George Akerlof: The Missing Motivation in Macroeconomics

Nobel prize winner in economics George Akerlof is delivering this year's presidential address at the American Economic Association meetings which are currently underway in Chicago. His speech as the outgoing president is an attempt to set economics on a new path, a path that departs from the theoretical foundations of modern macroeconomic models by including social norms for behavior into the theoretical structures. Here are some sections from the paper his speech is based upon, but first Louis Uchitelle of the New York Times gives a bit more background:

Encouraging More Reality in Economics, by Louis Uchitelle, NY Times: The annual meeting of the American Economic Association, which opened here [in Chicago] on Friday, is usually a pretty esoteric affair.

But this year it could resonate much more broadly as the departing president of the organization, which represents most of the nation’s academic economists, tries to push prevailing economic theory further away from the free market approach that has generally held sway for the last four decades.

The protagonist in this drama is George A. Akerlof, a Nobel laureate, who is using the same platform that the late Milton Friedman adopted in 1968. As president of the A.E.A. back then, Friedman laid out new theoretical justifications for a market system that he argued performs most favorably for nearly everyone when the government avoids tinkering with its operation.

The hundreds of economists who listened that day to Mr. Friedman’s memorable speech did not immediately embrace his ideas. Keynesian economics, with a big role for government, still held sway.

But over time the Friedman approach took hold, eventually having profound effects on politics and government policy far beyond the ivory tower. This was partly because of Mr. Friedman’s insistent, larger-than-life personality, and partly because Keynesian economics failed to adequately explain and respond to the simultaneous outbreak of higher inflation and rising unemployment that emerged in the 1970s.

Mr. Akerlof’s style, in contrast, is more diffident and modest. But he has already contributed significantly to a revamping of the economic theory that Mr. Friedman championed...

And he is doing so at the moment when income inequality, more concentrated wealth and upheavals from expanded globalization are straining faith in a relatively unfettered market system.

“I am trying to effect a return to sensible economics,” Mr. Akerlof said in an interview. “And what is sensible economics? It is very pragmatic. You think about problems in the world and you ask: can government do something about that? At the same time, you maintain your skepticism that government is often inefficient.” ...

In the text of his speech to be delivered on Saturday afternoon, Mr. Akerlof argues that the Friedman approach is based on false assumptions about human behavior. ... A result, Mr. Akerlof says, is misleading theory and misguided policy.

Mr. Akerlof is facing considerable criticism for his view that standard economics leaves out too much actual human motivation. What Mr. Akerlof sees as missing content, Mark Gertler, a New York University economist, describes as “frictions” that distort accurate theory.

“What Akerlof is doing is stepping out of line,” said Mr. Gertler... “A lot of people are correctly taking rational behavior as a baseline and are adding frictions, such as constraints on borrowing, that can lead to temporarily inefficient markets.” ...

In his speech, he encourages others to follow his lead, rejecting the focus on what he calls “parsimonious modeling” inspired by Friedman. Everyday experience and observation must be returned to a prominent place in the profession, he argues.

“The early Keynesians got a great deal of the workings of the economic system right in ways that are now denied,” Mr. Akerlof said in a study newly posted on the Internet that closely tracks the text of his speech. “They based their models, as Keynes put it, on ‘our knowledge of human nature and from the detailed facts of experience.’ ”

A lot of what Mr. Akerlof advocates in his speech is already under way, with Mr. Akerlof himself a major contributor. He shared a Nobel in economics in 2001 for his work on imperfect information... He was an early participant in behavioral economics, another assault on the rational, fully-informed behavior that Mr. Friedman counted on to make markets work efficiently without regulation or intervention.

People often do not behave rationally, the behaviorists found in their experiments. Most do not bother to sign up for a voluntary 401(k) plan, for example, but do not pull out of such a plan if an employer signs them up automatically.

Now Mr. Akerlof is taking a big step on his own. His ... research, much of it done with Rachel Kranton, a University of Maryland economist. They are trying to incorporate into theory, as Keynes once did, the great variety of “norms” that determine human behavior.

What Mr. Akerlof is trying to do, with Ms. Kranton’s help, is to reflect the variety of motivations that come from the sense people have of “what they are and how they should behave,” as Ms. Kranton put it.

Among the examples they cite:

A teacher in good standing among the parents of her students puts the preservation of that reputation ahead of attempts to maximize her pay. ...

Workers resist wage cuts even when unemployment is rising, despite standard theory that they will accept less pay to save their jobs.

The variations in norms and behavior are numerous and Mr. Akerlof, in his speech, calls on economists to incorporate this diversity into standard economic theory.

“If there is a difference between real behavior and behavior derived from abstract preferences, New Classical economics has no way to pick up those preferences,” Mr. Akerlof asserts. “A macroeconomics that incorporates observations regarding how people think they should behave combines the best of the two approaches.”

Here is the abstract, introduction, and conclusion to the paper:

The Missing Motivation in Macroeconomics, George A. Akerlof ABSTRACT The discovery of five neutralities surprised the economics profession and forced the re-thinking of macroeconomic theory. Those neutralities are: the independence of consumption and current income (given wealth); the independence of investment and finance decisions (the Modigliani-Miller theorem); inflation stability only at the natural rate of unemployment; the ineffectiveness of macro stabilization policy with rational expectations; and Ricardian equivalence. However, each of these surprise results occurs because of missing motivation. The neutralities no longer occur if decision makers have natural norms for how they should behave. This lecture suggests a new agenda for macroeconomics with inclusion of those norms.

I. Introduction

Macroeconomics changed between the early 1960's and the late 1970's. The macroeconomics of the early 1960's was avowedly Keynesian. This was manifested in the textbooks of the time, which showed a remarkable unity from the introductory through the graduate levels.[1] John Maynard Keynes appeared, posthumously, on the cover of Time Magazine.[2] Even Milton Friedman was famously—although perhaps misleadingly—quoted, “We are all Keynesians now.”[3] A little more than a decade later Robert Lucas and Thomas Sargent (1979) had published “After Keynesian Macroeconomics.” The love-fest was over.

The decline of the old-style Keynesian economics was due in part to the simultaneous rise in inflation and unemployment in the late 1960's and early 1970's. That occurrence was impossible to reconcile with the simple non-accelerationist Phillips Curves of the time.

But Keynesian economics also declined because of a change in economic methodology. The Keynesians had emphasized the dependence of consumption on disposable income, and similarly, of investment on current profits and current cash flow.[4] They posited a Phillips Curve, where nominal—rather than real—wage inflation depended upon the unemployment rate, which was used as an indication of the looseness of the labor market. They based these functions on their own introspection regarding how the various actors in the economy would behave. They also brought some discipline into their judgments by estimating statistical relations.[5]

But a new school of thought, based on classical economics, objected to the casual ways of these folks. New Classical critics of Keynesian economics insisted instead that these relations be derived from fundamentals. They said that macroeconomic relationships should be derived from profit-maximizing by firms and from utility-maximizing by consumers with economic arguments in their utility functions.

The new methodology had a profound effect on macroeconomics. Five separate neutrality results overturned aspects of macroeconomics that Keynesians had previously considered incontestable. These five neutralities are: the independence of consumption and current income (the life-cycle permanent income hypothesis); the irrelevance of current profits to investment spending (the Modigliani-Miller theorem); the long-run independence of inflation and unemployment (natural rate theory); the inability of monetary policy to stabilize output (the Rational Expectations hypothesis); and the irrelevance of taxes and budget deficits to consumption (Ricardian equivalence).[6] These results fly in the face of Keynesian economics. They undermine its conclusions about the behavior of the economy and the impact of stabilization policy.

The discovery of these five neutrality propositions surprised macroeconomists. They had not suspected that radically anti-Keynesian conclusions were the logical outcome of such seemingly-innocuous maximizing assumptions.

II. Neutralities and Preferences

How did macroeconomists react to the discovery of the five neutralities? On the one hand, the New Classical Economists viewed their neutrality results as a tell-tale: that Keynesian economists of the previous generation had been thinking in the wrong way. In their view, scientific reasoning was producing a newer, leaner, more precise economics.

On the other hand, Keynesian economists, for the most part, reacted differently. In due course they came to view the neutralities as logically impeccable. These New Keynesians accepted the methodological dictums of the New Classical economics: that constrained maximization of profit and utility functions is the appropriate microfoundation for macroeconomics. They also viewed the neutralities as having a certain sort of generality. The neutralities do commonly describe equilibria of competitive economies with complete information irrespective of people’s preferences—as long as those preferences correspond to economists’ typical descriptions of them. The Keynesians then resurrected some—but not all—of the Keynesian conclusions by adding a variety of frictions to the New Classical model. Those frictions include credit constraints, market imperfections, information failures, tax distortions, staggered contracts, uncertainty, and bounded rationality. This formulation preserves many (but not all) Keynesian conclusions regarding cyclical fluctuations and macroeconomic policy.

This lecture will suggest a new stance in regard to each of the five neutralities. Like New Classical and New Keynesian economics, it will derive behavior from utility and profit maximization. That captures the purposefulness of economic decisions. But this lecture will also question the generality of the preferences that lead to the five neutralities. There is a sense in which those preferences are very narrowly defined. They have important missing motivation—since they fail to incorporate the norms of the decision makers. Those norms reflect how the respective decision makers think they and others should or should not behave even in the absence of frictions. Preferences reflecting such norms yield a macroeconomics with important remnants of the early Keynesian thinking. They also yield a macroeconomics that, in important details, cannot be obtained only with frictions.

We shall see that with such preferences, even in the absence of frictions, each of the five neutralities will be systematically violated. Specifically:

—a realistic norm regarding consumption behavior will make consumption directly dependent on current income, in violation of the neutrality of consumption given wealth;

—a realistic norm will make investment directly dependent on cash flow, in violation of Modigliani-Miller;

—a realistic norm will make wages and prices dependent on nominal considerations and thus violate natural rate theory;

—a realistic norm will make income and employment dependent on systematic monetary policy, and thus violate rational expectations theory; and

—a realistic norm will make current consumption dependent on the current generation’s social security receipts, in violation of Ricardian equivalence.

Additionally, insofar as the behavior assumed by the early Keynesians differed from the behavior that produces the neutralities, there is likely to be a bias in favor of the Keynesians. The Keynesians based their models on their observation of motivations, rather than on abstract derivations. If there is a difference between real behavior and behavior derived from abstract preferences, New Classical economics has no way to pick up those differences. In contrast, models with norms based on observation will systematically incorporate such behavior — although, of course, as with any method, there is the possibility for error.

Inclusion of the “missing motivations in macroeconomics” then combines the observations of the Keynesians with the intentionality of economic decisions in New Classical economics. Such a synthesis yields the best of the two approaches.

Two disclaimers. Before beginning in earnest, let me offer two brief disclaimers. First, none of the behavior revealing of the norms that are introduced in this lecture will be new. On the contrary, I have purposefully chosen phenomena that have been emphasized since The General Theory by macroeconomists, who have followed Keynes in voicing their continuing doubts about classical interpretations of macroeconomic behavior.

Second, this lecture will discuss different norms that respectively correspond to the five neutralities. I shall assume that these norms are exogenous. Such assumptions of exogeneity are standard in economic analysis. In a given problem in a given time frame, some terms are assumed constant, while others are allowed to vary. I ask you, at least to the end of the lecture, to withhold your doubts regarding whether such exogeneity is a correct assumption or not. The incorporation of such endogeneity is the next step—not the first step—in the study of the effect of norms on macroeconomics, especially since such endogeneity may sometimes dampen, but will rarely nullify, the conclusions of this lecture.

...

XII. Conclusion

This lecture has shown that the early Keynesians got a great deal of the working of the economic system right in ways that are denied by the five neutralities. As quoted from Keynes earlier, they based their models on “our knowledge of human nature and from the detailed facts of experience.” They used their intuitions regarding the norms of how consumers, investors, and wage and price setters thought they should behave. There is systematic reason why such knowledge and experience is likely to be accurate: by their nature norms are generated and known by a whole community. They are known to those who abide by them, and those who observe them as well.

We have shown ways in which macroeconomic variables will be affected by norms. The neutralities say that consumption should have no special dependence on current income; investment should be independent of current cash flow; wages and prices should not depend on nominal considerations. The very construction of those neutralities denies the possibility that peoples’ decisions might be influenced by their views regarding how they, and how others, should behave. However, in practice, the neutralities are systematically violated. Insofar as economists have felt it necessary to explain these violations they have appealed to a variety of different frictions, such as myopia and credit constraint. In so doing they have failed to consider that those violations would occur even in the absence of those frictions: they will occur because of decision-makers’ norms.

The incorporation of norms based on careful observation imparts an appropriate balance to macroeconomics. The New Classical research program was correct in viewing models of the early Keynesians as too primitive. They had not been sufficiently attentive to the role of human intent in choices regarding consumption, investment, wages and prices. But that research program itself has failed to appreciate the extent to which the Keynesians’ views of macroeconomics were also reflective of reality, since they were based on experience and observation.

A macroeconomics with norms in decision makers’ objective functions combines the best features of the two approaches. It allows for observations regarding how people think they should behave. It also takes due account of the purposefulness of human decisions.

---

1 See for example Samuelson (1964), Dernburg and McDougall (1967), and Ackley (1961). The econometric model of Klein and Goldberger (1955) provides a useful synopsis of the variables that the early Keynesians thought most important for a macroeconomic model, and how they would be included.

2 Time Magazine, December 31, 1965. His appearance on the cover was especially remarkable because Time covers are rarely posthumous. Keynes had died in 1946.

3 But in a later disclaimer, Friedman said, almost surely correctly, that he had been quoted out of context. See http://www.libertyhaven.com/thinkers/miltonfriedman/miltonexkeynesian.html, which quotes Friedman (1968), Dollars and Sense, p. 15.

4 The treatment of consumption in The General Theory, as we shall see below, was typical of such thinking. Keynes first discusses the dependence of consumption on current income, which he clearly sees as the primary determinant of current consumption; but, in addition, he also makes a long list of other factors that will alter the relation between consumption and current income.

5 A good example of this methodology can be seen in Phillips’ (1958) mixture of light theory and statistical analysis in his estimation of the relation between wage inflation and unemployment.

6 Of course it took some time for the implications of these neutrality results to be fully appreciated. For example, life-cycle consumption and Modigliani-Miller were initially considered as nothing more than useful codicils to Keynesian thinking.

Comments

[reason]

Object oriented programming is the way to go. Build realistic models and see how they behave. Forget the redundant exactitude of mathematical models and go for behavioural and/or ecological models. We need models that properly allow for technological change, information costs, for power relationships and for asset price bubbles. Only then can we really start to understand the nature of the system, particularly those parts with positive feedback cycles.

[slink]

reason

math is a road way
not a road block

your science quest
is all admirable
still u must
leave room for
pragmatics

ie
what's the best policy
today for today
knowing what we don't know

----------------------------
nice post mark
covers
the macro policy counter revolution
very compactly

its
culmination
in "the bourbon restoration"
we now call
the reagan 80's
produced some seriously
harmful
mind games

harmful to ordinary folks
at any rate

a set of logical fantasy lands
wonderful nonsense
in themselves
worthy of
lewis carroll
and pastor abbott
certainly not in themselves
all that horrrid

(i like barro lucas et al)

but as policy considerations
rounds the bend
i just wouldn't listen to em
anymore then i'd listen to humpty dumpty

----------------------
now if i were a member in good standing
of the hi fi trans nat set ...

well
its really simple
if the electorate believes
uncle can't do anything helpful
for any one ...

hence the bourbon era
broadcasted models
worthy of
walt disney

if the logical implication
of your wonderland
when
applied to the real world
is as a practical matter
to "do nothing gubmint wise "
that can be
great for "profiteers" looking
to keep or generate "economic rents"
reduce wage share
escape whole society
righteous regs and rules
etc etc
---------------------------------------
as to
the deeper question
"but is it a true model"

that's a non questionif your some over grown babbit
and
the present set up is working for u

yup
and u tell the helots:

"just keep the beliefs
u got from ronnie
you jobbled rubes
while i keep the benefits "

[anne]

Excellent, excellent.

[pinus]

reason: "We need models that properly allow for technological change, information costs, for power relationships and for asset price bubbles."

No. We don't need such models. We have plenty of them. We need models that _explain_ all these phenomena. One thing is _allowing_ for a phenomenon, the other is explaining it. You can easily introduce a model that will allow for pretty much anything you can imagine. Will it be useful? No. The keyword is degrees of freedom.

I'm afraid Akerlof's suggestions go exactly in the way that was attempted by models with irrationality a decade or two ago. These were fantastic to describe a lot of what was going on. But were they able to explain something? Rarely. That's why these models somehow faded away.

Akerlof suggests introducing something called "norms". He wants to treat them, at least at the beginning, as exogeneous. Well, behavioral economists should tell us which norms are sensible. I am sceptical here, not only because behavioral economics is one of the branches of economics with (so far) worst results, and is least explored. Exogeneous norms are the same as introducing irrationality - they add so many degrees of freedom that the models' fit will be great, but explanatory power mediocre.

And endogenizing these norms will eventually bring us back to where we started - rational agents who act in their best interest, which will eventually lead to the birth of such "norms" in the model.

[slink]

pinus
i like your take away here
models must be simplificationsu really never escape analogy

models become more passive
and untracable
ie
useless
if they try to be
too comprehensively specified

a model as complex as the realityit models
would need a simpler model
to help us understand it

it would "explain"
nothing
it would simulate it

but a siumlator could
be judged by its forecasting strength

call it a rational oracle

who knows how or why
but it predicts the next step pretty damn well
or at least it has so far....

[pinus]

slink: But we are not trying to be oracles here, we want to be researchers. Research tries to understand, to explain. We do not need an oracle to tell us what happens tomorrow. Let's just wait one day and experience it. We want to know _why_ exactly this happened. Only then we can extrapolate for more general cases. A "forecasting oracle" cannot possibly deal with changing environment, with new situations.

[steve]

Pinus,

If you don't require that your model have some predictive ability then there will always be many models that fit the data -- just add degrees of freedom.

I keep hearing from economists about how their models are important for "understanding", even when they can't predict a thing. Well, how do you know you aren't fooling yourself - just making up a story (feature of your model) to fit the past, but which doesn't tell you anything about the future because, ultimately, your story didn't really capture the real cause of what happened.

"The bubble popped because the Fed raised interest rates. See, I have this simulation where the agents all start to sell when the riskless rate rises above their expected blah, blah..." So what? Your simulation is not worth the CPU time wasted to run it unless you rigorously test it - ideally by asking it to make predictions. Sure, it helps you "understand" - it tells a story like one of Aesop's fables, but does it capture reality?

In evolutionary biology they have a term for this - a "just so" story about how something evolved that may or may not be true. Until you can predict something you can never be sure you're not just playing games or telling stories.

[Mark Thoma]

Can't predict a thing?

That's quite an overstatement ... please take a closer look at the econometric work that's out there.

[js paine]

" A "forecasting oracle" cannot possibly deal
with a changing environment, with new situations"

yup
hence the shallow pragmatics
of say
greenspans for sale
"hand variable completed "
macro model of the early and mid 70's

imagine a model that spits out
a perfect fit function
that exactly duplicates
the sequence of numbers we call
" employment 1970 to 1986 "

surely we have no reason to assume
its 87 employment number
will be on the money
just because it was correct in producing
the prior 16 years numbers
thats reverse engineering so to speak

"We want to know _why_ exactly this happened"

why do we want to know why ????
when the future is our arena not the past

building larger frame works
that more deeply explain stylized facts
sure
but again why ??

is the added insight
part of a policy enlightening process???

if not
what did it explain and what for ????

lets not fall back on

"it describes "

okay i bet u agree
truer true models
are not always
generalizations

one's that can anticipate
"new situations"
ones that can turn developments
now crudely rendered as " random external shocks "
into
"internalized" features
ie
as a possible morph
anticipated by
the system's model

hopefully
that leads to a data hunt
for other confirmations
of the model's true ness
one looks to find
othermodel predicted "correlates "
to the anticipated morph's
own "inception" requirements

[steve]

Mark,

Sorry, my comment was too harsh. I mean the kinds of "high theory" involving either neoclassical optimization/equilibrium or simulations with agents. To me, the closer economic work is to the data, the more respectable!

I think the use of really sophisticated math like fixed point theorems is ridiculous. I once sat in on Debreu's class at Berkeley and could detect almost no connection to the real world.

To quote Einstein:

"Insofar as mathematical theorems refer to reality, they are not sure, and insofar as they are sure, they do not refer to reality.''

[Mark Thoma]

Thanks...

[evagrius]

A "forecasting oracle" cannot possibly deal
with a changing environment, with new situations"

The I-Ching?

[tom s.]

pinus - not sure if you have read the original papers that Akerlof is building on in this talk, but if not I recommend that you do. In particular, "Economics and Identity" by Akerlof and Kranton strikes me as having found a parsimonious way to introduce enough richness into economic behaviour to tackle broader problems, while keeping things simple enough that they can still explain many things about those problems, including the persistence of discrimination and apparently self-destructive behaviour in socially excluded groups. It's a fine read.

The paper is here: http://www.econ.umd.edu/~kranton/publications/economicsandidentity.pdf.

Also, I think Akerlof addresses your point about endogeneity -- just before the "Conclusions" in the extract posted here.

[kostas]

hey buddy let Aesopus out of it! Aesopus was as keen an observer of human nature insofar as it relates to maximizing behaviour as Adam Smith and Keynes. I like the one where the fox tries to get the grapes and cannot reach it. The fox then declares the grapes to be sour anyway. Dumb fox, right? Only an economist would say that, since the fox does (in this case merely say) something which is not going to help her increase her utility function. The neutrality theorem might be stated here like this: "no matter what the fox says it wont affect the quantity of grapes in her belly".

However, I tell my son that the fox is never dumb and always has a clever plan (as evidenced by every other Aesopus fable about the fox). I tell him the fox may be planning to bring a ladder tomorrow and says the grapes are sour just so others dont take it in the meantime. I also tell him that admitting she cannot jump is not a good idea either, prey might learn to climb trees when fleeing from her.

So, i guess Ackerlof is right about looking more closely into preferences. When important aspects of real behavior are not incorporated, maximizing behavior may not be apparent. He proposes to deal with this by introducing "norms", but i suspect that by the time we get to incorparating those indogenously it will just be the more sophisticated maximization behavior. Long live Aesopus!

[reason]

I think people are missing the point, because they have a body of study behind them and don't want to say the way we worked in the past is not necessarily the best way forward. Degrees of freedom if we are trying to map data points onto a model. But what if the parameters of the model (and the underlying model itself) are constantly changing. What I am suggesting is that we are trying to force the economic system to be something that it isn't (linear and mechanistic). It would nice to model the behaviour of the system as a whole and see how sensitive it really is to assumptions about the behaviour of individual agents (rather than trying to to do shifting short term linear approximations - maybe useful for forecasting but not for understanding). That is why I suggest moving to object oriented simulation and away from average statistical models.

[Oberbauer]

What kind of nonsense is this?
The introductory article statet:

... using the same platform that the late Milton Friedman adopted in 1968. As president of the A.E.A. back then, Friedman laid out new theoretical justifications for a market system that he argued performs most favorably for nearly everyone when the government avoids tinkering with its operation.

The hundreds of economists who listened that day to Mr. Friedman’s memorable speech did not immediately embrace his ideas. Keynesian economics, with a big role for government, still held sway.

Never mind that Keynes was NOT the inventor of big goverment, but quite certainly the inventor of modern monetary theory, friedman talked NOT AT ALL about gouverments interference but about what monetary policy can and cannot do.
This is quite a famous adress, i´ve got it in print on my bookshelf, and it deals NOT NOT NOT with big gouverment.

[johnchx]

Oberbauer wrote:

What kind of nonsense is this?

The sort of nonsense that usually materializes when journalists try to "summarize" an economics concept for a "general" audience. But don't fret, it's probably no worse than what physicists and biologists suffer at the hands of "science writers."

The real damage is done by those who insist on deriving their understanding of disciplines -- and the controversies within them -- from the 2-dimensional caricatures appearing in the popular press.