G.L.S. Shackle |

Commenting on a Lars P. Syll post, Brad
DeLong grants that there may be
something profound in the Post-Keynesian view that neither market participants
nor economists “know the data-generating process.” “OK,” Professor DeLong
allows, “suppose we decide to give up” this assumption, “what do we then do - what
kind of economic arguments do we make – once we have made those
decisions?"

I have a few suggestions, but
first let’s be more specific about the Post-Keynesian complaint.

*There is no “data-generating process”*if, by this, Brad means a stationary, ergodic process (See P. Davidson, John Hicks, and even K. Arrow). (For reference, here’s Wikipedia’s definition of “ergodic”: “A stochastic system is called ergodic if it tends in probability to a limiting form that is independent of the initial conditions”).
More generally, the application
of the probability calculus to the behavior of market participants isn’t wholly
satisfying (See J.M. Keynes [

*weight*of an argument], G.L.S Shackle [conceptual deficiencies of the probability calculus], and Taleb [Black Swans and Fat Tails]).
If one accepts these skeptical
propositions (at least provisionally), then the following analytical
suggestions seem worthy of consideration.
(I offer these without explanation, hoping the connections aren’t too
hard to figure out):

1. “History” should take its
place alongside “Equilibrium” in economic modeling and argument;

2. Hysteresis isn’t an
“add-on." Many (maybe even most) real-world outcomes are path dependent;

3. Disequilibrium is the normal
state of the “macroeconomy,” the ex post accounting world rarely corresponds
to the ex ante expectations that produced it, and you can't ignore "out-of-equilibrium behavior";

4. Agent-based modeling (Santa
Fe style) may prove useful in thinking about interactions among agents with
different “views of the world,” the bulls and the bears, etc.;

5. Leave Tobin’s “Liquidity as
Behavior Towards Risk” aside in favor of G.L.S. Shackle’s analysis of the
elemental need that’s satisfied by money in a contemporary economy the future
states of which can’t be known today; and

6. Coordination problems are important (if there are no rational expectations to replace the missing futures markets of Arrow and Debreu).

6. Coordination problems are important (if there are no rational expectations to replace the missing futures markets of Arrow and Debreu).