Ambiguous business cycles

There has been important work on the impact of risk and uncertainty on business cycles. Ilut and Schneider have presented an important paper on ambiguous business cycles published in the August 2014 American Economic Review. (I have provided the link to the working paper.) 

The representative household is averse to ambiguity which is Knightian uncertainty. Knightian uncertainty is that which is not measurable as opposed to risk which is measurable. Ambiguity averse agents lack the confidence to assign probability to all relevant events. 

Because of this lack of confidence, they act as if they evaluate plans using a worst case probability from a set of beliefs. When in doubt, choose the worst case because you cannot fully measure all of the alternatives and distinguish the correct one. If a loss of confidence is tied to increases in the width of forecasts, it makes the worse case worse. If agents then move to the worst case given their aversion to ambiguity, a loss of confidence is the same as receiving bad news. An increase in confidence which narrows the distribution of alternative is the equivalent to receiving good news. Confidence shocks are real and effects consumption and investment given agents have ambiguity aversion. It can explain the excess return on capital when there is a recession. Investors must be paid when there is a lack of confidence

Forecast dispersion is counter-cyclical so this ambiguity model can explain some of the behavior in business cycles. Confidence shocks will affect all major macro variables, so a loss of confidence can cause recessions. This falls into business cycle models that are based on pessimism but with more richness and detail. Agents come up with alternative beliefs that seem to fit the facts and then focus on the worst case in the dispersion of choices because they are adverse to this ambiguity. This type of model is consistent with time varying risk based on disasters, uncertainty, or forecast dispersion. 

So how do we use this type of model. It tells us that disperison of forecasts is important. Aggregate uncertainty is also important. Time varying risk premium are consistent with a lack of confidence. If investors do not know what is going on and cannot come up with consistent forecasts, macro aggregates will be decline, and by some measure asset prices will also be in decline. You need to incorporate sentiment in models to ensure that you capture this lack of confidence at critical times.