The Economist  this week led with this subheader: Action on climate is justified, not because the science is certain, but precisely because it is not. The underlying argument is that immediate action is akin to taking an insurance policy—you can’t wait until you have hard evidence in hand, because by that time, you can no longer protect yourself against a catastrophe.
This stresses an important point that nevertheless is often forgotten: Policy makers always make decisions under uncertainty. Only under special circumstances can they assume certainty equivalence , which would allow them to ignore uncertainty around central projections. But in most cases optimal decisions cannot be based solely on a central forecast. The character of decisions can change dramatically if uncertainty has to be explicitly taken into account. For example, it might be optimal to opt for policies that work relatively well in all possible scenarios. These are so-called no-regret policies. Or it can be optimal to follow cautious policies that prevent extreme scenarios. Taking uncertainty into account can also lead to the conclusion that you have to act soon (to prevent extreme scenarios) or, on the contrary, that it is optimal to postpone decisions (until uncertainty is reduced).
The challenges associated with uncertainty go far beyond these examples of optimal government policies. I realized that last week when I joined the final session of the workshop Economic and Environmental Consequences of Large-Scale Biofuels Expansion, organized by Dominique van der Mensbrugghe  and Govinda Timilsina . The workshop brought together research that was sponsored by the Knowledge for Change  program and based primarily on simulations using our ENVISAGE model (Environmental Impact and Sustainability Applied General Equilibrium Model).
The workshop focused on the economic viability of biofuel production, on the consequences for food prices, and how that all would be influenced by changes in policies and changes in oil prices. The closing discussion stressed the crucial role of uncertainty. Even if biofuel production is on average economically viable, there might be significant periods during which production is not profitable. Even if in the long run biofuel production does not threaten food supply, it can do just that in the short run.
I came away with two conclusions from that discussion. First, governments should not only take uncertainty into account when they design their policies, but they should also aim to reduce uncertainty for other agents by making their policies stable and predictable. Second, modelers should try to deal better with uncertainty. Not only by designing alternative scenarios, but also by exploring how unpredictable short-run volatility can influence individual behavior and market behavior.
So, uncertainty is a call for action, not only for policy makers as The Economist rightly observed, but also for modelers.