The models we use are the classical (Gaussian) linear regression model, which assumes that the errors (representing random fluctuations) are independent and identically distributed, with normal distribution, and the GARCH(1,1) model, in which the volatility is no longer assumed to be constant, but satisfying a recurrence equation with respect to its lagged values.
The classical linear regression model is usually estimated by the OLS (Ordinary Least Squares) method, whereas GARCH models are estimated by several variants of the ML (Maximum Likelihood) method.
We estimate autoregressions of the GDP growth rate for Romania, using, alternatively, both methods and comparing the results of estimations and prognoses.
We perform estimations and forecasts on the basis of yearly (1996-2009), as well as quarterly (2000-2009) data.
The source of the data is the National Statistical Institute of Romania (INS).
We also analyze the dependence of the GDP growth rate on the rate of innovation. We characterize innovation by the Summary Innovation Index (SII) provided by the European Innovation Scoreboard (EIS).