Scribd
Upload
502 views

Random Walk Model

Robert Hall developed the random walk hypothesis for consumption based on Friedman's permanent income hypothesis and rational expectations theory. The hypothesis states that changes in consumption are unpredictable as consumers use all available information to rationally assess their lifetime income and only unexpected events affect consumption. Hall (1978) empirically tests this hypothesis using postwar US data and finds that (1) consumption follows a random walk pattern and cannot be predicted by past consumption values, but (2) some recent past disposable income values have a small impact on current consumption, providing mixed support for the random walk hypothesis.

Uploaded by

Taruna Bajaj
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
502 views

Random Walk Model

Robert Hall developed the random walk hypothesis for consumption based on Friedman's permanent income hypothesis and rational expectations theory. The hypothesis states that changes in consumption are unpredictable as consumers use all available information to rationally assess their lifetime income and only unexpected events affect consumption. Hall (1978) empirically tests this hypothesis using postwar US data and finds that (1) consumption follows a random walk pattern and cannot be predicted by past consumption values, but (2) some recent past disposable income values have a small impact on current consumption, providing mixed support for the random walk hypothesis.

Uploaded by

Taruna Bajaj
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

RANDOM WALK HYPOTHESIS-ROBERT HALL

The theory that changes in consumption are unpredictable is based on the work of the
American economist Robert Hall. The theory is based on the Friedman’s permanent income
hypothesis and the theory of rational expectations. According to Friedman’s permanent
income hypothesis, consumption depends primarily on permanent income. At any moment in
their lifetime, consumers choose consumption based on their current expectations about
lifetime incomes. They would then change their consumption when they receive news that
causes them to change their expectations about their lifetime income. For example, a person
getting an unexpected promotion would revise his expectations about lifetime income
upwards and thus consume more. As long as consumers use all the available information to
assess their lifetime income, that is as long as they have rational expectations, then they
should only be surprised by events that were entirely unpredictable. Therefore, changes in
their consumption should be unpredictable as well.Hall (1978) tests this theory using
postwar aggregate US data .

Hall (1978) Hall (1978) uses the above specification to construct an empirical test of the permanent-
income hypothesis. In short, he runs a series of regressions to see whether the data support the PIH.
It is worth noting that some of the authors assumptions are critical to his empirical specification:
linear marginal utility and constant interest rates.

1. Does consumption follow a random walk?

Yes .

Hall (1978) uses quarterly data on real consumption per capita of nondurables and services,
1948-1977. Durable consumption is excluded because household purchases of these items are
problematic in terms of the model. These are large purchases that the household receives utility
from over time. In the model, consumption is time separable, so that the household receives
utility from consumption in period t and no utility from that consumption purchase thereafter.
Methodology Using the quadratic utility function allows for linear marginal utility. In the context
of the model, Hall (1978) regresses current marginal utility on its past value:7 Ct = βCt−1 + εt The
model tells us that β =1 , so that consumption follows a random walk. However, random walk
tests are notoriously difficult because by failing to reject the null hypothesis that β =1we know
very little. What we know is that the data are not inconsistent with the random walk, but
statistically, this isn’t the same as saying that β =1 . As Hall (1978) points out, this test is severely
limited because it does not allow us to distinguish this theory from other theories of
consumption. Results The data fail to reject the random walk specification. For the reasons
mentioned above, this is weak evidence of the PIH.

2. Can consumption be predicted based on past values of consumption?

No.

According to the model, consumption should be based solely on one lag of consumption.8 All of the
past values/information from further lags of consumption are already contained in the first lag. If
households observe a change in their income, they immediately adjust current consumption.
Therefore, if we include additional AR(p) terms, they should be statistically insignificant. Results An
F-test of the coefficients on the lags of consumption beyond the first lag fail to reject the null
hypothesis that they are jointly equal to zero. This is evidence in support of the PIH. Again, because
we fail to reject a null hypothesis, this evidence is at best weakly in support of the PIH.
Can consumption be predicted based on past values of disposable income?

3. Mixed
4. Hall (1978) uses quarterly data on real disposable income per capita (19481977) to see
whether past values of income matter for current consumption. He uses the same implicit
price deflator and population values used in constructing his consumption variables.
Methodology According to the model, consumption should be based solely on one lag of
consumption. All of the past values/information from further lags of disposable income are
already contained in the first lag of consumption. Lags of disposable income should be
statistically insignificant. Results In most specifications, the evidence fail to reject the joint
statistical significance of lags of disposable income. However, there is some evidence that
more recent lags of disposable income affect current consumption. For those lags that are
significant, they are relatively small in magnitude, and therefore explain a relatively small
portion of the variation in current consumption. However, joint tests longer lags of
disposable income are not statistically significant.

You might also like