200916 – Economic and Financial
Modelling
Week 12:
Modelling highly persistent time series
data
Unit Root and Cointegration
Text: Chapter 12
Stationary and Nonstationary Variables
The change in a variable is an important concept
– The change in a variable yt, also known as its
first difference, is given by
Δyt = yt – yt-1
• Δyt is the change in the value of the variable
y from period t - 1 to period t
Figure 1. U.S. economic time series
FIGURE 1 (Continued) U.S. economic time series
Formally, a time series yt is stationary if its mean
and variance are constant over time, and if the
covariance between two values from the series
depends only on the length of time separating the
two values, and not on the actual times at which
the variables are observed.
That is, the time series yt is stationary if for all
values, and every time period, it is true that:
E yt μ (constant mean)
var yt σ 2 (constant variance)
cov yt , yt s cov yt , yt s γ s (covariance depends on s, not t )
The First-Order Autoregressive Model
The autoregressive model of order one, the AR(1)
model, is a useful univariate time series model for
explaining the difference between stationary and
nonstationary series:
yt yt 1 vt , 1
– The errors vt are independent,
σ v2 with zero mean
and constant variance , and may be normally
distributed
– The errors are sometimes known as ‘‘shocks’’
or ‘‘innovations’’
Random Walk Models
Consider the special case of ρ = 1:
This model is known as the random walk
model
These time series are called random walks
because they appear to wander slowly upward or
downward with no real pattern
the values of sample means calculated from
subsamples of observations will be dependent on
the sample period
This is a characteristic of nonstationary series
Examples for random
walk realisations
The random
walks wander
around with no
clear direction
9
Example: 3-month bank bill yield
A random walk is a special case
of a unit root process
Unit root processes are defined
as the random walk but et may
be an arbitrary weakly
dependent process
From an economic point of view
it is important to know whether
a time series is highly
persistent.
In highly persistent time series,
shocks or policy changes have
lasting/ permanent effects.
In weakly dependent processes
their effects are transitory
10
Random walks with drift (adding a constant term)
In addition to the usual random walk
mechanism, there is a deterministic
increase/decrease (= drift) in each period
This leads to a linear time trend around which the series follows its random
walk behaviour. As there is no clear direction in which the random walk
develops, it may also wander away from the trend.
Otherwise, the random walk with drift
has similar properties as the random
walk without drift.
Random walks with drift are not
covariance stationary and not weakly
dependent.
11
Sample path of a random walk
with drift
Note that the series does not
regularly return to the trend
line
Random walks with drift may
be good models for time
series that have an obvious
trend but are not weakly
dependent.
12
We can extend the random walk model even
further by adding a time trend:
yt t yt 1 vt
13
Transformations on highly persistent time
series
Order of integration
Weakly dependent time series are integrated of order zero
(= I(0))
If a time series has to be differenced one time in order to
obtain a weakly dependent series, it is called integrated of
order one (= I(1)) After differencing, the
resulting series are weakly
Examples for I(1) processes dependent (because et is
weakly dependent).
Differencing is often a way to achieve weak dependence
14
Testing for unit roots
Dickey-Fuller Test 1 (No constant and
No Trend)
The AR(1) process yt = ρyt-1 + vt is
stationary when |ρ| < 1
But, when ρ = 1, it becomes the
nonstationary random walk process
We can test for nonstationarity by testing the
null hypothesis that ρ = 1 against the
alternative that |ρ| < 1
Or simply ρ < 1
Tests for this purpose are known as unit root tests
for stationarity
15
A more convenient form is:
yt yt 1 yt 1 yt 1 vt
yt 1 yt 1 vt
yt 1 vt
H 0 : 1 H 0 : 0
H1 : 1 H1 : 0
16
Testing for unit roots cont.
For the validity of regression analysis, it is crucial to know whether
or not dependent or independent variables are highly persistent
Dickey-Fuller test 2 (With Constant but No Trend)
The test is based on an AR(1) regression
Under the null hypothesis, the process
has a unit root. Under the alternative, it
is a stable AR(1) process
Use the t statistic to test the hypothesis, but under the null, it
has not got the t distribution but the Dickey-Fuller distribution
The Dickey-Fuller distribution has to be looked up in tables
17
Testing for unit roots (cont. 1)
Alternative Formulation of the Dickey-Fuller test
The alternative representation is
obtained by subtracting yt-1 from both
sides
Critical values for Dickey-Fuller test
The critical value is much
more negative than it
would be in a t
distribution
18
Testing for unit roots (cont. 2)
The t statistic is -0.714. As consequence,
the null hypothesis of a unit root cannot be
rejected
Include lagged
Augmented Dickey-Fuller test differences of
dependent
variable
The augmented Dickey-Fuller test allows for more serial
correlation
The critical values and the rejection rule are the same as
before
19
Testing for unit roots (cont. 3)
Dickey-Fuller test for time series that have a time trend
Under the alternative hypothesis of
no unit root, the process is trend-
stationary
Critical values for Dickey-Fuller test with time trend
Even more
negative
20
Critical Values for the Dickey–Fuller Test
21
Example
As an example, consider the two interest rate series:
The federal funds rate (Ft)
The three-year bond rate (Bt)
the inclusion of one lagged difference term is sufficient to
eliminate autocorrelation in the residuals in both cases
The results from estimating the resulting equations are:
F 0.173 0.045 F 0.561F
t t 1 t1
(tau ) ( 2.505)
B 0.237 0.056 B 0.237 B
t t 1 t1
(tau ) ( 2.703)
The 5% critical value for tau (τc) is -2.86
Since -2.505 and -2.703 < -2.86 (in magnitutue), we do not reject
the null hypothesis. The two interest rates series are non-
22 stationary.
Spurious regression
The main reason why it is important to know
whether a time series is stationary or nonstationary
before one embarks on a regression analysis is that
there is a danger of obtaining apparently
significant regression results from unrelated data
when nonstationary series are used in regression
analysis
– Such regressions are said to be spurious
23
Consider two independent random walks:
rw1 : yt yt 1 v1t
rw2 : xt xt 1 v2 t
These series were generated independently
and, in truth, have no relation to one another
Yet when plotted we see a positive
relationship between them
24
Time series and scatter plot of two
random walk variables
25
Time series and scatter plot of two
random walk variables cont.
26
A simple regression of series one (rw1) on
series two (rw2) yields:
17.818 0.842 rw ,
rw R 2 0.70
1t 2t
(t ) (40.837)
These results are completely meaningless, or
spurious
The apparent significance of the relationship is false
27
Spurious regression cont.
When nonstationary time series are used in a regression
model, the results may spuriously indicate a significant
relationship when there is none
Regressing one I(1) series on another I(1) series may lead to
extremely high t-statistics
Similarly, the R-squared tend to be very high
Including a unit root may generally lead to completely
misleading inferences
28
Cointegration
Not all regressions with I(1) variables are spurious
If there is a stable relationship between the time series that,
individually display unit root behaviour, these time series are
called ‘co-integrated’
There is an important case when et = yt - β1 -
β2xt is a stationary I(0) process
In this case yt and xt are said to be
cointegrated
Cointegration implies that yt and xt share similar
stochastic trends, and, since the difference et is
stationary, they never diverge too far from each other
29
The test for stationarity of the residuals is
based on the test equation:
eˆt γeˆt 1 vt
The regression has no constant term
because the mean of the regression residuals
is zero.
We are basing this test upon estimated
values of the residuals
There are three sets of critical values
Equation
Which set1: we use
eˆ y depends
bx on whether the
t t t
residuals are derived from:
Equation 2 : eˆt yt b2 xt b1
Equation 3: eˆt yt b2 xt b1 ˆ t
30
An Example of a Cointegration Test
Consider the estimated model:
Bˆt 1.140 0.914 Ft , R 2 0.881
(t ) (6.548) (29.421)
The unit root test for stationarity in the
estimated residuals is:
eˆt 0.225eˆt 1 0.254eˆt 1
(tau ) ( 4.196)
Comparing the calculated value (-4.196) with the critical value(-
3.37 at 5%), we reject the null hypothesis and conclude that (B,
F) are cointegrated
31
Critical Values for the Cointegration Test
32
