The impact of domestic R&D and North–South R&D spillovers on energy intensity in developing countries

Herzer, Dierk Article — Published Version The impact of domestic R&D and North–South R&D spillovers on energy intensity in developing countries Economic Change and Restructuring Provided in Cooperation with: Springer Nature Suggested Citation: Herzer, Dierk (2024) : The impact of domestic R&D and North–South R&D spillovers on energy intensity in developing countries, Economic Change and Restructuring, ISSN 1574-0277, Springer US, New York, NY, Vol. 57, Iss. 2, https://doi.org/10.1007/s10644-024-09591-3 This Version is available at: https://hdl.handle.net/10419/315252 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ Vol.:(0123456789) Economic Change and Restructuring (2024) 57:33 https://doi.org/10.1007/s10644-024-09591-3 1 3 The impact ofdomestic R&D andNorth–South R&D spillovers onenergy intensity indeveloping countries DierkHerzer1 Received: 29 June 2023 / Accepted: 1 December 2023 / Published online: 14 February 2024 © The Author(s) 2024 Abstract This study utilizes panel data between 1995 and 2015 for a cross section of 33 developing (lowand middle-income) countries to investigate the impact on domestic energy intensity both of domestic R&D and of possible spillovers from foreign R&D conducted in developed (high-income) countries. More specifically, it examines R&D spillovers from developed countries (North) to domestic energy intensity in developing countries (South) through disembodied channels, total goods imports, and imports of machinery and equipment. Our main findings, based on panel cointegration techniques, are as follows: First, domestic R&D in the long run does not contribute to reductions in energy intensity in developing countries; second, there is no evidence to suggest that disembodied North–South R&D spillovers affect the long-run level of domestic energy intensity; third, there are nevertheless significant spillovers from R&D conducted in industrial countries that reduce energy intensity in developing countries; and fourth, while many imported goods are not a channel for North–South R&D spillovers, such spillovers are transmitted through imports of machinery and equipment. Keywords Energy intensity· Domestic R&D· North–South R&D spillovers· Developing countries· Panel cointegration methods JEL Classification Q43· Q55· F18 1 Introduction Several studies find positive effects of domestic research and development (R&D) performed in developing countries and foreign R&D performed in industrial countries on total factor productivity (TFP) in developing countries (see, e.g., Coe etal. * Dierk Herzer [email protected] 1 Department ofEconomics, Helmut-Schmidt-University Hamburg, Holstenhofweg 85, 22043Hamburg, Germany Economic Change and Restructuring (2024) 57:33 1 3 33 Page 2 of 31 1997; Madsen etal. 2010; Herzer 2022a, b). An increase in TFP implies that a given output can be produced with fewer standard factors of production, such as labor and physical capital, as well as human capital, or that more output can be produced with the same quantity of factors of production. Thus, an increase in TFP can be interpreted as factor-saving technical change. Consequently, the available R&D-TFP literature suggests that both domestic R&D and foreign R&D conducted in developed countries generate new technologies that are factor saving in developing economies. Given this implication, and given that energy is necessary for the production of all kinds of goods, including energy itself, it is natural to ask: Do domestic and foreign R&D also generate energy savings per unit of output and thus reduce energy intensity (i.e., the ratio of energy use to GDP) in developing economies? The answer to this question is the subject of this study. A reduction in energy intensity in developing countries means that they can raise their living standards without a proportional increase in energy use, thereby reducing the growth of the environmental problems associated with increasing energy demand—such as air and water pollution, land disturbance, radioactive waste from nuclear energy production, and global climate change due to greenhouse gas emissions from fossil-fuel fired power plants. According to data from the World Development Indicators, the ratio of energy use to real GDP in the group of lowand middle-income countries (defined here as developing countries) exceeded that in the group of high-income countries by more than factor 1.4 in 2014 (the last year with available data for these country groups). Therefore, the answer to the above question is not only of academic interest but also directly relevant to policymakers concerned with both economic and sustainable development. However, the answer is theoretically unclear (as discussed in Sect.2), and the empirical evidence is scarce. There are only six studies on the impact of domestic R&D on energy intensity in developing countries (Yu 2012; Wang and Han 2017; Dong etal. 2018; Huang etal. 2018, 2020; Huang and Chen 2020), and only three on the impact of both domestic and foreign R&D on energy intensity in developing countries (Huang etal. 2018, 2020; Wang and Han 2017). All these studies are single-country studies for China, based on province-level panel data.1 The evidence from all of these studies suggests that domestic R&D has an energy intensity reducing effect.2 With respect to the effect of foreign R&D on domestic energy intensity, two studies find evidence of a negative (reducing) effect of both import-related and foreign direct investment (FDI)-related spillovers from R&D 1 A related study is that of Godil etal. (2021), who examine, among other things, the effect of R&D intensity (i.e., the ratio of R&D expenditures to GDP) on energy consumption per capita (not measured in logs) using time-series data for India. Their results suggest that R&D intensity has a negative effect on energy consumption per capita. However, their empirical model has the counterfactual implication that a doubling of R&D can reduce energy consumption per capita in an economy with only one dollar of R&D to the same extent as in an economy where R&D expenditures amount to one billion dollars. In addition, R&D intensity suffers from endogeneity because higher energy use may result in higher GDP (the denominator of R&D intensity). 2 Huang and Chen (2020) also consider different types of R&D and find that while industrial R&D contributes to reductions in energy intensity, independent R&D and higher education R&D have no significant effect on energy intensity. They also find that experimental R&D has a negative effect on energy intensity, whereas the effect of both basic R&D and applied R&D is insignificant. 1 3 Economic Change and Restructuring (2024) 57:33 Page 3 of 31 33 performed in high-income countries on domestic energy intensity (Wang and Han 2017; Huang etal. 2018); one study finds, somewhat surprisingly, that while the effect of foreign R&D spillovers through imports is insignificant, and while the effect of foreign R&D spillovers through FDI is negative, foreign R&D spillovers through exports increase domestic energy intensity (Huang etal. 2020). However, the majority of these studies (Wang and Han 2017; Huang etal. 2018, 2020; Huang and Chen 2020) do not control for (strong) error cross-sectional dependence due to unobserved common factors. Consequently, the results of the majority of studies may be biased in the presence of omitted common factors that are correlated with the included explanatory variables and the dependent variable.3 In addition, some studies (Yu 2012; Huang and Chen 2020; Huang etal. 2020) use methods that assume stationary data and thus can produce misleading results when the data are non-stationary. Moreover, most studies (Yu 2012; Wang and Han 2017; Huang and Chen 2020; Dong etal. 2018) utilize estimators that require strict exogeneity of the regressors, thus yielding potentially misleading results when the regressors are not strictly exogenous.4 Since all these studies suffer from at least one of these shortcomings, their results should be viewed with some caution. In addition, it is well known that findings from single-country studies cannot necessarily be generalized. Even if the findings of these studies are valid, it may therefore be that they apply only to China. Given the lack of general cross-country studies on the impact of domestic and foreign R&D on energy intensity in developing countries, this study aims to fill this gap. More specifically, we conduct a cross-country panel analysis using data from 33 developing countries (including China) spanning the years 1995 to 2015. It is worth noting that our study is the first to use panel data for a cross section of developing countries. In addition, this study differs from previous research by examining disembodied, non-trade-related R&D spillovers from developed to developing countries, as well as R&D spillovers through imports of all goods and imports of machinery and equipment. Furthermore, this study takes into account all the methodological problems addressed above.5 More specifically, we use panel cointegration methods to 3 Cross‐sectional dependence may be due to common factors that affect all panel units and/or spatial spillover effects across subsets of panel units. Cross‐sectional dependence due to common factors is also known as strong cross-sectional dependence; cross-sectional dependence due to spatial spillovers is also known as weak cross-sectional dependence. The presence of weak cross-sectional dependence does not affect the consistency of conventional panel data estimators, but the standard errors may be biased. In contrast, strong cross-sectional dependence, if not controlled, can lead to biased coefficient estimates (Chudik and Pesaran 2015). 4 If reductions in energy intensity imply GDP growth due to energy-saving technical change, and if firms respond to growth-induced increases in demand for variety by engaging in horizontal R&D to develop new varieties of existing products, then it is possible that reductions in energy intensity contribute to increased R&D activities via increases in GDP, at least in the short run. The implication is that domestic R&D is likely not strictly exogenous. 5 We note two things here. First, it is not possible to measure R&D spillovers through FDI over our sample period because complete time series data on bilateral FDI flows from developed source countries are not available for developing host countries over the period 1995–2015. It is perhaps interesting to Economic Change and Restructuring (2024) 57:33 1 3 33 Page 4 of 31 address the non-stationary nature of the data and analyze the long-run relationships between our variables of interest. As discussed in more detail in Sect.4.3, and as noted by Coe etal. (2009, p. 724), “[u]nder cointegration, parameter estimates are super consistent, and hence are robust to problems such as omitted variables, simultaneity, and endogeneity.” While we are not aware of theoretical reasons to suggest that foreign R&D is endogenous to domestic energy intensity, we thus account for the likely endogeneity of domestic R&D. In addition, we control and test for error cross-sectional dependence in the residuals of our models. It should be explicitly noted here that we use the World Bank classification of developing countries according to which lowand middle-income countries are classified as “developing countries” (World Bank 2007, 2012). Thus, the term “developing countries” includes post-communist countries. All countries in our sample (listed in Table2) that fall under the category of developing countries according to the World Bank classification are classified by the IMF as “emerging market and developing economies,” which also include post-communist countries. As a robustness check, we also use a sample of “developing economies” as classified by UNCTAD that does not include post-communist countries. We come back to this point in Sect.4.2. Here, we note for completeness that, following common practice, we use the term “North” as shorthand for developed or industrial economies and “South” as shorthand for developing countries. An important point is that our study also differs from previous work in that it also examines the impact of domestic R&D conducted in developed source countries of foreign R&D spillovers to developing countries on energy intensity within these source countries. If foreign R&D conducted in industrialized countries contributes to reductions in energy intensity in developing countries (through international R&D spillovers), it can be plausibly concluded that R&D conducted by industrialized countries tends to result in energy-saving technologies. If this conclusion is correct, then one should expect to find a negative effect of domestic R&D (conducted in developed countries) on energy intensity in developed countries. To our knowledge, there is only one cross-country panel study on the impact of R&D on energy intensity in developed countries: Alam etal. (2019). The authors analyze firm-level data for the G-6 countries (which include Canada, France, Germany, Japan, the UK, and the US) and find that R&D reduces energy intensity in these nations. Our study is the first both to examine the impact of foreign R&D conducted in industrial countries on energy intensity in developing countries and to conduct a plausibility check Footnote 5 (continued) note in this context that there are several studies on the impact of FDI on energy intensity in developing countries, which, however, do not explicitly examine the effect of FDI-related foreign R&D spillovers on energy intensity in developing countries, but focus on the broader impact of FDI. The evidence from these studies is mixed, with some indicating that FDI reduces energy intensity in developing countries, while others find no significant effect. For a review of this literature see Herzer and Schmelmer (2022). Second, we also examined R&D spillovers through exports of total goods and R&D spillovers through exports of machinery and equipment, but found little or no evidence of long-run spillovers from foreign R&D conducted in developed to domestic energy intensity in developing countries through exports of total goods and exports of machinery and equipment (from developing to developed countries). 1 3 Economic Change and Restructuring (2024) 57:33 Page 5 of 31 33 for our main results by investigating the impact of domestic R&D on energy intensity in 15 developed source countries of foreign R&D spillovers. To preview our main results, we find that while domestic R&D, in the long run, does not contribute to reductions in energy intensity in developing countries, foreign R&D performed in industrial countries reduces energy intensity in developing countries in the long run. Specifically, our results suggest that North–South spillovers occur mainly through imports of machinery and equipment rather than through imports of other goods and that the impact of foreign R&D varies with the share of machinery and equipment imports in GDP. However, we find no evidence of disembodied spillover effects. An additional result of this study is that there is evidence that domestic R&D performed in industrial source countries of R&D spillovers reduces energy intensity in these countries as well. The remainder of this paper is organized as follows. In Sect.2, we discuss the theoretical background. Section 3 presents the empirical model and defines the variables. Section4 describes the data, including the sample, and discusses some econometric issues and the empirical methodology. Section5 reports our results, and Sect.6 concludes and provides some policy implications. 2 Theoretical background We begin with a general energy-augmented aggregate production function of the form Y = Af(AK, AL, AE). In this equation, Y represents aggregate output, K stands for capital, L for labor, and E for energy. The multiplier A denotes the level of technology, which is the focus of our discussion here. In the terms AK, AL, and AE, A indicates that the technology augments capital, labor, and energy, respectively. If A appears in front of the function f, the technology is factor neutral. Technical change,  A , improves the productivity of K, L, and E, respectively. In the case of purely laboror capital-augmenting technical change, without energy-saving advances, technical change thus reduces the laboror capital-output ratio. It has no effect on energy intensity, the ratio of E to Y, provided both that there are no substitution effects between energy and labor or capital and that the growth of income due to technical change does not induce a shift in consumption patterns toward energy-intensive goods. Since laboror capital-saving technical change reduces the effective price of labor or capital, labor or capital will, however, be induced to substitute for energy. In addition, the reduction in the effective price of labor or capital should lead to lower prices for labor or capital-intensive products. The pattern of demand may therefore shift away from energy-intensive goods, so that less energy-intensive sectors expand relative to energy-intensive sectors. Thus, even purely labor or capital-augmenting technical change may, in the long-run, contribute to reductions in energy intensity. If, however, irrespective of relative prices, increases in income during industrialization are associated with a shift in consumption patterns toward energy-intensive goods and services (such as private vehicles, air conditioners, and flights), as argued and demonstrated by Hart (2018), then the increases in real income from rising productivity may result in increases in the relative size of energy-intensive sectors. It is therefore also possible, and likely, that labor or capital-augmenting technical change Economic Change and Restructuring (2024) 57:33 1 3 33 Page 6 of 31 leads to an increase in energy intensity in the long run, even if there is substitutability between energy and labor or capital. Analogously, purely energy-augmenting technical change implies that the same output can be produced with less energy and thus that the effective price of energy declines. The decline in the effective price of energy induces a substitution in favor of energy versus labor or capital, which offsets to some degree the initial reduction in energy intensity. The lower the elasticity of substitution between labor or capital and energy, the smaller the offsetting effect.6 In addition, the decline in the effective price of energy implies an increase in real income. If this extra income is spend on energy-intensive goods, the relative size of energy-intensive sectors increases. Thus, even energy-saving technical change does, in the long run, not necessarily lead to reductions in energy intensity. Finally, the concept of factor-neutral technical change implies that the ratio of L to Y, the ratio of K to Y, and the ratio of E to Y decrease, at least initially. Like above, increases in income as a result of factor-neutral technical change may, however, induce an increase in the relative size of energy-intensive sectors, and thus an increase in energy intensity in the long run. Thus, it can be assumed that energy intensity, EI, depends on the level of technology (which can be more or less energy-augmenting) using a function of the form EI = Aβ. Although the sign of the elasticity β is theoretically indeterminate, it is reasonable to assume that the more energy saving technical change is, the greater the likelihood will be that technical change will contribute to reductions in energy intensity in the long run. Assuming further a long-run relationship between R&D effort and the level of technology of the form A = R&Dφ, energy intensity can be expressed as where α ≡ φ × β is the elasticity of energy intensity with respect to R&D.7 Based on this equation and on the above theoretical considerations, it can be hypothesized (1) EI =R&D𝛼 6 If the elasticity of substitution of energy is less than one, then improvements in energy productivity will lead to reductions in energy intensity (holding income effects constant). If the elasticity of substitution is greater than one, then energy-augmenting technical change will induce increases in energy intensity. Koetse etal. (2008) find in a meta-analysis that the elasticity of substitution between capital and energy is less than one. Stern and Kander (2012), using historical data for Sweden, find that the elasticity of substitution between a capital-labor aggregate and energy ranges between 0.64 and 0.69. 7 The relationship between R&D effort and the level of technology of the form A = R&Dφ can be derived as follows. As discussed, for example, in Herzer (2022), semi-endogenous growth models assume a knowledge production of the form  A = δAϕR&Dλ, where  A is the flow of new knowledge or technical change; δ is a constant of proportionality; A represents the stock of existing knowledge or the level of technology; ϕ is a parameter that describes the nature of the returns to the stock of knowledge, R&D stands for R&D effort; and λ, where 0 < λ ≤ 1, is a parameter that captures the possibility of duplication in research (i.e., the possibility that a doubling of research effort less than doubles the production of new knowledge because of duplication). Assuming that the stock of knowledge grows in the long run at a constant rate g, the above equation can be solved for the stock of knowledge, yielding A = ( 𝛿 g A) 1 1−𝜙R&D 𝜆 1− 𝜙 . This equation predicts that, provided the growth rate of knowledge is constant over the long run, changes in R&D effort are positively associated with changes in the level of technology. For simplicity, setting the term ( 𝛿 gA) 1 1− 𝜙 , which is constant, equal to 1, the above equation corresponds to the equation A = R&Dφ, where 𝜑 ≡ 𝜆 1 − 𝜙 . 1 3 Economic Change and Restructuring (2024) 57:33 Page 7 of 31 33 that if more R&D is oriented more toward energy-saving technologies than toward laboror capital-saving technologies, it is more likely that R&D will contribute to long-term reductions in energy intensity. Unfortunately, data that allow the construction of proxies for R&D in laboror capital-saving technologies and/or R&D in energy-saving technologies are not available for a large number of countries, particularly developing countries.8 It is therefore not possible to quantify the relative amounts of R&D in energy-saving technologies and R&D in laboror capital-saving technologies in developing countries, and hence to assess a priori whether R&D in developing countries, in general, is oriented more toward energy-saving technologies than toward laboror capital-saving technologies. What can be said, however, is that the vast majority of worldwide R&D activity takes place in industrial nations.9 If R&D by industrial countries generates technologies that save more energy than those generated by R&D in developing countries, then it is possible that foreign R&D performed in industrial countries contributes more to reductions in energy intensity in developing countries through international R&D spillovers than domestic R&D. However, to the extent that R&D performed in industrial countries generates technologies that cannot be adapted to local conditions in developing countries, it may contribute less to reductions in domestic energy intensity than domestically performed R&D. Thus, the effects of domestic and foreign R&D on domestic energy intensity in developing countries are an empirical question. 3 Empirical model andvariable definitions We begin by taking natural logarithms of both sides of Eq.1. Then, we introduce country and time subscripts i and t, and add an error term εit. Additionally, we include country fixed effects ci to control for any unobserved time-invariant country characteristics. We also control for effects of unobserved time-varying common factors ρFt, which, if left uncontrolled, can induce cross-sectional dependence in the 8 OECD data (available at https:// stats. oecd. org/ Index. aspx? DataS etCode= GERD_ TORD) on total public and private energy R&D expenditures are available for only 25 countries, and all these countries have short and/or incomplete time series. The International Energy Agency reports data on government spending on energy R&D for 32 countries (available at https:// www. iea. org/ dataandstati stics/ dataprodu ct/ energytechn ologyrdand-dbudgetdatab ase-2), but data on total public and private energy R&D expenditures are reported for only three countries. 9 According to data from the UNESCO Institute for Statistics (available at available at http:// data. uis. unesco. org/ Index. aspx? DataS etCode= SCN_ DS), high-income countries accounted for about 68% of the total worldwide R&D in 2015 (the last year of our sample period), whereas middleand low-income countries were responsible for about 32% of worldwide R&D expenditures. Economic Change and Restructuring (2024) 57:33 1 3 33 Page 8 of 31 regression error and lead to inconsistent estimates. Our basic empirical model is thus given by where log EIit is the log of energy intensity in developing country i in year t, and log R&Dit represents the log of R&D effort. We estimate one specification with (the log of) domestic R&D effort in developing countries, log R&Dd it , and five other specifications with foreign R&D, which takes place in developed countries. The first of these five specifications is used to examine whether R&D spillovers from developed to developing countries occur through disembodied channels such as scientific journals, international conferences, and the internet. Following, among others, Coe etal. (1997) and Herzer (2022), we define the measure of foreign R&D spillovers in this specification as the log of the sum of the R&D efforts of N developed countries, log R&Df t , where R&Dd jt is the R&D effort of industrial country j. It is perhaps needless to say that the two country groups do not overlap. To estimate the impact of import-related R&D spillovers from North to South on energy intensity, we use four other specifications with further spillover variables. One of these spillover variables is log R&Df_T it , which, following the weighting scheme of Coe and Helpman (1995),10 is the log of the weighted average of the domestic R&D efforts of the N developed countries, with bilateral shares of (total) imports as weights, where IMT ijt stands for imports of total goods of developing country i from developed country j and IMT it denotes imports of total goods of country i from all N industrial countries, IM T it = N ∑ j = 1 IMT ijt. Coe and Helpman (1995) use total imports as their weighting factor and find evidence of spillovers from foreign R&D to domestic TFP in a sample of OECD countries. (2) log EIit = 𝛼log R&Dit + ci + 𝜌Ft + 𝜀it (3) log R&D f t≡log N ∑ j = 1 R&D d jt (4) log R&D f_T it ≡log N ∑ j = 1 IM T ijt IMT it R&Dd jt 10 Lichtenberg and van Pottelsberghe de la Potterie (1998) argue that the weighting scheme of Coe and Helpman (1995) is sensitive to a potential merger between countries, and suggest an alternative weighting scheme that is less sensitive to the level of aggregation. While Lichtenberg and van Pottelsberghe de la Potterie (1998) find that their weighting scheme yields somewhat better empirical results than the Coe and Helpman (1995) weighting scheme, Coe etal. (2009) find that the weighting scheme of Coe and Helpman (1995) performs somewhat better than the Lichtenberg and van Pottelsberghe de la Potterie (1998) scheme. We repeated the analysis using the latter scheme and found qualitatively similar results (available on request). 1 3 Economic Change and Restructuring (2024) 57:33 Page 15 of 31 33 Table 2 Sample countries and their classification during the period 1995–2015 World Bank classification IMF classification UNTAD classification Argentina Middle-income country Emerging market or developing economy Developing economy Armenia Middle-income country Emerging market or developing economy Developing economy Brazil Middle-income country Emerging market or developing economy Developing economy Bulgaria Middle-income country Emerging market or developing economy China Middle-income country Emerging market or developing economy Developing economy Colombia Middle-income country Emerging market or developing economy Developing economy Costa Rica Middle-income country Emerging market or developing economy Developing economy Croatia Middle-income country Emerging market or developing economy Czech Republic Middle-income country Emerging market or developing economy Ecuador Middle-income country Emerging market or developing economy Developing economy Egypt, Arab Rep Middle-income country Emerging market or developing economy Developing economy Estonia Middle-income country Emerging market or developing economy Hungary Middle-income country Emerging market or developing economy India Low-income country Emerging market or developing economy Developing economy Iran, Islamic Rep Middle-income country Emerging market or developing economy Developing economy Kazakhstan Middle-income country Emerging market or developing economy Developing economy Latvia Middle-income country Emerging market or developing economy Lithuania Middle-income country Emerging market or developing economy Mexico Middle-income country Emerging market or developing economy Developing economy Mongolia Low-income country Emerging market or developing economy Developing economy Panama Middle-income country Emerging market or developing economy Developing economy Poland Middle-income country Emerging market or developing economy Romania Middle-income country Emerging market or developing economy Russian Federation Middle-income country Emerging market or developing economy Slovak Republic Middle-income country Emerging market or developing economy South Africa Middle-income country Emerging market or developing economy Developing economy Tajikistan Low-income country Emerging market or developing economy Developing economy Thailand Middle-income country Emerging market or developing economy Developing economy Economic Change and Restructuring (2024) 57:33 1 3 33 Page 16 of 31 A third advantage associated with cointegration is that it implies long-run Granger causality in at least one direction (Granger 1988).16 Here we assume—and test the assumption—that long-run causality runs from log R&Dit to log EIit. Finally, a fourth advantage is that endogeneity does not lead to inconsistency in the regression coefficients in the presence of cointegration. However, although even the standard fixed-effects estimator is (super) consistent under panel cointegration even when the regressors are endogenous, it suffers from a second-order asymptotic bias due to endogeneity and serial correlation, and, as a consequence, its usual standard errors are not correct. Therefore, we use the panel DOLS of estimator of Kao and Chiang (2001), which allows for endogenous regressors and which has been shown to perform well in samples like the one used here (see, e.g., Kao and Chiang 2001; Wagner and Hlouskova 2009).17 A country is classified as a “middle-income country” [“low-income country”] if it is officially categorized as such by the World Bank in its “historical classification by income” (available at https:// datah elpde sk. world bank. org/ knowl edgeb ase/ artic les/ 906519) for more than half of the calendar years between 1995 and 2015. The World Bank classifies lowand middle-income countries as “developing countries” (World Bank 2007). A country is classified as an “emerging market or developing economy” if it is listed in the category “emerging market and developing economies” by the IMF in its World Economic Outlook reports (available at https:// www. imf. org/ en/ Publi catio ns/ WEO) for the years 2004 onwards. All countries classified as emerging markets or developing economies were previously categorized as either “developing countries” or “countries in transition” in the World Economic Outlook reports. A country is classified here as a “developing economy” if it is listed by UNCTAD as such in its classification (available at https:// uncta dstat. unctad. org/ en/ class ifica tions. html) Table 2 (continued) World Bank classification IMF classification UNTAD classification Trinidad and Tobago Middle-income country Emerging market or developing economy Developing economy Tunisia Middle-income country Emerging market or developing economy Developing economy Türkiye Middle-income country Emerging market or developing economy Developing economy Ukraine Middle-income country Emerging market or developing economy Developing economy Uruguay Middle-income country Emerging market or developing economy Developing economy 17 The DOLS method employs a parametric correction for endogeneity and serial correlation, based on lead, lag, and current values of the differenced regressors. An alternative estimation method for estimating cointegrating relationships is the (panel) FMOLS estimator, which uses a semi-parametric correction for endogeneity and serial correlation (based on the OLS residuals and the first differences of the regressors). Simulation evidence suggests that the DOLS estimator performs better than the FMOLS estimator in small samples (see, e.g., Kao and Chiang 2001; Wagner and Hlouskova 2009). Therefore, we prefer the DOLS estimator. The results (available on request) do not change qualitatively when the FMOLS estimator is used. 16 The concept of long-run (Granger) causality is to be distinguished from the more familiar notion of “Granger causality,” which refers to short-run forecastability and does not account for long-run causality through the error correction term in a cointegrated error-correction model. 1 3 Economic Change and Restructuring (2024) 57:33 Page 17 of 31 33 Table 3 Correlation matrix and summary statistics Log EIit (EIit) log R&Dd it ( R&Dd it ) log R&Df t log R&Df_T it log R&Df_M it mi T it ×log R&D f_T it mi M it ×log R&D f_M it A. Sample of 33 developing countries (according to World Bank classification) log EIit 1.000  log R&Dd it 0.183 1.000  log R&Df t − 0.256 − 0.028 1.000  log R&Df_T it − 0.412 0.024 0.293 1.000  log R&Df_M it − 0.365 0.024 0.292 0.986 1.000  mi T it ×log R&D f_T it − 0.202 − 0.491 0.003 − 0.122 − 0.149 1.000  mi M it ×log R&D f_M it − 0.049 − 0.386 0.412 − 0.192 − 0.193 0.578 1.000 Mean 11.810 (155,407.5) 7.669 (8899.64) 22.828 20.791 20.807 3.631 0.546 Maximum 13.113 12.740 23.002 22.101 22.093 19.136 3.807 Minimum 10.895 2.555 22.456 19.666 19.865 0.309 0.013 Std. Dev 0.468 1.818 0.134 0.594 0.545 2.969 0.554 B. Sample of the 15 developing countries that were used as sources of R&D spillovers Log EIit 1  log R&Dd it 0.254 1 Mean 11.631 (118,977.4) 9.931 (52,937.06) Maximum 12.402 13.05 Minimum 10.656 7.140 Std. Dev 0.332 1.308 Economic Change and Restructuring (2024) 57:33 1 3 33 Page 18 of 31 It should, however, be noted that the panel DOLS estimator is based on the assumption of error cross‐sectional independence.18 To account for weak cross-sectional dependence in the residuals of the DOLS models, we follow Bordo etal. (2017) and use Driscoll and Kraay (1998) standard errors; these standard errors are robust to heteroskedasticity, autocorrelation, and spatial correlation. To account for strong cross-sectional error dependence, we demean the data by subtracting the cross-sectional averages from the data and then use the demeaned data in place of the original data (which is equivalent to using the residuals from regressions of each variable on time dummies). In addition, to ensure that our results do not suffer from error cross-sectional dependence due to common factors, we test for strong cross-sectional dependence in the residuals from our DOLS regressions using the cross-sectional dependence (CD) test of Juodis and Reese (2022).19 5 Results 5.1 Main results Panel A of Table4 shows the panel DOLS results for the relationship in our sample between each of the R&D variables and the log of energy intensity. In Panel A, we also present the results of the Juodis–Reese test for strong cross-sectional dependence in the residuals of the DOLS regressions. The results of several panel cointegration tests are shown in Panel B. Regarding the results in Panel B, two things should be noted. First, for the Pedroni (1999, 2004) panel cointegration tests, which assume error cross-sectional independence, we report test statistics based on the demeaned data to control for error cross-sectional dependence. For the Gengenbach etal. (2016) test, which explicitly accounts for cross-sectional dependence via the use of cross-sectional averages of the variables, we report test statistics based on the raw data. Second, error-correction-based cointegration tests such as the Gengenbach etal. (2016) test incorporate in the alternative hypothesis the assumption that the dependent variable is not weakly exogenous with respect to the independent variables. Rejection of the null of no cointegration using an error-correction model with Δlog EIit as the dependent variable can therefore be interpreted as evidence that log EIit is not weakly exogenous to the independent variables and thus that the independent variables “cause” log EIit (provided that they are significant). 19 We use the Juodis and Reese (2022) test rather than the standard Pesaran (2021) test because the latter has no power to detect error cross-sectional dependence when the estimated models include time dummies (or cross-sectional averages) or are based on demeaned data. The Juodis and Reese (2022) test is a modified version of the Pesaran (2021) test that does not suffer from this problem. 18 We also experimented with the pooled common correlated effects (PCCE) estimator and the common correlated effects mean group (CCEMG) of Pesaran (2006). Both these estimators are specifically designed to account for error cross-sectional dependence. While the results from the PCCE estimator are significant (and negative) only for log R&Df t , the results from the CCEMG estimator are significant (and negative) only for log R&Dd it . Given, however, that these estimators are designed for large N and large T and that they require strictly exogenous regressors, the PCCE and CCEMG estimates are not reliable here due to the possibility of endogenous or weakly exogenous regressors and/or the relatively small number of countries and years in our sample. 1 3 Economic Change and Restructuring (2024) 57:33 Page 19 of 31 33 Turning to the results in column (1) of Table4, we find a weakly significant positive effect of domestic R&D on energy intensity. This effect appears not to be due to the presence of strong cross-sectional dependence in the residuals, as suggested by the Juodis–Reese test. Since, however, two of the cointegration tests do not reject the null of no cointegration, it cannot be ruled out with certainty that the observed effect is the result of spurious regression. We come back to this point when we discuss the results in column (1) of Table6. In column (2) of Table4, we see that while four of the five tests indicate cointegration, the coefficient on the log of the unweighted sum of the R&D expenditures of industrial countries is positive and statistically insignificant. However, the Juodis–Reese test indicates the presence of strong cross-sectional dependence in the DOLS residuals, and thus the estimation results should be viewed with caution. In column (3), the coefficient on log R&Df_T it is positive but insignificant, and only one test rejects the null hypothesis of no cointegration. We thus find no long-run evidence of uninteracted spillover effects of import-weighted foreign R&D expenditures using total imports as weights. For completeness, however, it should be said that we cannot rule out the possibility that the insignificant coefficient is the result of unobserved common factors in the DOLS residuals (as suggested by the Juodis–Reese test). Our evidence also does not support the existence of an uninteracted effect of capital goods import-weighted foreign R&D expenditures on the long-run level of domestic energy intensity. Column (4) shows that log R&Df_M it has a significant negative coefficient and that the Juodis–Reese test is insignificant, as in columns (1), (5), (6), and (7). However, none of the tests rejects the null hypothesis of no cointegration, suggesting that the regression result is spurious. Similarly, the coefficient in column (5) for the variable mi T it ×log R&D f_T it is negative and significant, but only two tests provide clear evidence (at the conventional 5% level or better) of cointegration. Thus, there is also no clear support for an interacted spillover effect of total import-weighted foreign R&D expenditures on the long-run level of domestic energy intensity. In contrast, the results in column (6) show clear evidence that R&D performed in industrial countries weighted by the bilateral share of machinery and equipment imports from the industrial countries reduces energy intensity in developing countries through its interaction with the machinery and equipment import share in developing countries’ GDP. All tests indicate cointegration between mi M it ×log R&D f_M it and log EIit at least at the 5% level; the Gengenbach etal. (2016) test suggests that log EIit is „caused‟ in the long run by mi M it ×log R&D f_M it 20; and the coefficient on mi M it ×log R&D f_M it is negative and statistically significant with a value of − 0.025. 20 We also computed the Gengenbach et al. (2016) t test statistic using the reverse regression with Δ mi M it ×log R&D f_M it on the left-hand side. The value of the test statistic is − 1.813, implying that the null of weak exogeneity cannot be rejected for mi M it ×log R&D f_M it . Weak exogeneity implies long-run Granger non-causality (see, e.g., Hall and Milne 1994). Thus, the evidence that mi M it ×log R&D f_M it is weakly exogenous and log EIit is not weakly exogenous means that mi M it ×log R&D f_M it has a long-run (causal) effect on log EIit, whereas log EIit has no long-run effect on mi M it ×log R&D f_M it . Economic Change and Restructuring (2024) 57:33 1 3 33 Page 20 of 31 Table 4 Main results (1) (2) (3) (4) (5) (6) A. DOLS estimates  log R&Dd it 0.030* (0.016)  log R&Df t 0.017 (0.029)  log R&Df_T it 0.028 (0.080)  log R&Df_M it − 0.312*** (0.073)  miT it ×log R&D f_T it − 0.027** (0.008)  mi M it ×log R&Df _ M it − 0.025*** (0.008) Juodis–Reese (p value) 0.234 0.000 0.050 0.338 0.776 0.866 No. of countries 33 33 33 33 33 33 No. of obs 531 532 532 532 532 532 Adjusted R20.951 0.951 0.951 0.954 0.951 0.951 B. Panel cointegration tests Pedroni (1999, 2004) Panel PP t-statistic − 1.045 − 5.496*** − 1.125 0.073 − 1.385* − 2.159** Panel ADF t-statistic − 1.270 − 6.540*** − 0.214 − 0.000 − 1.460* − 1.671** Group PP t-statistic − 1.955** − 5.360*** − 1.120 0.411 − 3.645*** − 2.312** Group ADF t-statistic − 4.058*** − 7.755*** − 0.988 − 0.686 − 3.197*** − 2.806*** Gengenbach etal. (2016) ECM t-statistic − 4.107*** − 1.764 − 4.202*** − 2.105 − 2.428 − 3.558*** 1 3 Economic Change and Restructuring (2024) 57:33 Page 21 of 31 33 The dependent variable in the DOLS regressions and the Pedroni (1999, 2004) tests is log EIit. The dependent variable in the Gengenbach etal. (2016) tests is Δlog EIit. All regressions (and tests) include country fixed effects. The DOLS regressions were estimated with one lead and one lag of the first-differenced regressors. The data on log EIit, log R&Dd it , log R&Df_T it , mi T it ×log R&D f_T it , log R&Df_M it , and mi M it ×log R&D f_M it for the DOLS regressions and the Pedroni (1999, 2004) tests were demeaned to account for (strong) error cross-sectional dependence due to unobserved common factors; log R&Df t is the same for each country and can be considered as an observed common factor (that cannot be demeaned). The Gengenbach etal. (2016) test accounts for strong error cross-sectional dependence via the use of cross-sectional averages. Juodis–Reese is the test for strong cross-sectional dependence of Juodis and Reese (2022) applied to the residuals from the DOLS regressions. The number of lags in the Pedroni (1999, 2004) (PP and ADF) and Gengenbach etal. (2016) tests was determined using the general-to-specific approach with a maximum of two lags. Two lags of the cross-sectional averages were included in the Gengenbach etal. (2016) tests. All tests reject for large negative values. The Pedroni (1999) test statistics are distributed as standard normal. The Gengenbach etal. (2016) critical value for one regressor at the 1% (5%) [10%] significance level is − 2.735 (− 2.601) [− 2.530] for N = 30. Numbers in parentheses are Driscoll and Kraay (1998) heteroskedasticity autocorrelation spatial correlation robust standard errors. *** (**) [*] indicate significance at the 1% (5%) [10%] level Table 4 (continued) Economic Change and Restructuring (2024) 57:33 1 3 33 Page 22 of 31 To provide a sense of the magnitude of the effect implied by this coefficient, consider that a one standard deviation increase in mi M it ×log R&D f_M it is associated with a decrease of 10.12 percent of a standard deviation in the energy intensity variable (− 0.025 × 1.977/0.4885), an effect that is economically significant. 5.2 Robustness checks In columns (1)–(6) of Table5, we check the robustness of our results with respect to the use of the sample of developing countries classified by UNCTAD. The results are very similar to those in Table4. The only worth mentioning differences are that the coefficients on log R&Dd it and log R&Df_M it are now insignificant, and that the coefficient on mi T it ×log R&D f_T it is significant only at the 10% level. Thus, the results in columns (1)–(6) of Table 5 once again suggest that domestic R&D does not contribute to reductions in energy intensity in developing countries. Furthermore, foreign R&D does not appear to affect domestic energy intensity through disembodied channels. Instead, we again find that foreign R&D conducted in developed countries reduces energy intensity in developing countries through imports, particularly imports of machinery and equipment, and that this effect depends on the share of machinery and equipment imports in developing countries’ GDP. In column (7) of Table5, we present results for the relationship between domestic R&D conducted in the 15 industrial source countries of R&D spillovers and energy intensity in those countries, as a plausibility check. Four of the five cointegration tests suggest that there is a long-run relationship between log R&Dd it and log EIit, and the DOLS coefficient on log R&Dd it is negative and highly significant. Thus, we find evidence that domestic R&D contributes to reductions in energy intensity in developed source countries of foreign R&D spillovers, which supports the plausibility of our finding that there are significant spillovers from R&D conducted in industrial countries that reduce energy intensity in developing countries. In Table 6, we once again use the sample of 33 developing countries, classified according to the World Bank, and assess the robustness of our results to various specifications involving multiple R&D variables in columns (1)–(6). In column (1), we report DOLS results of a regression that includes both log R&Dd it and mi M it ×log R&D f_M it . The coefficient on mi M it ×log R&D f_M it remains negative and statistically significant, and the coefficient on log R&Dd it is positive and significant at the 10% level, like in column (1) of Table4. However, the evidence for cointegration between log R&Dd it , mi M it ×log R&D f_M it , and log EIit in column (1) of Table6 is weaker than the evidence for cointegration between mi M it ×log R&D f_M it and log EIit in column (6) of Table4. If (as discussed in Sect.4.3) there is an integrated regressor that is not cointegrated with other cointegrated variables in an equation, the residuals of such an equation will tend to be non-stationary, and the evidence of cointegration may therefore be weak (or even absent). Thus, the results of the cointegration tests in column (1) of Table6 together with those in column (6) of Table4 can be interpreted as an indication that while there is a long-run relationship between mi M it ×log R&D f_M it and log EIit, there is no long-run relationship between 1 3 Economic Change and Restructuring (2024) 57:33 Page 23 of 31 33 Table 5 Results based on the subsample of developing countries classified by UNCTAD (columns (1) – (6)) and results using the source countries of R&D spillovers as the sample (column (7)) (1) (2) (3) (4) (5) (6) (7) A. DOLS estimates  log R&Dd it 0.010 (0.012) − 0.136*** (0.033)  log R&Df t 0.016 (0.033)  log R&Df_T it 0.027 (0.073)  log R&Df_M it − 0.042 (0.059)  mi T it ×log R&D f_T it − 0.022* (0.012)  mi M it ×log R&D f_M it − 0.023*** (0.007) Juodis–Reese (p value) 0.199 0.312 0.002 0.153 0.456 0.766 0.466 No. of countries 21 21 21 21 21 21 15 No. of obs 324 325 325 325 325 325 270 Adjusted R20.950 0.950 0.950 0.950 0.951 0.951 0.977 B. Panel cointegration tests Pedroni (1999, 2004) Panel PP t-statistic − 1.574* − 4.261*** − 0.896 − 0.039 − 0.671 − 3.902*** − 0.816 Panel ADF t-statistic − 1.772** − 4.823*** − 1.237 − 0.139 − 1.450* − 2.505*** − 2.447*** Group PP t-statistic − 1.099 − 3.979*** − 1.093 − 1.041 − 2.361*** − 4.287*** − 2.911*** Group ADF t-statistic − 2.463*** − 5.375*** − 0.375 − 1.138 − 1.943** − 2.808*** − 6.153*** Gengenbach etal. (2016) ECM t-statistic − 1.869 − 1.886 − 1.745 0.320 − 2.172 − 3.394*** − 2.805** Economic Change and Restructuring (2024) 57:33 1 3 33 Page 24 of 31 Table 5 (continued) The dependent variable in the DOLS regressions and the Pedroni (1999, 2004) tests is log EIit, The dependent variable in the Gengenbach etal. (2016) tests is Δlog EIit. All regressions (and tests) include country fixed effects. The DOLS regressions were estimated with one lead and one lag of the first-differenced regressors. The data on log EIit, log R&Dd it , log R&Df_T it , mi T it ×log R&D f_T it , log R&Df_M it , and mi M it ×log R&D f_M it for the DOLS regressions and the Pedroni (1999, 2004) tests were demeaned to account for (strong) error cross-sectional dependence due to unobserved common factors; log R&Df t is the same for each country and can be considered as an observed common factor (that cannot be demeaned). The Gengenbach etal. (2016) test accounts for strong error cross-sectional dependence via the use of cross-sectional averages. Juodis–Reese is the test for strong cross-sectional dependence of Juodis and Reese (2022) applied to the residuals from the DOLS regressions. The number of lags in the Pedroni (1999, 2004) (PP and ADF) and Gengenbach etal. (2016) tests was determined using the general-to-specific approach with a maximum of two lags. Two lags of the cross-sectional averages were included in the Gengenbach etal. (2016) tests. All tests reject for large negative values. The Pedroni (1999) test statistics are distributed as standard normal. The Gengenbach etal. (2016) critical value for one regressor at the 1% (5%) [10%] significance level is − 2.796 (− 2.653) [− 2.530] for N = 20. The Gengenbach etal. (2016) critical value for one regressor at the 5% significance level is − 2.698 for N = 15. Numbers in parentheses are Driscoll and Kraay (1998) heteroskedasticity autocorrelation spatial correlation robust standard errors. *** (**) [*] indicate significance at the 1% (5%) [10%] level 1 3 Economic Change and Restructuring (2024) 57:33 Page 31 of 31 33 Huang J, Chen X (2020) Domestic R&D activities, technology absorption ability, and energy intensity in China. Energy Policy 138:111184 Huang J, Hao Y, Lei H (2018) Indigenous versus foreign innovation and energy intensity in China. Renew Sustain Energy Rev 81:1721–1729 Huang J, Lai Y, Hu H (2020) The effect of technological factors and structural change on China’s energy intensity: evidence from dynamic panel models. China Econ Rev 64:101518 Juodis A, Reese S (2022) The incidental parameters problem in testing for remaining cross-section correlation. J Bus Econ Stat 40:1191–1203 Juselius K (2006) The cointegrated VAR model: methodology and applications. Oxford University Press, Oxford Kao C (1999) Spurious regression and residual-based tests for cointegation in panel data. J Econom 90:1–44 Kao C, Chiang M (2001) On the estimation and inference of a cointegrated regression in panel data. In: Baltagi BH, Fomby TB, Carter Hill R (eds) Nonstationary panels, panel cointegration, and dynamic panels (advances in econometrics), vol 15. Emerald Group Publishing Limited, Bingley, pp 179–222 Koetse MJ, de Groot HLF, Florax RJGM (2008) Capital-energy substitution and shifts in factor demand: a meta-analysis. Energy Econ 30:2236–2251 Lichtenberg FR, van Pottelsberghe de la Potterie BP (1998) International R&D spillovers: a comment. Eur Econ Rev 42:1483–1491 Madsen JB, Saxena S, Ang JB (2010) The Indian growth miracle and endogenous growth. J Dev Econ 93:37–48 Metcalf GE (2008) An empirical analysis of energy intensity and its determinants at the state level. Energy J 29:1–26 Mimouni K, Temimi A (2018) What drives energy efficiency? New evidence from financial crises. Energy Policy 122:332–348 Pedroni P (1999) Critical values for cointegration tests in heterogeneous panels with multiple regressors. Oxf Bull Econ Stat 61:653–670 Pedroni P (2004) Panel cointegration: asymptotic and finite sample properties of pooled time series tests with an application to the PPP hypothesis. Econom Theory 20:597–625 Pedroni P (2007) Social capital, barriers to production and capital shares: implications for the importance of parameter heterogeneity from a nonstationary panel approach. J Appl Econom 22:429–451 Pesaran MH (2006) Estimation and inference in large heterogeneous panels with a multifactor error structure. Econometrica 74:967–1012 Pesaran MH (2007) A simple panel unit root test in the presence of cross-section dependence. J Appl Econom 22:265–312 Pesaran MH (2021) General diagnostic tests for cross-sectional dependence in panels. Empir Econ 60:13–50 Sadorsky P (2013) Do urbanization and industrialization affect energy intensity in developing countries? Energy Econ 37:52–59 Stern DI, Kander A (2012) The role of energy in the industrial revolution and modern economic growth. Energy J 33:125–152 UNCTAD (2023) Classifications. Retrieved September 28, 2023, from Classifications https:// uncta dstat. unctad. org/ en/ class ifica tions. html Wagner M, Hlouskova J (2009) The performance of panel cointegration methods: results from a large scale simulation study. Econom Rev 29:182–223 Wang D, Han B (2017) Effects of indigenous R&D and foreign spillover on energy intensity in China. J Renew Sustain Energy 9:03590 World Bank (2007) World development report 2007: development and the next generation. World Bank, Washington DC World Bank (2012) World development report 2012: gender equality and development. World Bank, Washington DC Yu H (2012) The influential factors of China’s regional energy intensity and its spatiall inkages: 1988– 2007. Energy Policy 45:583–593 Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.