It’s been slow going, but I’ve finally finished the draft chapter of my book-in-progress that looks forward to a new research program for macroeconomics, an absurdly ambitious task, but one that needs to be tackled. Of course, what I’ve written isn’t fundamentally new – it’s a distillation of points that Old Keynesians, post-Keynesians and some behavioral economists have been putting forward for a while. But I hope I’ve got some positive contribution to make. More than ever, comments are much appreciated.
Update In response to comments, mostly at Crooked Timber, I’ve fairly substantially revised the section on “avoiding stagflation”. While I don’t back away from the points I made previously, I took for granted some things that I’d mentioned in other places in the book. The result made for a fairly unbalanced treatment with an excessive focus on the role of labor militancy. I’ve now tried to put this into proper context. I don’t expect that will satisfy everybody, but this is closer to what I meant to say all along.End update
What next ?
The economics of the textbooks seeks to minimise as much as possible departures from pure economic motivation and from rationality. There is a good reason for doing so – and each of us has spent a good portion of his life writing in this tradition. The economics of Adam Smith is well understood. Explanations in terms of small deviations from Smith’s ideal system are thus clear, because they are posed within a framework that is already very well understood. But that does not mean that these small deviations from Smith’s system describe how the economy actually work. Our book marks a break with this tradition. In our view, economic theory should be derived not from the minimal deviations from the system of Adam Smith but rather from the deviations that actually do occur and can be observed. Animal Spirits, Akerlof and Shiller
It was reading this passage in Animal Spirits, and posting about its implications for macroeconomics in the Crooked Timber blog, that led to the writing of this book. A commenter suggest that this, and some earlier posts, would make a good book, Brad DeLong of UC Berkeley picked the idea up and the result is before you.
Animal Spirits was mostly written before, or in the early stages of, the Global Financial Crisis, but the Crisis has made its central point more important than ever. For many years economists have worked like the anecdotal drunk who searches for his dropped keys under a lamppost because the light is better there. In the future, and particularly in macroeconomics, economists need to start looking where the keys are, and try to build tools that will improve the chances of success.
This does not mean abandoning all the work of the past thirty years and returning to old-style Keynesianism. But it does mean starting from the traditional Keynesian perspective that a general macroeconomic theory must encompass the reality of booms and slumps, and, particularly of sustained periods of high unemployment that cannot be treated as marginal and temporary deviations from general equilibrium. We must model a world where people display multiple and substantial violations of the rationality assumptions of microeconomic theory and where markets depend not only on prices, preferences and profits but on complicated and poorly understood phenomena like trust and perceived fairness.
First, the program needs more realistic microfoundations. As Akerlof and Shiller observe, we need to look at how people actually behave, and how this behavior contributes to the performance of the economy as a whole.
Second, we need to reconsider the concept of equilibrium. The whole point of Keynes “General Theory” was that the market-clearing equilibrium analysed by the classical economists, and central to DSGE models, was not the only possible equilibrium. An economy can settle for long periods in a low-output, high-unemployment state that may not meet the neoclassical definition of equilibrium, but does match the original concept, borrowed from physics of a state in which the system tends to remain and to which it tends to return. More importantly, perhaps, we need a theory which encompasses crises, and rapid jumps between one kind of equilibrium and another. Ideally this will combine ‘old Keynesian’ analysis of economic imbalances with a Minsky-style focus on financial instability.
Between these two levels, we need to consider the fact that the economy is not a simple machine for aggregating consumer preferences, and allocating resources accordingly. The economy is embedded in a complex social structure, and there is a continuous interaction between the economic system and society as a whole. Phenomena like ‘trust’ and ‘confidence’ are primarily social, but they affect, and are affected by the performance of the economic system.
Finally, now that Keynesian macroeconomic policy has re-emerged as a practical tool, we need to reconsider the real and perceived failures of the past, and in particular the emergence of stagflation in the 1960s. If the revival of Keynesian policy is to be sustained, it must provide not only an emergency response to the present crisis but a set of tools that can deliver sustained non-inflationary growth.
Better microfoundations ?
It is now generally accepted that people are not, and cannot be, the infinitely foresightful, unbounded rational utility maximizers described by the axioms of dynamic stochastic general equilibrium theory. On the contrary, economic behavior, even that of highly sophisticated actors like the ‘rocket scientists’ who design financial instruments for investment banks, is inevitably driven by a partial view of the world, with heuristics and unconsidered assumptions inevitably playing a crucial role. For finite beings in a world of boundless possibilities, nothing else is possible.
The problem for a new macroeconomics is not so much a failure of economists to understand this point as an embarrassment of riches. Several decades of research in behavioral economics, non-expected utility decision theory and other fields have demonstrated, to anyone willing to look, a wide variety of ways in which real economic behavior differs from the neoclassical ideal. The problem is to focus on behavioral foundations that are most relevant to the problems of macroeconomics.
An obvious place to start is with attitudes to risk and uncertainty. Keynes himself wrote extensively in this topic, and was highly sceptical of the ideas that led to the emergence of the now-dominant expected utility theory (the first formal exposition, von Neumann and Morgenstern’s classic Theory of Games and Economic Behavior was published in 1994 only two years before Keynes’ death. The starting point for expected utility was the idea that people can, and should, reason about uncertainty on the basis of their perceived probability of relevant events such as an increase in interest rates or a slump in exports.
“By `uncertain’ knowledge, let me explain, I do not mean merely to distinguish what is known for certain from what is only probable. The game of roulette is not subject, in this sense, to uncertainty…The sense in which I am using the term is that in which the prospect of a European war is uncertain, or the price of copper and the rate of interest twenty years hence…About these matters there is no scientific basis on which to form any calculable probability whatever. We simply do not know.” (J.M. Keynes, 1937)
Post-Keynesian economists like Davidson and Shackle argued, that this fundamental uncertainty was central to Keynes’ thought and that it had been ignored as part of the development of the Keynesian-neoclassical synthesis. But, as with so many ‘heterodox’ schools of economic thought, the post-Keynesians were much stronger on critique than on the development of a coherent and usable alternative. Shackle in particular ended up denying that we can know anything about probability, even in such simple cases as the toss of a coin, a nihilistic view that was never likely to convince many.
Davidson took the critique in more productive directions and did some valuable work on the way in which attitudes to uncertainty affect individuals demand to hold money, which play a crucial role in Keynes theory of the ‘liquidity trap’, a situation where even at interest rates of zero, investors and households would prefer to save rather than invest.
Mainstream Keynesians, such as James Tobin, had also developed the idea that liquidity preference could be seen as a reflection of risk attitudes. Tobin’s analysis, was developed using the standard financial portfolio analysis based on the idea that the investment involves trading off mean returns against measures of riskiness such as the variance, which depend on the assumption that we can always formulate sensible probabilities for events. Although Tobin himself was always highly critical of the irrational behavior of financial markets, his analysis was easily restated in terms of expected utility theory and absorbed into models based on the efficient financial markets hypothesis.
Over the past thirty years, however, a huge body of research has shown that people do not always make choices in line with the requirements of expected utility, and a great many models of choice under uncertainty have been developed over the past thirty years to produce more realistic representation of behavior. Probably the most famous is the prospect theory of Kahneman and Tversky, put forward in 1979, which earned Kahneman a Nobel prize in economics and Tversky a rare posthumous mention.
My own academic career got its start with a paper published a couple of years later, giving a tweak to the idea of probability weighting by showing that the model worked better if low-probability extreme events (large gains and large losses) were overweighted, while events leading to intermediate outcomes were overweighted. Kahneman and Tversky incorporated this idea in a revised version of their original model, called cumulative prospect theory.
What specific features of a more general and realistic model of choice under uncertainty might contribute usefully to a renewal of Keynesian macroeconomics? There are at least two obvious examples. First, there is the problem of unknown unknowns, which is also, and not coincidentally, a critical problem for the efficient markets hypothesis. An obvious feature of economic crises is that people are forced to consider contingencies they might previously have disregarded, such as the possibility that their employer, or their bank might fail, or that currency might rapidly lose its value. When such a contingency suddenly enters the minds of many people, large macroeconomic shocks may result.
Second, as I’ve already mentioned, although people fail to consider some low-probability extreme contingencies, they tend (perhaps in compensation) to overweight those they do consider. It is this fact that keeps the sellers of lottery tickets and air crash insurance in business. In the macroeconomic context, a ‘normal’ situation in which people disregard or at least do not account for the risk of a serious recession may suddenly be replaced by a far more pessimistic outlook in which the same people place a high weight on the possibility of total economic collapse. Unsurprisingly, such a change in ‘animal spirits’ may represent a self-fulfilling prophecy. If a lot of people expect a recession and try to increase savings and reduce investment, these defensive actions may bring about the recession against which they are designed to guard.
Of course, awareness of this fact will do nothing to moderate the potential impact; if anything the reverse. People who are suddenly worried about a recession will not, if they are looking to their own well-being, keep spending in the hope that others will do likewise and thereby keep the economy afloat. Rather they will reason that others are likely to think as they do, and that a recession is even more probable than the objective evidence would suggest.
Keynes talked about such phenomena in terms of ‘animal spirits’. Such notions seemed hopelessly old-fashioned in the light of the development of rigorous models of choice under uncertainty based on the elegant axioms of expected utility theory, the apparent success of the theory in explaining a wide range of economic behavior and the dominance of the efficient markets hypothesis. But as the evidence against these models has mounted, the pendulum has swung. The idea of animal spirits has been revived in by George Akerlof and Robert Shiller in a recent book of the same name.
Akerlof and Shiller consider five deviations from the standard model of rational maximization (confidence/trust, fairness, corruption, money illusion and stories) and argue that some combination of these can be used to explain a range of economic outcomes inconsistent with the standard model. Their discussion makes a compelling case that macroeconomics needs new, and more realistic foundations.
If the prospects for a macroeconomic analysis based on alternatives to expected utility theory are so promising, why has so little work been done along these lines? In part, perhaps, this simply reflects the effects of specialisation. Decision theorists focus on individual choices, and when they seek economic applications, this leads them naturally to look at microeconomic problems (that’s certainly true in my own case).
But there is a more fundamental problem. Individuals who satisfy the conditions of expected utility theory display a property called ‘dynamic consistency’ which, as the name suggests is of fundamental importance in dynamic stochastic general equilibrium models. Dynamically consistent economic agents never change their view of the world in any fundamental way. They respond to new information by changing their subjective probabilities for particular events, but they never change their underlying prior beliefs and preferences about the world. That means, in particular, that they can fully anticipate how they will respond to any possible future situation, and would never wish to change their mind about this, or to ‘lock themselves in’ to a course of action they might be unwilling to carry through when the time comes.
Such consistency is admirable (at least in the eyes of decision theorists) and makes it much easier to obtain well-defined solutions for dynamic stochastic general equilibrium models. But it is far from realistic. It turns out, however, that the decisions predicted by such models always display dynamic inconsistency under certain circumstances. This problem has been the subject of considerable controversy on the rare occasions when economists have sought to introduce non-expected utility preferences into macroeconomic theory (as with the robust control theory of Hansen and Sargent.
From the neoclassical viewpoint that dominates modern macroeconomics, the absence of a coherent dynamic equilibrium concept seems like a fatal objection. But from a Keynesian perspective, and on the basis of real world experience, this is a positive, indeed necessary, feature of a sensible macroeconomic model. The fundamental macroeconomic problem is precisely that an economy that seems to be enjoying an equilibrium path of steady growth can suddenly crash or veer off into an unsustainable boom.
Aggregate models and equilibrium
If there is one thing that distinguished Keynes’ economic analysis from that of this predecessors it was the rejection of the idea of a unique full employment equilibrium to which a market economy will automatically return when it experiences a shock. Keynes argued that an economy could shift from a full-employment equilibrium to a persistent slump as the result of the interaction between objective macroeconomic variables and the subjective ‘animal spirits’ of investors and other decisionmakers. It is this perspective that has been lost in the absorption of New Keynesian macro into the DSGE framework.
The revival of notions like ‘animal spirits’ by leading economists such as Akerlof and Shiller offers the potential to revive these fundamental Keynesian insights. But this is not simply a matter of modifying the way we model individual behavior Phenomena like animal spirits, social trust and business confidence can’t be reduced to individual psychology. They arise from economic and social interactions between people.
It’s precisely for this reason that such social aspects of individual psychology are likely to be associated with multiple equilibria in the real economy. The aggregate level of trust and confidence in an economy cannot be derived by simply adding up individual values in the way in which DSGE models aggregate consumer preferences.
As long as particular assumptions are implicitly taken for granted in a given social group, such as the business community, few members of that group are likely to consider the possibility that these assumptions might fail. Evidence against those assumptions will be ignored or explained away. So, for example, the spectacular examples of market irrationality and business corruption exhibited during the dotcom boom and bust did almost nothing to shake the faith of business and political leaders in the efficiency and stability of financial markets. This faith remained strong even as the evidence of fundamental problems grew through 2007 and early 2008. Then, in the space of a few months this confidence collapsed to be replaced by a panic in which even the most reputably financial institutions would not lend to each other, and instead threw themselves on the protection of the national governments they had previously dismissed as obsolete relics.
A realistic macroeconomics requires the incorporation of variables like trust and confidence in explanatory models. Fluctuations between ‘irrational exuberance’ and equally irrational ‘panics’ (this old term for a financial crisis is in many ways more useful than the technical language of ‘recessions’) give rise to bubbles and busts, which in turn drive much of the macroeconomic cycle. The insights of behavioral economics provide good reasons to expect such fluctuations, but they do not, at least as yet, admit the kind of rigorous derivation of aggregate values from individual preferences that is referred to in the standard demand for ‘microfoundations’.
Expressed in the language of systems theory, the traditional Keynesian approach treated macroeconomic behavior as an emergent property of the economic system, to be analysed in their own terms rather than being derived from supposedly more ‘fundamental’ microeconomic explanations.  In a world of boundedly rational economic decisionmakers, and, for that matter, boundedly rational economists, we need to simplify and the simplifications that are appropriate for doing macroeconomics may not be the same as those that are appropriate in microeconomics.
Obviously, it’s much easier to announce a new program for macroeconomics than to actually implement it. To give some more concreteness to the general proposals presented here, it’s worth thinking about some specific problems, such as bubbles and the ‘Minsky moments’ in which they burst.
Bubbles and Minsky moments
Macroeconomists working in the micro-economic foundations framework did not ignore bubbles. Far from it. Dozens of papers were written on the possibility or otherwise of self-sustaining bubbles in asset markets. But, characteristically, the central concern was to determine whether or not bubbles could arise in markets with market participants who were perfectly rational, or nearly so. This focus on microeconomic foundations diverted attention from the real issues.
There was a rather smaller policy oriented literature, concerned with the question of whether central banks should intervene to prevent the emergence of bubbles, or to burst them early, before they became too damaging. Most of this literature followed the lead of Alan Greenspan, who initially showed some sympathy for the idea of intervention, but eventually became the strongest advocate of the view that central banks should not second-guess markets. But even interventionist participants in the discussion took it for granted that an anti-bubble policy had to be implemented within a policy framework of inflation targeting using interest rates as the sole policy instrument. With these constraints, the conclusion that nothing could or should be done was largely inevitable.
A realistic theory of bubbles would start with the observation that every bubble has a story to explain why, in the words of …, ‘this time it’s different’. And, for particular assets and markets, sometimes it is different. Those who got in early with shares in Microsoft or Google, or with land in … in … multiplied their money many times over. And although the days of spectacular growth came to an end in each case, there was no bursting of the bubble ending in losses all around.
So a theory of bubbles designed to inform a policy of bubble-pricking must begin with an attempt to understand how ‘this time it’s different’ stories emerge and come to be believed and how to distinguish true, or at least plausible, stories from those that involve a collective abandonment of reality. The story-telling aspect of animal spirits discussed by Akerlof and Shiller is important here.
Given a better understanding of bubbles it may be possible to develop an analysis of the costs and benefits of pricking putative bubbles. Such a policy reduces the damage from spectacular busts such as the one we have just seen, but it would require a willingness on the part of central banks to explicitly over-ride the judgements of capital markets, rather than merely ‘leaning against the wind’ by raising interest rates.
An uncontrolled bubble must eventually burst, and the bursting of a bubble is a prime example of a ‘Minsky moment’, when euphoria suddenly turns to panic. In Minsky’s model there are three classes of financial enterprises – conservative ‘hedge’ financiers whose operations generate sufficient income to service their capital costs, speculative financiers who rely on rising asset prices to service debt and who drive the market further upwards, and ‘Ponzi financiers’ cover their costs in either the short term or the long term, but who can conceal their insolvency long enough to reap substantial gains. Ponzi operators fail from time to time, but, in periods of growth, these failures are seen as isolated events of no general significance. However, in the later stages of a bubble, when a large proportion of economic activity has been devoted to speculative finance, the failure of a Ponzi financier can bring about a sudden shift in sentiment, as investors fear that the associated corruption is widespread. The rush to withdraw extended credit brings about more failures, not only of Ponzi financiers but of the speculative finance firms that relied on continued growth.
The last Keynesian golden age ended in stagflation. The causes of this breakdown are many and complex, but they must be addressed if we are to avoid repeating them. In particular, it is important to avoid relying on easy excuses, such as the 1973 oil shock and to face the fact that the stagflationary breakdown reflected serious failures in the dominant version of Keynesian macro theories, and in the political and industrial strategies of the social democratic, left and labour movements. These failures were amplified by the expansion, from very small beginnings, of a global financial system that broke down the institutional framework of the Bretton Woods agreement.
The discovery of the Phillips Curve around 1960, and the general success of Keynesian macroeconomic policies in the postwar period produced increasing support for policies of fiscal expansion aimed at reducing already low levels of unemployment even further, and an acceptance of higher rates of inflation and sustained budget deficits as a reasonable price to pay. This intellectual atmosphere fitted in neatly with the political needs of the Johnson Administration in the US, which sought to implement both an expensive (but initially quite successful) set of welfare programs dubbed the War on Poverty and an actual, if undeclared, war in Vietnam, while avoiding the political opprobrium of raising taxes. There were similar developments in other countries as pressure to expand the welfare state ran into the first elements of resistance that would later become the Tax Revolt of the 1970s.
In the short run at least, expansionary fiscal policies resolved these problems, and an expansionary fiscal stance became accepted as the norm. This contrasted with the older Keynesian approach where expansionary policies used to stimulate the economy out of recessions and depressions were balanced by contractionary policies aimed at controlling overheated booms.
From the late 1960s onwards, rates of wage and price inflation rose steadily. Throughout society, the combination of (seemingly permanent) full employment and economic growth with inflationary pressure led to the abandonment of attitudes of restraint that had, until then, been engendered by memories of the Great Depression and fears of a new one. Business leaders ceased to be the sober, socially-minded, technocrats described in works like JK Galbraith’s New Industrial State and started on the path that would lead to the lionization of figures like ‘Chainsaw Al’ Dunlap and Jack Welch.
Financial markets shook off the memories of the Great Crash and became, once again, places where vast fortunes could be made from abstruse transactions. Most attention was focused on stock markets, which went through their first real boom since the 1920s. More significant in the long run was the (re)emergence of an uncontrolled global financial market. This began, with the creation of the ‘Eurodollar’ market, in which mostly European banks located outside the regulatory control of the US dealt in dollar-denominated securities, with liquidity provided by the shift of the US balance of payments from a century old pattern of surpluses to an almost equally durable string of deficits. in one of history’s ironies, the most important single player in the early years was Moscow’s Narodny Bank, which faced increasingly pressing needs for access to Western financial markets and an equally pressing imperative to avoid the control of US authorities.
But the most striking manifestation of the inflationary breakout took place in labour markets. There was an explosion of labour militancy, reflected in an upsurge in strikes and in wage demands that could not be met except through continuing inflation. Even without the militant push, low unemployment would have strengthened the bargaining power of unions and put upward pressure on wages. But the revolutionary utopianism of the 1960s, exemplified by the events of May 1968 in Paris, produced an atmosphere where any kind of restraint became impossible. Unions that sought to focus on realistic and sustainable demands were pushed aside by their own members.
By 1973, after the breakdown of the Bretton Woods system and a failed attempt by the Nixon Administration to halt inflation through a wage-price freeze, the era of Keynesian dominance was drawing to a close. The coup de grace came in October of that year when, in response to US support for Israel in the Yom Kippur war, the members of the Organization of Petroleum Exporting Countries first cut off oil supplies to the West, then raised prices fourfold.
The oil shock was a consequence, not a cause of the inflationary upsurge. Commodity prices were rising sharply across the board well before this event. However, the structure of the oil market, with a small group of oil companies (called the ‘Seven Sisters’) facing an increasingly well-organized OPEC meant that, when the price shift came, it took the form of a single dramatic leap. And, having been caused by stagflation, the oil shock amplified and entrenched it in the economic system, leading to decades of high unemployment and persistent inflation.
The stagflationary outbreak took a heavy toll on the Keynesian social-democratic welfare state and the organizations and ideas associated with it. In particular, the great wage push was disastrous for both for the unions and for the Keynesian/social democratic system. The seemingly-continuous strikes of the 1970s undermined popular support for unions and paved the way for a series of ever-more brutal assaults by governments and employers. Margaret Thatcher’s crushing victory over the National Coal Miners and Ronald Reagan’s equally successful action in firing striking air traffic controllers en masse brought an end to the idea that strikes represented a reliable route to improved wages and conditions, let alone to the collapse of the capitalist system. Particularly in the English-speaking world, union membership dropped rapidly as new laws made it easy for employers to keep unions out.
Keynesian economists were discredited and driven from positions of power by monetarist and new classical rivals. Only by making the massive theoretical and policy concessions involved in New Keynesianism were they able to regain a seat at the table.
Meanwhile, the financial sector, which had precipitated the crisis claimed victory, as did the economists who extolled its merits. Stagflation was seen as a demonstration that attempts to resist the logic of the market must ultimately fail. It took several decades to relearn the Keynesian lesson that an uncontrolled financial system will fail even more disastrously
The inflationary surge that began the late 1960s has some important lessons that must be learned if we are to avoid similar failures in the future. First, it is important to maintain a focus on keeping inflation rates low and stable as well as on maintaining full employment. Once inflation rates get signficantly above 3 per cent per year, the risk of embedding inflationary expectations, and the eventual cost of lowering those expectations, becomes greater. It is therefore important to maintain a commitment to low inflation and to adopting the policies necessary to contain and reduce inflation when some shock to the system produces a significant increase in the price level.
At a theoretical level, this does not involve huge modifications to the standard Keynesian view. The idea of a stable long-run trade-off between unemployment and inflation, represented by the Phillips curve, was a relatively late addition, and quickly abandoned. But the problem of how to deal with inflation remains largely unresolved.
In policy terms, inflation can’t be reduced unless macroeconomic policy acts to constrain excess demand and liquidity. So Keynesian policies must be used consistently throughout the cycle, to reduce excess demand in boom periods as well as stimulating demand during recessions.
This still leaves the problem of what to do if high inflation becomes established. A number of countries showed, in the 1980s and 1990s, that a co-operative approach could reduce inflation and unemployment simultaneously. In Australia, following a deep recession in the early 1980s, the newly elected Hawke Labor government reached an agreement with the trade unions referred to as ‘The Accord’. Under the Accord, unions agreed to reduce the rate of growth of wages in return for an increase in the social wage, most notably the introduction of a national system of health insurance, called Medicare.
At about the same time, and facing similar problems, unions and employer groups in the Netherlands negotiated the Wassenaar agreement. In this case, the trade-off for wage moderation was a reduction in working hours and the adoption of a range of measures designed to promote employment growth. The Wassenaar approach survived the stresses of the early 1990s and, according to the ILO was “a ground breaking agreement, setting the tone for later social pacts in many European countries.”
The co-operative approach that motivated these policies was ultimately swept away by the ever-growing power of the financial sector. But, if a Keynesian policy framework is to be successful, it must be revived. Hopefully, the memory of past disasters will promote a more cautious and co-operative approach in future.
 Unfortunately, discussion of these ideas tends to get bound up in more or less mystical claims and counterclaims about reductionism and holism. But nothing of that kind is intended here. In principle, without doubt, all social phenomena are determined by interactions between individual people, whose behavior is in turned determined by their genes and the environment in which they grew up. Genes are collections of DNA molecules which in turn are made up of atoms made up of subatomic particles behaving according to the laws of quantum physics. If we were the unboundedly rational individuals posited in the DGSE literature, , such we would presumably be doing quantum physical calculations whenever we made economic decisions.