On November 5, 2008, the Telegraph famously reported, During a briefing by academics
at the London School of Economics on the turmoil on the international
markets the Queen asked: Why did nobody notice it? Good question. Since then,
many economists and others have responded. What is wrong with (macro)economics?
Actually, the question should be, What is wrong with the modeling economists do?
Peter Spiegler takes up this issue in Behind the Model: A Constructive Critique of
Economic Modeling. The books cover shows a square peg forced into a round whole.
Spiegler wonderfully addresses the problem.
To be clear, this is not a critique along the lines of the one made by Deirdre
McCloskey, that standard economic modeling (Max U is her preferred label for what
she calls Samuelsonian economics) is sterile and not able to address the big
questionsfor example, How is it that some nations are poor and some comparatively
rich? Todays mainstream economics, heavily engaged in mathematical economics,
steers clear of these interesting questions because it has nothing to say about how
cultures evolve or how we adopted bourgeois virtues.
Others have worried that economics took a wrong turn, going the way of physics.
Spiegler poses the mismatch question (between the principles discerned by the scientist
and the phenomenon under investigation [p. 10]) early in the book. Physicists
model how particles interact. But particles do not have mood swings that baffle us.
Spiegler addresses mainstream economics on its own terms. He begins with the
point made by Milton Friedman some years ago that models are best judged by their
predictions. Thats not good enough, writes the author; we would learn too little about
the validity of relations (paths) inside the black box. How would we identify the
important causal factors? Are the models that economists adopt plausible and proper metaphors? Spiegler refines this question with great care.
He looks at conventional mathematical modeling by economists, noting that it
requires two transitions (apt representations, correspondences) between ordinary
language and mathematical language (twice crossing a significant linguistic divide
[p. 47]). The mathematical language is metaphorical. Are the mathematical models
economists work with appropriate, plausible, and illuminating metaphors for the
human social activity being studied? Spiegler relies on R. I .G. Hughess framework,
which emphasizes denotation, demonstration, and interpretation (DDI) (Models and
Representation, Philosophy of Science 64 : S32536). In the denotation phase,
we move from delimitation, posing the question in plain English, to a mathematical
statement of a theoretical model. We solve the model and then interpret the solution,
finally answering the original motivating question in plain English. Both language
transitions are fraught and not quite as scientific as the modeling project suggests.
Spieglers discussion is outstanding in elaborating this problem.
In another chapter, Spiegler addresses the empirical tests economists rely on. The
rules of econometrics guide us as we go from data to parameter estimates. But then
what? Having seen that model construction is fraught, what can we really claim to have
found once parameter estimates are in hand? We have tested representations with
questionable aptness. Besides, Spiegler notes, the ability of an economist to produce an
effective empirical model is largely an art (p. 86). How many times have we plowed
through a theoretical presentation only to find that a multiple regression is presented
and, often as though pulled from a hat, with the claim that the regression equation
represents a proper reduced formand with a bunch of control variables thrown in for
To illustrate his arguments, Spiegler singles out several well-known economic
models in current economic literature and challenges the aptness of the metaphors that
the models authors have chosen. Among the models scrutinized is the Solow growth
model, the ShapiroStiglitz paper on efficiency wages, and the new new institutional
economics (NNIE). I focus on the discussion of NNIE because it is an attempt by
economists to model big questions. Have economists helped us to better understand
how institutions emerge? How do institutions make a difference? There are other big
questions NNIE also considers.
Having read this far, readers will not be surprised by Spieglers conclusion that the
work of NNIE modelers has set back economists understanding of institutions by
overstating the applicability of formal models (p. 96). But this conclusion is not simply
opinion. Spiegler has arrived at it via serious argument. He cites the leaps required from
ordinary-language descriptions of the social phenomena being modeled to the proto-model (mixture of languages, informal statement of the formal approach) to the
formal model in mathematical form. These leaps are inevitably difficult. Do the
modelers give up more than they get?
Testing is usually conducted as solutions of the formal models are compared to
some slice of historical experience. Spiegler cites examples of such attempted matchings
by modelers whom he finds unconvincing. The formal models are not actually tested.
The comparison is done not with the actual elements of the formal model but rather
with more nuanced and ambiguous versions of them, i.e,. the proto-model (p. 110).
We are back to language transitions that perhaps inevitably come up short.
Later in the book, Spiegler looks at the performance of dynamic stochastic general
equilibrium (DSGE) models in light of their inability to foresee (or explain) the financial
crisis of 20078 and the run-up to the Great Recession. Specifically, he wants economists
to supplement mathematical models with methods of investigation of the target
that cast a wider net (p. 119). DSGE requires three radical simplifications: the use of
a representative agent, the assumption that financial markets are efficient, and the
assumption that a linear system (log-linear) is an adequate representation. We now
know that what was once seen as a modeling accomplishment did not end well.
Spieglers big point about the postcrisis situation is the concern that there is
virtual unanimity . . . in the view that there is no need to look outside economists
current toolbox for answers (p. 133).He briefly sketches four areas of postcrisis DSGE
criticism: do nothingfor example, current models did as well as could be expected
in light of policy errors; model finance, or make more of the financial sector endogenous;
add complexityfor example, simulate complexity via agent-based
computational models; and take a fringe position to reconsider formalism. The
author strongly suggests that these approaches do not go far enough.
The worlds middle class gained several billion members in the past twenty to
thirty years. There would be new demands for and supplies of credit. There would be
new credit markets and instruments. Could standard DSGE models be elaborated to
include shadow banking as well as all of the novel assets and liabilities that were devised
and traded? Funds from around the world found their way into U.S. housing. We got
the housing bubble and its collapse. Spiegler notes that economists saw the run-up to
the collapse but left the DSGE apparatus alone. And even if DSGE had kept producing
decent forecasts, would that have been a good thing in light of its gross
misrepresentation of the economy?
The author dismisses postcrisis work to fixDSGE as inadequate and argues instead
for a new field of interpretive economics: look at the world before you look at the
models. This is where the author gets behind the model. This is where compatibility
between formal models and the phenomena being modeled would be checked. The
standard modeling Spiegler criticizes in the earlier chapters is too detached from the
real-world phenomena involved. Before proceeding on the standard research path,
economists should get out more and get to know something about the real-world
dimensions of the problem, including some knowledge of real-world decisions as well as of those who make them. Interview them. Get to know what they do, how, and why.
The commonsense nature of Spieglers suggestion may bring a smile to noneconomists,
but Spiegler has a point. He gives Truman Bewley (Why Dont Wages Fall during
a Recession? [Cambridge, Mass.: Harvard University Press, 1999]) a great deal of credit.
Here is a traditional theorist who chose to detour from standard modeling and take
a cue from anthropologists approach. Why are wages rigid? Because those who would
cut wages reveal their concern for workers morale. Spiegler reviews four recent papers
that apparently succeed because the authors took the time to go out and look.
I had thought that macromodels are inadequate because they deal in aggregates
and thereby miss the interesting adjustments that market participants are routinely
prompted to think hard about and act on. Some of us also think that the omission of
public choice from so much of economic modeling leaves the models mute when it
comes to the crony capitalism that ever more defines the capitalism we have. The recent
recession came about after a concocted housing boom collapsed. The boom was
supported by interest groups with powerful lobbies that never rest. These two themes
do not appear in Spieglers book. But at near two hundred pages, the volume under
review offers the best ratio of good ideas per pound this reviewer can recall seeing.
Complex materials are presented carefully and clearly. The book is a joy to read and
Concerns over the limits of mainstream economics from on high (including keynote addresses and Allied Social Science Association panels) have been around for some years. But has the core of what is taught in most universities changed much? What is taught is still mostly mathematical modeling. In a better world, Ph.D. students in economics would read (and discuss!) the volume under review in their first year of study. It would give them perspective; it would get them to think hard about what they might spend a lifetime doing.