Research on regional differences and influencing factors of green technology innovation efficiency of China's high-tech industry (Q2293632)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Research on regional differences and influencing factors of green technology innovation efficiency of China's high-tech industry |
scientific article |
Statements
Research on regional differences and influencing factors of green technology innovation efficiency of China's high-tech industry (English)
0 references
5 February 2020
0 references
The paper studies the green technology innovation (GTI) efficiency shown by high-tech industry in China between the years 2008 and 2016, as well as, the factors which are preponderant to explain such efficiency. The study is done in 4 steps. At Step 1, taking into account that development level on high-tech industry is highly correlated to the GTI efficiency and China's development is very heterogeneous in its different regions, 30 of its provinces are divided in 4 clusters, according to their development levels, using K-means technique. At Step 2, a measure of GTI efficiency is computed for each province, at each year. Slacks based measure (SBM) joint with data envelopment analysis (DEA) techniques are used in such computation. At Step 3, 13 factors are considered in each of the 4 clusters to identify which ones are important (significant) for explaining the GTI efficiency, using Lasso regression. \(C_p\) criterium is used to indicate that 4 is a reasonable number of non-zero parameters to consider in the regression at each cluster, if it is true the assumption that variable selection in each cluster is similar to the one for the whole country, as assumed. Two comments are to be done at this point. One, the foreign direct investment (FDI) factor, being a proportion, exhibits an odd 4.466 maximum value in Table 1. Two, the observations are done during 9 consecutive years. In that sense, the independence assumption for the errors in the regression model seems to be inappropriate. At Step 4, the analyses of factor influences are deepened in two ways. First, by using quantile regression model, it is studied how the factors coefficients in the regression change at three different quantile levels of the GTI efficiency, in each one of the 4 different clusters. Second, by using DEA-Tobit model, more robust estimations of the regression parameters are obtained for the 4 different clusters. The estimation results obtained by the 3 regression models look to be coherent.
0 references
innovation efficiency
0 references
regional differences
0 references
influencing factors
0 references
Lasso regression
0 references
K-means
0 references
quantile regression
0 references