optimal forecast
is 40 minutes.
If the only information available to Joe before he leaves for work that would
have a potential effect on his driving time is that he is leaving during the rush hour,
what does rational expectations theory allow you to predict about Joe s expecta-
Adaptive Expectations
FYI
The adaptive expectations hypothesis can be
stated as follows
x
t
e
x
t
1
e
u
(
x
t
x
e
t
1
), 0
u
1
(7)
where
x
t
is the value of
x
at time
t
and
x
t
e
is
the expected value of
x
for period
t
1, with
the expectation held at time
t
. Equation 7
says that the change in the expected value of
x
,
x
e
t
x
e
t
1
, is proportional to the forecast
error, which is the discrepancy between the
current actual and last period s expected
value of
x
.
if expectations realize (i.e.,
x
t
x
e
t
1
),
then
x
e
t
x
e
t
1
if
x
turns out to be surprisingly high (i.e.,
x
t
x
e
t
1
), then
x
e
t
x
e
t
1
if
x
turns out to be surprisingly low (i.e.,
x
t
x
e
t
1
), then
x
e
t
x
e
t
1
Clearly, the formulation in Equation 7
expresses the ability of economic agents to
learn from their past mistakes. This is why it is
also known as the
error-learning hypothesis
.
Equation 7 can be written as
and by continuous back substitution yields
In short, the adaptive expectation hypothe-
sis implies that the expected value of
x
at time
t
,
x
t
e
, is a weighted average of current and past
values of
x
. The weighting scheme can be
thought of as a memory. In fact,
as
u
*
0, the weights decline slowly and the
economic agent is said to have long mem-
ory, in the sense that information from the
distant past influences significantly the for-
mation of expectations for the future.
as
u
*
1, the weights decline quickly and the
economic agent is said to have short mem-
ory, in the sense that only information from
the recent past influences the formation of
expectations for the future.
x
t
e
x
t
(1
)
x
t
1
(1
)
2
x
t
2
. . . .
x
t
e
x
t
(1
)
x
t
1
e
2
John Muth, Rational Expectations and the Theory of Price Movements,
Econometrica
29 (1961):
315 335.
tions of his driving time? Since the best guess of his driving time using all available
information is 40 minutes, Joe s expectation should also be the same. Clearly, an
expectation of 35 minutes would not be rational, because it is not equal to the opti-
mal forecast, the best guess of the driving time.
Suppose that the next day, given the same conditions and the same expecta-
tions, it takes Joe 45 minutes to drive because he hits an abnormally large number
of red lights, and the day after that he hits all the lights right and it takes him only
35 minutes. Do these variations mean that Joe s 40-minute expectation is irrational?
No, an expectation of 40 minutes driving time is still a rational expectation. In
both cases, the forecast is off by 5 minutes, so the expectation has not been per-
fectly accurate. However, the forecast does not have to be perfectly accurate to be
rational
it need only be the
best possible
given the available information; that is,
it has to be correct
on average
, and the 40-minute expectation meets this require-
ment. Since there is bound to be some randomness in Joe s driving time regard-
less of driving conditions, an optimal forecast will never be completely accurate.
The example makes the following important point about rational expectations:
Do'stlaringiz bilan baham: |