Excellent thread by @JoMicheII explaining core Keynesian dogma via Joan Robinson. This idea is important and worth understanding. However, the follow-up critique of competing dogma seems invalid (and unnecessary) to me. Let me explain. https://twitter.com/JoMicheII/status/1343890019291955202

Everything JR says up to this point is something every macroeconomist will agree with. This is b/c she's mainly describing accounting to this point -- little or no theory is involved. https://twitter.com/JoMicheII/status/1343893509393305606?s=20

So, for a closed economy, the equation S = I must hold b/c its an identity. But as she mentions, saving is not the same as investment. What does she mean by this?

I think what she means is there are behaviors (decisions that people make) that can have *separate & distinct* influences on S and I and that theses decisions must ultimately be reconciled by S = I. Much of the debate in macro boils down to how this reconciliation takes place.

In the language of economists, what are the "equilibrating forces?" In reality, there are likely millions of dimensions along which adjustments take place to ensure S = I. Can we identify the most important of these?

Traditionally, macroeconomists have focused on income (Y) and interest rates (R) as the most important "equilibrating" variables. Much of the Neoclassical-Keynesian divide rests on which of these two variables is more important.

N-view places much faith in ability of markets to coordinate activity to point where "full employment" occurs. So, with Y essentially determined in this manner, only thing left to adjust is R. K-view flips this around, with R determined by policy and Y adjusting to ensure S = I.

Clearly, some middle ground here. Also, a need to move beyond just Y and R as equilibrating variables. But JR claims there is no middle ground. Only one of these views is correct; the other is logically invalid.

E.g., an increased desire to postpone consumption *can only* result in a decline in Y. The confidence with which this statement is made is breathtaking--there's not one ounce of humility in it. This is too bad, b/c some humility is certainly in order https://twitter.com/JoMicheII/status/1343895509518462977?s=20

Note, I am not saying JR is wrong in suggesting how an increase in the propensity to save can (and mostly likely would) lead to recession. Here is a paper that formalizes the argument in the language of game theory: https://research.stlouisfed.org/wp/more/2020-019 I think @farmerrf might like it.

But what my paper shows is the *possibility* of such an outcome & not its *necessity.* Other things could happen too--it depends. The conditions may seem unlikely to hold, but this is irrelevant from demonstrating logical outcomes (which is what JR is trying to do).

So, how might one counter the argument JR makes here? https://twitter.com/JoMicheII/status/1343897743182483458?s=20

How might an increase in the desire to increase future provisions actually stimulate current and future income/production?

One mechanism is often explained using the "Robinson Crusuoe" metaphor. RC works hard to harvest coconuts (so, full employment) but chooses not to consume harvest today (desire to save). Instead, he chooses to consume it in the future. How? By planting the harvest (investment).

In this example, the composition of current output (b/w C and I) changed, resulting in a higher future capital stock (more coconut trees) and larger future harvests (so, more C and I in the future). This is perfectly logical. The only question is whether it applies empirically.

The superficial objection to an RC economy is that it consists of one person and so does not apply to modern economies. But this result is not sensitive to there being only one person. It can also hold for economies populated by many heterogeneous people.

The key (objectionable) assumption made by the textbook RC model is that all economic activity is perfectly coordinated. This is trivial for an economy consisting of N = 1 people. But it probably also holds empirically for N = 2, 3, 4, ... ?

It's not entirely clear how large an economy must get before coordination failure becomes a concern. In any case, if an economy is at (or near) full employment, then the RC mechanism may very well be relevant.

The result is likely more general though: it's likely to hold for *fixed* employment (i.e., even if employment is for some reason fixed at a level less than full employment).

Do I think this possibility is likely? No. Do I think it's important to study this (and other) possibilities? Yes. More generally, I think we need to be clearer in making a distinction between the logic of an argument and how it's applied to interpret the phenomena we observe.

And just because the logic of an argument does not seem to fit what we're looking at empirically today, does not mean it might not fit the world tomorrow. As one of my professors once told me "the real world is just an example."

There is value in having academics mull over hypothetical worlds because it helps inform policymakers about the type of contingency plans that might be desirable.

To conclude, I'd like to see less time spent in boring debates b/w different schools of thought and more time devoted to the discovery of "robust" policies--that is, policies that can be expected to work well in a variety of hypothetical worlds, including the one we live in!