Why is spending so resilient?
The resilience of spending can be understood through limit cycle view of economic fluctuations.
A while back Paul Krugman posed what I consider the currently most intriguing macroeconomic question:
More recently, Tyer Cowen was asking something similar in a post:
One of the current macro puzzles is that we keep on receiving good labor market reports during a time of monetary and credit tightening. Which is the missing “dark matter” variable that helps to explain this? [emphasis mine]
Now this is a hard question, and there is no point in pretending that one can provide answer in which one would have high degree of confidence. Better than trying to provide the answer, I think this question calls for a speculative answer. I think it is a question which is supposed to make us reconsider our model of the macroeconomy. And in this spirit, I will propose one possible answer here: Macroeconomic cycles are much more self-reinforcing and correspondingly resistant to shocks and hence longer-lasting than what is commonly thought.
To appreciate this view, I think it is best to contrast a standard macro model of the world with an alternative postulated going back to limit cycle theory and recently revived by Beaudry, Galizia and Portier in their paper ‘Putting the Cycle Back into Business Cycle Analysis’.
View of modern macro models
Modern macro models of the DSGE family are built around a steady-state growth path that is disturbed by various shocks. A key thing to realize is that effects of these shocks are relatively short-lived because the models feature relatively weak propagation of shocks. What do I mean by “relatively short-lived” and “relative weak”? I mean shocks with half-life of several quarters, say 3-5 quarters. In other words, a one-time shock will cause a deviation from steady-state growth path half of which will die after 3-5 quarters.
Now this does not sound too short-lived, and indeed it took DSGE models quite some time to get here. For example my reading of 1990s macro literature is as a shift away from models with shocks that had AR(1)-like effect with half-life of 1-3 quarters to models that featured shocks that followed more of a AR(2) profile and had half-life slightly longer. The key difference is hump-shaped effect of shocks, with shock effects featuring amplification, not just propagation. As far as I know Cogley and Nason raised the shape and persistence of impulse responses as a problem for RBCs (see left picture; solid is data, dotted is RBC model). Meanwhile, motivated by Cogley and Nason, Carlstrom and Fuerst showed that model with financial frictions can yield shock responses that are hump-shaped (right picture; black is standard RBC, blue is their model with financial frictions).[1]
So 3-5 quarters is not too short. But it is still counted in terms of quarters. Shocks with such nature mean that positive shocks can be easily overpowered by subsequent negative shocks, given that the positive shocks loose power in span of few quarters. And this seems to be hard to reconcile with the resiliency of the economy in last one-and-half year: We got pretty much unprecedented energy and food price shocks due to the invasion of Ukraine; we got clearly unprecedented tightening in global monetary policy, and more broadly defined tightening in financial conditions with stock prices dropping by 20+%.; and on top of this the second largest economy underwent recessionary-large swings in activity due to Covid-19 lockdowns. And yet, the global economy continues humming along, and that even includes the European economy, which saw all these shocks plus geopolitical shock on top of it.
In other words, all these shocks combined should have sent us into recession, insofar as modern macro models are concerned. Indeed, some central banks had forecasts that were either outright recessionary (BoE) or recession-like (Fed). Same applies to many private forecasters, with recession being talked about endlessly over last 12 months. This disconnect between the real world and what economic models (and economic analysts) would suggest is behind the soul-searching questions from beginning of this post.
Limit cycle view
Alright, if modern macro models are failing us now, what is the alternative? Enter the (stochastic) limit cycle view of economic fluctuations of Beaudry and co-authors. The key idea is that economic developments feature strong endogenous propagation mechanisms which yield endogenous cyclical behavior ala limit cycles. Of course, we know that the economy does not feature deterministic limit cycles (left picture), but that can be addressed by including shocks yielding stochastic limit cycles (right picture), which makes the cycle length depend on timing of shocks and hence random.
What is at heart of this behavior? There are three factors that the authors point towards. First is strategic complementarity between economic agents’ behavior, such as the fact that firms want to invest when other firms are investing. Second, you need inertia in individual behavior, so that rapid changes in behavior are costly. And finally, you need dependence on some stock variable such as aggregate amount of capital or durable goods. All of these requirements seem very realistic assumption about the real world.
What are the implications of such model? Among other things this means that in certain periods of time the economy is very robust to negative shocks, because these shocks do not have the power to overturn the forces (resulting from the strategic complementarity between agents) that are pushing economic activity higher. They might push us slightly lower for some period of time, but they don’t turn things completely around. And once their effect fades the economy continues in its previous trajectory. In a sense, this view suggests that economic cycles are more like a titanic.
It also means that there is certain dichotomy in terms of shocks. Shocks ranging in size from “small” to “large-but-not-gigantic” proportions do not change the underlying trajectory of the economy, unless we are already close to turning point in the limit cycle. Meanwhile, truly gigantic shocks, like the global financial crisis, are not just deviation from the medium-term trend defined by the limit cycle, but a change in the medium-term trend.
And I believe that this is a fitting description of current economic environment. We had large shocks, all of which we can imagine that in other times they could have caused recession. But in the view of limit cycles they came at a time when the economy was in the upswing stage of economic cycle. And given that they were large-but-not-gigantic, they did not have the power to change this underlying direction of the economy. They caused a slowdown in economic growth, but did not shift us into full-blown recession. Instead, the economy continued growing along, albeit at slower pace.
Another way to think about this is through the perspective of firm behavior. Following the large shocks firms clearly became nervous. But they did not switch to cautious mode of behavior. They did not slash their inventories and they did stop investing. Most importantly, they did not stop hiring, and much less did they start firing people. Why did they continue operating as normal despite their nervousness? Because throughout their micro-markets they still observed demand coming in, which was because of the powerful propagating mechanism at play. Simply, the upward momentum embedded in the strategic complementarity was too strong to be derailed by large-but-not-gigantic shocks.
All this then raises the follow up question: Why are we in the strong upswing phase of an economic cycle? Is it just luck, or does it have to do with something else? And I believe that here the limit cycle view can again provide clarity. Specifically, we know that the pandemic recession and recovery are characterized by a rapid decline followed by a rapid rebound, as economy was shut down and then re-opened. And in the view of limit cycles, this rapid rebound caused by re-opening provides a strong enough impetus that puts us on the self-reinforcing part of the economic cycles. Simply, the speed of the re-opening is carried through the mechanisms embedded in limit cycle model so as to create a powerful momentum that is carried into the future.
Conclusion
To summarize, the limit cycle view suggests that a powerful enough shock, such as the rebound when economies re-opened after the pandemic recession, can put us on a upward spiral, and that such spiral features such a strong self-reinforcing mechanisms that even large negative shocks cannot derail us. This in contrast to standard DSGE models that feature only relatively weak propagation of shocks, and, crucially, do not feature any medium-term cyclical behavior.
The limit cycle view can also explain the puzzling combination of empirical data we observe in some countries like Czechia, where some data scream recession, but overall the economy does not look like undergoing recession. As I pointed out in my earlier post, it feels like economy can record contraction but avoid entering the recessionary self-reinforcing spiral. Again, this is entirely plausible in limit cycle view.
Still, I don’t know how much I believe the limit cycle view. I consider it as interesting alternative, but I don’t think it is my main model of the macroeconomy. What I know for sure is that the last year and a half made me consider this view much more seriously that I would have before.
[1] Cogley and Nason, Output dynamics in real-business-cycle models, 1995.
Carlstrom and Fuerst, Agency costs, net worth, and business fluctuations: A computable general equilibrium analysis, 1997.