Go back

Decision Theory

Decision Theory in | IE Building Resilience

In this unprecedented health crisis, policymakers all over the world are trying to reduce the spread of the coronavirus while minimizing the impact on economic growth and people’s general well-being. Decision theory could help with that.

By Jeeva Somasundaram, Assistant Professor of Decision Sciences at IE Business School.

 


 

From measuring disease curves to managing people’s emotions, decision theory can help us think strategically as we respond to the COVID-19 crisis on multiple fronts.

Differentiating risk and ambiguity

The common cold and the seasonal flu are examples of risky situations where the likelihood of consequences such as death are well understood, thanks to numerous scientific studies. In the case of COVID-19, however, the mortality rate, incubation period, and symptoms are not yet clear. The probabilities of the disease’s consequences are changing every day as new information comes to light. On the basis of such partial information, it would be erroneous to equate COVID-19 (an ambiguous situation) with an already well-understood disease (a risky situation). The unknown unknowns of COVID-19 could be potentially dangerous.

Interpreting partial information (use Bayes’ rule)

In such ambiguous situations, it is important to understand and analyze information properly. We typically encounter conditional probabilities when dealing with such situations. For instance, according to a recent study, 98.6% of COVID-19 patients experience fever. However, we need to understand that this does not mean that if someone experiences fever, the probability of having COVID-19 is 98.6%. To do a rough calculation, assume 0.5% of Spain’s population (approximately 230,000 people out of 46 million) is infected with COVID-19. Assume that, at any point in time, 0.75% of Spain’s population has a fever due to flu or other reasons (apart from COVID-19). Then, applying Bayes’ rule, the probability of a random person with fever having COVID-19 is only 39.8%. We get such a low probability because of the low base rate of COVID-19 in the population. Recent articles suggest that a majority of COVID-positive patients are typically men with the A-positive blood type. We need to understand, however, that this does not mean that men in the A-positive blood group are more susceptible to the disease. Bayes’ rule should be used to calculate the correct conditional probabilities.

Dynamic reactions to regime shifts

One of the major challenges for policymakers and the general public is to understand whether or not the disease has progressed to the next stage in transmission. Overreaction during such crises might be costly, but underreaction might be disastrous. Studies in judgment and decision-making have discovered that people underreact to unstable environments with precise signals. COVID-19 is an example of an unstable environment that transitions quickly from one stage to the other. In such situations, even though policymakers get precise signals on the number of positive cases, it is human nature to underestimate the exponential growth rate and underreact. It is important to recognize such biases very early.

Managing emotions

Managing people’s emotions during such crises is extremely important. On the one hand, anxiety might lead symptomatic high-risk individuals to avoid information (e.g. testing or doctor consultation). On the other hand, anxiety might lead asymptomatic low-risk individuals to seek information (for reassurance). This unnecessary information-seeking behavior might increase the load of the already stretched healthcare system. Policymakers should ensure that people don’t panic and feel reassured.

Avoid framing losses narrowly

In such ambiguous situations, the typical objective would be to minimize the maximum possible loss. The loss in this case should be framed broadly (i.e. not just deaths due to COVID-19). For instance, an overstretched healthcare system might not be able to cater to patients suffering from other chronic illnesses. This might lead to deaths that could have been prevented. In addition, extreme measures such as lockdowns over a prolonged period might lead to starvation and psychological disorders. In such situations, measures such as quality-adjusted life years (QALYs) are useful to quantify losses across different domains and make decisions.