Stationarity and Endogeneity

Slides by Soleil Frederick, 2021

1
 Stationarity
• The stationarity or otherwise of a series can strongly influence its
behaviour and properties.
• A stationary variable is one that is mean reverting.

2
Non-Stationarity: Random Walk and Random Walk
with Drift
70

60
Random Walk
50
Random Walk with Drift

40

30

20

10

0
1 19 37 55 73 91 109 127 145 163 181 199 217 235 253 271 289 307 325 343 361 379 397 415 433 451 469 487

-10

-20

3
Definition of Stationarity
• The time series yt is stationary if for all values, and every time period,
if it is true that;
• (constant finite mean, not dependent on t)
• (constant finite variance, not dependent on t)
• (covariance is a function of s, not dependent on t)

• This definition of stationarity is known as weak stationarity or

covariance stationarity.

4
 Box and Jenkins solution
• Box and Jenkins suggested that series can be differenced to make them
variable stationary.
• The weakness of this proposal is that it resulted in a loss of the long term
memory: loss of variable trend.
• Equation brings about stationarity at the expense of trend information.
This is done through variable stationarity.

5
 Types of trends
There are two types of trends in time-series data: deterministic and
stochastic.

1. A deterministic trend is a non-random function of time

(no constant)
(with constant)
In this model, shocks to the system have a transitory effect in that the
process always reverts to its mean (βt or α + βt).

6
 Types of trends
2. By contrast, a stochastic trend is random:
(no constant)
(with constant)
The effect of shocks to variables accumulate over time, with the result
that the best view of where the next value of series will be is either the
current value (random walk) or the current value plus some constant
(random walk with drift) and not some deterministic mean value.

7
 A Deterministic Trend Process

30
25
20
15
10
5
0
-5 1 40 79 118 157 196 235 274 313 352 391 430 469

8
 Two types of Non-Stationarity
• Various definitions of non-stationarity exist
• Here, we are really referring to the weak form or covariance
stationarity.
• There are two models which have been frequently used to
characterise non-stationarity: the random walk model with drift:
yt = λ + yt-1 + ut (1)
and the deterministic trend process:
yt =  + t + ut (2)
where ut is iid in both cases.
9
 Stochastic Non-Stationarity
• Note that the model (1) could be generalised to the case where
yt is an explosive process:
yt = λ + yt-1 + ut
where  > 1.
• Typically, the explosive case is ignored, and we use  = 1 to
characterise the non-stationarity

10
 Definition of Non-Stationarity
• Consider again the simplest stochastic trend model:
yt = yt-1 + ut
or yt = ut

• We can generalise this concept to consider the case where the series
contains more than one “unit root”. That is, we would need to apply the
first difference operator, , more than once to induce stationarity.

11
 Definition of Non-Stationarity cont’d
• If a non-stationary series, yt must be differenced d times before it
becomes stationary, then it is said to be integrated of order d. We
write yt I(d).

So, if yt  I(d) then dyt I(0).

An I(0) series is a stationary series
An I(1) series contains one unit root,
e.g. yt = yt-1 + ut

12
 Characteristics of I(0), I(1) and I(2) Series
• I(0) series should cross the mean frequently.
• An I(2) series contains two unit roots and so would require
differencing twice to induce stationarity.
• I(1) and I(2) series can wander a long way from their mean value and
cross this mean value rarely.

13
 Informal tests of nonstationarity
• In general, nonstationarity can be detected by simply plotting the
series over time.
Mean
• If a series shows no tendency to drift upwards over time, then it is
stationary in the mean. Otherwise it is non-stationary in the mean.
Variance
• If a series starts to gyrate such that over time the amplitudes of the
peaks and the troughs increase, then the time series is non-stationary
in the variance. Otherwise it is stationary in the variance.

14
 Testing for a unit root
• The early and pioneering work on testing for a unit root in time series
was done by Dickey and Fuller.
• The basic objective of the test is to examine the null hypothesis that
φ=1

against the one-sided alternative φ < 1.

• If the error term is white noise, then this is a random walk with φ = 1

15
 Dickey Fuller
• In practice, the following regression is employed, rather than the
previous one, for ease of computation and interpretation

• so that a test of φ = 1 is equivalent to a test of ϕ = 0 (since φ − 1 = ϕ).

• The alternative hypothesis of the Dickey-Fuller test is therefore ϕ < 0.

16
 Dickey Fuller Test
• Dickey-Fuller (DF) tests can be conducted allowing for an intercept, or
an intercept and deterministic trend, or neither, in the test regression.
The model for the unit root test in each case is

• The tests can also be written, by subtracting yt−1 from each side of the
equation, as

17
 Dickey Fuller Test
• The test statistics do not follow the usual t-distribution under the null
hypothesis, since the null is one of non-stationarity, but rather they
follow a non-standard distribution.
• The null hypothesis of a unit root is rejected in favour of the
stationary alternative in each case if the test statistic is more negative
than the critical value.

18
 Problem of Dickey Fuller
• The DF test is only valid only if ut is white noise. In particular, ut is
assumed not to be autocorrelated, but would be so if there was
autocorrelation in the dependent variable of the regression (Δyt )
which has not been modelled.
• If this is the case, the test would be ‘oversized’, meaning that the true
size of the test (the proportion of times a correct null hypothesis is
incorrectly rejected) would be higher than the nominal size used.

19
 The Solution
• The solution is to ‘augment’ the test using p lags of the dependent
variable. The alternative model in case is now written

• The lags of Δyt now ‘soak up’ any dynamic structure present in the
dependent variable, to ensure that ut is not autocorrelated. The test is
known as an Augmented Dickey-Fuller (ADF) test and is still conducted on
ϕ, and the same critical values from the DF tables are used as before.

20
 Criticism of Stationarity Tests
1. The most important criticism that has been levelled at unit root tests is that their power is
low if the process is stationary but with a root close to the non-stationary boundary,
especially with small sample sizes.

2. The null distribution is only available asymptotically i.e., the distribution for small samples is
argued to differ substantially.

3. Criticism of testing nonstationarity in the null hypothesis.

21
Endogenous variables
• Recall is a violation of one of our classical least squares assumptions.
• Explanatory variables that are correlated with our error term are
called endogenous variables.

22
Reasons for a Relationship between x and e
• Measurement Error
• Simultaneous Equation Bias
• Omitted Variables

23
Questions?

24