Deciding how to decide

A recent issue of the Harvard Business Review highlighted an article "Deciding how to Decide" by Hugh Coutney, Dan Lovallo, and Carmina Clark. It plots the choices of tools available for making a good decision. It is a nice piece on the what is available to executives who have to make difficult decisions. Their framework can also be very useful within investments. Finance seems to rely too heavily on quantitative methods which often are not appropriate given the type of uncertainty faced.  In fact, setting up the problem may be the most useful part of decision analysis. You have to know what you know and know what you do not know. If you don't have the right information or data, or you cannot frame the problem within specific bonds, you are limited in your choice of tools. They break the world into five types of situations or contexts based on whether the outcomes can be defined and the amount of knowledge you have about the problem. If you can define it and measure it, then you should move in the direction of quantitative tools and if you cannot, then you should focus on cases or analogies. The context through these two lens will determine the type of tool that should be used. 

If you understand the model which describes the situation and you can predict the outcome from a decision with a high degree of certainty then you can use some simple tools. If there is little uncertainty about what the environment will be like, then focus on just discounting cash flows in a straight-forward manner. Unfortunately, there are view of these situations, but you can think about a repeatable event where we have a lot of past information to help us look at this type of situation. This would be a classic textbook problem.

A second situation is when the causal model or links are well understood and there is some uncertainty and it can be easily bounded. If the information or knowledge required is not disperse than quantitative tools with scenario analysis should be used. This would be the classic type of work that is undertaken by an investment quant. If knowledge is more disperse, there is a need for some sort of further data analysis or information aggregation to provide context for alternative scenarios. 

If decision outcomes are not easily known, the use of quantitative scenario analysis is not enough to provide an answer. There has to be some form a qualitative analysis coupled with cased-based decision analysis to handle this situation. This would be fundamental analysis with expertise from past knowledge or experiences. 

If the model for describing the situation is not known, the decision outcomes become even more unclear. If set of possible outcomes are known and knowledge can be controlled then, forecasting tools can be applied with case-base decision analysis. This is similar to the second situation but we are not sure of the true model. When faced with new central bank policies we may not understand the causal model that can be applied but we are able to bound what are possible outcomes. 

When the causal model is not known and and the outcomes are not well understood, the only choice is to use a case-base decision making. Look through past similar experiences in order to find situations that may fit the current events. The use of analogies is critical when faced with significant uncertainty.

If I strip away the usual management jargon, decision-making can be broken into two major choices. One, problems where we have a well established model and data to analyze that can tell something about market reactions, and two, problems s where there is more uncertainty about the model to be used and limited data to analyze. In the first case, we use quantitative tools that can be used repeatedly and in the second case where our best option is to look for past situations or cases to compare to the current event. 

When you have a model and data, use it in quantitative approach.

When you don't have a model and limited data, use a qualitative approach and look for past cases.

This seems simple but finding the right framework is critical for decision success.