Tip:
Highlight text to annotate it
X
in this video we are going to see it how to make an informal analysis
of stationarity by using time series plot
and correlograms. First of all you have to open a database
in this case we're going to use an excel file
so we click on the "file", "open data", then
"import", "Excel" and then double-click
on the Excel file. Gretll
asks if it has to import at first row and first column
we say okay as it is our case
we start the informal analysis of stationarity
by looking at that time serious plot we can look at each of their
plot by right click on a variable
and select time series plot
you can also plot
all the time series on a single graph
to be more quick or to look for co-movements among the variables
to do that just select all the variables
right click and plot time series
on a single graph
from this graph
it seems that both the variables are non-stationary as they rarely cross
their mean
we expect all the series to have a constant
the first two variables, the red and the blue line
seems also to have a downward trend
another informal method that to test for stationarity
is to look at correlograms; to see the correlogram
you have to right click on a variable
and then the click on correlogram; 12 lags
is good, since we have quarterly data
ACF stands for auto-correlation function and the horizontal bars
represent that the auto-correlation between a variable and
a lag of itself,
Gretl displays also the specific value of its auto-correlation
we can see that autocorrelations are
statistically significant at 1 % up to
7 lags, as there are three asterisks
lag eight to ten are significant at five percent, because we have 2
asterisks
and lag 11 at 10%
a slow decline in the horizontal bars suggest a long memory of the
process
PACF stands for partial auto-correlation functions
and describes the amount of correlation between a variable
and a lag of itself which is not explain by
auto-correlations at lower lags order
a partial autocorrelation value close to 1
is a hint for a possible unit root. This strong and persistent temporal
dependence
confirms the non-stationary of the process
this concludes the informal analysis
of stationary; in the next video
we are going to to see how to make a formal analysis
of stationarity with ADF and KPSS test