2 Need for central risk management (continued)

2.5  Central risk management is difficult

Several practical problems make central risk management especially difficult for the Crown.

First, concerns are sometimes expressed that opening the door to central direction of risk taking could encourage political involvement in decisions that the government had previously decided were best delegated. For example, if ministers are encouraged to direct the NZSF to take a particular approach to foreign-exchange risk, they may be less inhibited about directing it to invest in particular businesses, for reasons unrelated to risk management.

A second problem is the inevitable approximation in any measure of the Crown's exposure to risk. Risk measurement is difficult even for organisations much smaller and simpler than the Crown. Future variances and correlations are unknown. They may be estimated from samples of historical data, but the samples may not be representative. And variances and correlations change over time. During a crisis, they may suddenly increase, aggravating an organisation's exposure to risk. Other assumptions introduce further approximations. We assume that the annual changes in the values of assets and liabilities are normally distributed. This simplifies the analysis, but the distributions of actual changes in values of financial assets are often found to have fatter tails than those of a normal distribution. That means that extreme events are more likely. Our modeling may therefore underestimate the risk to which the Crown is exposed.[6]

These problems apply to any organisation. Further problems arise for organisations such as the Crown that hold many untraded assets and liabilities. To estimate the risk in the Crown's portfolio, the government needs to estimate the value of assets such as schools, roads, and the present value of future tax revenue. It also needs to estimate variances and correlations for these assets. Estimates can be made, but in some cases they may be no more than partially educated guesses.

And even accurate risk measurement reveals only probabilities, not a prediction of the future. Analysis might conclude, for example, that the probability of the Crown's comprehensive net worth declining by more than, say, $50 billion next year under current policies was only 1 in 20 (see below for our estimates). Even if the measurement is correct, it does not imply that the Crown could lose no more than $50 billion next year.

For these reasons, it is sometimes argued that traditional risk measurement is worse than nothing: not only are the results false, they create a false sense of security (Mandelbrot and Hudson, 2005). A more conventional view is that rough estimates of exposure to risk are better than no estimates. Decisions have to be made, and in the absence of estimates from a model they will be made on the basis of intuition alone, which may be very poor. Yet it is clear that results of risk measurement should be treated with great caution.

A third limitation on the extent of feasible risk management is the difficulty of precisely specifying the Crown's risk-measurement objective. If the objective is vague—keeping the chance of financial distress low and maintaining fairly stable tax rates and spending plans—a broad range of exposures to risk may be more or less equally acceptable: measurement may not identify any policy changes that clearly improve the Crown's position. Being precise about the Crown's appropriate objective is, however, difficult.

It is common to frame choices about risk in terms of the expected return and the risk of a portfolio.[7] Increasing the expected return of a portfolio generally implies accepting more risk, so the owner of a portfolio must generally trade off risk and expected return. In a simple world, the owner can say which of the feasible combinations of risk and expected return is best, and a risk manager can then adjust the portfolio to achieve the desired combination. This would be hard to do for the Crown. Typical Cabinet ministers are unlikely to have a numerical view about the risk to which they are prepared to expose the Crown. They are unlikely to want to answer questions such as, ‘Is it better for the Crown's portfolio to have an expected return of 8% and volatility of 15% or an expected return of 10% and a volatility of 20%?'

There is also a more fundamental problem. The Crown's exposure to risk should not be chosen in isolation from New Zealanders'. Indeed, if the Crown's goal was simply to reduce its exposure to risk, it could adopt a tax system that calculated tax due not as a proportion of income but as a proportion of the Crown's spending. Because this just transfers risk from the Crown to New Zealanders, it is unattractive as risk management.

Some insight into the formulation of the Crown's risk-related objectives can be found in the practice of asset management by pension funds and insurance companies.[8] These organisations have liabilities to their customers in the form of insurance contracts and pension obligations, and they have assets that help them meet their customer liabilities. The difference between their assets and liabilities is their net worth. Their objective can be expressed in terms of the risk and expected return of their net worth. Moreover, risk management can sometimes be framed as choosing the assets to achieve satisfactory levels of risk and expected return, taking the customer liabilities as given.

The government's choice might be conceived similarly. That is, risk management for the Crown might be viewed as selecting certain assets and liabilities to achieve desired risk and expected-return objectives, taking other assets and liabilities as given. For example, the composition of debt and financial assets could be varied, while other assets and liabilities, including the present values of future tax and spending was taken as given. Doing so would help distinguish between risk-reduction policies that involuntarily expose New Zealanders to greater risk and those that do not.

But taking other assets and liabilities as given is not always desirable. For example, the Crown may be exposed to too much risk, even after the composition of debt and financial assets have been optimally chosen. If so, the government must think about changing taxes and spending plans. Taxes might be shifted to less volatile bases; eligibility criteria for spending might be changed to make spending more predictable.

These problems do not, however, mean that risk management is impossible or that risk measurement is futile. Even in the absence of consensus about a precise goal, it may still be clear that the Crown's exposure to risk is too great—or, alternatively, that it can safely be increased. Rough estimates of the Crown's exposure to risk, and of how the exposure varies with policy changes, can help inform those judgements.

2.6  But improvement is possible

Given these problems, radical departure from the government's current approach to risk management should not necessarily be expected. But there appears to be room for improvement.

First, the government should routinely measure the Crown's exposure to risk and estimate the effect of major risk-related policy changes on the exposure. Second, the government should seek to clarify the Crown's risk-related objectives. It should not necessarily aim for numerical precision. But refinement and clear expression of those goals would be desirable. Together, better measurement and a clearer view of the goal of risk management should allow the Crown to decide whether to change the Crown's exposure to risk.

Changes could be made in several ways. One option would be for the government to direct the managers of the Crown's main financial assets and liabilities to change their portfolios in certain ways—for example, to hedge or not hedge foreign exchange-rate risks; to hold or not hold domestic shares or foreign shares; to try to match their assets or liabilities with other of the Crown's liabilities or assets. A second would be to centralise management of those portfolios, either by a direct transfer of the assets in the portfolios or by requiring the managers of the portfolios to invest only in specific assets. A third—perhaps less likely option—would be to continue to allow those managers to choose their own approaches to risk management, but for a central risk manager to use derivative contracts to change the Crown's exposure when that was deemed desirable. Such an approach could involve ‘undoing' certain risk-management actions of the managers of portfolios, if the interests of the portfolio diverged from the interest of the Crown. Lastly, spending plans, tax rates, and tax bases could be changed with a view to changing the Crown's exposure to risk.