The Most Important Points of Criticism for Both Adaptive and Rational Expectations

One of the most relevant questions in the economic theory is about the capability of the government to affect the real state of the economy and by what means this can be done. The main debate takes place between views that are opposite in their nature: laissez-faire and the necessity of authorities to intervene. The question is whether monetary authorities are able to manage the economy and what are the best ways to do this, whether there is a tradeoff between economic variables, e.g. between rates of inflation and real growth together with employment as proposed by Phillips curve. The controversy of the problem which economic policy should be adopted was even intensified as the theory of expectations evolved. Although the notion of expectations applied in economic theories is quite broad and not new, in my essay I will concentrate on two main hypotheses of expectations, namely on adaptive and rational expectations. After giving an overview of the evolvement of expectations in economic thought, illustrating the essence of adaptive and rational expectations, I will try to find the explanation for conundrum why after completely substituting adaptive with rational expectations the economics profession turned again to former after some time. I will also reflect on the issue and express my own view on the question which type of expectations is more relevant depending on certain conditions.

In modern economics, expectations have taken a central place. Their importance can be explained by the fact that economic and econometric models heavily depend on the assumptions that they rely on. Moreover, expectations regarding future are one of the most significant factors that influence the decisions and behavior of economic agents. Broadly speaking, if certain expectations prevail in society, this will affect the way in which way regulative actions of monetary authorities will influence the economy. Thus the outcomes of introduced policies to a large extent depend on this factor. As policymakers try to choose what policy to adopt, they rely on forecasts proposed by models.

The theory of expectations tries to explain in what way economic agents form their anticipations about future. The utilization of the expectations in explaining economic phenomena is not new, although the peak accounts for modern economics. The first time expectations were used in economic theory by Emile Cheysson in 1887. A further contribution to theory was made by Alfred Marshall as he introduced the concept of short- and long-run to classical economics and static expectations hypothesis. Mordecai Ezekiel was the first who deeply analyzed the influence of expectations on the stability of economic equilibrium (Ezekiel 1938). Expectations were more frequently used in the 1930s as a relevant tool for constructing macroeconomic models, e.g. Fisher hypothesis that explains inflation rate as the difference between nominal and real interest rates. Another economist of that time, Gunnar Myrdal, studied the role of expectations in business cycles. However, the biggest influence on the theory of expectations of that time had John Maynard Keynes and especially his work The General Theory of Employment, Interest, and Money (1936). By distinguishing between short- and long-term expectations, he emphasized the importance of the latter regarding prospective investment returns and asset prices as the main source of volatility in the economy. Although Keynes assigned a central role to expectations in predetermining the level of output and employment, he did not provide a coherent theory of how agents’ anticipations are formed. First models of expectations date back to 1940s. In 1941 Lloyd Appleton Metzler constructed macroeconomic models of inventory cycles that included expectations. In the 1950s and 1960s expectations were commonly used in macroeconomics regarding consumption, investment, inflation, and employment.

The main period of interest in the economic history for this essay starts with the wide exploitation of adaptive expectations. In his famous book, A Theory of Consumption Function (1957) Friedman asserts that consumer spending depends on the long-term expected income rather than on current income. This theory explains the decision-making process of agents in the consumption-saving problem and is also known as a permanent-income hypothesis. In 1968 Friedman and Edmund Phelps independently came to the conclusion that expectations of inflation affect current rate of inflation. While analyzing short-run and long-run Phillips curve, Milton Friedman came up with the natural rate hypothesis. According to it, inflation is already embedded in expectations and therefore in order to avoid accelerating inflation over time unemployment rate must be high enough so that actual inflation equals expected inflation. The attempts of monetary authorities to peg unemployment below its natural rate will lead to ever rising inflation. The economics profession adopted the opinion that expected the rate of inflation is the most important factor affecting actual inflation, more important than for example unemployment level.

In his reasoning, Friedman employed adaptive expectations concept. This promoted adaptive expectations hypothesis which became mainstream in the economics of that time. The main idea of this hypothesis is that economic agents form their expectation of the future value of some economic variable (e.g. inflation) basing solely on its past values. There are several forms in which adaptive expectations hypothesis might be formulated. The most popular formulation which gives better-fitted equations is the assumption that expected changes are equal to an average of past changes.

For example, adaptive price expectation means that agent revises his expectation of future price taking into account difference between his former expectations of current price and the actual current price. To give an example, according to this theory if during several previous years the rate of inflation was 2% and this year monetary authorities adopt expansionary policy and inflation rate increases to 4%, this creates brief gap between reality and perception as in the short run people expect inflation to be 2% based on their previous experience. Aggregate demand will temporarily increase which in turn will increase GDP level. According to adaptive expectations, all this is possible because the increase in inflation was unexpected and therefore there is a tradeoff between inflation and output level in the short run. However, after people realize what happened, they will demand higher wages, production costs will increase and output level will return to its previous potential level at higher prices and unemployment to its natural level. Therefore, it is possible to delude agents temporarily, as they look at the past values of the variable and then they try to adapt if there is a mistake in their expectations of that values. Agents with adaptive expectations cannot react immediately to the current events and have to wait until they observe their mistake in order to adjust their expectations. They are just passive participants who do not expect future changes in the economy.

This was one of the main points of criticism of this theory. Rational agents should be able to adapt their expectations and hence their decisions and behavior not only basing on past events but also by observing current changes. Individuals do not form their expectations only by looking over their shoulder at past values, but by also taking an active part in the economy, by monitoring current events and announcements and building anticipations also on their basis. For instance, just by knowing what policy government is going to introduce (e.g. it is a common knowledge that expansionary policy will lead to higher level of inflation) rational individuals can update their expectations for future. One more drawback of adaptive expectations is that according to this hypothesis agents commit systematic errors. This is completely at odds with the concept of rationality.

Some authors even argue that adaptive formalization of expectations contradicts the very purpose of building a theory of expectations because according to this attitude what influences the future is affected by history only, not by expectations; forward-looking attitude of agents is completely lost (Gertchev 2007).

The understanding and the role of expectations evolved over time. Monetarist theory gave rise to a new classical school of macroeconomic thought in the 1970s. New classical economists disagreed with Friedman and basing on weak points of adaptive expectations hypothesis they elaborated on the rational expectations concept.

In 1973 the oil crisis occurred and US economy experienced stagflation. This was a trial for existing theories and ability to exist economic approaches to make predictions. To a great surprise of proponents of monetarist theories, these methods failed. In response to this, Lucas introduced rational expectation hypothesis basing on empirical research of Jan Tinbergen and on theoretic elaborations of John Muth (1961). He announced that existing economic models could not predict the crisis because they were based on misleading and unrealistic assumptions of adaptive expectations. Lucas argued that rational agents are active participants who are able to anticipate and adjust their anticipations in accordance with changes in the real economy. They do not react passively to actions of government post factum, but in turn, try to predict them. One more important elaboration of this theory contrary to adaptive expectations is that agents do not make systematic mistakes while forming their expectations.

Following the previous example, if monetary authorities announce that they are up to introduce expansionary policy, individuals who act accordingly to rational expectations hypothesis can figure out that this means higher level of inflation in the next period and therefore they will adjust their expectations without waiting for inflation to actually increase, they will anticipate this in advance. Consequently, if authorities are to increase the money supply, there will be no tradeoff between inflation and output at all even temporarily, aggregate demand will not increase and the economy will immediately end up with the same level of GDP at higher prices.

Both adaptive and rational expectations hypotheses, despite their differences, are still quite similar in this respect and lead to the same overall conclusions regarding what kind of policies government should pursue. According to both theories government should not intervene in the economy by enacting expansionary policy as it will only lead to higher prices in the long run.

Both theories are similar and still, they are ultimately different in their essence. Still, there is a wide and ongoing debate which hypothesis is more realistic and should be utilized in economic models. This discussion is also sparked by the importance of underlying assumptions for the final results and predictions provided by economic models. And these, in turn, are widely used by policymakers to predict what impact this or that movement of monetary authorities will have on the real economy and at what magnitude. Based on the above-mentioned considerations on the process of the evolvement of these theories and ideas behind them, it seems reasonable that rational expectations hypothesis is more advanced and realistic compared to advanced expectations. In the light of current technological advances, this seems even more plausible. Information becomes more and more easily accessible; speed of information dissemination rises, informational space transforms in such a way that information available to agents converges to perfect information concept. However, as time elapsed after the adoption of rational expectations as a better alternative, they were heavily criticized. The main conundrum is why after some time economics profession started to deny rational expectations that were designed to eliminate erroneous assumptions that were commonly used before.

There are two versions of rational expectations: “weak” and “strong”. “Strong” version assumes that individuals have access to all information and meet only rational decisions basing on the whole scope of available knowledge. Any mistakes might occur only due to unpredicted events. The subjective expectations are almost identical to the objective. It is like if people had “correct model” in their heads that give unbiased predictions. “Weak” version assumes that economic agents have limited scope of information based on which they form their expectations and make decisions. This is more realistic since very often people use the rule of thumb to take some routine activities like buying groceries.

The majority of criticism was targeted at “strong” version. Rational expectations were mostly attacked for the ambiguity concerning the way in which individuals receive information that allows them to act unmistakably as “strong” version assumes. To have this become reality requires static world with typical transactions and predictable actions of other market participants on the basis of perfect information (Garbicz 2008). However, the real world is very dynamic and obtaining information is costly. This raises a question whether agents can make correct predictions, all in the same manner. All individuals differ in their background, personal characteristics, circumstances that they find themselves in and access to information. Therefore, the formations of their expectations differ as well. Another criticism concerns the fact that rational expectations hypothesis does not take into account costs of acquiring information used to form expectations (Mucha 2009). In addition to the fact that all agents cannot be equally-well informed in principle as discussed above, it is also necessary to keep in mind that although the way information can be accessed was simplified by virtue of technological advances, it still requires some costs. As already mentioned, in reality, people do not remind perfectly behaved homo economicus and on contrary tend to simplify decision-making process regarding routine tasks.

Moreover, rational expectations assume not only that all individuals share the same information, but also the same capability to make use of it, that being the fundamental deficiency of this theory (Gomes 1982). Wible (1982) expressed similar critique that all agents are assumed to be experts in economics and able to use available information in the most proficient way.

In this essay, I highlighted the most important points of criticism for both adaptive and rational expectations. In fact, there is much more to mention. The debate itself what assumptions should be adopted and which hypothesis – of adaptive or of rational expectations – reflects reality better lasts to the present day. The outcomes of the discussion differ. Basing on imperfections of both hypotheses new theories and approaches were developed. Among these is, for example, learning behavior models with explicit theories concerning information collection and formation of expectations (Evans and Honkapohja 2001). Economics professional who first of all disagree with assumptions of rational expectations that people make their forecasts and decisions basing on complete and perfect information and have utility maximization as fundamental objective deny rationality paradigm by emphasizing the limited cognitive capabilities of human beings. They are inclined to base economic assumptions on psychological peculiarities and limitations of human nature, tend to imperfect knowledge economics introduced by Frydman and behavioral economics. They propose bounded rationality assumption as an alternative to previous ones (Mikolajek-Gocejna 2014). Another group of economists argues that agent make decisions and form their expectations based on emotions (Loewenstein et al. 2001) and introduced risk-as-feelings hypothesis model, which unlike others provides an explanation for economic bubbles.

In contrast, some economists defend adaptive expectations and assert that despite their simplicity and seeming remoteness from reality they perform decently in economic models. Mills (1961) argues that stability model with adaptive expectations will produce an expected time path similar to real values in its dynamic characteristics. Figlewski and Wachtel (1981) by constructing regression and applying it to data come to the conclusion that adaptive expectations explain inflationary expectations in a better way. Gregory Chow (2011) argues that rejection of the adaptive expectations hypothesis in favor of rational expectations lacked sufficient scientific reason and empirical basis. In his study, he explicates that the adaptive expectations hypothesis is supported by theoretical statistical reason and empirical econometric evidence. His reasoning boils down to the fact that people put more weight on more recent historical data while estimating the future value of the specific economic variable. Chow also conducts an econometric study in order to support his arguments, which shows the consistency of the model based on adaptive expectations with real data. He also refers to his earlier researches (Chow 1989 and 2007, cited in Chow 2011) that also provide strong empirical evidence on the coherence of adaptive expectations with reality.

In my opinion, adaptive expectations were substituted with rational expectations too quickly. That is true that latter has some advantages in comparison to former and is more elaborated. Indeed, it is necessary to account for the ability of agents to analyze the current environment and adjust their behavior accordingly, not only passively wait until their fallacy will be supported by real evidence in order to account for their mistakes and adjust their expectations until new mistakes are evident – like robots without analyzing and learning abilities. However, one needs complete homogeneity of individuals in order for these elaborations to correspond with the reality. In fact, the overwhelming majority of individuals has no economic background and cannot make use of available information even if they encounter it since an average person does not trace economic news. For this part of the population, adaptive expectations explain behavior in the best way. At the same time, there are professionals who actively try to make use of all information that they can get, e.g. investors, labor unions, banks etc. For these economic agents, adaptive expectations cannot be used and the rational expectation hypothesis corresponds to their process of decision-making better. Therefore, I believe that the whole dispute which theory is more relevant misses the main point, namely: individuals are ultimately heterogeneous. Both approaches have right to exist, each suiting better specific group of agents. Regarding more advanced assumptions and theories like learning behavior, bounded rationality, risk-as-feelings hypotheses and others, I believe that these are too difficult to employ. Economic models and assumptions that they are based on should stay as simple as possible as long as they bring satisfactory results. This is what adaptive and rational expectations are about. In my view, the best solution would be not to substitute one with another, but find a way to utilize them together, in proportion and in the balance to account for underlying reality.