Scribd
Upload
23 views

Assignment 3

Uploaded by

Luca Vanz
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
23 views

Assignment 3

Uploaded by

Luca Vanz
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Luca Vanz

Assignment 3
A brief examination of export trends over the past five decades involves creating three
graphs for the variable (EXPGSC1) and three for its logarithmic function (LEXPGSC1). In
the initial set of graphs, stationarity is not evident. While in the set of first difference, the
outcomes show significant improvement, particularly noticeable in the logarithmic
function, which appears less impacted by the COVID-19 period. However, the variance
notably increases in the second differencing graph, indicating a more pronounced effect
during the pandemic. Therefore, employing the I(1) model of the logarithmic function
appears to be the most suitable approach for depicting stationarity.
By employing the correlogram, we can identify the most suitable model for describing
stationarity at each level of difference. For data at level I(0), the autocorrelation
consistently decreases, indicating that the mean is stable at zero. Significant partial
autocorrelation is observed only at the first lag, except for the fifth, which is beyond
consideration due to its distance. This suggests the potential presence of a linear trend
in an ARIMA (1,0,0) model. In the case of first-order differencing, both autocorrelation
and partial autocorrelation exhibit patterns consistent with a random walk. Although the
second lag in autocorrelation is statistically significant, its magnitude is negligible.
Thus, an ARIMA (0,1,0) model with a linear behavior is plausible. As for the second-order
differencing model, pinpointing a singular optimal fit is challenging. The autocorrelation
reveals significant negativity in the first and second lags, with the latter just exceeding
the confidence interval. This implies the inclusion of one or two moving average terms
(MA = 1 or 2). Meanwhile, the partial autocorrelation extends beyond the confidence
interval for the first three lags, decreasing for the fourth, increases again before and
becomes insignificant beyond the fifth. This suggests the potential inclusion of
autoregressive terms 3, 4, 5, or 6.

You might also like