The Scientific Wealth Of Nations With Special Reference To Algeria : A Cross-country Productivity Analysis Of Academic Research

Résumé

This article aims at evaluating and explaining cross-country academic research activities through two-stage DEA methodology. A sample of 161 countries including 19 from MENA region is analyzed. In the first stage, DEA is applied to measure research productivity. The inputs "Population" and "Gross Domestic Product" (GDP) are used to construct Scientometric Indicator Scores (SIS) for the full sample while the inputs “Human Capital” and “GERD” are used to obtain Scientometric Efficiency Scores (SES) for 116 countries. In the second stage, as postulated by the New Institutional Economics literature, “Quality of Institutions” is treated as a “fundamental” factor explaining scientific research indicators through “Human Capital” channel. Throughout, an IV approach is applied to correct for potential biases where both “Quality of Institutions” and “Human Capital” are treated as endogenous variables determined mainly via appropriate exogenous historical and cultural factors. With a mean of 1.8 Scopus documents (per thousand people) for Algeria during 1996-2019 and 4.5 for MENA region against 13.2 documents worldwide, MENA countries and particularly Algeria seem to be trailing behind. As shown by DEA scores, the mean constant return to scale (CRS) SIS score, estimated at 11.7% for Algeria and 17% for MENA countries against 28% for the remaining 142 countries, confirms that MENA countries are significantly less productive. However, the mean CRS SES score, estimated at 16.6% for Algeria and 27.3% for 13 MENA countries against 34.5% for 103 countries worldwide, indicates no significant difference suggesting that MENA countries are not significantly less efficient. This last result indicates, as expected, that low performance of scientific research for many developing countries including Algeria, could be explained, at least partly, through low levels of both "Human Capital" and “GERD”. At the second stage, for SIS model, OLS estimations confirm the previous result even if the human capital coefficient is small and barely significant while IV estimations show that human capital coefficient is more than double and highly significant suggesting that the size of the bias due to endogeneity is important. However, due to the high level of inefficiency in many non-HIOCED countries, a sizable productivity advantage for HIOCED countries is documented beyond what can be attributed to human capital and GERD. Moreover, IV estimations confirm the key role played by the factor “Quality of Institutions” in explaining SIS (and SES) scores differences. Furthermore, “Cultural Zone” and “Identity of Colonizer” as well as “Natural Resources Rents” prove to be significant exogenous variables. These findings seem to be quite robust with respect to alternative measures of both “Quality of Institutions” and “Human Capital”