In a correlogram, for a purely white noise process the autocorrelations at various lags hover round zero. Thus, if the correlogram of an actual (economic) time series resembles the correlogram of a white noise time series, we can say that time series is probably stationary, (Gujaratti, 2004)
For the PDI the autocorrelation coefficient starts at a very high value at lag 1 (0.9670) and declines very slowly. Thus it seems that the PDI time series is non-stationary.
PCE correlogram
For the PCE the autocorrelation coefficient starts at a very high value at lag 1 (0.9696) and declines very slowly. Thus it seems that the PCE time series is nonstationary.
PROFITS correlogram
For the Profits the autocorrelation coefficient starts at a very high value at lag 1 (0.9539) and declines very slowly. Thus it seems that the Profits time series is non-stationary. DIVIDENDS correlogram
For the Dividends the autocorrelation coefficient starts at a very high value at lag 1 (0.9718) and declines very slowly. Thus it seems that the Dividends time series is non-stationary. 21.18) The ADF test includes time lags therefore it is appropriate for when the error term is serially correlated than the DF test. This test is conducted by augmenting the equations by adding the lagged values of the dependent variable.
The null hypothesis is that = 0; that is, there is a unit rootthe time series is non-stationary. The alternative hypothesis is that is less than zero; that is, the time series is stationary. Reject null hypothesis when t value is greater than 5% critical value If the null hypothesis is rejected, it means that Yt is a stationary time series with zero mean, (Gujaratti, 2004)
For PCE Test statistic = -1.646 5% Critical value = -3.488, Do not reject null hypothesis, the time series are non-stationary
For Profits Test statistic = -2.725 5% Critical value = -3.488, Do not reject null hypothesis, the time series are non-stationary
For Dividends Test statistic = -3.175 5% critical value = -3.488, Do not reject null hypothesis, the time series are non-stationary
For PDI Test statistic = -2.405 5% critical value = -3.488, Do not reject null hypothesis, the time series are non-stationary
21.19 a) When you regress two time series that are non stationary, you will get a spurious regression, therefore it is to be expected
b) There is no cointegration between dividends and profits. We can test using the Cointegrating Regression Durbin Watson test (CRDW) whereby the 1, 5, and 10 percent critical values to test the hypothesis are 0.511, 0.386, and 0.322, respectively. Thus, if the computed d value is smaller than, say, 0.511, we reject the null hypothesis of cointegration at the 1 percent level, (Gujaratti, 2004) In our case computed d = 0.071, it is less than the critical d values at 1%, 5% and also 10% therefore we reject the null hypothesis of cointegration, therefore there is no cointegration
We can also use the Johansen test; Johansens testing procedure starts with the test for zero cointegrating equations (a maximum rank of zero) where the null hypothesis for the first is there is no cointegration for zero equations which you reject because trace statistic (47.1196) is more than critical value(15.41). But if there is no cointegration for zero equations then we have to look at the next if there is cointegration for at least one linear equation, that will be our null hypothesis, in our time series, the trace value (0.0296) is less than the critical (3.76) which means we need to accept the null hypothesis that at least one linear equation are cointegrated in our model. However the weakness of this test is that it relies on asymptotic properties and is therefore sensitive to specification errors in small samples. Our time series only has 25 realizations which is limited. In the end some judgment in combination with some economic and statistical model building is unavoidable.
c) Using the ECM the dividends in relation to profit exhibit no long run relationship because the p = 0.265, it is greater than 0.05 which means it is insignificant, which shows there is no long run relationship. The series are also not cointegrated therefore they will have no long run relationship.
d) Dividends has a stochastic trend, this can be seen by using the graphical analysis test
Profits also has a stochastic trend, this can be seen by using the graphical analysis test
e) When the time series are cointegrated it does not matter which is the regressor and which is the regressand but when you refer to the theory it is obvious that dividends depend on profit. 0 5 0 1 0 0 1 5 0 D I V I D E N D S 1960q1 1965q1 1970q1 1975q1 1980q1 Quarter 5 0 1 0 0 1 5 0 2 0 0 2 5 0 P R O F I T S 1960q1 1965q1 1970q1 1975q1 1980q1 Quarter