I have a 40-year time series with numerous dependent variables and just one explained variable to investigate the relationship between them. Which time series model can therefore be used in this scenario? Please provide any references to this study.
You may need to explain better. I know a time series always involves one lagged variable. How can a time series has many dependent variable? How can one variable be explained to account for the relationship between dependent variables?
The most commonly used time series regression model is the Autoregressive Integrated Moving Average (ARIMA) model. ARIMA is a powerful tool for time series forecasting and can be used to analyze a wide range of time series data, including non-stationary and stationary data.
Other models that can be considered for this type of data include Vector Autoregression (VAR) and Vector Error Correction Model (VECM). Both of these models are useful for analyzing multivariate time series data and can be used to examine the relationship between multiple variables over time.
It's important to note that before running any model, it is crucial to check the stationarity of the data and possibly taking the necessary steps to make the data stationary if it is not. Additionally, it's important to check the autocorrelation and partial autocorrelation of the data and adjust the model accordingly.
Sir, in my pov, though I have not much researched about this, but in my guess the modelling through a independent variable and multiple dependent variables is quite vague. In Simultaneous Equation Method I think this can be quite considered when endogenous variable are taken as as dependent variable. But they are endogenous not really a dependent one.
In my own opinion, I think the pre-tests should suggest the best model for analyzing such numerous dependent independent variables and one explanatory or independent variable. If the pre-tests suggest that the variables are all I(1)s & cointegrated, then apply VECM or ARDL model depending on the objectives of your study. However, my worst fear here is that, you might end up having numerous equations. Also, you can decide to build a system of simultaneous equations for such datasets since it is not perturbed by order of integration of time series. I strongly believe that it can be modelled if you can endure the cumbersome processes.