UNIT 5 RATIONAL EXPECTATIONS

Structure
5.0 Objective
5.1 Introduction
5.2 Concept of Rational Expectations
5.3 Assumptions of Rational Expectations
5.4 Algebraic Expression of Rational Expectations
5.5 Implications of Rational Expectations Hypothesis
5.6 Limitations of Rational Expectations Hypothesis
5.7 Policy Ineffectiveness Proposition
5.7.1 Lucas Supply Curve
5.7.2 Lucas’ Imperfect Information Model
5.7.3 Assumptions of the Lucas Model
5.8 Let Us Sum Up
5.9 Answer/Hints to Check Your Progress Exercises

5.0 OBJECTIVES
After studying this unit you will be able to:
 explain the concept of rational expectations;
 interpret rational expectations algebraically;
 identify the scope and limitations of rational expectations;
 explain the policy ineffectiveness proposition (PIP) concerning the Lucas
supply curve.

5.1 INTRODUCTION
The rational expectations hypothesis is widely used in macroeconomics.
According to this hypothesis, economic agents use all available information to
make predictions about economic variables. In addition, this hypothesis says that
economic agents, along with available information and their experience, use their
human rationality to predict the future value of an economic variable. They are
well aware of it that predictions may not be correct. However, they learn from
mistakes and improve their predictions for the future. This hypothesis not only
applies to formulate expectations about inflation and income but also explains the
formation of a wide range of economic variables.
Rational expectations hypothesis was proposed by John F. Muth in his seminal
paper, “Rational Expectations and the Theory of Price Movements,” published in
1961 in the journal, Econometrica. In this Unit we will discuss the hypothesis
and its criticisms.


Dr. Tarun Manjhi, Sri Ram College of Commerce, University of Delhi
67
 Expectations, Inflation
and Unemployment
5.2 CONCEPT OF RATIONAL EXPECTATIONS
In the middle of the twentieth century many economists were of the view that
theories based on rational behaviour were inadequate to explain observed
phenomena. The argument of Muth was the exact opposite of this, i.e., existing
economic models did not assume enough rational behaviour. The rationality of
economic thinking can be ensured by introducing the expectations of economic
variables in models used to explain human behaviour.
Given the economic model, expectations are rational if actual values of variables,
on average, are equal to the expected values of variables. For example, suppose
there is a producer with rational expectations and (s)he performs the following
thought experiments: what price should I expect, which is equal to everyone’s
expected price? The producer takes into account various factors for this exercise.
These factors could be the anticipated supplies by others, behaviour of other
producers, inflation, etc. After consideration of all these, (s)he computes the
price that will prevail in future.
Milton Friedman emphasised that economic agents act as if they are maximising
profit/ utility. According to Muth, people do not work with the system of
equations that economists use for maximisation of profit or utility. Further,
individuals do not have similar expectations; they differ in their beliefs.
However, individuals’ expectations should be distributed around the actual value
of the variable to be forecasted. In this sense the anticipations of an average
individual should be the expected value of the variable.
There are two versions of the rational expectations hypothesis: weak and strong.
In the weak version, it is assumed that people have access to limited information;
but they make best use of the information. Let us take a concrete example. You
buy wheat flour (atta) every week for household consumption. You do not know
the relative prices and nutrient levels of all the brands of wheat flour available in
the market. With limited information available to you, however, you usually stick
to the same brand (and may be the same shop, without knowing that other shops
are charging a lower price!). Individuals however vary in their decision-making.
They do not stick to the same brand. Thus there is no systematic error in their
choice. When we take the expected value (that is, the average value) of a
variable, it is usually not different from its actual value.
In the strong version of rational expectations hypothesis, it is assumed that people
have access to all information. Decisions taken are based on all information.
Thus, expected value of a variable is equal to its actual value. Any error in
forecast is due to unexpected developments.

5.3 ASSUMPTIONS OF RATIONAL EXPECTATIONS

To have a logical understanding of rational expectations hypothesis, you should
have clear understanding of probability theory, particularly conditional
probability and the expectation operator (You should have gone through Units 9,
68
10 and 11 of BECC 107: Statistical Methods for Economics). The basic premises Rational
Expectations
under which the rational expectations hypothesis is developed are given below.
(i) Economic agents have full and perfect information to predict the value of a
future event.
(ii) Event/variable should be quantifiable to facilitate data collection and its
analysis.
We know that it is difficult to quantify many variables. Changes in economic
environment are difficult to quantify. We assume that probability distributions of
the events are known. We are in a position to find out at least the first two
moments (mean and standard deviation) of the probability distribution.
(ii) Economic agents (firms, household and government) are assumed to be
rational.
They compare among available alternatives. They have the cognitive ability, time
and resources to evaluate each alternative against the others. Households
maximise utility while firms maximise profits. People are consistent in their
choices.

5.4 ALGEBRAIC EXPRESSION OF RATIONAL

EXPECTATIONS
The rational expectations hypothesis argues that people will use all available
information related to the determination of the expected value of any variable.
Let us assume that there is an economic variable Y and its expected value is
determined by its own lagged values and by lagged values of other variables (X
and Z) in any time period ‘t’.
𝑌 =𝑎 +𝑎 𝑌 +𝑎 𝑋 +𝑎 𝑍 … (5.1)
where X, Y, and Z are variables
𝛼 , 𝛼 , 𝛼 are fixed coefficients
Equation (5.1) is an expression of general hypothesis about formulating
expectations. Here we are neither defining any variable nor saying anything
about the values of the coefficients in this equation. It is only an algebraic
representation of a process.
Let us assume that there is a person who is formulating expectations about the
value of Y in time period ‘t’ at the end of time period t-1. He knows the process
of determining the value of Y is given by equation (5.1). Let us also assume that
the person knows the values of all the lagged values of X, Y, and Z by the end of
time (t-1). If he is rational, his expectations about the value of Y in time period
(t) will be based on his information set at the end of time period t-1. The process
of determining Y will be
𝐸 𝑌 =𝑎 +𝑎 𝑌 +𝑎 𝑋 +𝑎 𝑍 … (5.2)
69
 Expectations, Inflation where 𝐸 𝑌 is the expected value of Y in time period ‘t’ and it is formed on the
and Unemployment basis of the information set available at the end of time period t-1.
Formally, 𝐸 𝑌 is equal to 𝐸(𝑌 /𝐼 ) where E is the mathematical expectation
operator and 𝐼 is the information set available at time period (t-1). The rational
expectation of Y at time (t) is formulated on the basis of available information at
time period (t-1). ‘E’ is rational expectations operator for expectations anticipated
based on information of time period (t-1).
If Y is following the process mentioned in equation (5.1), then the implication of
equation (5.2) is that expectations will be accurate. In other words, forecasting
error will be zero. Forecasting/prediction error is the difference between the
actual and the expected values of a variable.
You should note that the value of prediction error will not always zero, i.e., the
expected value of a variable is not always equal to it true value. It is true when
the economic process of formulating the expected value of a variable is
deterministic. However, in a real sense, most of the economic processes are
stochastic, not deterministic. A stochastic process includes an element of
unpredictability or uncertainty. As you know, economics deals with the
unpredictable/ random behaviour of human beings. This element of
unpredictability in the process of formulating the rational expectations can be
explained by adding a random variable in equation (5.1). That makes this process
more realistic.
𝑌 =𝑎 +𝑎 𝑌 +𝑎 𝑋 +𝑎 𝑍 +𝑣 … (5.3)
Here, 𝑣 is a random variable and its value may be positive or negative. The
variable 𝑣 represents a large number of random variables that affect human
behaviour. Therefore, a smaller value of 𝑣 is better. It implies that probability
distribution of the stochastic variable 𝑣 is concentrated at zero. In other words,
the expected value of 𝑣 is zero.
The assumptions pertaining to 𝑣 are
(i) 𝑣 can be positive or negative.
(ii) It has a constant and finite variance (𝜎 )
(iii) The smaller values of 𝑣 are supposed to occur more frequently
than large values of 𝑣 so probability distribution of 𝑣 is
centred at 0.
(iv) 𝑣 is unknown at the end of tth period. It is also not the part of
Information Set [It-1].
The process of rational expectations of Y in time period (t) is based on the set of
information at the end of time period (t-1). It is formed in (5.3) equation as
𝐸 𝑌 =𝑎 +𝑎 𝑌 +𝑎 𝑋 +𝑎 𝑍 +𝐸 𝑣 … (5.4)
where (𝐸 𝑣 ) is the expectation of (𝑣 ). That is formed based on a set of
information available at the end time period (t-1).

70
The rational expectations hypothesis must assume that the formulated Rational
Expectations
expectations of this period’s value of (𝑣) by a rational individual is made on the
basis of the process determining (𝑌 ), given the set of information for time period
(t-1). Further, assume that the rational expectations of 𝑣 in time period (t) based
on a set of information in time period (t-1) is zero. It can be written as:
𝐸 𝑣 =0 … (5.5)
Now, based on the available information in time period (t-1), the rational
expectations of Y in time period (t) can be written as follows:
𝐸 𝑌 =𝑎 +𝑎 𝑌 +𝑎 𝑋 +𝑎 𝑍 … (5.6)
If the value of Y is determined as per specifications of equation (5.3), then the
value of prediction error is given by following equation:
𝑌 −𝐸 𝑌 = 𝑣 … (5.7)
In stochastic models the prediction error is equal to the actual value of 𝑣 . The
value of 𝑣 is known only after it has occurred. If 𝑣 happens to be large, it
implies that error is large because it is difficult to predict the actual 𝑣 . In rational
expectation hypothesis, there is no systematic pattern of 𝑣 . Thus, forecasting
error also does not show any pattern.

5.5 IMPLICATIONS OF RATIONAL

EXPECTATIONS HYPOTHESIS
It is understood from the above discussion that if the process of determining the
value of rational expectations of the variable Y remains unchanged, there will
some components of randomness in Y which is represented by 𝑣 . The
implications of above are given below.
(a) The mean or average of the error term is zero - Once the random
variable 𝑣 is included in the process of determining Y, the rational
expectations of Y will not be perfectly accurate. The best thing which a
rational forecaster can do is to assume that the value of 𝑣 is zero. It means
that error made by forecaster in each period will be equal to value of 𝑣 in
that period. Sometimes this error will be positive, at other times it will be
negative. In certain rare cases the error will be zero. However, over
several periods, negative errors will cancel out with positive errors. Then
the average of error will be zero.
(b) Errors of rational expectations exhibit no pattern - the hypothesis of
rational expectations rule out any pattern in forecasting errors because of
the assumption that random element itself exhibits no pattern. It cannot be
predicted based on the available information at the time of the forecast.
However, what if the random element (𝑣) shows some pattern? For
example, if the current value of (𝑣) is linked to its previous period’s value
as:
𝑣 = 𝛽𝑣 + 𝜖 … (5.8)

71
 Expectations, Inflation where 𝜖 is a random error with a zero mean and which cannot be
and Unemployment predicted based on any information available at the end of time period (t-
1). The value of parameter 𝛽 varies between –1 and +1.
If (𝑣) is determined according to the process mentioned in equation (5.8),
then the rational economic agents will formulate expectations about the
present value of (𝑣) by following that process. Since the value of (𝑣) in
period (t-1) will be a part of the information at the end of time period (t-
1), the forecast value (𝑣) will deviate from the actual value of (𝑣) by an
unknown, unpredictable element (𝜀 ).
This element (𝜀 ) shows no pattern and its mean value is zero. Therefore,
even if (𝑣) shows a pattern, the rational forecast of Y will still be correct
on an average and forecasting error will show no pattern.
(c) Rational Expectations are the most accurate expectations - the
forecasted value of Y follows the process discussed above and it is based
on available information in time period (t-1). Uncertainty about expected
the value of Y arises because of the presence of random element (𝑣).
Although the mean of the random variable is zero, it can be positive or
negative in any time period. There is the variance (𝜎 ) that tells the value
of (𝑣) that will occur. If the variance (𝜎 ) is very high then the value of
𝑣 will be high and vice versa. Therefore, the variance of 𝑣 is a
measurement of the inherent unpredictability of Y. Higher the value of
the variance higher will be the unpredictability of Y. If the value of
variance is zero then the predictability of Y is perfect. Therefore, the
range of the forecasting errors is in the same range of the unpredictable
component of the process determining Y.
In other words, rational expectations are the most efficient method of
forecasting because the variance of the forecasting errors (due to random
element) will be lower under rational expectations than the use of any
other method for forecasting or formulating expectations.

5.6 LIMITATIONS OF RATIONAL EXPECTATIONS

HYPOTHESIS
Criticism of the rational expectations hypothesis can be understood through the
following points:

(a) The logic of rationality - This rational expectations hypothesis assumes

that an individual is a rational human being. However, in reality, is it
plausible to assume that a typical individual is sensible enough to use all
available information to forecast the value of any variable? Is it not that
the people are more often ignorant about the economic phenomenon?
For example, how many people would be able to give a reasonably
precise definition of money supply? So, if an individual is forming
rational expectations about the rate of inflation, he is expected to know
what the money supply is and how it is growing. In a normal situation,
72
 rationality in economic theory implies that a person compares the cost and Rational
Expectations
benefit of any activity and carries out that activity up to the point where
marginal cost is equal to the marginal benefits. For example, a firm will
produce up to the point where marginal revenue from producing and
selling an additional unit of a good is equal to the marginal cost. If we
apply to the hypothesis of rational expectations, the individual
(forecaster) will compare the marginal cost of acquiring information
about the process of determining a variable and the marginal benefits of
making more accurate forecasts. However, the point at which marginal
cost and marginal benefits are equal does not necessarily correspond to
the point at which the forecasting error is equal to the purely random
component of the determining process.
(b) The availability of information - The rational expectations hypothesis
assumes that the process of determining Y is known and the values of
other variables in that process are known at the end of time period (t-1).
But what if the value of any of these variables used in the determination
of Y is not known at the end of time period (t-1)? How will a rational
forecaster determine the value of Y in time period ‘t’? While there may be
mathematical solutions to the above questions, lack of knowledge about
information will adversely affect the accuracy of rational expectations.
(c) Unrealistic elements - It is unrealistic to say that the expectations of
every individual are precisely the same, as every individual cannot track
the data of every variable necessary for predictions. Information
collection and processing is a costly affair which is not possible for all
individuals. If an individual is not using all relevant information for
predication then there is a higher chance of formulating wrong rational
expectations.
(d) Flexible price and market clearing mechanism- Rational expectations
hypothesis assumes that price is flexible and there is continuity in market
clearance. It is not true because of the prevalence of stickiness in prices
and wages.
Check Your Progress 1
1. Briefly discuss the concept of rational expectations.
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………….........…...…

73
 Expectations, Inflation 2. Discuss the role of a random variable 𝑣 in rational expectations.
and Unemployment ……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………….........…...…
3. Briefly discuss the implications of rational expectations hypothesis.
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………….........…...…
4. Briefly discuss the limitations of rational expectations hypothesis.
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………….........…...…

5.7 POLICY INEFFECTIVENESS PROPOSITION

As per the rational expectations hypothesis, there is no impact of predictable
macroeconomic policy on output and employment. Unpredictable policy actions
have some impact on the real GDP and employment level. The rational
expectations is a cause that negates the change in the AS and AD. There is
increase in price level but output remains constant.
If people forecast changes in policy correctly, then there will be an increase in the
price level. With an increase in prices, there will be a demand for higher wages
and it will lead to a decrease in AS.
If people forecast changes in policy incorrectly, then they cannot forecast change
in prices caused by policy. Consequently, there will be an increase in output and
increase in aggregate supply with the increase in prices.
5.7.1 Lucas Supply Curve
Robert Lucas was the first economist who highlighted the importance of public
expectations in macroeconomic policymaking and forecasting. According to
Lucas, the anticipations of economic agents (public, firms, and government) are
more important than that of the policymakers. Hence, policymakers have to know
how the economy works and to understand the expectations of economic agents
(households, firms, and government).

74
In the early 1970s, Robert E. Lucas developed an alternative model of Phillips’ Rational
Expectations
Curve by assuming the rational expectations hypothesis. Lucas showed that a
positive relationship between output and inflation can arise because of imperfect
information about the price level in the economy. The Lucas Model says that
with rational expectations, only unanticipated changes in the money supply make
an impact on real output. On the other hand, all anticipated changes in money
supply only affect the price level. It is known as the Policy Ineffectiveness
Proposition.
5.7.2 Lucas’ Imperfect Information Model
As per the rational expectations hypothesis, subjective expectations are made
according to the following:
𝑋 = Ε[(𝑋 |𝐼 )]
𝑋 = Ε[(𝑋 |𝐼 )], i.e., the expectations of a variable ‘x’ in this period (t) is a
conditional mathematical expectation on all the information available till this
period (t). In the rational expectations hypothesis, the value of mean forecast
error (MFE) ≡ 0.
There are three characterisations of rational expectations:
(i) Mean forecast error (MFE) = 0
(ii) There is no systematic pattern in the forecast error
(iii) It is the most accurate forecast; since by definition, we are using all the
information and we cannot have a better forecast.
Therefore, the critical assumptions underlying the Policy Ineffectiveness
Proposition are:
(i) Prices and wages are perfectly flexible (perfect competition set up)
(ii) Expectations are rational
If prices are sticky, anticipated changes in the money supply will affect real
output, even under rational expectations.
5.7.3 Assumptions of Lucas Model
The main idea behind the Lucas Model is as given below.
(i) When a firm observes a change in the price of the product, the firm does
not know whether this change in the price of the product is caused by the
change in the aggregate price level or a change in the product’s relative
price level. Any change in the relative price will change the optimal
quantity of the good that the firm produces. This is the Signal Extraction
Problem.
(ii) Since prices and wages are assumed to be flexible in this model, a firm
produces according to the rule: P = MC (perfectly competitive set up).
(iii) The firm produces more only if there is a rise in the price of the good it
produces relative to the prices of other goods. When the aggregate price
level (say, CPI) rises, not necessarily the relative price, the firm’s output
75
 Expectations, Inflation will not rise. Firms are assumed to have imperfect information on the
and Unemployment prices. These imperfections are due to informational barriers. Firms tend
to confuse overall price movements with the relative price movements,
which lead to them to deviate from their optimal production, in the short
run.
(iv) Lucas also assumes that people make decisions according to the rational
expectations.
The Imperfect Information Model of Lucas has the following three
structural/behavioural equations:
1. AS equation:
𝑌 = 𝑌 + 𝛽 (𝑃 𝑃 ); 𝛽 > 0 … (5.9)
The AS equation implies that the output in this period is the sum of the full
employment output.
2. AD equation:
𝑀 + 𝑉 = 𝑃 + 𝑌 (taking log of QTM equation) … (5.10)
The AD equation is the usual Quantity Theory of Money (where the velocity of
money, V, is assumed to be constant).
3. Monetary Feedback Mechanism:
𝑀 = 𝛼(𝑌 ) + 𝜀 ; 𝜀 ∼ Ν(0, 𝜎 ); where 𝛼 < 0 … (5.11)
If output is more than the full employment level of output, the use of
expansionary monetary policy and hence, the policy parameter is negative. Here,
money supply is some function of the actual output in the previous period (t-1)
plus some stochastic error (𝜀 ) component which follows a normal distribution
(with mean zero and constant variance).
In effect, 𝛼 or the policy parameter is the anticipated part of the money supply
(since it depends on the actual output of the previous period 𝑌 ). While 𝜀 is
the unanticipated part of the money supply (say, due to unforeseen situations
such as oil price shock, war with neighbours, drought, etc.), on average, the
stochastic error term is zero.
We will prove that the policy is ineffective, i.e., anticipated changes in policy
have no effect on real variables and only unanticipated changes can affect real
output.
Mathematical Derivation of the Model:
𝑃 = (𝑎𝑌 + 𝑒 ) + 𝑉 − 𝑌 … (5.12)
We assume rational expectations, i.e., 𝑃 = Ε[ 𝑃 | 𝐼 ] The expected prices are
prices conditioned upon all the information available till the point the
expectations are made.
This means: 𝑃 = Ε[(𝛼𝑌 +𝜀 )+𝑉−𝑌|𝐼 ]

76
𝑖. 𝑒. , 𝑃 = Ε[𝛼𝑌 |𝐼 ] + Ε[𝜀 |𝐼 ] + Ε[𝑉 | 𝐼 ] − Ε⌈𝑌 |𝐼 ⌉ Rational
Expectations
(as the expectation is a linear operator).
𝑖. 𝑒. , 𝑃 = 𝛼𝑌 + +𝑉 − 𝑌 (𝑡ℎ𝑖𝑠 𝑓𝑜𝑙𝑙𝑜𝑤𝑠 𝑏𝑦 𝑑𝑒𝑓𝑖𝑛𝑖𝑡𝑖𝑜𝑛)
𝑖. 𝑒. 𝑃 = 𝛼𝑌 + 𝑉 − 𝑌 (𝑢𝑛𝑑𝑒𝑟 𝑅𝐸) …(5.13)
Summarising, we have: 𝑃 = (𝛼𝑌 +𝜀 )+𝑉−𝑌
and, 𝑃 = 𝛼𝑌 +𝑉−𝑌
that is, 𝑃 − 𝑃 = 𝑌 − 𝑌 + 𝜀 … (5.14)
Substituting (5.14) in (5.9):
𝑌 = 𝑌 + 𝛽(𝑌 − 𝑌 + 𝜀 )
that is, (𝑌 𝑌)(1 + 𝛽 ) = 𝛽(𝜀 )
𝑌 = [𝛽(𝜀 )/(1 + 𝛽)] + 𝑌 … (5.15)
In equation (5.15), we see that the policy parameter (α) does not appear in 𝑌 . It
implies that the anticipated part of monetary policy is ineffective. Further, the
term α does not affect output. Only unanticipated part of the money supply (εt)
will have an impact on output in this model. This is the Policy Ineffectiveness
Proposition.
Also, it can be proved that in this model, the Mean Forecast Error (MFE) for
Output (𝑌 ) and Prices (𝑃 ) = 0. Hence, this model is consistent with rational
expectations hypothesis.
According to Lucas (Signal Extraction Problem), all unemployment is voluntary
because the workers speculate about leisure, over time. They work more in the
present if wage rate is higher, with a belief that they will enjoy leisure when
wage rate low. There is uncertainty in such speculation, as there is imperfect
information. Temporary changes induce certain actions and we attach a
probability to the change being temporary.
Check Your Progress 2
1. Explain the underlying idea behind policy ineffectiveness proposition.
…………………………………………………………………………...…
…………………………………………………………………………...…
…………………………………………………………………………...…
…………………………………………………………………………...…
2. Point out the factors that affect output as per Lucas’s understanding of
supply.
…………………………………………………………………………...…
………………………………………………………………………..….…
…………………………………………………………………………...…
…………………………………………………………………………...…
77
 Expectations, Inflation
and Unemployment
5.8 LET US SUM UP
Rational expectations hypothesis is an improvement over the adaptive
expectations hypothesis. It says that people use all available information for
formulating predictions about economic variables. It also says that people use
their human rationality, available information and their experience to predict the
future value of any economic variable. People know that predictions may not be
correct always. However, they learn from mistakes and improve upon their
predictions for the future. This hypothesis is widely used in macroeconomics. It
not only applies to formulate expectations about inflation and income but also
explains the formation of a wide range of economic variables.
Lucas pointed out that under the rational expectations hypothesis there is no
impact of anticipated macroeconomic policy on output and employment.
Unanticipated policy actions, however, have some impact on output and
employment.

5.9 ANSWERS/HINTS TO CHECK YOUR

PROGRESS EXERCISES
Check Your Progress 1
1. Go through Section 5.2. Provide the intuitive idea behind rational
expectations hypothesis.
2. Go through Section 5.5. You should write the implications of the
stochastic error term.
3. Go through Section 5.5 and answer.
4. Refer to Section 5.6 and answer.
Check Your Progress 2
1. Refer to Section 5.7 and answer.
2. Refer to equations (5.12), (5.13) and (5.14).

78