Great video! Your videos are always so well explained and very detailed. Well done. Please if I may ask - do we have to check for trend? Is there an implication if we leave it at the default constant? Thank you once again.
Very helpful video. Can I ask why you leave the trend term (+ constant) in the PUR test at First Differences - should not the trend term only be included for the PUR at LEVELS ?
What if cointegrating equation (speed of adjustment/ error correction term) value is positive and insignificant (In Short run equation panel of ARDL results)? kindly explain
In econometric analysis, particularly when using the Autoregressive Distributed Lag (ARDL) model or an Error Correction Model (ECM), the error correction term (ECT) represents the speed at which the dependent variable returns to equilibrium after a change in the independent variables. This is often derived from the long-run relationship between the variables, which is assessed through cointegration tests. Now, if the error correction term is positive, this is unusual because it suggests that if there is a deviation from the long-run equilibrium, the variable will diverge further from the equilibrium in the next period, which is counterintuitive. Typically, we expect a negative ECT so that any deviation from the long-run path is corrected over time, bringing the variable back towards equilibrium. Moreover, if the error correction term is statistically insignificant, it indicates that there is no error correction mechanism at play. This could mean that: There might not be a long-run equilibrium relationship between the variables (i.e., they are not cointegrated). The model may be misspecified. This could involve issues like omitted variables, incorrect functional form, or measurement errors. The data may not have enough power to detect the speed of adjustment. For policy analysis and forecasting, an insignificant and positive error correction term suggests that adjustments to deviations from the long-run path are either not occurring or not behaving as theory would predict. This means that short-run fluctuations may not be informative about long-run relationships, and relying on such a model for policy implications or forecasting could be misleading. In such cases, it's important to review the model specification, consider alternative estimators or models, and ensure that the data is suitable for the analysis at hand.
Yes, you can use the panel Autoregressive Distributed Lag (ARDL) model. Given that you have 22 firms and 10 years of data, this gives you 220 observations. While this isn't a massive dataset, it's sufficient for many econometric analyses, including panel ARDL, especially if the relationships in your data are strong.
A brilliant video!. You made it so simple and made my day. Kudos and a tonne of thanks for sharing your econometric expertise with us. Phil.
Obezip, thank you so much, this was very helpful, provides so much clarity.
Thanks for the video on Panel ARDL, did you have any other video which covers the diagnostics of panel ARDL model
Thanks for watching. Unfortunately, I've not made a video yet on panel ardl diagnostic test... Thou I've made for linear ardl
Great stuff, Sir.
Great Sir.. Stay happy...
Great video! Your videos are always so well explained and very detailed. Well done. Please if I may ask - do we have to check for trend? Is there an implication if we leave it at the default constant? Thank you once again.
Very helpful video. Can I ask why you leave the trend term (+ constant) in the PUR test at First Differences - should not the trend term only be included for the PUR at LEVELS ?
great clip. thanks a lot
What if cointegrating equation (speed of adjustment/ error correction term) value is positive and insignificant ? kindly explain
please reply, how to get long run coefficient for individual banks?????
@@knowledgebulb6232 One of the assumptions of PMG is that the long run parameters are the same (homogenous) across groups
Only the short run parameters vary
Can diagnostic test be conducted on the result?
Good job
What if cointegrating equation (speed of adjustment/ error correction term) value is positive and insignificant (In Short run equation panel of ARDL results)? kindly explain
In econometric analysis, particularly when using the Autoregressive Distributed Lag (ARDL) model or an Error Correction Model (ECM), the error correction term (ECT) represents the speed at which the dependent variable returns to equilibrium after a change in the independent variables. This is often derived from the long-run relationship between the variables, which is assessed through cointegration tests.
Now, if the error correction term is positive, this is unusual because it suggests that if there is a deviation from the long-run equilibrium, the variable will diverge further from the equilibrium in the next period, which is counterintuitive. Typically, we expect a negative ECT so that any deviation from the long-run path is corrected over time, bringing the variable back towards equilibrium.
Moreover, if the error correction term is statistically insignificant, it indicates that there is no error correction mechanism at play. This could mean that:
There might not be a long-run equilibrium relationship between the variables (i.e., they are not cointegrated).
The model may be misspecified. This could involve issues like omitted variables, incorrect functional form, or measurement errors.
The data may not have enough power to detect the speed of adjustment.
For policy analysis and forecasting, an insignificant and positive error correction term suggests that adjustments to deviations from the long-run path are either not occurring or not behaving as theory would predict. This means that short-run fluctuations may not be informative about long-run relationships, and relying on such a model for policy implications or forecasting could be misleading.
In such cases, it's important to review the model specification, consider alternative estimators or models, and ensure that the data is suitable for the analysis at hand.
very informative and very well explained..thanks@@obezipacademy
if I have 22 firms with 10 years, can I use this technique?
Yes, you can use the panel Autoregressive Distributed Lag (ARDL) model. Given that you have 22 firms and 10 years of data, this gives you 220 observations. While this isn't a massive dataset, it's sufficient for many econometric analyses, including panel ARDL, especially if the relationships in your data are strong.
Video is interesting but when you were estimating panel ARDL, why you didn't take the ist difference of CAR as it was stationary at ist difference.
Kindly expatiate further
Hello sir. Before estimating panel unit root, do we need to log the data first or no need?
if you are using it as logged values in your final model, then you must take log first before stationarity test.
How can you find ECT?
Look at COINTEQ01