I am studying on how the interaction of domestic credit growth and macroeconomic variables changed after a certain event in a country by an econometric analysis thus the data (quarterly) starts from 2012:Q3 till 2017:Q1 corresponding to 19 terms. I also took the changes in each variable to see the differences according to the previous term.
Is it alright if I conduct causality, cointegration and multiple regression analyses for this data?
I would be pleased if you could suggest a proper analysis method for this limited data or any other relevant suggestions.
Thanks a lot in advance..
Hey, apparently you need to use a multivariate time series approach, as a suggestion you first need to rely upon large sample in order to get reliable and meaningful results. On top of that most of the macro variables are frequently reported as I (1) variables meaning that cointegration or causality analysis of them is not as straightforward as for stationary series. For example you will need to differentiate them so you really need to have a large sample. Why not using simulation? Read carefully Hamilton (1994) for a good modelling of your data. Good luck
if you want to examine the impact of a certain event to the changes of the relationship between domestic credit growth and macroeconomic variables (assuming if this is what you are trying to achieve in you objectives), maybe you can do difference-in-difference (double difference) econometric analysis or maybe just a time-series cointegration analysis, but you need to have a s dummy variables for two periods "before" that event (i.e coded = 0) and "after" that event (coded = 1). This is just a suggestion that you could explore.
Best regards,
Lemuel Preciados
Dear Yvonne Umulisa,
Thank you very much for your reply. I'll check for the study you suggested:)
Best wishes..
Dear Lemuel Preciados,
Thank you very much for your reply. I think I am going to use dummy variable as well regarding the event. But I also check for difference-in-difference (double difference) econometric analysis for the future research :)
All the best..
You can check the structural change by applying chow test. Many others are there which i am not thorough with but literature are available. U can also try dummy variable model (ANCOVA) to understand the reason for structural change if any
Thank you Feroze Sheikh,
I decided to add dummy variable instead.
Best wishes..
Hi,
The DID is a very good approach provided your data is stationary. To test for that, use zivot-Andrews test with endogenous structural break. If the data is non-stationary, then use VEC with weak exogeneity tests.
Good luck