Part One
Animal Spirits
ONE
Confidence and Its Multipliers
ONE OF US (Akerlof) remembers a dinner conversation a few years ago. During the housing boom a distant relative from Norwayâby marriage by marriage by marriage, known only from a brief encounter at a family weddingâhad reportedly bought a house in Trondheim, for more than $1 million. That seemed like a lot of moneyâperhaps not for New York, Tokyo, London, San Francisco, Berlin, or even for Osloâbut certainly for Trondheim, up the Norwegian coast, on the edge of settlement, and vying for the title of worldâs most northern city. Nor was it a mansion. This thought remained quietly parked in Akerlofâs brain, classified along with other observations that property values were high in Scandinavia.
Recently Akerlof told his co-author, Shiller, that he had been wondering if he should have given more thought to the Trondheim story. We discussed the matter. This seems to have been a mental lapse, accepting this story of the high price as nothing more than an insignificant oddity. On the contrary, Akerlof should have seen it as an incongruity requiring active thought, to be resolved within the context of a larger view of the markets.
We decided that this little story is worth pondering at greater lengthâfor the insight it offers into the thought patterns that underlie the booms and busts that characterize the business cycle, and, notably, the twin crises of confidence and credit that currently envelop much of the world.
Confidence
The newspapers and the pundits tell us when the economy goes into recession that it is necessary to ârestore confidence.â This was J. P. Morganâs intention after the stock market crash of 1902 when he put together a bankersâ pool to invest in the stock market. He employed the same strategy in 1907.1 Franklin Roosevelt analyzed the Great Depression in similar terms. âThe only thing we have to fear,â he declared in his first inaugural address in 1933, âis fear itself.â Later in the same speech he added: âWe are stricken by no plague of locusts.â Ever since the founding of the U.S. republic, business downturns have been proclaimed as the result of a loss of confidence.
Economists have a particular interpretation of the meaning of the term confidence. Many phenomena are characterized by two (or possibly more) equilibria. For example, if no one rebuilds his house in New Orleans after Hurricane Katrina, no one else will want to rebuild. Who would want to live in desolation, with no neighbors and no stores? But if many people rebuild in New Orleans, others will also want to. Thus there may be a goodârebuildingâequilibrium, in which case we say that there is confidence. And there may also be a badânon-rebuilding âequilibrium, with no confidence. In this view there is nothing more to confidence than a prediction, in this case regarding whether or not others build. A confident prediction is one that projects the future to be rosy; an unconfident prediction projects the future as bleak.
But if we look up confidence in the dictionary, we see that it is more than a prediction. The dictionary says that it means âtrustâ or âfull belief.â The word comes from the Latin fido, meaning âI trust.â The confidence crisis that we are in at the time of this writing is also called a credit crisis. The word credit derives from the Latin credo, meaning âI believe.â
Given these additional shades of meaning, economistsâ point of view, based on dual equilibria or rosy versus bleak predictions, seems to miss something.2 Economists have only partly captured what is meant by trust or belief. Their view suggests that confidence is rational: people use the information at hand to make rational predictions; they then make a rational decision based on those rational predictions. Certainly people often do make decisions, confidently, in this way. But there is more to the notion of confidence. The very meaning of trust is that we go beyond the rational. Indeed the truly trusting person often discards or discounts certain information. She may not even process the information that is available to her rationally; even if she has processed it rationally, she still may not act on it rationally. She acts according to what she trusts to be true.
If this is what we mean by confidence, then we see immediately why, if it varies over time, it should play a major role in the business cycle. Why? In good times, people trust. They make decisions spontaneously. They know instinctively that they will be successful. They suspend their suspicions. Asset values will be high and perhaps also increasing. As long as people remain trusting, their impulsiveness will not be evident. But then, when the confidence disappears, the tide goes out. The nakedness of their decisions stands revealed.
The very term confidenceâimplying behavior that goes beyond a rational approach to decision makingâindicates why it plays a major role in macroeconomics.3 When people are confident they go out and buy; when they are unconfident they withdraw, and they sell. Economic history is full of such cycles of confidence followed by withdrawal. Who has not taken a hike and come across a long-abandoned railway lineâ someoneâs past dream of a path to riches and wealth? Who has not heard of the Great Tulip Bubble of the seventeenth-century Netherlandsâ a country famous, we might add, for its stalwart Rembrandt burghers and often caricatured as the home of the worldâs most cautious people. Who does not know that even Isaac Newtonâthe father of modern physics and of the calculusâlost a fortune in the South Sea bubble of the eighteenth century?
All of which takes us back to Trondheim. Akerlof had stored the observation prompted by his relativeâs million-dollar home in the wrong place in his brain. He should have seen that home prices in Trondheim were not merely indicative of curiously high real estate prices in Scandinavia; they were part of a worldwide real estate bubble. He had been too trusting.
But that takes us even further back, to Keynesâ passage about animal spirits. When people make significant investment decisions, they must depend on confidence. Standard economic theory suggests otherwise. It describes a formal process for making rational decisions: People consider all the options available to them. They consider the outcomes of all these options and how advantageous each outcome would be. They consider the probabilities of each of these options. And then they make a decision.
But can we really do that? Do we really have a way to define what those probabilities and outcomes are? Or, on the contrary, are not business decisionsâand even many of our own personal decisions about which assets to buy and holdâmade much more on the basis of whether or not we have confidence? Do they not involve decision making processes that are closer to what we do when we flip a pancake or hit a golf ball? Many of the decisions we makeâincluding some of the most important ones in our livesâare made because they âfeel right.â John F. âJackâ Welch, the long-time CEO of General Electric and one of the worldâs most successful executives, claims that such decisions are made âstraight from the gut.â (We shall revisit him later.)
But at the level of the macroeconomy, in the aggregate, confidence comes and goes. Sometimes it is justified. Sometimes it is not. It is not just a rational prediction. It is the first and most crucial of our animal spirits.
The Confidence Multiplier
The most basic element of Keynesian economic theory is its notion of the multiplier. The concept, originally proposed by Richard Kahn as a sort of feedback system, was adopted by Keynes and became the centerpiece of his economic theory.4 Within a year of the publication of Keynesâ General Theory, John R. Hicks published a quantitative interpretation of Keynes that emphasized a rigid multiplier and the interaction of its effects with interest rates. Hicksâ version soon superseded Keynesâ original as the authoritative embodiment of Keynesian theory.5 Keynes was ruminating, discursive, disjoint, impenetrable, but nevertheless provocative and amusing; Hicks was orderly, efficient, and logically complete. Hicksâ version won the day. He is not as famous as Keynes, for he is often viewed as a mere interpreter of Keynesâ genius. But in terms of the history of thought, the âKeynesian revolutionâ was just as much a âHicksian revolution.â
But we believe that the Hicksian embodiment of Keynesâ notions is too narrow. Instead of the simple multiplier that Hicks focused on, we should look at an allied concept, which we call the confidence multiplier.
The Keynesian multiplier, taught for generations to millions of undergraduates, works as follows. Any initial government stimulus, say a program of increased government expenditure, puts money into peopleâs hands, which they then spend. The initial government stimulus is the first round. Each dollar spent by the government ultimately becomes income to some people, and, once it has been put into their hands, they spend some fraction of it. That fraction is called the marginal propensity to consume (MPC). Thus the initial increase of expenditures feeds back into a second round of expenditures, made by people, not the government. This then feeds back again into income for yet more people, in an amount equal to the MPC dollars. These people in turn spend a fraction of the MPC, called the MPC squared dollars. This is the third round. But the story is not over yet. Round after round of expenditure follows, and so the sum of the effects of the initial expenditure of a single dollar by the government may be represented as $1 + $MPC + $MPC2 + $MPC3 + $MPC4. . . . The sum of all these rounds is not infinite; it is in fact equal to 1/(1 â MPC), a quantity that is called the Keynesian multiplier. But the sum may be much larger than the original government stimulus. If the MPC is, say, 0.5, the Keynesian multiplier is 2. If the MPC is 0.8, the Keynesian multiplier is 5.
That idea was captivating for many people when Keynes articulated it in his 1936 book, and it was seized upon by Hicks in 1937. It was interpreted as explaining the mystery of the Great Depression. The Depression had been so puzzling because people could see no readily comprehensible cause for such an important event. The multiplier theory explained that a small dip in expenditure could have greatly magnified effects. If there were a small but substantial decline in consumption expenditures because people overreacted in fear to a stock market crash, such as the one of 1929, then this would act just like a negative government stimulus. For each dollar that people cut their consumption, there would be another round of expenditure cuts, then another and then another, resulting in a much larger decline in economic activity than would be attributable to the initial shock. A depression could come about over the course of several years, as the multiple rounds of negative expenditure hits put businesses further and further into the red. The theory won widespread acclaimâif not immediate policy implementationâfor it sounded like just what was happening to businesses as the Depression increasingly deepened from 1929 to 1933.
Keynesâ multiplier theory also won popularity among econometricians because it could be quantified and modeled. Authoritative statistics on national consumption and income became available at around the same time as Keynesâ General Theory and Hicksâ interpretation of it were first published, and they provided the data sets for their analysis. The first estimates of national consumption data were published by the Brookings Institution in 1934.6 The U.S. National Income and Product Accounts were developed and put into a framework amenable to Keynesian-Hicksian theory by Milton Gilbert in the early 1940s.7 To this day the U.S. government, like the governments of other major countries, still produces national income and consumption data in accordance with the demands of this theory. Surprising as it may seem, given the huge volume of economic literature, no other macroeconomic model after that of Hicks has had such authority to dictate major changes in the way national data are collected. In a sense it is the data that dictate the theory that serves as the basis for formal modelingâfor the data we have today were generated with but one theory in mind.
The creation of the data sets led to the development of large-scale computer simulation models for the economies of the countries of the world. This modeling started when Jan Tinbergen developed an econometric model of the Dutch economy in 1936 and a forty-eight-equation model of the U.S. economy in 1938. In 1950 Lawrence Klein developed another model of the U.S. economy, which grew over subsequent decades into the enormous Project Link, which linked together econometric models of every major country of the world, composed of thousands of equations. Such models have only a minimal role for animal spirits, and Keynes himself was skeptical of them.8
But it is possible to conceive of a role for confidence within these models. We usually think about multipliers only with respect to conventional variables that can be easily measured. But the concept applies equally well to variables that are not conventional and that cannot be measured so easily. Thus there is not only a consumption multiplier, an investment multiplier, and a government expenditure multiplier, which represent the change in income that occurs when there is, respectively, a $1 change in consumption, investment, or government expenditure. There is also a confidence multiplier. That represents the change in income that results from a one-unit change in confidenceâhowever it might be conceived or measured.
We can also think of the confidence multiplier, like the consumption multiplier, as resulting from different rounds of expenditure. Here the feedbacks are more interesting than in our earlier simple example of rounds of consumption expenditure. Changes in confidence will result in changes in income and confidence in the next round, and each of these changes will in turn affect income and confidence in yet further rounds.
For a long time now there have been survey measures of âconfidence.â The best known of these is the Michigan Consumer Sentiment Index, but there are others. Some statisticians have developed models that test for feedback from confidence to gross domestic product (GDP) using these data. There is little doubt that such measured confidence is a predictor of future expenditure. Causality tests for several countries suggest that current measured âconfidenceâ does feed future GDP, and this result would seem to confirm the feedback implicit in the confidence multiplier.9 Other statisticians have performed similar analyses using credit quality spreads, measured as the difference between interest rates on risky debt and interest rates on less risky debt, interpreting these as measures of confidence and testing whether they feed into, and help predict, GDP.10 But we believe that such tests are actually of limited value. Even when such results are obtained strongly, that does not necessarily imply that animal spirits are playing a role. Why not? Because the measure of confidence may not be measuring them. Instead they may only be reflecting consumersâ expectations regarding current and future income.11
And of course we would expect them to be predictive of future expenditure and income. It is also difficult to measure the effects of confidence on income bec...