Customer Reviews

Shaping the Next One Hundred Years: New Methods for Quantitative, Long-Term Policy Analysis

5 star:		(4)
4 star:		(0)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

I first got interested in the long-term prediction problem when I came across a funny story in Taleb's The Black Swan: The Impact of the Highly Improbable about this planner trying to forecast the oil price ten or so years ahead. The point is that such estimates are required to price certain asset classes, among other things. For instance, the present value of a pension fund's liability can depend on the interest rates that take place 30-70 years from now. How is one supposed to estimate that? Apparently, any quantitative model based on the current data will have to make predictions that are too out of range to be trusted. Besides, how can one incorporate a variety of political and environmental events that are likely to affect the interest rates? Taleb's book asks such questions, but does little to answer them.

I like the book of Lempert et al because, first of all, they admit that "conventional" modeling for the purpose of prediction is useless when applied to long-range problems. A model may generate a set of plausible scenarios, but an attempt to assign more or less precise probabilities to each of them is bound to fail. In that case, choosing a strategy based on that questionable probability distribution (e.g., maximizing the expected utility) is futile. Instead, they offer to choose a strategy that is robust in the sense that it works "well enough" for a large proportion of possible scenarios.

Generating the scenarios themselves involves the elements of "conventional" parametric modeling and, importantly, a good deal of human intervention. The authors readily admit that the conventional part inherits a number of limitations. Also, the model cannot be validated in the standard sense because there is no possibility to test it "out of sample" when the range is 100 years or so. The purpose of such model is not to forecast but to generate a rich set of scenarios to find a robust decision rule.

Getting back to the pension fund example above, the approach of Lempert et al is clearly unable to provide a Holy Grail answer as to the precise value of long-duration liability. A "robust" range of possible values can be obtainable, but, again, do not expect it to possess an easily determined "confidence level" like in a standard prediction problem. As the authors say, they "do not put pressure on science and experts to seek more certain judgments that are warranted or possible". Therefore, I would recommend this book to anyone interested in the limitations of quantitative modeling.

The visionary ideas put forward by Lempert, Popper and Bankes work not only for business analysis but provide thought -provoking models for many different types of long term planning. Of particular interest to the general reader are the discussion of conditions of deep uncertainty and the necessity of seeking robust strategies.

The collaboration of Robert J. Lempert, Steven W. Popper and Steven C. Bankes, Shaping The Next One Hundred Years: New Methods For Quantitative, Long-Term Policy Analysis offers complex mathematical analysis techniques for making long-term policy predictions. Sample models, ways to analyze sustainable development, robust and adaptive strategies, and more fill the pages of this technical guide focusing upon quantitative approaches to long-term policy analysis. An excellent guide for forward-thinkers and strategists in business who must take into account such diverse considerations as environmental protection, political terrorism, genetic technologies, international trade developments, and more, Shaping The Next One Hundred Years is very strongly recommended reading.

"Shaping the Next One Hundred Years" by Lempert, Popper and Bankes is as an exciting new book on strategic thinking and is comparable in importance to Michael Porter's "Competitive Strategy ". The book describes a methodology that mirrors how successful business leaders think while using novel software to expand and codify the decision making process. Lempert et al accomplish this by using exploratory modeling to evaluate the myriad possible futures a decision maker faces and then comparing different strategies against the desired metrics. The results are new insights and a strategy that works regardless of unpredictable events the future. The methodology is revolutionary and will become the preeminent way important decisions are made.