The role of R&D and economic policy uncertainty in Sri Lanka’s economic growth

Financial Innovation

Table 3 Long-run elasticities

Panel A: Long-run elasticities
Variable	Coefficient (p value)	Variable	Coefficient (p value)
lnY	2.388*** (0.000)	lnAL	2.481*** (0.000)
lnA	− 1.685* (0.064)	lnA	5.492*** (0.000)
Constant	− 7.571*** (0.000)	Constant	− 25.882*** (0.000)
Adjusted R²	0.995	Adjusted R²	0.993

Panel B: Piecewise long-run elasticities
Variable	Coefficient (p value)
lnA	10.755*** (0.000)	–	–
lnY	–	2.086*** (0.000)	–
lnAL	–	–	4.886*** (0.000)
Constant	13.676*** (0.000)	− 4.926*** (0.000)	− 64.324*** (0.000)
Adjusted R²	0.956	0.991	0.953

The table shows the long-run elasticity estimates obtained using the dynamic ordinary least squares (DOLS) approach. Panel A estimates the specification \(lnX_{t} = \mu lnQ_{t} + \kappa lnA_{t} + e_{t}\), where \(\kappa = \left( {1 - \phi } \right)/\sigma\). Schumpeterian growth theory holds true if \(\kappa = 0\) and \(\mu = 1\), while semi-endogenous growth theory holds true if \(\kappa > 0\) and \(\mu = 0\). \(X\), \(A\), and \(Q\), denote, respectively, R&D expenditure, TFP, product quality (i.e. \(Y\) or \(AL\)). Panel B estimates the piecewise regressions \(\nu_{t} = lnX_{t} + \left( {\left( {\phi - 1} \right)/\sigma } \right)lnA_{t}\) and \(\varsigma_{t} = lnX_{t} - lnQ_{t}\). \(lnX_{t}\) is the dependent variable. In the Schumpeterian growth model, the predictor is either \(lnY\) or \(lnAL\), whereas in the semi-endogenous growth model, it is \(lnA\). We used fixed lags and leads of one and estimate the Newey-West fixed bandwidth based long-run variance. * and *** indicate, respectively, statistical significance at 10% and 1% levels. Coefficients and p-values are, respectively, outside and inside the parentheses. Our sample is from 1980 to 2018