Today I was reading a pretty interesting article called Diagnosing Recession, written by Òscar Jordà which deals with an improved recession determination model. The disadvantage of the traditional National Bureau of Economic Research (NBER) Business Cycle Dating Committee’s model is that it informs about the start of a recession with a twelve-month delay. Due to the NBER’s admittedly accurate, but rather reactive, not proactive determination of cyclical turning points a more in-time determination of the start of a recession is needed. Policymakers, managers, and consumers will all benefit from a timelier recession determination because they can actually make adequate decisions that apply to the current economic situation and not to the situation twelve months before when the NBER publishes its findings. To counteract this delay, the press is using a very inaccurate approach of determining a recession when two consecutive quarters show a negative GDP growth.
A Better and Timelier Approach
Jordà applied medical diagnostic evaluation procedures to the Chicago Fed National Activity Index and the Philadelphia Aruoba-Diebold-Scotti Index to come up with an in-time diagnostic test and determine economic expansions and recessions. The idea of this model is to access the publicly available indexes and to determine a threshold level for action for both indexes. Once either index passes this threshold, one would then know in-time that a recession is happening. While determining this threshold, one has four possible outcomes: true positives, which is calling out a recession when there actually is a recession. False positives, which is calling out a recession when there is no recession and the economy is expanding. True negatives and false negatives can be explained symmetrically. All those four outcomes come with specific costs and benefits.
To determine the best threshold, one has to consider several factors such as the rate of true and false readings, the underlying incidences of recession and the specific costs and benefits of all four options. While the threshold level affects the true/false ratio, it never affects the real incidence of a current recession happening. Hence, the net benefits for choosing a specific threshold consist of the benefits of correctly predicting a positive (calling out a recession and implementing a stimulus), true negatives (not implementing a stimulus because the economy is expanding), false positives (applying a stimulus to an expanding economy), and false negatives (not implementing a stimulus when there is a recession).
By applying the calculation of medical practitioners that they use to determine whether to do a biopsy, one can get to a threshold at an optimal level. The net benefits of a threshold for a recession will then be calculated by multiplying the expected costs and benefits of each of the four possible outcomes times the frequency of those outcomes and the incidence of the underlying condition. After the calculation, one would just have to run some trial and error attempts to arrive at the optimal recession threshold level. For the Chicago index, the optimal threshold is -0.72 and -0.80 is the optimal threshold for the Philadelphia index (see pictures below). Once either index falls below this threshold, the likelihood of a recession is very high. By applying the calculations of medical practitioners to the Chicago Fed National Activity Index and the Aruoba-Diebold-Scotti Index, Jordà came up with an accurate in-time model for determining a recession.
Limitations of the Method
One limitation of this model is that costs and benefits of all four possible outcomes can be different for policymakers, managers, or consumers. For example, considering that a policymaker’s costs of implementing a stimulus during a misdiagnosed expansion are not symmetric to the costs of not implementing a stimulus during a misdiagnosed recession, the policymaker’s threshold will be different than the one calculated above. The same accounts for the manager and the consumer that both have different cost-benefit calculations than the policymaker. Hence, the actual application of in-time recession diagnostic tests is then more than just calculations and one has to “guess” specific cost and benefit triggers individually. This means that there is no perfect threshold for every person considering their different cost and benefit triggers, but the application of medical practitioners’ calculations to the Chicago and Philadelphia index provide an accurate model for determining a recession in time.
If you are interested in the full article, click here.