The rational expectations hypothesis was the gold standard in macroeconomics for many years. It assumes that people take a realistic view of the future. In fact, people often attach too much importance to short-term trends, which may directly impact a central bank’s policy.
  |   Philipp Kartaev

In the second half of the 20th century, macroeconomics mainly rested on the powerful and elegant idea of rational expectations. According to this hypothesis, economic agents make forecasts using all available information, not only from the past, but also from analysis of the present situation and possible developments, and generally do not make predictable mistakes. The rational expectations hypothesis does not claim that economic agents are ‘always right’, but rather suggests that mistakes people make are random, and not systematic, in nature.

However, data accumulated over the past several decades increasingly demonstrate that expectations often depart from this ideal in the real economy. Households and firms may systematically overestimate the latest news and over-extrapolate recent events into the future.

It is the attempts to understand these biases that gave rise to the new research area of behavioural macroeconomics. It allows for deviations from rationality due to different reasons.

Over the past two decades, several research areas have developed that seek to expand the standard model of rational expectations.

  • The rational inattention theory argues that information processing is costly. People are unable to take into account all data available, so they distribute attention selectively. Therefore, expectations may be delayed or shifted not because of human irrationality, but because of constraints on information processing.
  • Heterogeneous expectations models assume that different groups of agents make forecasts differently. Rational, adaptive, and other types of expectations may co-exist in an economy, an overall trend is the result of their interaction.
  • Behavioural macroeconomic models attempt to incorporate psychological mechanisms, such as limited attention and overreaction to news, into standard DSGE models, preserving their analytical discipline (DSGE models – dynamic stochastic general-equilibrium models – are a modern standard for modelling monetary policy implications based on the behaviour of economic agents having rational expectations).
  • The diagnostic expectations theory suggests that people systematically overestimate new information, perceiving it as more ‘typical’ of the future than is statistically justified.

The common idea of these areas of research is not to destroy the basic theory, but to make it more data-driven. The diagnostic expectations theory detailed in this article is one of the most influential in academic literature on this topic.

What are diagnostic expectations?

Diagnostic expectations are based on the psychological effect of representativeness, i.e. a cognitive bias, when the likelihood of an event is assessed based on its most pronounced features fitting a certain stereotype (e.g. a person wearing a business suit and carrying a briefcase will be perceived as a manager, rather than a rock musician). This helps promptly classify information, saving cognitive efforts, but may also cause systematic errors and fallacies.

According to this approach, people are often inclined to attach too much weight to new, ‘striking’, information, believing that it is more representative of the future than it actually is. This means that people mistake a short-term effect for a sustainable trend. If inflation accelerates, this is perceived as a ‘new norm’. If shares rise quickly, this seems to be a consistent tendency.

This is not a random error but a systematic distortion of a probabilistic assessment. In mathematical terms, this means that the relative weight of fresh signals is overestimated compared to long-term statistics.

Over recent years, diagnostic expectations were used to explain a whole range of economic phenomena observed. For example, Pedro Bordalo (University of Oxford) and Andrei Shleifer (Harvard University), with co-authors, showed that the forecasts of stock market analysts are systematically shifted by the diagnostic distortion, which results in return predictability, contradicting the efficient market hypothesis. Later on, the researchers generalised their approach to demonstrate that models incorporating such expectations provide a good description of the business cycle asymmetry observed in empirical data. An article by Peter Maxted (Hass School of Business) explains how combining the concept of diagnostic expectations with specific features of the financial market helps better describe the consequences of macroeconomic shocks.

This approach gained popularity as it provides a natural explanation of the empirical data observed. For example, inflation in the US accelerated sharply in the 1970s, driven by changes in oil prices and errors in economic policy. Households’ inflation expectations also surged, remaining elevated for a long time even after the actual triggers disappeared.

Can this be explained without resorting to diagnostic expectations? Of course, there are alternative interpretations:

  • people witnessed steady price growth, logically expecting it to persist;
  • oil shocks did contribute to persistent inflation; and
  • the policy was not tight enough, with expectations adapting to facts.

This means that higher expectations could be a rational response to the dynamics observed.

However, diagnostic expectations offer a simpler explanation of the response amplitude, i.e. expectations increased faster and more notably than suggested by the fundamental parameters because a new inflation spike was perceived as extremely ‘typical’ of the future. That is to say, temporary price growth was regarded as a new persistent trend.

In other words, the diagnostic expectations hypothesis explains oversensitivity to news without necessarily assuming sharp changes in fundamental economic characteristics.

Alternative explanations are also possible. However, the diagnostic interpretation often proves to be more natural and effective in terms of the model’s structure.

Is the diagnostic expectations theory applicable to Russia?

In our recent study, my co-authors Sofya Kolesnik and Timur Magzhanov and I used Russian data to assess the parameters of the New Keynesian DSGE model incorporating diagnostic expectations instead of the familiar rational expectations.

We discovered that the model featuring diagnostic expectations ensured better statistical correspondence to Russian data than the model assuming complete rationality of economic agents. Specifically, it better reproduces inflation dynamics and describes fluctuations of inflation expectations more accurately.

This means that in the Russian economy, expectations may systematically be more sensitive to recent news than assumed by common macroeconomic general equilibrium models. Since expectation dynamics impact decision-making by firms and households, many other variables will change more significantly. For example, in response to a key rate hike, output and inflation drop more sharply and rebound more slowly than in the case of rational expectations.

If expectations do respond to news more strongly than assumed by standard models, a natural question arises: what does this mean for monetary policy?

To find an answer, we used Russian data to assess the optimal monetary policy rule (within the framework of the model) for the cases of diagnostic and rational expectations. We discovered that the optimal value of the monetary policy inertia coefficient was higher for the model incorporating diagnostic expectations than for that featuring rational expectations. Therefore, amid diagnostic expectations, a central bank should opt for a smoother key rate change.

This is because economic agents show more ‘anxiety’ than assumed by rational expectations, which amplifies the effect of inflation expectations channel, making firms and consumers attach too much importance to recent changes.

This behaviour alone may cause even more significant changes in actual inflation. For example, fast monetary policy easing may result in a far more notable inflation rise than predicted by traditional models relying on the rational expectations hypothesis. A smoother key rate change weakens such impact of diagnostic expectations, ensuring lower variability of both inflation and output.

It is worth mentioning that the diagnostic expectations theory does not exclude the assumption of economic agents’ rationality. However, if people are inclined to ‘dramatize’ the latest news, the central bank should probably not only try to convince them of the contrary, but also use special tools to adsorb the ‘anxiety’. In this case, a smoother key rate path will not be evidence of the regulator’s indecision, but will instead help it offset the economy’s oversensitivity, minimising costs.

Philipp Kartaev
Philipp Kartaev
Chair of the Department of Micro- and Macroeconomic Analysis of the Economics Faculty, Lomonosov Moscow State University