Over the last few weeks, you have likely heard that we are either in or are definitely heading towards a recession. But, within the last few days, various news outlets have started reporting that recessionary fears are overblown. You may have even thought… how could there be such a quick change in reporting? Is there not a consensus prior to uttering the “R word”? Well, aside from the usual clickbait that comes with reporting, part of the problem lies in the fact that there are multiple ways to model / define recessions. Actually… there are technically an infinite number of ways!

This is because no one person’s definition has managed to satisfy all of the economic sensitivities that households and firms experience in our economy. So, economists and financiers continue to invent new ways of defining recessions to satisfy the sensitivities that matter most to them and their stakeholders. The chart below displays just five of the most heavily referenced definitions:

As shown, some definitions appear far more frequently over time while others seem to almost overlap. The latter point is expected, however, since many definitions share variable inputs. But, there is definitely an issue with consistency which leads to a lack of consensus. Moreover, there is an issue with whether or not a definition is data-driven (e.g. Hamilton – a clearly-defined mathematical definition assigns the indicator) or data-informed (e.g. NBER, the most famous indicator – a professional decision to assign “recession” is made by a committee that observes changes in variable importance over time). This is all rather convoluted, I know. It is no wonder why reporters seem to flip-flop on the topic – different subject-matter experts will likely have different go-to definitions!

At Market Theory AI, we follow the data-driven recession modelling route – and we have our own way of defining recessions.

Our biggest differentiator is that we model more than one type of recession, rather than broadly or near-arbitrarily categorising a quarter as “recessionary”. Our approach is inspired by the guiding principles of general equilibrium theory, which positions us to model against varied economic agents – Firms and Households. Each agent in our model works with or against two economic sensitivities that are bucketed into two categories – Investment vs Income:

Households invest their time by engaging the labour market, while firms set-up an infrastructure to make that labour more efficient. And both receive an income for their efforts or existence.

Prior to classifying a period as recessionary, Market Theory AI’s approach first identifies whether there are any deviations-from-trend across the four sensitivities. If there are deviations, our model classifies the period as either “Firm Sensitive” or “Household Sensitive”:

While this categorisation is generally much broader than other recessionary measures observed, this initial condition provides us with a warning map since economic sensitivities often precede recessions. When we begin to look at recessions from this perspective, we are able to more easily identify where there are stress points in the macroeconomic structure. From here we can be proactive instead of reactive.

But, as you can see, there is a lot of colour here… actually, there are more coloured bars than empty slots. So to call each of these economically sensitive quarters a “recession” would be the professional equivalent of “the boy who cried wolf”! This is why the sensitivity map is a necessary but not sufficient condition to our recession modelling.

For a period to be properly classified as either a “Firm Recession” or “Household Recession”, both sensitivities need to be met for either economic agent:

This brings us from 66% of quarters being labelled as “sensitive” (since 1969) to 22% of quarters being labelled as “recessionary”, of either type. And from here we can classify a period as an “Economic Crisis” when both recession types happen concurrently  (representing 10%). By comparison, the NBER classified 14% of quarters as recessionary, Piger 15%, Hamilton 18%, Sahm 24% and the OECD 47%.

Through the Market Theory AI model, 2020 experienced two firm-recessionary quarters – and we have not been in a proper recession since Q4 2020. We have, however, experienced several economic sensitivities since Q4 2019, and we are experiencing a household sensitivity now!

This approach has helped many of our clients to “hope for the best, but be prepared for the worst”, and we look forward to sharing these results with our viewership every quarter (as well as forecasts on the topic).

Thank you for taking an interest, see you next week and please feel free to reach-out with any questions.