These respondents were pervasively pessimistic. This may well be a sign of the times. But I suspect that it is generally easier to imagine what can go wrong than what can go right. Personally I would add a scenario entitled something like, “Greying boomers continue to work and play hard”. Under such a scenario work would blur into retirement and boomers in their seventies would continue to make a huge contribution to economic activity through paid and unpaid work, and would consume hard in pursuit of active leisure. The next step is to forecast asset class returns for each asset grouping under each scenario. This can be done using simple determinist return models. Don’t worry about risk at this point. The idea is to imagine what return could be generated by an asset class if that scenario were to eventuate. Here is an example for Australian equities in a scenario that worries me – one in which China’s demand for Australian resources diminishes as its pace of building infrastructure and housing decelerates. In this scenario the world generally recovers from today’s economic malaise, and China itself grows – but in industries that are not as resource intensive.

This has an especially negative impact on Australia. The long mining boom falters, with a flow-on to other sectors, such as housing and banking. Real earnings per share decline over five years, with concomitant impact on reinvestment and dividends. Moreover, the value of Australian shares is rerated downwards as a reflection of pessimism and/or a better understanding of the risks inherent in this narrow economy. Using a simple model with simple assumptions, it is not difficult to arrive at a nominal total return on Australian equities in this scenario of 3 per cent per annum over five years. An awful situation, but one that is entirely possible. Similar calculations are also performed for all other asset classes and for all scenarios. This requires a bit of work, but it is not difficult. It is best to keep the number of scenarios small, and to use aggregated asset groupings. The key is to ensure that the narratives for each scenario are clearly and obviously evident in the derived returns.

A test of this is to ensure that everyone in the team can approximately reproduce any rate of return using the back of a small envelope and a calculator. One point of difference between this scenarios model and the approach promoted by Peter Schwartz is in the number of scenarios. Schwartz recommends that only a handful of scenarios be considered, even suggesting that three is sufficient. But in doing so he is at pains to emphasise that you should not fall into the trap of thinking in terms of base case, good and bad scenarios. These are not narratives, just perturbations on narrow thinking. Instead, go to the effort of developing a narrative, using the tools of fiction writers, including plots and even characters and location. On the other hand, the scenarios modelling technique developed by Susan Gosling will often use 20 to 40 scenarios. The richness of this approach is lost if the number of scenarios is too small. But I do believe that the benefits of parsimonious use of scenarios (that is, keeping it real and tractable) outweigh any potential benefits of greater granulation. Similarly, the number of asset groupings should also be kept as small as possible.