2.1 Behavioral Finance

Behavioral economists come in two flavors: those who view behavioral considerations as substitutes for traditional economic analysis based on the assumption of rational decision makers and those who view them as complements.Footnote ⁸ I am in the later camp which I believe is now the majority view. This approach needs not only to investigate various possible deviations from efficiency, rational or not, but also the conditions under which different types of behavior are more likely to occur. These will interact with the standard economic rational reasons for inefficient outcomes such as moral hazard and principal–agent problems, which will be discussed in Section 3.

We can identify at least three important complementary strands of a broad-based behavioral approach. The first investigates various possible deviations from the types of rational decision making assumed in most traditional economic analysis and the implications of these various deviations.

The second strand emphasizes real-world constraints such as imperfect information, decision making costs, and limitations to humans’ cognitive powers. These can lead to behavior that would be irrational in the context of the standard economic models but can become rational in more complex environments. Kay and King (Reference Kay and King2020) refer to such considerations as behavior in small versus large worlds. An important early contribution to this approach was Simon’s (Reference Simon1991) concept of bounded rationality and this has given rise to large literatures on the economics of information, including the effects of perverse incentive structures such as principal–agent and moral hazard problems.Footnote ⁹ Complexity economics is also an important element of this strand.

One important aspect recently stressed by Kay and King is Knight’s (Reference Knight1921) important distinction between risk and uncertainty, also emphasized by Keynes (Reference Keynes1921).Footnote ¹⁰ Much of economics has tended to treat imperfect information as a matter of decision making under risk, where probability distributions are known, even as the particular outcomes from those distributions are not. Kay and King show that this approach can lead to serious difficulties when the true situation is one of uncertainty and the probability is not known. A prime example was putting too much faith in value at risk (VaR) models which lured many financial decision makers into taking on more risk than they realized. The standard VaR models assumed a normal distribution which gives too little weight to tail events such as crises. To relax this assumption, current technologies such as the bootstrap, Monte Carlo methods, and machine learning models involve conducting thousands of resampling calculations to simulate and derive probability distributions which can then be used to calculate VaR. While there have been substantial improvements, these methods are still subject to the problem of structural shifts.

Coping with uncertainty can make behaviors, such as emphasis on narratives, stories, or mental models, and rules of thumb, which would be quite irrational in a world of traditional economic models, rational.Footnote ¹¹ The potential for efficient use of simple narratives under conditions of uncertainty has also recently been emphasized by Shiller (Reference Shiller2020), as have the problems generated by the spread of false narratives.

Although it can be argued that this approach should not be considered part of the behavioral approach per se, I see it as very kindred in spirit in terms of helping to explain why simple models of rationality sometimes do not work. Thus, I include it under a broad concept of the behavioral approach. Alternatively, one could consider it as a complementary approach to a narrower concept of behavioral finance.

A third strand of literature has emerged that has begun to look for explanations in terms of evolutionary theory. These writers stress that evolution operates over a long time scale and that many of the habits that cause problems for behavior in financial markets today stem from behavior that was beneficial in the past. Thus, “Homo sapiens haven’t had time to adjust to the new realities of life” (Lo Reference Lo2019, p. 20). The financial sector is an area where the need for adjustment to modern conditions is particularly great. Drawing on recent advances in neuroscience, this approach stresses how the structure of the brain itself limits its ability to instantly adapt. While the evolution of the brain offers new capacities and powers of analysis, this new capacity was built on top of, rather than replacing, our primitive instincts developed at an earlier stage of evolution in the amygdala, sometimes called the “lizard” part of our brain.Footnote ¹² These two portions of the brain are the basis for Kahneman’s important contribution, Thinking, Fast and Slow (Reference Kahneman2011), which stresses the differences between what he calls System 1 and System 2 thinking, one operating quickly based on instincts and the other operating more slowly based on more analytical thought.

Drawing on this type of analysis, Lo (Reference Lo2019) suggests that “investors and financial markets behave more like biology than physics,” and that “the principles of evolution are more useful for understanding the inner workings of the financial industry than the physics like principles of rational economic analysis” (p. 2). While the question of which approach is more useful is quite interesting, we need not reach a conclusion on the issue to acknowledge that both are important.

2.2 Efficient Market Theory and Its Critics

There are two major propositions associated with efficient market theory. One is that individual agents cannot beat the market. There are likely exceptions to this conclusion but a looser form – that it is very hard to beat the markets – has a huge amount of support, at least for well-developed financial markets.Footnote ¹³

The second and much stronger conclusion is that markets quickly set the correct or equilibrium price. While this second conclusion implies the first, contrary to what has often been assumed, the first proposition need not imply the second. The considerable evidence supporting the first proposition need not provide equal support for the second one.Footnote ¹⁴

A key assumption of efficient markets theory is that there are no serious limits to arbitrage, so that efficient speculators can take advantage of irrational ones and in the process will make profits and force prices to their efficient levels. Only when these arbitrage conditions are not met will the cognitive biases discussed in the behavioral finance literature become relevant for market prices. Even in many well-developed financial markets, these assumptions sometimes do not hold. Many of the opportunities to take advantage of perceived mispricing of assets are in fact quite risky. The current popularity of the label risk arbitrage is actually a contradiction in terms since it is not risk-free arbitrage. This label has perhaps become preferred as it sounds more socially acceptable than calling it speculation, which it actually is.Footnote ¹⁵ We turn next to a discussion of the major types of conditions, both rational and behavioral, why markets may not produce efficient outcomes.

3.1 Reasons Consistent with Rational Behavior

3.2 Behavioral Biases

Behavioral biases constitute the area most associated with the technical literature on behavioral and neuroeconomics, and finance. A standard argument often made by economists against those arguing for irrational behavior is that there are almost an infinite number of ways in which people could be irrational, so it is difficult to undertake useful analysis based on this. Developments in cognitive psychology and neuroscience have made this much less of a problem by focusing on a more limited set of possibilities.

The number of such potential biases is quite large, but behavioral economists and finance experts have been able to focus on a limited number that appears likely to be particularly relevant for economic and financial behavior, and which can often be formulated in ways that can be subject to empirical testing. Here I focus on an even smaller set that I have found to be highly useful for international monetary and financial analysis.

3.3 Faulty Mental Models

A frequent cause of market failures such as bubbles, crashes, and currency and financial crises is the widespread adoption by market actors of narratives or mental models which turn out to be faulty.Footnote ³⁰ The world can be complex, so it is not necessarily a sign of irrationality that people sometimes hold views that turn out to be incorrect. However, potential irrationality may appear under circumstances such as the following:

• Doing insufficient homework before adopting a particular view, especially as a result of going with the herd and adopting fashionable narratives such as “this time is really different”;

• Adopting a narrative because one would like it to be true, an example of wishful thinking and confusing positive with normative analysis;

• Placing excessive faith in a particular narrative to the exclusion of considering other possibilities. Potential causes for this are hubris, leading to the belief that those with other views aren’t as smart and just don’t “get it,” and confusing views that do have some explanatory power with thinking that they are all that is relevant;

• Excessively discounting evidence that contradicts one’s view, confirmation bias.

It is sometimes difficult to decide unambiguously whether sufficient analysis has gone into the selection of a particular narrative or mental model. Likewise, it is often impossible to take every potential contingency into account when making decisions. There are questions about which reasonable people may often differ in particular cases. As has been pointed out by authors such as Kay and King (Reference Kay and King2020), rationality is difficult to apply to narratives. They suggest it is more useful to use concepts such as “productive” or “counter-productive,” or from an evolutionary perspective, adaptive or maladaptive.

As with formal theories, it is important to distinguish between narratives that are fundamentally wrong and those which are accurate yet leave out major aspects of the question. It is usually impossible to take every aspect and potential contingency into account when making decisions. Thus, few theories can be complete. Faulty decision making comes in where agents treat the story as if what it focuses on are the only things that matter. This tendency becomes particularly counterproductive when the narrative selected focuses only on a minor, rather than major, aspect of the issue.

Given the lack of objective criteria to label particular narratives as faulty, I leave it to the readers to make their own judgments about particular narratives. In the following section I briefly discuss a series of major financial crises where faulty narratives or mental models contributed importantly to the crisis.

4.1 Kernel of Truth: The Laffer Curve and the UK Crisis

“Kernel of truth” fallacies are true narratives which are carried too far. For example, in stock market bubbles, arguments that “this time Is different” often have some truth, in the sense that the developments which spark them will differ at least somewhat from the innovations that sparked past bubbles, but this does not mean that everything is different, as investors frequently find out when the bubble crashes. One of the most important of these involves the Laffer curve.

In this example, the perfectly valid argument that tax increases will tend to reduce supply was taken too far and transformed into the belief that tax cuts would increase supply so much that they would pay for themselves. Just as there are revenue maximizing prices for firms, there will be revenue maximizing level of taxes and tariffs. When prices or taxes are above these levels, reducing them will increase revenues.

This is a perfectly sound economic concept but has frequently been misused in the political arena. Conservatives had traditionally been viewed as facing a dilemma between their objectives of cutting taxes and avoiding large fiscal deficits. Economist Art Laffer famously introduced the idea into American politics that cutting taxes would increase rather than reduce revenues and hence reduce fiscal deficits, thus offering a way out of the Republican dilemma. While this idea was already well known by economists, it became widely known as the Laffer curve and was quickly embraced by Republicans.

Where there are cases where cutting specific taxes has increased revenue, these were generally cases where supply curves were quite elastic. However, supply schedules for broad-based changes such as income taxes will tend to be much less elastic. Several presidents, most famously Mr. Reagan, tried this strategy, but their income tax cuts did not increase revenues.Footnote ³²

This, however, did not keep President Trump from ignoring the evidence and pushing through an income tax cut which he claimed would pay for itself. It did not, and the US deficit soared. Undeterred, President Trump continued to pursue large tax cuts in his second term despite the already huge fiscal deficits. The only surprising thing about this episode is that, while many economists worried about the sustainability of these budget deficits, the financial markets have shown only moderate concern.Footnote ³³

However, this was not the case with the budget proposals of the UK’s new prime minister Liz Truss in September 2022. Despite all of the accumulated evidence that tax cuts of the levels seen in the US and UK would not pay for themselves, with the UK already in a dire fiscal position, she went ahead and proposed a Laffer curve type budget with substantial tax cuts. Given her previous economic writings, it seems entirely possible that she truly believed in the Laffer curve and had not really looked at or not been convinced by the evidence that the cuts would not pay for themselves. In any event, reactions in both the long-term financial and foreign exchange markets were swift and powerful. Many financial institutions had not hedged against possible increases in interest rates, and the substantial increases in rates following Truss’s proposal threatened the solvency of many of them. A full-scale financial crisis was a real possibility and forced Truss to withdraw her proposal, after which the markets fairly rapidly calmed and soon after she was forced to resign.

4.2 The False Belief That if the Macro Fundaments Are Sound, Then the Financial and External Sectors Will Take Care of Themselves

4.3 “Countries Don’t Go Broke,” and the Latin American Debt Crisis

This adage, strongly associated with former head of Citibank Walter Wriston, became popular in the 1970s, and belief in it likely played an important role in generating the flood of bank lending to Latin American governments during that period. On the face of it, Wriston’s statement is technically true. As he explained, a country’s assets include all of its resources, and these are bound to exceed its international liabilities. These resources do not go away, so countries cannot go out of business as firms would when they become insolvent. Governments do not have all of these resources at their disposal, however, so their assets can frequently fall far below their liabilities. Up to a point this is not a problem, but continued large deficits can lead to debt levels that become unsustainable. History is littered with examples of government defaults, and this is just what happened with a number of Latin American governments, beginning in 1982, and leading to a major threat to the solvency of the US banking system and to a lost decade of growth in Latin America.Footnote ⁴¹ Heavy initial borrowing may have been quite rational since interest rates were quite low. However, many of the loans were denominated in dollars at variable interest rates, and when US interest rates began to rise substantially, the burdens of these debts increased in step.Footnote ⁴² At the same time there were major slumps in the prices of the major commodities that many of these countries exported. The crises soon followed.

4.4 Moral Hazard and Countries and Institutions Too Important to Default

We are perhaps most used to hearing arguments about moral hazard in the case of private sector institutions that governments consider too economically or politically important to allow them to go bankrupt. These situations can indeed contribute to excessively risky activities, given the assumption that the companies or individuals will keep any profits generated but will be protected at least to some degree if things go bad. Such policies can result in perverse liberalization, where connected firms are given more profit opportunities but feel protected against many of the risks they take. Such considerations were quite important in causing the financial sector problems which contributed to the Asian crisis.

In other cases, large amounts of private sector lending to governments were motivated by the belief that, for a variety of reasons, these countries were considered too important by members of the international community to be allowed to default. Such considerations also gave rise to considerable lending by international institutions when troubles began to arise. In a number of cases, however, such official lending served only to postpone, not avoid, defaults. Three of these cases are briefly discussed in the following sub sections.

5.1 The Interrelationships between Uncertainty and the Prevalence of Behavioral Biases

Under conditions of uncertainty, it is usually much more difficult to tell which narratives or mental models are more appropriate. As Lo (Reference Lo2019) stressed in his adaptive markets hypothesis, markets tend to learn over time, and in stable environments will also tend toward efficiency. But changing environments generate uncertainty, and learning behavior may need to begin all over again, with the added complication of first requiring some unlearning.

Rapid, unambiguous feedback, as well as open minds are both conducive to learning. These conditions apply in many areas, but often not in financial markets. There can be considerable periods of time between some decisions and their consequences. In the meantime, individuals may continue to stick with their initial views. In a complex world, unforeseen developments can make decisions which were ex ante efficient turn out to be ex post inefficient. In complex situations, extracting the signal from the noise can often be a highly difficult and sometimes impossible task.

Given the strong reluctance of many people to admit mistakes, it is often not difficult to convince ourselves, and perhaps also others, that we made the correct decisions ex ante and that our mental models or views of the world remain sound. Hindsight bias can be an important contributor to this process, as can hubris, confirmation bias, and wishful thinking. Some people are so committed to ideas that almost no evidence can convince them that they are wrong. The continued adherence by some to the Laffer curve argument that income tax cuts will pay for themselves remains a major plank of Republican thinking in the US, despite considerable evidence against it.

5.2 Estimating Risk and the Cost of Errors

Important examples of how psychological tendencies or biases may sometimes interact involve the incentives for herding into investment opportunities and excessive responses to lower interest rates. Let us consider first the tendency of investment managers to herd into taking excessively risky investments. Of course, one possible and not completely rational psychological reason for following the leader is the feeling of wanting to belong, a desire which has been ingrained in humans over the ages. A second type is the inheritance of behavior which was quite rational in a previous environment that no longer obtains. In the state of nature in which humans lived for eons, the cost of not following a suddenly running group member could be death by being eaten by a tiger, while the comparative costs of running in response to a false signal was relatively low. What was highly adaptive in that environment may not be so in the environment of today’s financial markets.

There are also strictly rational reasons for such responses today, in an environment which itself has important inefficiencies. As Kahneman (Reference Kahneman2011) discusses, overconfidence in one’s forecasting ability often encourages individuals to underestimate risks, and this applies to professional experts as well as amateurs. Indeed, he reports on studies finding that excessive risk-taking in the financial sector is generated more by underestimating risks than by decision makers being overconfident of expected returns, or by the incentives generated by playing with other people’s money. While professionals in many fields often tend to substantially understate the likely range of possible outcomes, those who are more accurate about the risks involved, instead of being rewarded for having more accurate forecasts, tend to find themselves replaced by more confident individuals even though their greater apparent confidence is not justified.

An important contributor to such behavior is another possible bias, the tendency to have short time horizons, that is, high rates of time discount. Rewards and punishments often tend to be handed out over fairly short time periods, so that the conservative manager may not still be around when they are proved to be right. There are many stories of risk managers who warned of the greater risks being accumulated by many financial institutions leading up to the US subprime crisis being shunted aside or being replaced. This is a case where more rational individuals tend to lose out because of the environment generated by the less rational behavior of others. And, of course, confirmation bias makes these other actors less likely to pick up on information that support the views of the less overconfident actors.

At first glance, such a tendency of excessive risk-taking would seem to conflict with the emphasis placed by Kahneman and others on the importance of risk aversion. This would seem to yield high risk aversion and, if anything, lead managers to exaggerate the risk of large losses such as typically occur in crises. Behavioral science’s emphasis on the importance of framing may help explain this seeming contradiction. For an individual investor in isolation, this would indeed be a contradiction if it were not for the tendency to underestimate risks, which could have a stronger effect. However, where a manager’s reputation and financial rewards are heavily influenced by the behavior of others in his “tribe,” loss aversion may contribute to going along with the herd. In this framing, the most relevant loss, in this case, is the manager’s reputation and perhaps job.Footnote ⁴⁴

5.3 Learning and Forgetting and Time inconsistency Problems

A central insight of Lo’s (Reference Lo2019) adaptive markets hypothesis is that while market participants are subject to the types of biases and cognitive limitations emphasized in the traditional literature on cognitive psychology and behavioral finance, such behavior occurs particularly strongly in rapidly shifting environments where there is great uncertainty. As discussed Section 3.1.3, such environments make it easy for behavioral biases and faulty mental models to thrive. With stable environments there tend to be more consistent feedback from actions. It helps a great deal if such feedback occurs fairly quickly, and the interpretation is unambiguous. It also helps to evaluate the effects of actions if they occur fairly frequently so that there are more data points. Of course, confirmation bias can substantially impede learning. Still, if the environment stays stable long enough, market behavior should tend to evolve toward efficiency. However, as Lo stresses, environments can change frequently so that steady progress toward greater efficiency is not assured.Footnote ⁴⁵

Learning that old narratives have become defective in the new environment, and that developing new more productive ones is necessary, is what Lo means by evolution at the speed of thought, as opposed to the slow operation of physical evolution. While Lo is fairly optimistic that good narratives will tend to drive out bad ones, Shiller (Reference Shiller2020) is much less so. We would expect that narratives about trading strategies would be subject to more useful feedback than those concerning the effects of many types of loans and government policies.Footnote ⁴⁶ It took quite a long time for the results of the risky lending and investments during the housing bubble in the US to demonstrate their true riskiness. Likewise, riding up a bubble can give good feedback until it bursts.

Government policies are also likely to be subject to weak feedback systems. Policies often change only infrequently, yielding fewer data points, and their effects frequently take many years to be fully felt. And even then, interpretations may differ greatly.

Memory can be an important source of learning. Unfortunately, people tend to learn much more from their own mistakes than from those of others. For example, Asian countries tended to learn much more from the Asian crisis of 1997–98 than did the advanced economies, and many of the mistakes leading to the Asian crisis were repeated in the run-up to the global financial crisis in the 2000s.

Agents often give too little weight to likely future developments if they are not expected to occur fairly quickly, as people often have excessively short time horizons, that is, too high time rates of discount. This contributes importantly to another major source of crises, time inconsistency problems. These occur where the effects of the costs and benefits of actions occur disproportionately at different times. People then tend to be excessively biased toward actions that have most of their good effects first and bad effects later, and against actions that have more of their bad effects first and good effects later.

This gives rise to the famous case of political business cycles where, to increase their chances of reelection, governments adopt expansionary policies shortly before elections to benefit from the initial predominantly good effects on output and unemployment, while the bad effects on inflation appear mainly after the election is safely over. Such actions by governments may be quite rational, but to be successful, they rely on the majority of the public not having sufficient information or rational expectations to anticipate the later bad effects. Some new classical economists argued that political business cycles would not be a continuing problem because the public would catch on and punish governments who tried it.

This learning mechanism has often been weak. Memories may be short. Furthermore, such ignorance may be quite rational. In a complex world, it does not pay for voters to be highly informed on a vast range of issues; thus “rational ignorance” is an important concept for political economy analysis. This is overlooked by some proponents of rational expectations macro models, leading to assumptions of much better knowledge than many economic agents actually have. This can, in turn, lead to the conclusion that most changes in policies are anticipated and hence have little impact on the real economy, and that there are only very limited incentives for politicians to generate political business cycles.Footnote ⁴⁷ In the financial sectors, many decision makers have much stronger incentives to be well informed but even there it is often argued that financial markets frequently operate on the basis of excessively short time horizons.

Such time inconsistency problems can also contribute to tendencies for governments to be too slow to adopt anti-inflation policies and undertake timely adjustments to avoid currency and financial crises. Such policies are usually costly in the short run while the benefits occur later, and there is always the chance that future developments would reverse balance of payments deficits. The combination of uncertainty about whether the crisis would actually occur if strong actions are not taken and wishful thinking can be powerful forces leading to failure to take adjustment actions in a timely fashion. Such tendencies will be reinforced if an election is nearing.

In some cases, the meaning of stable environments can be unclear. Take for example the insightful adage associated with Herman Minsky (Reference Minsky1975, Reference Minsky1986), among others, that crises are bred in the good times that proceed them. From a shorter-term perspective, the environments of good times and crisis times differ dramatically. But from a longer-term perspective, the tendency of good times to end in crises may be fairly stable in the sense that these are frequently recurring patterns. The time horizons of actors become quite important here. With long memories and foresight, participants in the financial sector, including regulators, should catch on to this pattern and take actions to avoid it. For example, very low volatility might be taken as a signal that the markets are becoming dangerously complacent, rather than that risk is very low.

On the other hand, with short time horizons looking both forward and backward, risks may begin to be substantially underestimated based on recent developments – the “recency” bias explored in behavioral finance. While one major problem with human decision making is the tendency to be shortsighted, also important is the tendency to have short memories. Furthermore, there is a tendency to learn more from one’s own mistakes than from those of others, as when the countries hit by the Asian crisis in the late 1990s tended to learn that strong macroeconomic conditions are not sufficient to avoid financial sector problems and engaged in a great deal of productive financial sector reforms, while this episode had much less impact on those in the advanced economies. These lessons became widely recognized only in the wake of the global financial crisis of 2008.

A further problem is that feedback can be ambiguous. Again, the 2008 crisis provides a cogent example. After almost twenty years, the major causes of the crisis are still under debate. While there is a consensus among mainstream economic and financial experts about some of the most important lessons, for example, the limitations of standard value-at-risk models of risk management and the dangers of complicated derivatives that do reduce risk in good times but multiply them during crises, there is still considerable disagreement among many of the far right and left who, consistent with powerful confirmation biases and the coloring of positive analysis by normative considerations, tend to place major blame alternatively on government policies to increase home ownership among lower income families versus the deregulation of financial markets.

6.1 Bubbles and Capital Flow Surges

Domestic asset market bubbles, credit booms, and capital flow surges are natural areas for behavioral analysis. However, contrary to those who view markets as wildly irrational, seldom if ever do these processes start from purely psychological factors. They typically start for good reasons, such as the development of new technologies, an improving economy, or economic reforms in developing countries.

They are almost always aided by easy money. As Vivek Moorthy has pithily put it in private correspondence, “strict credit limits can curb animal spirits.” Monetary and financial officials often fall prey to the same types of overoptimism that can plague the private sector. The faulty mental model discussed earlier, that if inflation is low the financial sector will take care of itself, also often plays a major role in causing monetary and financial officials to play an enabling role in the generation and continuance of bubbles and capital flow surges through the provision of easy credit.

The problem is that these initially rational reactions sometimes take on a life of their own. The initial success of investments based on recognition of these new developments can become self-propelled by a host of behavioral factors. One is the tendency for false extrapolation. Despite the standard warnings from financial institutions that past performance is no guarantee of future performance, for some individuals, recent substantial price increases can lead to expectations that they will continue, attracting more investors in the markets and increasing both the demand for loans and the willingness of financial institutions to supply them. Wishful thinking can be important here since investors would like the price increases to continue.

Especially with respect to asset prices, stories or narratives may start to spread that such price increases can continue for long periods because “this time is different.” Old criteria for evaluating equilibrium prices are no longer relevant. Those who urge caution “just don’t get it.” Those who do “get it” can feel superior, believing that they are considerably above average in intelligence. Those less sure of themselves may join the herd because they think others know something that they themselves do not. This type of herding can sometimes even be rational in the presence of asymmetries in information.

As in the housing price bubble which led to the subprime crisis in the US, confirmation bias can lead to warning signs that conflict with the dominant narratives being heavily discounted. Indicators such as price–earnings ratios may be considered no longer relevant in the new situation, and the history of past bubbles and crashes overlooked. Fear of missing out can draw others in and keep the bubble going. With respect to real estate, another consideration can be the fear that if one doesn’t buy now, they will soon be priced out of the market.

The developments described so far, while typically not fully rational, are not wildly irrational. However, at some point as the bubble continues, some investors may become carried away in a frenzy of greed and euphoria, an example of the wisdom of crowds turning into the madness of crowds.Footnote ⁴⁸ Perhaps the most important factor influencing whether crowds display wisdom or madness is the extent to which individual decisions are made independently or not. It is when markets or crowds become dominated by the same idea or narrative, or euphoria, that they are most likely to display madness.Footnote ⁴⁹

Some participants in bubbles may have rational reasons although they likely suffer from hubris. The idea is to ride the bubble up, making profits along the way, but then pulling out shortly before it crashes.Footnote ⁵⁰ They expect to profit from the irrationality of other participants. This strategy can be quite risky, however, since getting the timing right may be extremely difficult.Footnote ⁵¹ Overconfidence likely plays an important role for the fund managers who try this strategy.Footnote ⁵²

Cognitive limitations, as distinct from psychological biases, certainly account for part of the tendency of decision makers to acquire insufficient knowledge about facts and the appropriate frameworks for analyzing them. Acquiring information can be costly, and it is often difficult to determine when the costs of obtaining more information begin to exceed the benefits.Footnote ⁵³ Likely much more important than underinvesting in acquiring factual information, however, is the tendency to put too much faith in single conceptual frameworks or narratives to interpret the available facts.Footnote ⁵⁴

Extreme examples of hubris are typically outliers, but in milder forms they are not unusual. Furthermore, the conviction that one is right, combined with the desire to not be wrong, generates one of the most dangerous of behavioral biases, confirmation bias.Footnote ⁵⁵ The combination of these traits is an important explanation for why crises often generate such strong responses. The new signals generated by a crisis can be so dramatic that they often override even strong cases of confirmation bias. The resulting realization that one’s narrative was seriously defective would typically prompt much stronger reactions than that the crisis was just a bad draw from a known probability distribution.Footnote ⁵⁶

Recognition that one no longer understands the situation can understandably lead to extreme risk aversion, resulting in investors running for the exits. This helps explain why initial mild contagion from crises is often widespread, with later contagion becoming stronger and much more focused as investors begin to formulate new views.

In a complex world characterized by considerable uncertainty, the adoption of a view that turns out to be wrong is no sure sign of irrationality, and it can sometimes be difficult to judge whether a particular view was plausible ex ante or more the result of oversimplified or wishful thinking. What we can say, however, is that in a world of uncertainty it is not rational to base decisions only on a single perspective. Hubris and overconfidence are major contributors leading to beliefs in one’s superiority in selecting the correct view and ignoring others.

Capital flows tend to generate much more complex effects and have a broader range of motivations than domestic asset purchases, and thus would seem likely to be less influenced by price dynamics that feed on themselves.Footnote ⁵⁷ However, this is not to say that there are no such effects. As with bubbles, they seldom if ever begin from mood swings. Frequently there will be developments in the recipient countries which justify substantial capital inflows. New governments which undertook financial liberalization policies and adopted sound economic policies attracted large capital flows into East Asia, Mexico, and Argentina in the 1990s, while joining the euro area with the concomitant required policy reforms was the primary cause of the large capital inflows into several Southern European countries. The strong domestic economic conditions likewise induced domestic corporations and financial institutions to borrow more from abroad to obtain greater leverage.

On the push side, loose financial conditions in the advanced economies contribute to the size of such flows and may even be initiating forces, as low interest rates stimulate institutions to stretch for yield and increase investments in assets with higher expected returns and, of course, higher risks. This makes investing in emerging markets more attractive, while the easy financial conditions make more funds available to invest.

While less the result of price dynamics in asset markets, such capital flows are often increased by herding in popular narratives or mental models. This is particularly dangerous when markets tend to converge on a single paradigm. One of the reasons that such flows may become excessive is that, while it is relatively easy to see developments that make investments more attractive, it is generally more difficult to analyze when a country’s capacity has been reached or when its stock has become overvalued.

Stock market booms do not always end in crashes. Likewise, many capital flow surges do not end in disruptive sudden stops and reversals.Footnote ⁵⁸ When they do, it is seldom due primarily to investor mood swings. Often there are developments such as the elections of populist governments or unexpected crises in the recipient countries that lead to major reevaluations of views. While such events in themselves should lead to a decline in capital flows, often these reevaluations of views lead to much larger outflows than if the initial situation had been in equilibrium. While some of these reactions may well be due to investor and borrower panic, often these large outflows may be the result of recognition that a disequilibrium situation had developed where their financial positions are much riskier than originally thought, just as a small spark can set off a crash of an overvalued stock market.Footnote ⁵⁹

6.2 Limits to Arbitrage and International Capital Mobility

Limits to arbitrage are an important condition for making behavioral finance relevant. An essential assumption of efficient market theory is that, unlike in political voting where irrational votes are not offset by rational votes, in financial markets irrational market participants create profit opportunities. Taking advantage of these, rational speculation would offset the effects of the irrational participants, but as discussed earlier there are risks and costs to engaging in arbitrage, and not all rational investors may act on their assessments of the market. Another possibility would be if all participants became irrational, as could occur in the case of the madness of crowds leading to widespread panic.

For international monetary and financial analysis, an obvious implication is that, if behavioral biases are operative, then expectations in the foreign exchange market may display a wider range than just the static expectations assumed in the well-known Mundell–Fleming (MF) model or the rational expectations often assumed in more recent open-economy macro models. A review of the different capital flow assumptions of major types of open economy macro models will be discussed in the following section. A further implication is that international capital mobility will be less than perfect even if there are no formal restrictions on capital flows. This has major implications for a number of important issues, such as the effects of monetary and fiscal policies, and of sterilized intervention in the foreign exchange market.

One key implication that is often overlooked is that excessive exchange rate volatility may result not only from destabilizing speculation but also from a lack of stabilizing speculation. Because of lags in adjustments in real sectors, such as exports and imports, various types of shocks may create differences between short-term and longer-term equilibrium prices in the absence of speculation. In such circumstances, there are profit opportunities to take advantage of expectations of how prices will adjust from the short-term to the long-run equilibrium, although uncertainty may greatly reduce the willingness of speculators to take advantage of these expected price and exchange rate movements. Such speculation would move the price or exchange rate adjustments forward to the longer-run equilibrium level and thus reduce price volatility.

For example, large sudden outflows of capital during the Asian crisis of 1997–98, caused in part by the wake-up call generated by the devaluation of the Thai baht, led to substantial exchange rate depreciation for a number of Asian economies. A considerable portion of these depreciations were afterwards reversed, leading to many charges that these over-depreciations were caused by panic and destabilizing speculation (Willett et al., Reference Willett, Nitithanprapas and Rongala2004). However, this overshooting of the exchange rates is also consistent with an insufficiency of stabilizing speculation resulting in part from the huge amount of uncertainty that had been generated. Trade elasticities are often fairly low in the short run, so the longer-term adjustments of the trade balance to the capital flows would take some time.Footnote ⁶⁰ In the absence of stabilizing speculation, the result is overshooting of exchange rate.

Analytically, the degree of capital mobility refers to the amount of capital that will flow in response to a unit change in incentives.Footnote ⁶¹ For different types of capital flows, the relevant incentives may differ. For example, for highly risk averse investors who wish to avoid the risk of exchange rate fluctuations, the relevant incentive is deviation from covered interest rate parity, that is, the interest rate differential adjusted for the cost of forward cover.

On the other hand, for risk-neutral international investors, the relevant incentives would be determined by deviations from uncovered interest rate parity (UIP), the interest differential adjusted for expectations of changes in the value of the currency over the life of the investment. If changes in interest rates are accompanied by changes in the exchange rate expectations, the incentives for capital flows may not change.Footnote ⁶²

While some empirical studies have investigated the determinants of international capital flows, they tend to focus on the comparative importance of different push and pull factors such as financial conditions in the advanced economies and inflation and policy reforms in developing countries, rather than the degree of capital mobility per se.Footnote ⁶³ Much more testing has focused on the efficiency of the foreign exchange market by looking at whether profit opportunities generated by interest differentials are arbitraged or speculated away.

These studies generally find that between countries with well-developed financial markets and no capital controls, covered interest parity (CIP) generally holds. In general, capital flows coupled with the purchase of forward contracts work to eliminate the risk of exchange rate changes.Footnote ⁶⁴ This is especially true of the euro currency markets. However, contrary to what is sometimes argued, this does not necessarily mean that capital mobility is perfect. Deviations from covered interest rate parity offer clear evidence that capital mobility is imperfect, but the converse does not hold. If there is a great deal of uncertainty and market participants are hesitant to take on uncovered positions, then a fairly small capital flow may be sufficient to close the covered differential, but it would take much larger capital flows for UIP to hold. Thus, CIP is not an appropriate measure of the degree of capital mobility.Footnote ⁶⁵

Tests of uncovered interest differentials give a much better test of exchange market efficiency than those of covered arbitrage. With UIP, any interest rate differentials would be matched by expected equivalent changes in exchange rates. What would be gained by investing in higher interest rate assets abroad would be offset by losses because of depreciation of the exchange rate. While we often do not have good measures of exchange rate expectations, this condition can be tested by whether the interest rate differential is an unbiased predictor of the future spot rate.

There have been many tests of UIP across a wide range of countries and time periods. While some of these studies find that UIP holds for particular currencies and time periods, most do not.Footnote ⁶⁶ As Albagli et al. (Reference Albagli, Ceballos, Claro and Romero2024) write: “While the uncovered interest rate parity (UIP) has long been at the core of international macroeconomics, few relationships have received a starker rejection in empirical work” (p. 1). This indicates that the simple efficient market hypothesis often does not hold in this context. However, the simple version assumes risk neutrality. The more general theory of efficient markets drops this assumption and allows for efficient risk premia. In this case, higher expected returns are due only to higher risks.

There were conflicting interpretations of the findings that simple UIP often did not hold. One school took this as an indication that the market was not efficient, and others that it must mean that there is a time-varying risk premium. It is not easy to clearly distinguish between these two interpretations, but the failure to find that various proxies for risk explain a considerable part of the deviations suggests that many foreign exchange markets, including some for the advanced economies, are not fully efficient. In the absence of capital controls this could be due to nonrational expectations or limited capital mobility, that is, limits to arbitrage, or both.

While tests of UIP focus particularly on flows of liquid shorter-term instruments, there is also evidence of limited capital mobility for other types of capital flows, such as the home bias in portfolio investments. For portfolio investments, many studies have demonstrated substantial home bias. Likewise, international correlations between domestic output and consumption conflict with the theory that full advantage is being taken of the opportunities for international capital flows to improve the intertemporal optimization of consumption.

We still do not fully understand all of the reasons why arbitrage is limited, even in domestic financial markets. The fact that arbitrage can be highly risky at times is certainly one of the most important reasons. Prohibitions on short selling also often limit the extent of speculation against bubbles. Furthermore, access to assets and borrowing power of those willing to bet against the market is likely to be limited.

When it comes to international capital flows, the possible reasons for limitations become even stronger. To many economists, government regulations like capital controls first come to mind as the reason why all international profit opportunities are not exploited, but there are a number of other reasons as well. Desired information about international investments is often limited and costly to obtain. While basic macroeconomic data for most countries is now fairly easily available, these may sometimes give a seriously misleading picture of the true situation. This was true in the run-up to the Asian crisis. The perception that foreign investment is inherently riskier than domestic investments is still widespread and likely has a substantial impact on retail investors. Many still do not understand the full benefits of diversification. This, of course, does not apply to the fund managers, but it does limit the amount of retail funds they have to manage.

In summary, there is considerable evidence that capital mobility, while often quite high, is frequently less than perfect. Markets often behave with neither static nor rational expectations, as is assumed in most popular international monetary and macro models. This has important implications for international monetary and financial analysis, which will be explored in the following section.

7.1 Mundell–Fleming Models

Probably the most well-known international macro model is the traditional MF model. While it is used a good bit less today in advanced economic research, and there are updated versions that make alternative assumptions about wage and price stickiness and expectations, the traditional model is still the workhorse of most international and macroeconomics texts.

The traditional version is based on a simple short-run fixed price Keynesian macro model that is extended to take into account trade balances, exchange rates, and international capital flows. It is particularly famous for its analysis of how the degree of capital mobility and fixed versus flexible exchange rates affect the strength of domestic monetary and fiscal policies. For example, monetary policy is strongest under flexible exchange rates and loses all of its potency under fixed rates with perfect capital mobility. Under fixed rates, fiscal expansions can generate either balance of payments surpluses or deficits, depending on whether the effects on capital flows are greater or less than the effects on the current account. Likewise, under flexible rates, fiscal expansion can lead to either appreciation or depreciation.Footnote ⁶⁸

Other important contributions include its focus on the importance of how fiscal deficits are financed, its analysis of the international transmission of various shocks, and the delineation of optimal monetary, fiscal, and exchange rate polices in response to them. It is important, however, to keep in mind that the MF model focuses only on short-term demand effects and assumes that governments are creditworthy. This short-run focus means that the model does not deal well with economic growth and with the medium as well as longer-term effects of monetary and fiscal policies as wages and prices become more flexible over time.Footnote ⁶⁹ Its neglect of the role of expectations limits its appropriate applications to cases of unanticipated policy changes which are expected to be permanent. In other words, it assumes static expectations. Another crucial limitation is that the model assumes a flow theory of capital flows, which is inconsistent with modern finance theory. As a result, its usefulness is limited to the short run. This will be discussed in the following section on the international monetary trilemma.

The model also has important implications for the degree to which international financial markets provide discipline over domestic macroeconomic policies. In the model, under fixed exchange rates, high capital mobility provides strong discipline over domestic monetary policy. However, contrary to the popular view that financial markets provide healthy discipline over domestic macroeconomic policies in general, it indicates that under fixed rates international capital mobility makes it easier to finance fiscal deficits, since the capital inflows limit the degree to which interest rates rise in response to larger fiscal deficits. This case of anti-discipline fits well the case of Argentina’s currency board, where capital flows provided strong discipline over monetary policy but helped finance the continued financing of large fiscal deficits. The Greek fiscal crisis provides another example.

Besides static expectations, the types of herding discussed in the section on capital flow surges can also help explain cases where international financial markets do not provide discipline. For financial market discipline to work, expectations must be forward looking, reasonably rational, and based on reasonable information. The MF model assumes that the solvency of countries is not at issue, while concerns over future solvency are often a primary factor in the cases where financial markets do provide discipline by providing early warning signals of emerging problems. These different possibilities offer important examples of how different mechanisms of expectations formation can have extremely important implications and why markets may behave differently at different times.Footnote ⁷⁰

Another difficulty with the MF and many other international macro models is that they focus only on net capital flows. For some purposes this is quite important, but for a number of international financial issues a focus on gross capital flows is essential.Footnote ⁷¹ For example, the study of capital flow surges and reversals has shown that there are often important differences in the behavior of gross and net capital flows. Likewise, a major part of the spread of the US subprime crisis to Europe was through the interdependence of two-way gross flows that would not show up in an analysis focusing only on net flows. European banks had invested heavily in US subprime securities but had financed them largely by borrowing from US money market funds. Thus, despite having little net exposure to US financial institutions, they were highly exposed to the US financial crisis.

Despite the many criticisms that the model has drawn, it has proven to be quite useful in explaining many short run developments. However, neglect of expectations limits its appropriate applications, as it assumes static expectations.Footnote ⁷² As emphasized in later literature, in many cases this assumption is seriously misleading.Footnote ⁷³ However, often

policy changes are largely unanticipated, and agents may have diverse expectations, so that policy changes that are anticipated by some may still be unanticipated by others.

Expectations formation is also extremely important for the extent to which international financial markets provide discipline with respect to monetary and fiscal policies. With static expectations, as assumed in the MF model, high capital mobility would help governments finance deficits with lower interest rates and hence help them continue to run excessive deficits. This is not just a theoretical curiosity but has occurred on many occasions, such as the Greek fiscal crisis.Footnote ⁷⁴ It would require fairly good information and forward-looking expectations for markets to recognize these problems and begin to demand higher risk premia which in turn would serve as early warning signals. Markets have been quite variable in the extent to which they have provided such signals.

The real effects of monetary and fiscal policies will be determined not only by the degree of capital mobility and exchange rate regime, as in the original MF model, but also by the extent to which economic agents are forward looking, the degree to which policy actions are anticipated, and the extent of wage and price flexibility. As stressed in the early new classical macro models, with flexible wages and prices, monetary policies will only have real effects if they are unanticipated.Footnote ⁷⁵ However, there is often a good deal of wage price stickiness in the short run. Furthermore, while empirical research shows that there are some regularities in policy reaction functions, their unexplained variance is often quite considerable.Footnote ⁷⁶

7.2. The International Monetary Trilemma

One of the most frequently used concepts in international monetary analysis is the international monetary trilemma.Footnote ⁷⁷ It states that countries cannot have at the same time fixed exchange rates, independent monetary policies, and the absence of capital controls.Footnote ⁷⁸ The trilemma is often justified in terms of the MF model, which shows that with perfect capital mobility and fixed exchange rates, monetary policy is ineffective. Any domestic efforts at undertaking independent monetary policy actions would be fully offset by international capital flows. It has not always been sufficiently recognized, however, that as is discussed in the section on capital mobility, capital controls are not the only reason why capital mobility may be limited. In such cases sterilization of capital flows can allow countries to operate outside of the trilemma constraints in the short run. Thus, the international monetary dilemma provides long-run constraints on countries’ policy options but provides short-run constraints only in the case of very high capital mobility.

The trilemma is an application of the theory of economic policy, which demonstrates that, in general, to achieve a given number of policy objectives one must have at least that number of independent policy instruments. In the international monetary context, it was generally assumed that countries have two major policy objectives: internal and external balance.Footnote ⁷⁹ External balance was defined as the need for balance of payments equilibrium over the medium or long term, while internal balance was defined as the best feasible combination of inflation and unemployment. Three main policy instruments could be used to achieve needed balance of payments adjustments: exchange rate changes, capital controls, and changes in monetary policy. With fixed exchange rates ruling out the use of exchange rate changes, and a prohibition on controls, domestic monetary policy would be the only remaining policy instrument. However, using monetary policy to restore external balance would often conflict with its use to achieve internal balance. For example, if a country had high inflation but a balance of payments surplus, tightening monetary policy to reduce inflation would also increase the balance of payments surplus.

The Bretton Woods system was based on having exchange rates fixed in the short run but adjustable in situations such as just described, where there was a conflict between the uses of domestic policy to achieve internal and external equilibrium. The IMF called such situations ones of fundamental disequilibrium, and recommended exchange rate adjustments.Footnote ⁸⁰ In the case of payments imbalances that were not likely to be persistent, the Bretton Woods philosophy was to keep the exchange rate fixed and use international reserves to finance the payments imbalances. As the MF model shows, such delaying of adjustments would be feasible under fixed rates only if capital mobility was less than perfect. However, in the early days of Bretton Woods, there were still extensive exchange controls that had been put on during WWII, so the assumption of imperfect capital mobility was quite reasonable.

With imperfect capital mobility, the effects of international capital flows on the domestic money supply can be offset by domestic open market operations to sterilize the effects of the capital flows on the domestic money supply, thus giving short-term monetary independence. Such sterilization provides an additional policy instrument. This works, however, only in the short run since it allows payment imbalances to continue and they cannot continue to do so forever.Footnote ⁸¹ The ability to sterilize capital flows in the short run has the benefit of giving countries an extra degree of freedom to operate outside of the trilemma constraints in the short run. If used well, this can be welfare enhancing. However, it also creates the problem, discussed in the section on currency crises, that countries may delay needed adjustments for too long, resulting in currency crises. As we will discuss, behavioral biases tend to worsen this problem.

As the Bretton Woods era progressed, beliefs that capital flows were inherently unstable declined, and the reduction in capital controls combined with improvements in technology, led to much larger capital flows. As a result, the size of capital flows during crises increased substantially and the extent to which countries could successfully postpone adjustments decreased. The resulting frequency of major currency crises increased, and these were a major contributor to the breakdown of the Bretton Woods system of adjustably pegged exchange rates.Footnote ⁸²

The frequency with which adjustable peg exchange rate regimes became associated with currency crises gave rise to the popularity of the bipolar or unstable middle hypothesis that, in conditions of high capital mobility, countries needed to move away from adjustable peg regimes, which offered what was called a one-way speculative option.Footnote ⁸³ Private agents didn’t know whether an exchange rate would be adjusted at any particular moment, but they could be sure of the direction in which it would be moved if it were changed. A deficit country would only devalue its currency, while a surplus country would only revalue it. The higher the international capital mobility, the quicker crises would occur if currencies remained overvalued. The tendency to delay adjustments too long can be explained in substantial part by the combination of time inconsistency problems, wishful thinking, and political economy considerations, and are likely to be more important in situations of considerable uncertainty.Footnote ⁸⁴

The validity of the trilemma analysis has been challenged from several directions. In the 1960s, Mundell argued that fiscal policy could offer an additional policy tool. For example, with tighter monetary policy and easier fiscal policy, the resulting increase in interest rates would attract greater capital inflows to reduce a payments deficit, while the expansionary fiscal policy would offset the contractionary effects of tight money and keep the economy at full employment. With the flow theory of capital flows contained in the MF model, these increased capital inflows would continue indefinitely. However, this would be inconsistent with the portfolio balance considerations emphasized in modern finance theory. Investors would adjust their portfolios to reflect the change in the interest differential, but once they had done this, investors would not have incentives to continue adjustments and generate continuous capital flows.

To generate persistent large capital inflows, continuously increasing interest rates would be required. A given level of monetary tightening would only improve the balance of payments in the short run. Thus, adjustments in the monetary fiscal mix should be viewed as a way of financing a deficit, similar to running down reserves, rather than as a method of balance of payments adjustment. In other words, from the standpoint of the theory of economic policy, monetary and fiscal policy would be independent instruments only in the short run, just as in the case of using international reserves to finance payments imbalances.Footnote ⁸⁵

From the other direction, Rey (Reference Rey2013) has argued that because exchange adjustments are ineffective, there is only a dilemma, with independent monetary policy requiring capital controls. While gaining a good deal of attention, her argument that exchange rate flexibility generally is not effective in giving greater monetary independence has not been widely accepted, although it is likely true for very small open economies, as shown in the theory of optimal currency areas. See, for example, the analysis and references in Willett (Reference Willett, Salvatore and Dean2003).

A key issue is the meaning of the effectiveness of exchange rate flexibility for the independence of monetary policy. It is well known that flexible rates cannot insulate countries completely from foreign shocks. Thus, no economy can be fully independent of what is going on in the world economy. However, exchange rate adjustments can give countries the ability to use monetary policy to respond to both domestic and international shocks in a way that is not possible under fixed exchange rates.

7.3 Monetary Models

The earliest Keynesian models were seriously deficient in terms of giving attention to monetary policy. This was partially corrected in the MF model since it explicitly included monetary policy. From the perspective of monetarists, however, monetary policy in the MF model was still formulated in a deficient way.

The simple monetary models provide a very different perspective from the Keynesian-based MF model.Footnote ⁸⁶ They assume a good deal of wage and price flexibility, and, as a result, purchasing parity holds over the long run. Some models assume it also holds in the short run. While the MF models assume that interest rate increases will be associated with currency appreciation, the monetary models imply just the opposite. Currencies with higher interest rates will have depreciating currencies.

Which model predicts correctly? In truth, both are part right and part wrong. Sometimes the relationship goes one way and sometimes the other. However, this does not reflect a fundamental difference in the economic theory involved. The models vary both in the structures of the economy that underlie them and the nature of the important shocks that are assumed to face them. The MF model assumes that an increase in nominal interest rates is due to an increase in real interest rates, for example, due to an unanticipated tightening of monetary policy and will be associated with currency appreciation.

The monetary models usually make the Fisher (Reference Fisher1930) assumption that the real interest rate is constant, so that any increase in the nominal interest rate is due to an increase in inflationary expectations. In this case, we would expect the currency to depreciate. The Fisher open version assumes that the real exchange rate is constant as well as the real interest rate. Therefore, purchasing power parity (PPP) also holds, as does UIP.Footnote ⁸⁷ In all the monetary models, this relationship holds in the long run, but in some they may not hold in the short run, especially for unanticipated changes in monetary policy.Footnote ⁸⁸

The evidence indicates that in the short run interest rates change in response both to changes in inflationary expectations and to factors affecting real interest rates. Furthermore, because expectations are often hard to pinpoint, at times the causes of changes in interest rates may be subject to considerable uncertainty, giving rise to differing narratives. This offers another illustration that we should not expect any one model to be appropriate for all situations.Footnote ⁸⁹

7.4 Intertemporal Optimization Models of the New Classical Macroeconomics

Much of the effort in both domestic and open economy macroeconomic models in recent decades has gone into developing theories based on explicit microeconomics-based optimizing analysis of rational agents.Footnote ⁹⁰ While older macroeconomic models tended to be based on assumptions which were often plausible from a microeconomic perspective, such as if income goes up people will consume more and if interest rates go up investment will fall, these were generally not based on analysis of fully optimizing behavior. Thus, these newer efforts make important contributions. However, they face the difficulty that to be manageable mathematically, they can focus on only a limited set of considerations. This is true of all models, but there has been an unfortunate tendency by some users of these models to treat considerations outside of the model as being irrelevant. This reminds one of the comment that it is usually better to be roughly right than precisely wrong.

Of course, a main feature of behavioral economics is that people’s behavior is not always rational and they often face considerable uncertainty. While macroeconomics has paid a good deal of attention to factors such as imperfect information and uncertainty, so far there has been much less explicit introduction of behavioral factors into macroeconomic and exchange rate models than into financial economics.Footnote ⁹¹ Still, it can be useful to take into consideration the less rigorous but often pertinent insights from earlier macro models and behavioral considerations. They provide powerful critiques of some of the conclusions drawn by a number of the new classical macroeconomics.

Much of the new classical literature focused on intertemporal optimization of consumption streams. Early Keynesian analysis made the simple assumption that consumption depends on the level of current income. As a starting point, this seems to be quite a reasonable but incomplete first step for short-run analysis. Unfortunately, some of the prominent early Keynesian economists overlooked this limitation and assumed that this simple assumption would work for the long term as well. This led to the very pessimistic conclusion that since the marginal propensity to consume was less than one, over time consumption would not keep up with the growth in income, and an increasing glut of savings would occur.

This potential problem, which was called the secular stagnation hypothesis, had similar predictions to standard Marxist analysis that workers would be kept at subsistence levels so that there would be increasing savings gluts, resulting in cumulative capitalist crises. Alvin Hansen (Reference Hansen1953) of Harvard (Reference Hansen1953) was the most prominent advocate of this secular stagnation view. While he put forward a number of arguments for this conclusion, the implications of the standard Keynesian consumption function were the most influential.Footnote ⁹² Keynesian economics provided a solution to this problem. Increasing savings gluts could be offset by expansionary fiscal policies. Fortunately, most workers’ incomes did not remain at subsistence levels, nor did the ratios of consumption to income tend to increase over time. Indeed, the problem for many countries became too little rather than too much savings.

Modern macroeconomics highlights the crucial importance of factors other than current income that influence consumption, such as wealth or permanent income that shift up the Keynesian consumption functions as income grows over the longer term. The classic reference is Friedman (Reference Friedman1957), which developed the permanent income hypothesis. Since then a number of other expansions of the simple Keynesian consumption function have been developed. (See any major macroeconomics text.) Modern macroeconomics also emphasizes the important distinctions between anticipated and unanticipated shocks, the degree of credibility of government debt, and permanent versus temporary changes in tax policies. Thus, for example, measures that offer temporary income tax cuts are likely to have considerably smaller effects on spending than permanent tax cuts. An example of this is President Johnson’s temporary tax cut during the Vietnam War, which had very limited effects.

An extreme version of such analysis became popular among new classical economists. In this view, changes in taxes have no effect on aggregate demand. One could think of this Ricardian equivalence as an extreme version of such an analysis, where all changes in taxes are considered to be temporary.Footnote ⁹³ The evidence does not generally support this extreme view.

Uncertainty and behavioral considerations help explain this result. Few, if any, individuals are sufficiently farsighted to weigh heavily events that may not take place for many decades, nor can individuals be certain of the structure of taxes they will face in the future. Thus, while the distinction between temporary and permanent tax cuts is important, its relevance is likely to be largely limited to cases where the reversal is expected to occur not too far in the future and where there is less uncertainty about the incidence of future taxes.

7.5 Capital Flows and Currency Crisis Models

While economists have analyzed the causes of currency crises for centuries, in the 1990s a formal literature devoted to currency attacks was developed and proved to be extremely influential among economists and researchers in international monetary and political economy.Footnote ⁹⁴ These models have generally been categorized as falling into three generations. The first-generation models focused on cases of fundamental disequilibrium in which crises were inevitable and only the timing was in question.

Second-generation models expanded the range of the fundamentals from just good or bad to include an intermediate category of fundamentals that were neither so strong that a crisis was unthinkable, nor so bad that one was inevitable. With fundamentals in this intermediate category, a country entered a vulnerable zone with multiple equilibria where crises were possible but not inevitable.

While these first two generations focused on flow disequilibria, the third generation drew on the literature of banking crises, focusing on the financial sector problems of inadequate liquidity and the potential for runs on currencies where liquid liabilities were high relative to countries’ and financial institutions’ reserves.Footnote ⁹⁵ These considerations help to explain crises that hit countries that did not have initial balance of payments difficulties.

In a number of cases where there was confidence in financial sectors, large capital inflows financed current account deficits, such as in Indonesia and Korea in the mid 1990s. However, when wake-up calls led to the discovery of serious problems in the financial sectors, the capital inflows turned into outflows, and currencies which had not been overvalued quickly became so, leading to still further capital outflows.

The initial first-generation model was presented by Paul Krugman (Reference Krugman1979). While highly influential in the academic community because of its attractive theoretical formulation, it was of little practical value. Because it assumed a small open economy with perfect capital mobility, the only warning signs of the looming crisis in the model were large, persistent fiscal deficits and losses of international reserves. The higher interest and inflation rates and monetary expansion that usually also accompany cases of fundamental disequilibrium are absent, and when the peg is forced to be abandoned there is no sharp depreciation, only the beginning of a gradual decline.Footnote ⁹⁶

The second-generation models are considerably more attractive in two respects. First, they consider cases where fundamentals are neither so bad that a crisis is inevitable nor so strong that a crisis is highly improbable. Many countries frequently fall into such zones of vulnerability where, even with fully rational speculation crises are possible but not inevitable. This gives rise to the possibilities of multiple equilibria and self-fulfilling speculation.

The second major improvement is that political economy considerations are introduced. Unlike the first-generation models, it is assumed that governments have the power, but not necessarily the willingness, to undertake domestic adjustments to protect the currency. Thus, perceptions of likely government reactions to possible capital outflows become an important aspect of fundamentals.

While yielding important insights, tractability in modeling required the initial versions of these, and the third-generation models as well, to assume homogenous speculators and high capital mobility. These assumptions imply that crises would occur quickly and that speculative attacks will succeed. In practice, because of uncertainty, differences in expectations, and imperfect capital mobilities, crises often, although not always, extend over a considerable period of time and often do not succeed.Footnote ⁹⁷

Note that self-fulfilling speculation need not be irrational or unjustified, as Krugman (Reference Krugman1996) wrongly argued. Some have gone so far as to incorrectly say that these models showed that even countries with strong fundamentals could face speculative attacks. This might occur, but not within the context of standard second-generation models. The formal crisis models generally assume rational expectations. Many of the claims of destabilizing speculation come from officials practicing blame avoidance. Such claims are also often made by private sector agents to explain why they did not see crises coming, another example of the trait of blame avoidance.

In such models, as countries enter a vulnerable zone, the markets would force up their interest rates to reflect the possibility of crises. In many cases, however, such increases did not occur, challenging the assumption of rational expectations. In these models, countries in the vulnerable zone who were attacked would be unlucky, but not innocent victims. This term should apply only to attacks based on destabilizing speculation or foreign shocks.

A second aspect of these models is that they do not specify what causes shifts in expectations. This is quite understandable from the standpoint of the feasibility of modeling, but what is unfortunate is that while causes of the shifts in expectations are outside of the model, the literal implication is that the shifts in expectations which generate the speculative attacks are purely arbitrary. While this is certainly a possibility, it is seldom the case. In the many crises that I have studied, the spark that sets off the crisis is usually a shock that generates wake-up calls, not arbitrary shifts in the expectations of private sector actors.Footnote ⁹⁸

It should be stressed that in particular crises, more than one of these types of models may give useful insights. For example, large fiscal deficits associated with reunification in Germany, combined with tight monetary policies, led to large capital flows from the other European economies into Germany. These policies generated a first-generation type situation of fundamental disequilibrium in the European monetary system in the 1990s as long as macroeconomic policies remained largely unchanged.Footnote ⁹⁹ However, it was not clear for a good while whether this assumption of no major changes in the policies of the capital exporting and importing counties would be correct. There were substantial negotiations attempting to get Germany to take actions to lower its interest rates, and it did so to a small degree, but not nearly enough to reduce the major disequilibrium. As stressed in the second-generation models, policy reactions were crucial. If Germany had agreed to cut interest rates further, in exchange for substantial tightening in the other countries, the breakdown of the pegged rate system might have been avoided. This is an example of how political economy expectations could have an important effect, as is highlighted in the second-generation models.Footnote ¹⁰⁰

Another example of the applicability of multiple crisis models was the Asian Crisis. The Thai devaluation which set off the crisis was of the first-generation type, an overvalued currency. Countries like Korea did not obviously have this problem. However, many of the East Asian economies were in vulnerable zones because of problems in their financial sectors. As long as these problems were not widely recognized, it was reasonable to consider their fundamentals to be strong, although it was likely not reasonable to not have investigated their financial sectors more closely despite their lack of transparency. However, the Thai devaluations generated a wake-up call that led to the discovery of many second- and third-generation-type problems, especially in the financial sectors of a number of countries, including Thailand itself.Footnote ¹⁰¹

7.6 Contagion

The term “contagion” has been frequently used in different ways. The basic idea is clear. Contagion occurs when developments in one market, such as currency or financial crises or sharp drops in asset prices, lead to similar developments in other markets and often other countries.Footnote ¹⁰² However, contagion may have a number of different causes, transmission mechanisms, and types of effects.

Failure to specify what type of contagion one has in mind has often led to confusion, with evidence of contagion frequently taken as evidence of irrational behavior of the types emphasized in behavioral finance. Thus, for example, the contagion that occurred during the Asia crisis in 1997–98 has often been described as being due to irrational panic, but much of the panic was justified by conditions that had not been previously recognized, such as the severe problems in domestic financial sectors, combined with high levels of debt denominated in foreign currency, and the shattering of the idea that there could be no major depreciations, triggered by the huge depreciation of the Thai baht.Footnote ¹⁰³ This was a classic case of a crisis generating a wake-up call for the private sector to reevaluate their financial positions.

Contagions can also vary considerably in terms of strength, and the failure to distinguish these varying strengths sometimes exaggerates the extent of damage that contagion does to the world economy. For example, the Asia crisis is sometimes discussed as having been felt across the globe. This was true, but the effects on most countries were very mild and of little consequence for policy. On the other hand, a number of East Asian countries were hit with major speculative attacks and capital flight, which had devastating consequences. Between these extremes are episodes which have substantial effects on foreign exchange markets and asset prices but which fall short of major crises. An example is the Russian crisis of 1998 where relatively few countries suffered major speculative attacks, but a number of emerging market countries lost their access to borrow from the global financial markets for a considerable period of time.

A full analysis of a particular episode of contagion needs to specify the events which stimulated the contagion, the types of causes of the contagion, the channels through which it operated, its strength and the factors which influenced its strength. These can vary greatly across episodes. Also important are the policies adopted to try to reduce its effects. For example, there is considerable evidence that the extremely hesitant policies adopted by the euro zone officials to deal with the Greek crisis caused it to spread further and have much stronger negative effects than was necessary. However, although not all-powerful, market reactions seem to have played a role in forcing governments to make policy adjustments more rapidly than they would have otherwise. Likewise, many have argued that the failure to bail out Lehman Brothers caused much greater contagion during the US subprime crisis than was necessary.Footnote ¹⁰⁴

The most common way of testing for international contagion has been to look for whether there are substantial increases in the correlations of the behavior of financial markets across countries.Footnote ¹⁰⁵ This is consistent with the old adage that in crises the only things that go up are correlations. However, such testing does not allow us to tell how much of the contagion was unjustified, nor whether the increases in correlations were due largely to common shocks. Furthermore, the welfare effects of contagion may be affected much more by the size of the total fall of asset prices than by the level of their daily covariance once the crises have hit. The effects of movements in the prices of financial assets also can vary greatly. Thus, the contagions of currency crises are much more harmful if they are also associated with domestic banking and financial crises.

The major channels through which the effects of crises are spread are trade and capital flows and changes in perceptions and expectations. The latter are usually accompanied by changes in international capital flows, but the transmission may go beyond actual flows. Shocks will often change prices directly even before funds flow.

Crises may often generate “wake-up” calls which lead actors to reevaluate their perceptions of preexisting situations in other countries. The Asian crisis was an important example of this. Wake-up calls usually involve looking more closely for possible problems in other institutions and countries. They may also cause reevaluations of the types of narratives or mental models that had been used to evaluate situations. For example, in the cases of both the Mexican and Asian crises in the 1990s, many investors had focused too heavily on the strong domestic macroeconomic fundamentals in these countries and paid too little attention to exchange rate overvaluation and financial sector problems.

When investors or borrowers suddenly discover that they had been using faulty analysis, it does not require irrational panic for them to react initially by pulling back on their investments. There may initially be a fairly general pullback which then becomes more focused as market actors have more time to evaluate the new situation. An example is the Mexican crisis of 1994–95 where there were widespread, though somewhat mild, effects across most of Latin America, but major crises hit only a few countries which had weak economic and financial fundamentals.

Of course, rational explanations for contagion should not be taken as strong evidence that no excessive reactions also occurred, Behavioral finance points to a number of factors that may lead markets to overreact. Panic and tendencies to follow the herd are prominent among them. Such rational explanations, however, do offer a valuable perspective that we should not assume that all of the contagion is irrational and should investigate to what extent it may have been due to fundamental factors.

Contagion based on fundamentals is often called interdependence to distinguish it from possible irrational contagion. For example, if a crisis leads to a recession, the country will import less, directly leading to declines in income for its trading partners. Real effects may also occur indirectly. The severe recessions associated with the Asian crisis led to a substantial fall in the demand for oil, and the resulting fall in prices was a major cause of the substantial fall in Russia’s export proceeds. This was an important contributor to the Russian crisis in 1998. Consequently, the crisis in Asia is sometimes described as leading to rolling crises across the globe.Footnote ¹⁰⁶ This is indeed true, but this spread was due primarily to interdependence rather than irrational contagion.

Kaminsky et al. (Reference Kaminsky, Reinhart and Vegh2003) make an important distinction between mild contagion, which they call “ripple effects,” and major contagions, which they label “fast and furious.” They find that the most important factors putting a country at risk of such contagion are large previous capital inflows, common leveraged lenders, and the initial crisis being largely a surprise. They also emphasize that most crises do not lead to fast and furious contagions, suggesting limits to how strongly the international financial system is subject to irrational panic.

Irrational contagion resulting from panic should occur quickly, so that crises that spread slowly are unlikely to be caused primarily by irrational reactions. This focus helps us realize that the Asian crisis was really two crises. The contagion from Thailand to Southeast Asia was fast and furious, but it was not until months later that major crises hit the Northeast Asian countries of Hong Kong, Korea, and Taiwan. There is little question that the crises in Southeast Asia made the later speculative attacks in Northeast Asia more likely, but it seems highly doubtful that they occurred primarily from unjustified panic.

The evaluation of the importance of irrational speculation can be particularly difficult in the case of the pullback of investments in countries not directly connected to crises. There is likely something to the argument that some investors viewed the East Asian economies as one investment bundle, and thus with the crisis in some Asian countries pulled back their investments in all of them, and of course this would be the case with regional Asian funds.

However, large losses in one market may quite rationally require portfolio reallocations and reductions in investments in other areas. Indeed, some risk management programs call for doing this automatically. This is particularly true when the initial investments were highly leveraged, as often occurs with international investments. Resulting margin calls stimulated by falling asset values may require reducing funds in a number of countries. Somewhat ironically, to minimize losses from forced sales in thin markets, these cutbacks are often concentrated in markets which were the most developed and liquid, giving the impression that irrational speculation was hitting countries with the strongest fundamentals.

In terms of second- and third-generation crisis models, such developments can be the event that tips countries with vulnerable fundamentals from good to bad equilibria, generating even greater capital outflows. And where high leverage is involved, all such investments are in the vulnerable zone, irrespective of the countries’ economic fundamentals.

These considerations strengthen the case for official lenders of last resort. In closed economies, lender of last resort functions are the responsibility of national governments and central banks, but in open economies with large crises, countries may not have sufficient international reserves to keep outflows of capital from creating major currency depreciations that in turn will further worsen the balance sheets of many financial institutions. While not a true international lender of last resort, the IMF is the international institution which most closely serves that function. Emergency support from the IMF is often supplemented by bilateral arrangements, such as Federal Reserve swaps and regional financial arrangements.Footnote ¹⁰⁷

In such cases, the distinction between illiquidity and solvency can be important. Institutions with sound but longer-term investments and with insufficient short-term liquidity could be forced by a crisis to liquidate assets at fire sale prices, which can cause initially solvent enterprises to become insolvent. The concept of solvency is much more difficult to apply operationally to countries than to businesses, but the basic concept still applies. This implies that even countries with strong macroeconomic fundamentals may be in a vulnerable zone if their short-term liquid liabilities substantially exceed their international reserves and easy borrowing power. This is especially true where a high proportion of foreign debt is denominated in foreign currency. This was definitely a problem in the Asian crisis and led these countries to substantially build up their holdings of international reserves after the crisis.Footnote ¹⁰⁸

In summary, contagion may take many forms and is not always primarily associated with overreactions and irrational panic. Often the more serious market failures occurred before the crisis when various actors took on financial positions that were much riskier than they realized. In this process, behavioral factors such as faulty mental models or narratives, herding, hubris, and confirmation bias often played important roles. Often the contagion resulting from crises has a substantial rational component, and it is not unusual for behavioral factors to have been more important in the prelude to crisis than during the crisis itself. In a number of cases, such as Asia in the 1990s and Greece in the 2000s, substantial financial imbalances were generated by excessive capital inflows during good times. In each case, substantial inflows were justified by the fundamentals, but considerations such as overoptimism, insufficient homework, and the herding of narratives led the inflows be excessive, leading to bubbles in financial assets and real estate prices. In these cases, international financial markets acted as facilitators of financial crises, rather than as providers of discipline.

7.7 Implications for the Effectiveness of Official Intervention in the Foreign Exchange Markets

There has been considerable controversy about the effectiveness of official intervention policy in the foreign exchange markets. Most of the debate over this issue has focused on the effects of sterilized intervention. With sterilization, the effects of the changes in international reserves associated with the intervention are not allowed to affect the domestic money supply. If the interventions are not sterilized, then they will directly affect the domestic money supply since international reserves are a part of the monetary base. Such interventions are essentially an open market operation which is carried out in the foreign exchange market instead of the domestic money market. It is generally agreed that such monetary operations will affect the exchange rate. With respect to unsterilized intervention, the policy issue is the degree to which monetary policy should respond to domestic versus external developments.Footnote ¹⁰⁹ Note that an important implication of the MF model is that with perfect capital mobility, effective sterilization is not possible.

The effectiveness of sterilized intervention has two different aspects. One is whether it will actually affect exchange rates. The second is whether, assuming that it can affect the exchange rate, the intervention improves welfare, such as by offsetting the effects of destabilizing or insufficiently stabilizing speculation.

Whether sterilized intervention can affect exchange rates has often been debated in terms of an unqualified yes or no. However, this is not a useful way to look at the question. But as is emphasized in behavioral finance, markets can behave differently in different situations and at different times. A more fruitful framing of the issue asks under what conditions can sterilized intervention affect the exchange rates and when not.

A foreign exchange market may be quite thin in some situations but quite thick in others. The size of capital movements can vary greatly depending on the degree of development of foreign exchange and domestic financial markets in different countries, the breadth and effectiveness of any capital controls, expectations and the amount of uncertainty. Even for the same country, the degree of capital mobility may vary a great deal from one situation to another.

Even for advanced economies with open capital markets, sterilized intervention can have an effect on exchange rates when speculative expectations are not strongly held, so that the arbitrage and speculative schedules are inelastic. This is particularly likely to happen when developments generate a great deal of uncertainty. In such cases rates may be forced away from equilibrium levels, not only by destabilizing speculation but also by a lack of stabilizing speculative funds. In such cases, sterilized intervention has the potential of improving efficiency.Footnote ¹¹⁰

At other times, capital mobility may be quite high and sometimes asymmetrical. Consider the case of a country with an exchange rate that the market views as being substantially overvalued. This is likely to lead to large outflows even for countries where the response of capital flows to changes in interest rates is usually fairly low. This is the classic case of not knowing whether there will be an exchange rate change in the near future, but knowing the direction of change if one does occur, the famous one-way speculative option. In such situations capital mobility may be quite low with respect to inflows responding to increased interest rates in the crisis country. In such cases, the main effect of the initial interventions is to allow some individuals to get their money out at the old exchange rate.

In the aftermath of major depreciations, the market may become thin for a while. Both the difference between short-term and longer-term elasticities of trade flows, and the possibility of initial excessive panic could lead exchange rates to initially over-depreciate, because of the great uncertainty, only a limited number of speculators are willing to take advantage of the resultant potential profit opportunities. This may have occurred with the exchange rate overshooting that developed for some countries during the Asian crisis. This is a case where greater sterilized intervention may have been both feasible and useful.

However, it may often be difficult in real time to know whether or not the market is pushing the exchange rate away from equilibrium. It is doubtful that officials will always have better estimates of equilibrium exchange rates than will private sector agents. And apart from such technical issues, government officials may have exchange rate objectives other than promoting exchange rate equilibrium. Many of the kinds of biases that we have discussed with respect to financial sector actors can also apply to government officials. Such biases may lead officials to assume that any market pressures that move rates away from their objectives must be destabilizing, an example of wishful thinking, allowing normative objectives to influence their positive views. Considerable uncertainty and short time horizons can add to these tendencies since one can be sure of the costs of taking actions now, but the future crisis may not happen, at least not until later.Footnote ¹¹¹ And confirmation bias may lead officials to be slow in giving appropriate weight to any evidence suggesting that their assumptions that there was destabilizing speculation were not correct.

These are not just theoretical possibilities. Thus, sadly, it is not unusual to face situations where we must trade off between imperfect markets and imperfect governments.Footnote ¹¹²