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

Amarasekara, Chandranath; Iyke, Bernard Njindan; Narayan, Paresh Kumar

doi:10.1186/s40854-021-00322-5

Financial Innovation

Table 5 Impact of EPU on TFP growth based on alternative R&D intensity indicators

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

	(1)	(2)	(3)	(4)
Panel A: RD/Y
Constant	0.185* (0.057)	0.134 (0.238)	0.016 (0.893)	− 0.027 (0.839)
\(\left( {\Delta lnX_{t}^{d} } \right)*\left( {\Delta EPU_{t} } \right)\)	− 0.346* (0.085)	− 0.304 (0.118)	− 0.301* (0.075)	− 0.310 (0.105)
\(\left( {\Delta lnX_{t - 1}^{d} } \right)*\left( {\Delta EPU_{t - 1} } \right)\)		0.295** (0.042)	0.437*** (0.002)	0.450*** (0.002)
\(\left( {\Delta lnX_{t - 2}^{d} } \right)*\left( {\Delta EPU_{t - 2} } \right)\)			0.422*** (0.001)	0.435*** (0.002)
\(\left( {\Delta lnX_{t - 3}^{d} } \right)*\left( {\Delta EPU_{t - 3} } \right)\)				0.017 (0.898)
\(\Delta lnX_{t}^{f}\)	0.018 (0.797)	0.011 (0.883)	− 0.020 (0.746)	− 0.016 (0.801)
\(ln\left( {X/Q} \right)_{t}^{d}\)	0.034 (0.347)	0.042 (0.235)	0.050 (0.104)	0.054 (0.111)
\(ln\left( {X/Q} \right)_{t}^{f}\)	0.052* (0.069)	0.038 (0.260)	0.004 (0.903)	− 0.008 (0.831)
\({\text{ln}}\left( {A^{JPN} /A^{LKA} } \right)_{t - 1}\)	0.062 (0.164)	0.033 (0.483)	− 0.015 (0.734)	− 0.029 (0.571)
\(\delta\)		− 0.008 (0.973)	0.559** (0.045)	0.592* (0.051)
\(R^{2}\)	0.151	0.252	0.478	0.482
Panel B: RD/AL
Constant	− 0.325* (0.092)	− 0.268 (0.250)	− 0.049 (0.834)	− 0.005 (0.987)
\(\left( {\Delta lnX_{t}^{d} } \right)*\left( {\Delta EPU_{t} } \right)\)	− 0.284 (0.132)	− 0.249 (0.174)	− 0.252 (0.115)	− 0.235 (0.187)
\(\left( {\Delta lnX_{t - 1}^{d} } \right)*\left( {\Delta EPU_{t - 1} } \right)\)		0.293** (0.046)	0.438*** (0.002)	0.442*** (0.003)
\(\left( {\Delta lnX_{t - 2}^{d} } \right)*\left( {\Delta EPU_{t - 2} } \right)\)			0.424*** (0.002)	0.424*** (0.003)
\(\left( {\Delta lnX_{t - 3}^{d} } \right)*\left( {\Delta EPU_{t - 3} } \right)\)				− 0.015 (0.905)
\(\Delta lnX_{t}^{f}\)	0.020 (0.772)	0.014 (0.858)	− 0.030 (0.655)	− 0.030 (0.666)
\(ln\left( {X/Q} \right)_{t}^{d}\)	0.020 (0.568)	0.031 (0.378)	0.043 (0.168)	0.041 (0.217)
\(ln\left( {X/Q} \right)_{t}^{f}\)	0.021* (0.083)	0.017 (0.240)	0.003 (0.836)	0.000 (0.988)
\({\text{ln}}\left( {A^{JPN} /A^{LKA} } \right)_{t - 1}\)	0.050 (0.223)	0.028 (0.523)	− 0.014 (0.725)	− 0.020 (0.665)
\(\delta\)		0.045 (0.852)	0.610** (0.028)	0.615** (0.047)
\(R^{2}\)	0.138	0.238	0.464	0.461

The table shows estimates of the TFP growth regression, which uses alternative measures of R&D intensity: \(\Delta lnA_{t} = \gamma_{0} + \gamma_{1} (\Delta \ln X_{t}^{d} )*(\Delta EPU_{t} ) + \gamma_{2} (\Delta \ln X_{t - 1}^{d} )*\left( {\Delta EPU_{t - 1} } \right) + \gamma_{3} (\Delta \ln X_{t - 2}^{d} )*\left( {\Delta EPU_{t - 2} } \right) + \gamma_{4} (\Delta \ln X_{t - 3}^{d} )*\left( {\Delta EPU_{t - 3} } \right) + \gamma_{5} \Delta lnX_{t}^{f} + \gamma_{6} ln\left( \frac{X}{Q} \right)_{t}^{d} + \gamma_{7} ln\left( \frac{X}{Q} \right)_{t}^{f} + \gamma_{8} ln\left( {\frac{A}{A}_{LKA}^{JPN} } \right)_{t - 1} + e_{t} .\) In this regression model, EPU influences TFP growth (\(\Delta lnA\)) through domestic R&D (\(X^{d}\)). This impact is captured by the interaction terms between \(EPU\) and \(X^{d}\). We include the contemporaneous interaction term as well as lagged interaction terms up to three lags, to capture the persistent impact of EPU on R&D and TFP growth. From the second (2) to fourth (4) regressions, we estimate the joint impact of the interaction terms, and this is captured by the parameter, \(\delta = \mathop \sum \limits_{i = 0} \gamma_{i + 1}\). The variables, \(X^{f}\), \(X/Q\), and \(A^{JPN} /A^{LKA}\), denote, respectively, Japan’s R&D, Sri Lanka’s R&D intensity, and technology frontier. In Panel A, R&D intensity is proxied by R&D per GDP (Y), while in Panel B it is proxied by R&D per the product of TFP and labour (AL). Coefficients and p-values are, respectively, outside and inside the parentheses
*, **, and *** statistical significance at 10%, 5%, and 1% levels. Our sample is from 1980 to 2018

Back to article page