The model used an optimal level of 4 lags, and i had 43 observation. however, dus to the number of lags chosen 39 observations are used in the results. kindly assist with insignificant coefficients for variables in the long-run, because F-statistic for the bound test proves that there is a long run relationship. Furthermore, the error correction term is -1.316439 but still the coefficents of variables in the long-run are insignificant i do not know what to do as the variables were stationary at different levels ranging from 1(0) to 1(1) so what could have poosibly went wrong?
The ARDL (Autoregressive Distributed Lag) model is commonly used for analyzing the long-run relationships between variables. When the coefficients of variables in the ARDL model are not significant, it typically indicates several possible reasons:
To address these issues, it's important to conduct diagnostic tests such as tests for multicollinearity, check the stationarity of variables, consider alternative specifications, and ensure the model includes all relevant variables. Additionally, increasing the sample size if possible or using alternative estimation techniques may help improve the robustness and significance of coefficients in the ARDL model.
Surely there is feedback between your explanatory variables and GDP. For example, an increase in GDP will impact Bank deposits, Investment, government expenditure, and possibly other explanatory variables. Then the explanatory variables are not weakly exogenous and the ARDL methodology is invalid.
I would add the following points
If coefficients in the ARDL Long Run Form and Bounds Test are not significant, several factors may contribute:
To resolve, verify lag length appropriateness, refine model specifications, improve data quality, re-evaluate theoretical justifications, address multicollinearity, and adapt estimation methods as needed.
The ARDL Long Run Form and Bounds Test coefficients for variables may not be significant due to insufficient data points, multicollinearity among variables, misspecification of the model, or the presence of structural breaks or non-stationarity in the data.
- Badges
- Science topic
- Similar topics
- Psychology
- Cognitive Science
- Mind