottingham University Business School
Insurance companies run the risks that other firms and individuals don’t want, but how do general insurers know how to run the right amount of risk?
This question is sometimes expressed, how do firms set their risk appetite? It is not clear that we can easily answer a question such as how much do firms like risks? We might say that an individual likes risks (or apples) but the concept may not easily transfer to a firm run by a board of directors and with multiple stakeholders.
To answer the question, we need to look to the firm’s objectives. An insurer decides its pricing policy, say, as that which will best contribute to its objectives; its risk policy should use the same principle. In the case of a proprietary company the objective may well be shareholder value, although recognising that the interests of others - in particular employees and customers - impact on that. Therefore, the simple answer is that the insurer has an appetite for all risks that can add to shareholder value.
It is therefore important for the Chief Risk Officer to understand the way that risks affect shareholder value. Financial theory suggests that managing risks can enhance shareholder value in a number of ways. For example, if the firm’s cash flows are more predictable, then the costs of planning and co-ordinating resources can be lower. Risk management can also increase value even if the expectation is that profits are likely to be lower: lesser risks can mean a lower probability of insolvency, with a saving on the costs of potential financial distress, and the loss of goodwill if the firm eventually ceases to operate. Lesser risks, especially if supported by a better credit rating, can improve the potential for profitable business. On the other hand, a lower likelihood of insolvency means that the value to shareholders of the put option to default is reduced.
One role for the CRO is to recognise that decisions about risks are taken by managers, and can therefore reflect managers’ views about risk. These may differ from the perspective of shareholders. For example, there is plentiful evidence that remuneration structures such as the use of shares or share options influence decisions about how much risk to take (e.g. Tufano, 1996).
It is also recognised that managers’ understanding of risk may be inadequate: they may treat high probability events as certain and ignore low probability outcomes (Helliar et al, 2001). Risk professionals need to do their best so that they do not succumb to these biases. To make sound business judgements, a good manager recognisesthe range of possible outcomes in a situation, including low-probability events. He or she is good at picking up signals that an outcome is beginning to look adverse, and responding accordingly before the worst happens.
However, can we say what the ‘right’ probabilities are? When the floods hit the Cumbrian town of Cockermouth in 2009, Hilary Benn, the Environment Secretary, said that whilst the area's flood defences had been built to withstand a "one in 100 year" flood, "what we dealt with last night was probably more like one in a thousand”. It is not always easy to get a good grip on probabilities; we hope actuaries can do better than others.
As regards financial risks, the global financial crisis has alerted us to the highly improbable. An example of the problem is that the Financial Services Authority set capital requirements for with-profit life insurers in 2004 with stress tests based on a 99.5% likelihood of being solvent in a year’s time. These included share values, property prices and interest rates changing by 20%, 12.5% and 17.5% respectively. In 2008 they changed by 33%, 26% and 17.6% (O’Brien, 2010). It is clearly difficult to estimate low frequency events, and as Aviva (2009) said, “Over the last century it could be argued that the economy … has suffered 6 one in two hundred year events”.
A lesson from the global financial crisis is that banks’ use of Value at Risk was not a sound way to measure risk. There are inevitably questions about how Solvency II, based around 99.5% VaR, will fare. The regulators will doubtless challenge insurers about their models; internally, the CRO has to adopt a critical (but not negative) approach to the model.
Stress tests avoid the problem about probabilities being questionable by defining a scenario without saying how likely it is: on the other hand, there is subjectivity in saying whether it is of the right strength to set capital requirements. The ‘realistic disaster scenarios’ (RDS) of Lloyd’s are one approach to setting scenarios that can stimulate discussion on what risks a general insurer should run. Amlin records, in its annual accounts, the largest RDS as a proportion of its risk appetite.
We cannot make sensible decisions without some idea of the probabilities of alternative scenarios which, after all, are encompassed in some way in shareholder value. Woo (2002) showed that deterministic approaches of probable maximum loss had severe limitations, and probabilistic modelling has been an important step forward. But it is not a complete solution; Amlin (2009), writing in the context of the uniqueness of Hurricane Ike, wrote that “Therefore to rely on models as the key risk management tool is folly”. Yes, probabilities are important, but we must also recognise their limitations.
Amlin (2009). Annual report and accounts 2008.
Aviva (2009). Policyholders’ future security and risk appetite: Policyholder advocate’s report, para 16.12.
Daily Telegraph (2009), http://www.telegraph.co.uk/news/6616952/Cumbria-floods-once-in-a-thousand-years-deluge-swamps-defences.html
Helliar, C., Lonie, A., Power, D. & Sinclair, D. (2001). Attitudes of UK managers to risk and uncertainty. Institute of Chartered Accountants of Scotland.
O’Brien, C. (2010). “Insurance regulation and the global financial crisis: a problem of low probability events”, Geneva Papers, 35, 35-52.
Tufano, P. (1996). “Who manages risk? An empirical examination of risk management practices in the gold mining industry", Journal of Finance 51, 1097-1137.
Woo, G, (2002). “Natural catastrophe probable maximum loss”, British Actuarial Journal, 8, 943-959.
|
