We are looking at conducting a study to measure the impact of private loan on GDP growth in Cambodia, using yearly time series data covering a rage of 16 years (we have data only about this range). What model and tests do you suggest? Help us by sending some relevant literature which will highly be appreciated.
Measuring the impact of private loan on GDP growth can be done by using linear regression, which alsoinclude forecasting for the future periods.
the preliminary observation of your data allowed to denote that are macroeconomics data. In fact, I think that you have many choices depending on the nature of the data:
-- cointegration (johnson test) for long run and ECM for short run.
-- Panel if the study includes many countries
-- time series if the data are stationary
Measuring the impact of private loan on GDP growth can be done by using linear regression, which alsoinclude forecasting for the future periods.
I agree with Balazs and Mahfuz, as mentioned by Pahlaj, the lag length of the data is 15 and his study concerns only one country which is Cambodia, hence we can use only the linear regression where economic growth is explained by private loan but the author should include other independent variables in the model. As mentioned by Balazs, the author can use quarterly data. In the case of 15 years, the author cannot use the time series econometrics (stationarity, cointegration and causality) because the lag length is too short.
Vector Error Correction model together with Johansen Co-integration test; and Granger causality test.
Dear, you can not answer such a general question. However you. You can not help but try general linear models or mixed models depending on claiming to be explanatory variables.
Dear, for time series you have to test whether your data are stationary or not. If your data is stationary use VECM, if non-stationary use ECM. I used that to test whether trade openness promote growth for one single country.
Consider Multivariate Polynomial Regression (MPR) - multilinear regression (MLR) on steroids! It has the fitting capability of artificial neural networks with resistance to overfitting, the transparency and ease of use of MLR but includes nonlinearity and complex interactions. It can handle time-series, too. www.simetrica-llc.com
Dear, I suggest the approach based on state space time series analysis. By private I can send you a text book, or e-mail address.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling