BLOOMBERG'S MARKET WRAP reports indicate that short-term movements in a wide array of fundamental factors—from corporate earnings to the price of oil—underpin participants' trading decisions in equity and other asset markets. Participants interpret the impact of movements of fundamentals on the prospects and returns of companies over the near and longer term, causing their relative prices and thus their access to financial capital to change. As this allocative process unfolds over time, individual stock prices and broad price indexes also tend to undergo swings of unequal magnitude and duration away from and toward estimates of common benchmark levels.
These observations suggest that swings in asset prices and risk arise from the way that financial markets adjust relative prices and allocate capital. And because such swings depend on how market participants interpret fundamentals in forecasting outcomes, they are ultimately driven by short-term movements in fundamentals. But to account for swings in asset prices and risk on the basis of fundamental factors, we must jettison fully predetermined models. After all, profit-seeking participants revise their forecasting strategies in nonroutine ways, owing to news about fundamentals and to psychological considerations, such as confidence and optimism (which depend on fundamental factors but do not move in lockstep with them).
Whereas nonbubble Rational Expectations models ignore psychological considerations and select one fully predetermined fundamental relationship to account for asset prices, the fully predetermined Rational Expectations and behavioral bubble models recognize the importance of psychological considerations for understanding price swings but largely ignore the role of fundamentals. The problem is not only that both fundamental and psychological factors matter for asset-price fluctuations but also that they matter in ways that cannot be fully prespecified with mechanical rules. It is not surprising, therefore, that neither contemporary approach has been able to account for the swings in prices and risk that we actually observe in financial markets.
Imperfect Knowledge Economics eschews fully predetermined models and the presumption that market participants steadfastly rely on one overarching forecasting strategy. As we have argued throughout this book, in coping with ever-imperfect knowledge, profit-seeking market participants revise their forecasting strategies at times and in ways that they themselves, let alone an economist, cannot fully foresee—especially given that institutions, economic policies, and other features of the social context that underpin movements of fundamentals also change in nonroutine ways.
Although movements in fundamentals and revisions of forecasting strategies are nonroutine, there may be qualitative regularities that characterize them. Moreover, these regularities characterize change during the periods of time that no one can fully foresee. Our IKE account of asset-price swings and risk is based on such qualitative and contingent regularities.
One of the regularities that we formalize was emphasized by Keynes (1936, p.152): regardless of whether participants in financial markets are bulls or bears, they tend to assume that the “existing state of affairs will continue indefinitely, except in so far as we have specific reasons to expect a change.” Even when a participant does “have specific reasons to expect a change,” it is entirely unclear what new forecasting strategy, if any, she should adopt. Faced with this uncertainty, participants tend to revise their thinking about how fundamentals matter in “guardedly moderate ways”: there are stretches of time during which they either maintain their strategies or revise them gradually. Such revisions do not generally alter, in substantial ways, the set of fundamentals that participants consider relevant and/or their interpretation of their influence on future outcomes. As we shall see, a price swing arises during stretches of time in which market participants on the whole revise their forecasting strategies gradually and fundamentals trend in unchanging directions.1
But like price swings themselves, the tendency toward guardedly moderate revisions are not only qualitative but also contingent. A participant's decision to revise her forecasting strategy depends on many considerations, including her current strategy's performance, whether she has “specific reasons to expect a change” in how fundamental factors are trending or how they are influencing prices, and the “confidence with which we…forecast” (Keynes, 1936, p. 148).
Moreover, market participants on the whole may not revise their strategies in guardedly moderate ways. There are occasions when news about fundamentals and price movements leads participants to radically revise their forecasting strategies. Such revisions can have a dramatic impact on prices and can spell the end of a price swing in one direction and the start of a new one in the opposite direction. Shifts in how fundamentals are trending can also lead to such reversals in price movements.
By formalizing Keynes's insight as a qualitative and contingent regularity, we can account for the importance of trends in fundamentals for driving price swings, and yet remain open to the nonroutine ways in which fundamentals and psychological factors matter over time. Moreover, because our IKE model fully pre-specifies neither when the trends in fundamentals will change nor when market participants will revise their strategies in guardedly moderate ways, it does not predict sharply when a swing in asset prices may begin or end. This feature of the model is what enables it to account for asset-price swings of irregular duration and magnitude and to remain compatible with the presumption that profit-seeking participants largely forecast in reasonable ways.
In Chapter 8, we discussed how the behavior of short-term speculators can distort relative prices. But we also argued that, because their trading decisions are based on fundamental factors, they help markets to allocate capital across companies and projects. Our IKE account of markets shows how both types of actors—short-term speculators and longer-term value speculators—also underpin the tendency of asset prices to undergo swings of irregular duration and magnitude. When speculators tend to revise their strategies in guardedly moderate ways and short-term movements in fundamentals unfold in unchanging directions, asset prices undergo a swing in one direction or the other.
BULLS, BEARS, AND INDIVIDUAL FORECASTING
Market participants' trading decisions determine whether an asset price rises or falls. The central factor behind these decisions, of course, is an individual's forecast of future prices and risk. If, for example, individuals in the aggregate raise their forecasts of next month's price, they will buy the asset today and bid up its price. To account for asset-price swings, then, we must characterize how participants' forecasting behavior unfolds over time.2
When forming her forecast of the future price and risk of a stock or other asset, a participant must choose which fundamental considerations are relevant and how much weight to place on each, as well as how much weight to place on an asset's current price. As Keynes emphasized, because forecasting “cannot depend on strict mathematical expectation” (Keynes, 1936, pp. 162-63) other considerations, such as one's confidence and intuition, play a role. But these other considerations are also, in part, related to movements in fundamental factors. We can thus portray a participant's forecasting strategy with a simple linear relationship that connects future prices and risk to a set of fundamentals with a corresponding set of weights.
News about the fundamentals that participants think are relevant drives their forecasts. But short-term movements in these factors can lead to changes in a participant's forecast without any change in her forecasting strategy. For example, suppose an individual interprets a fall in interest rates as positive news for a company's prospects over the near or longer term. If such news was reported, and she maintained her assessment of the impact of interest rates, then, other things being equal, she would raise her forecast of the company's stock price.
News can also influence a participant's thinking and intuition about the fundamentals that are relevant and the weights that she attaches to them in forming her forecasts. For example, the uptrend in the inflation rate that began in 2004 was no larger than the downtrend that prevailed between 2001 and 2003. Nonetheless, Bloomberg News reported that the importance of inflation as a main driver of stock prices (measured by the average proportion of days each month that this factor was mentioned in Bloomberg's stories) changed dramatically, rising from below 5% during the earlier period to 45% by 2005 (see Chapter 7). This change suggests that during this period, many market participants revised their thinking about the importance of inflation for forecasting market outcomes, leading to movements in their price forecasts.
Of course, every market participant, regardless of whether she is a short-term or value speculator, formulates a forecasting strategy that reflects her own knowledge and intuition. Some participants look at recent trends in fundamental factors and, drawing on other considerations as well (for example, an understanding of certain historical economic episodes or information about companies' research projects), predict rising prices. Other participants, with different knowledge, intuitions, and thus forecasting strategies, look at the same trends in fundamentals and predict falling prices. Indeed, given imperfect knowledge, bullish views about the future may be no less reasonable than bearish ones (and vice versa). What matters, then, for how asset prices move over time is how fundamentals—and the aggregate of participants' bullish and bearish forecasts based on them—move over the same period.
PERSISTENT TRENDS IN FUNDAMENTALS
We have already seen that corporate earnings can trend in one direction or the other for extended periods (see Figure 7.1). In fact, economists have long observed that many basic fundamental factors that market participants rely on to form their forecasts of asset prices and risk, such as overall economic activity, employment, and interest rates, exhibit this pattern.3 No one can fully foresee how long or steep such trends may be. At any point in time, changes in the economy can cause these trends to reverse, beginning a countertrend in the other direction.
The tendency of many basic fundamental factors to trend in one direction for stretches of time is a key reason that asset markets are prone to price swings. To see this, consider a period during which the trends that participants consider to be relevant for forecasting the price of a particular stock (say, interest rates and company earnings) remain unchanged.4 Suppose each quarter during this period, interest rates and company earnings tend to rise—negative and positive news, respectively, for a stock's price. Some participants attach greater relative weight to higher interest rates, leading them to lower their price forecasts, while others, attaching greater relative weight to higher earnings, raise theirs.
And, of course, no one knows with certainty beforehand how trends in these variables will unfold. Some participants might conclude that a rise in earnings over the most recent quarter implies their subsequent fall and thus attach a negative weight to higher earnings when forecasting prices. For these individuals, higher earnings and interest rates would lead them to lower their price forecasts.
However, regardless of the diversity of participants' views, the unfolding of their price forecasts may, for a time, entail no revisions in their forecasting strategies. During such periods, the impact of developments in fundamentals on individuals' forecasts would not change. If the trends in fundamentals also remained unchanged, these forecasts would tend to move one specific direction during the period. The aggregate of these forecasts, and thus the stock price, would also tend to move in one direction—that is, undergo a swing.
GUARDEDLY MODERATE REVISIONS
Keynes's insight that market participants tend to assume that the existing state of affairs will continue suggests that they tend to stick with a forecasting strategy for stretches of time. Indeed, it is often unclear whether an individual should alter her strategy. A quarter or two of poor forecasting performance may be the result of ephemeral chance events and not an indication of a failing strategy. So unless an individual has specific reasons to expect a change in the market, she may leave her current strategy unaltered, even if its performance begins to flag over several periods.
If a market participant does have reasons to suspect or anticipate a genuine change, she cannot be sure about her beliefs, let alone about the precise nature of the change. When and how she decides to revise her strategy depends on her intuition about how market relationships may be unfolding and on her confidence in this intuition. In such uncertain situations, it is often reasonable for an individual to reinterpret the implications of recent trends in fundamentals or prices for future outcomes in guardedly moderate ways, in the sense that the impact of these revisions on her forecast does not outweigh the influence stemming from the trends themselves.
Continuing with our example, consider an individual who interprets the trends in interest rates and earnings in such a way that would lead her to raise her price forecast over the period without any revisions in her forecasting strategy. If she decided to revise her strategy at some point, the impact of her revisions on her forecast would either reinforce or impede the positive change stemming from the direct effect of trends in fundamentals.
In our example, reinforcing revisions would, by definition, lead an individual to increase her price forecast, as they would tend to reinforce the positive impact of fundamentals. This is not generally the case for impeding revisions in forecasting strategies. These revisions could be quite substantial: a participant may alter her set of relevant fundamentals or her interpretations of them so significantly that she updates her price forecasts in ways that outweigh the impact of trends in fundamentals. Nevertheless, owing to her imperfect knowledge and uncertainty about the future, she is reluctant to revise her strategy so radically. Consequently, we would expect stretches of time during which impeding revisions remain guardedly moderate—smaller—in their impact on a participant's forecast. During such periods, the impact of trends in fundamentals would dominate and lead the individual's price forecast to move on average in one direction.
As noted earlier, we do not assume that such guardedly moderate revisions characterize participants' forecasting strategies at all times. Indeed, that revisions can sometimes be radical plays a key role in our IKE account of price reversals. However, our model's ability to account for persistent swings in asset prices crucially depends on whether guardedly moderate revisions adequately characterize forecasting during periods that are comparable with the irregular duration of price swings.
Direct empirical evidence that would support viewing guardedly moderate revisions as a qualitative regularity that holds during the upswing (or downswing) still needs to be gathered and examined.5 But psychologists have uncovered experimental evidence indicating that when individuals change their forecasts about uncertain outcomes, on average they tend to do so gradually, relative to some baseline.6 This observation is consistent with Keynes's insight and our characterization of guardedly moderate revisions: the impact of participants' revisions on their forecasts is evaluated relative to the baseline change that would have resulted if they had left their strategies unchanged and updated their forecasts solely on the basis of trends in fundamentals. Because it is qualitative, this portrayal of individual decisionmaking is open to myriad possible nonroutine ways in which a market participant might revise her forecasting strategy.7
Our qualitative and contingent characterization of individual decisionmaking is not only open to nonroutine changes in bulls' and bears' forecasting strategies, but is also consistent with the diverse ways in which this change occurs. Regardless of the precise way in which a participant revises her strategy or whether she is a bull or a bear, her price forecast tends to move in an unchanging direction for as long as her revisions are guardedly moderate and trends in fundamentals remain unchanged.
What matters for whether an asset price undergoes a swing is how participants' diverse price forecasts, in the aggregate, change over time. If trends in fundamentals lead the market's forecast to rise, say, during some stretch of time, and participants on the whole revise their strategies in guardedly moderate ways, the asset price will also tend to rise, regardless of whether bulls become more bullish or bears become less bearish.
PRICE SWINGS IN INDIVIDUAL
STOCKS AND THE MARKET
Some of the basic fundamentals on which market participants rely are company specific, such as corporate earnings and sales. Short-term trends in these factors vary across companies and provide clues to their differing prospects and returns over the near and longer term. Other fundamental factors, such as overall economic activity and interest rates, are interpreted more broadly to have an impact on many companies. But trends in these fundamentals also have differing implications for companies' prospects and future returns. Consequently, periods characterized by persistent trends in the basic fundamentals and guardedly moderate revisions of forecasting strategies will not only involve swings in individual stock prices but will also entail changes in relative prices between companies—and thus their relative access to capital.
However, as this allocative process unfolds, the prices of many companies often undergo swings in the same direction, because market participants interpret many of the basic fundamentals' influence on future outcomes in a qualitatively similar way. For example, positive trends in overall economic activity and employment are often viewed bullishly for many companies, whereas rising interest rates are usually interpreted bearishly. Moreover, trends in company-specific fundamentals tend to be related to overall economic activity. For example, the earnings and sales of many companies generally rise and fall along with the economy. These observations, together with persistent trends in economy-wide macrofundamentals and guardedly moderate revisions, imply that broad stock price indexes will tend to undergo swings.
PRICE SWINGS, GENUINE DIVERSITY,
AND RATIONALITY
Beyond accounting for the pattern of asset-price swings in real-world markets, the contingency of IKE models renders them compatible with the coexistence of bulls and bears in asset markets and the rationality of both positions, despite their contradictory predictions of price movements.
Our IKE model explains the presence of both bulls and bears in the market at every point in time by the fact that no participant can predict with certainty when trends in fundamentals may switch directions, or when other participants may cease to revise their forecasting strategies in guardedly moderate ways. Because a price swing may continue or end at any time, betting one way or the other does not involve any obvious irrationality on the part of participants who hold either bullish or bearish views.
SUSTAINED REVERSALS
Even if trends in fundamentals continue in the same broad direction, the qualitative regularity of guardedly moderate revisions in forecasting strategies may cease to hold at moments that neither market participants nor anyone else can fully foresee. Participants may revise their set of relevant fundamentals, or the weights that they attach to them, so drastically that the effect of their updated forecasts outweighs the impact of trends in fundamentals. If a sufficient number of participants revises their forecasts in such radical ways and bets their wealth on such a reversal in their forecasts, stock and other asset prices themselves will also undergo a reversal.
Non-moderate revisions of forecasting strategies are often proximate to points at which trends in fundamentals also reverse or there is a major change in economic policies or institutions.8After all, as Keynes pointed out, recent or impending changes in the way fundamentals are trending provide specific reasons to expect a change, which is why profit-seeking individuals pay a great deal of attention to them.
If there are indications that fundamentals may unfold in new directions, participants will decide on what weights to attach to them in forecasting the future. But once they do, they are likely to resume revising their strategies in guardedly moderate ways. Again, if sufficient wealth is bet on such forecasts, a sustained reversal in the upswing or downswing ensues. The swing in the new direction continues until market participants again lose confidence that the new trends in fundamentals will persist or until they have reason (or intuition) to believe that others are about to revise their strategies in radical ways. At such points in time, the market experiences another reversal, which may or may not turn into a sustained movement of prices in the opposite direction.
We thus expect that the duration and magnitude of asset-price swings will vary in irregular ways, as the stretches of time in which fundamentals trend persistently and revisions are guardedly moderate also vary in similarly irregular ways. Figures 5.1 and 5.2 not only reveal such irregular price swings but also show that instability in asset prices is bounded: price swings sometimes end too late, but they do eventually undergo sustained reversals. We next show how this excess can be explained and develop an IKE model of risk that explains why price instability is bounded.
1Whether a price swing continues or ends also depends on the degree of diversity in how trends in fundamentals affect participants' price forecasts. See Frydman and Goldberg (2010b).
2For a rigorous exposition of our approach to revisions of forecasting strategies and an account of swings in asset prices and risk based on Imperfect Knowledge Economics, see Frydman and Goldberg (2007, 2010b). Using a novel approach to connect theory with empirical analysis, Frydman et al. (2010b) show that an IKE model of currency swings accords significantly better with empirical evidence than a large, widely relied-on class of Rational Expectations models.
3See Juselius (2006) and Johansen et al. (2010) and references therein, who find that the persistence of trends in many macroeconomic variables is much greater than commonly believed. Although discussing why this is so is beyond the scope of this book, we suspect that imperfect knowledge plays a key role.
4For now, we consider a situation in which participants' assessments of risk do not vary. In Chapter 10, we explain how movements in these assessments play a key role in the ultimately self-limiting character of asset-price swings.
5For example, one could survey market participants with qualitative questions about how much they revised their forecasting strategies from one point to another during price swings.
6See Edwards (1968) and Shleifer (2000) and references therein.
7For a rigorous formulation of qualitative and contingent conditions characterizing guardedly moderate revisions, see Frydman and Goldberg (2010b). In sharp contrast, formalizations by behavioral economists of the experimental evidence on so-called “conservatism” presume that market participants never revise their forecasting strategies. Instead, individuals are assumed to underreact to new information in a mechanical way relative to what the economist's overarching probability model would imply. See Barberis et al. (1998).
8For econometric evidence related to this point, see Frydman and Goldberg (2007, chapter 15).