1 Background of the Study

Anthropogenic greenhouse gas emissions, especially carbon dioxide (CO2), produced from economic and human activities have contributed immensely to global warming, and has become an international menace. Striking the balance between economic growth and environmental performance has been the catchphrase of most recent studies (Song et al. 2018). Although many scholarly discourses and parleys have recently been focused on environmental efficiency evaluations, there still exists more room for comprehensive studies to investigate the possible effects of economic growth on the environment to help in the conception and formulation of robust sustainability policies. The economy of Asia Pacific includes the ever-expanding and develo** nations like India and China, and plays pivotal roles in the increasing global warming. Recent reports indicate that five countries of the top ten highest emitters of CO2 in the world are in Asia (China, India, Japan, South Korea, and Indonesia) (Li et al. 2022; Mohmmed et al. 2019). The world’s fuel-based CO2 emission continued to escalate, with China being responsible for 27% of the world’s total greenhouse gases according to a Global Carbon Project report in 2021 (Global Carbon Project 2021). The Asia Pacific region has plans of transitioning to the use of renewable energy and electricity generation via nuclear and natural gas, however, fossil fuel, particularly coal burning continues to be the main source of carbon emissions (Droege 2011).

The increasing interest of people in the use of resources and its resultant eco-problems led to Fare et al.’s proposition of environmental efficiency (Färe et al. 1989). The concept is considered to be of theoretical value and practically significant. Environmental efficiency means simultaneously maximizing good outputs whiles minimizing the use of inputs such as energy, labor, etc., and bad outputs such as CO2 (Long et al. 2017). A number of studies have been conducted on environmental efficiency in Asia and other parts of the world including the effect of renewable energy consumption (Jiang et al. 2022), relationship between carbon dioxide emissions, financial development and energy consumption (Chunyu et al. 2021), and the correlation between ICT, renewable energy, financial development, and CO2 emissions (Batool et al. 2022). Previous studies have used the Malmquist Luenberger Productivity Index and Data Envelopment Analysis (DEA) to estimate environmental efficiency in China (Piao et al. 2019), as well as the SBM Metafrontier to estimate the effect of fertilizer usage intensity on environmental efficiency (Long et al. 2018). Using the SBM DEA model, Liu et al. (2020) evaluated environmental efficiency in the transport industry while other researchers have employed the Super efficiency DEA model to estimate the environmental efficiency of Asia Pacific and the Organization for Economic Co-operation and Development (OECD) economies respectively (Twum et al. 2021; Zhang et al. 2021). However, most of these studies were conducted from single country’s perspective, and the few that considered multi countries did not look at environmental efficiency convergence. Moreover, human capital was not explored as this study seeks to do.

Moreover, researchers have emphasized the importance of human capital in economic growth and environmental efficiency. Economic growth is basically driven by human capital accumulation over time and human capital is found to encourage environmental performance in Asia pacific (Twum et al. 2021). Empirical studies established that people with formal education showcase a more pollution-free behavior comparatively, and human capital is favorably influenced by climate change (Kurowski et al. 2022; Tolppanen et al. 2022). For example, in the investigation of the relationship between human capital and environmental quality, using a sample size of 3900 adults, a study found that persons who have higher education tend to grow more environmentally concerned over global warming and exhibit willingness to take environmental actions than those with none (Chankrajang and Muttarak 2017). Moreover, consumer behavior is positively driven by knowledge about climate change (Kurowski et al. 2022), thus human capital remains the primary source of innovation.

Literature has extensively discussed convergence in efficiency from a diverse perspective. Criado et al., (2011) confirmed convergence when they studied growth and pollution convergence in European countries within the period of 1980–2005. Convergent clubs, showing different energy intensities are identified among 193 cities within China (Zhu and Lin 2020), where they discovered that foreign direct investment (FDI) and industrial structure drive the formation of these convergent cubs. In the exploration of environmental efficiency and convergence in OECD economies, Camarero et al. (2013) used Nitrogen Dioxide, Carbon Dioxide, and Sulphur dioxide as proxies for environmental pollution, and observed improvement in environmental performance in the OECD countries. They also established a club convergence among the efficient economies. Xu et al. (2020), and studied the dynamic convergence of eco-efficiency in China. Through the application of the Super Efficiency SBM model as well as the dynamic spatial Durbin Model (DSDM), they found a positive autocorrelation in inter-regional Eco-efficiency and established steadiness in convergence over time as environmental efficiency narrowed (Long et al. 2017). A similar study investigated the convergence of environmental efficiency in over 104 countries from 1980 to 2016 and established conditional convergence based on globalization, energy price, and industrial structure (Sun et al. (2020). While Barassi et al. (2011) investigated the stochastic convergence of CO2 in OECD economies, other researchers have examined the same stochastic convergence in per capita CO2 emissions in other emerging economies (Awaworyi Churchill et al. 2020; Yirong 2022). Barassi et al. (2011) showed that CO2 emissions converge over time, although quite slowly. Long et al. (2017) assessed convergence from a sector base and analyzed the environmental efficiency of the cement producers in China, and confirmed convergence in the East, West, and Middle of China. This current study considers a 29-year period, and shows a possibility of convergence among the countries in the Asia Pacific region.

This study contributes to the literature on environmental efficiency based on a panel of 14 economies in the Asia Pacific region. The Asia Pacific has made attempts to transition into low-carbon economies, hence, the relevance in assessing environmental efficiency and convergence. Unlike extant studies on environmental efficiency, the novelties of this study are as follows. Firstly, this study analyzes the relationship between trade competitiveness and environmental efficiency. Secondly, the study combines stochastic, absolute β-convergence, and conditional β-convergence to explore the different convergence tendencies within the Asia Pacific regions. Finally, the effect of green technology on environmental efficiency convergence in the Asia Pacific will be analyzed, considering endogenous growth.

The knowledge gaps of this study are as follows. First, environmental efficiency is explored across the Asia Pacific region via integrating stochastic, absolute β-convergence, and conditional β-convergence. Stochastic convergence via panel unit root test can overcome the demerits of conventional β-convergence, which might have invalid inference. Furthermore, we also analyze how green technology and human capital impact environmental efficiency from the perspective of endogenous growth. Third, we also analyze effect of trade competitiveness impact environmental efficiency. Last, environmental efficiency is evaluated via metafrontier super EBM efficiency. Metafrontier has the merits to consider the heterogeneity of environmental technology. Super efficiency has the advantages of discriminating different DMUs with the same efficiency value. EBM can address the demerits of conventional CCR and SBM(slacks-based measure). CCR has the demerits of ignoring input excesses or output shortfalls. SBM has the demerits that cannot allow for the proportionate changes of inputs and outputs.

The remaining sections are demarcated as follows: section two captures the data and model, section three entails the methodology, whiles sections four and five feature the results, discussion, conclusion, and possible policy recommendations based on findings.

2 Data and Model

2.1 Data

Following existing literature on environmental efficiency and according to availability, data is collected from 1990 to 2018 and the study region is divided into a Whole panel, South Asia, East Asia, Southeast Asia, and Oceania. A total of five variables, of which three, (labor, capital and energy) and two, (GDP and CO2) were used as input, good and bad output respectively, in estimating environmental efficiency. Patent from WDI is used as a proxy for green technology (GT), trade competitiveness index (TC) is calculated by authors (Export–Import)/(Export + Import), industrial structure (indst) and urbanization (urb) from World Development Indicator whiles human capital (hc) is collected from Penn World Table. These variables are employed as conditions for environmental efficiency convergence.

2.2 Econometric Model

This paper examines the role of human capital in environmental efficiency convergence in 14 selected countries within the Asia Pacific region. The study also looks at the influencing factors of green technology, industrial structure, urbanization, and trade competitiveness on environmental efficiency. Applying the truncated regression as well as the Tobit regression, this study goes farther to examine the non-linear relationship between trade competitiveness and environmental efficiency, assuming that the correlation between the two would not always be linear.

Following Extant studies by Mankiw et al. (1992), Peng and Wang (2005), Long et al. (2017), the absolute β-convergence analysis can be econometrically indicated by the formula (1):

$$\Delta EE_{{it}} = \alpha _{ + } \beta _{1} EE_{{i(t - 1)}} + \mu _{{it}}$$
(1)

where the regressor indicates CO2 emissions during the period \(t\) against the base year. A negative co-efficient of the base year presupposes a narrow deviation in environmental efficiency, hence the indication and confirmation of absolute β-convergence. Meanwhile, the \({u}_{it}\) indicates the random disturbance.

A lot of studies have shown different ways of measuring the individual impacts of a panel data, Luo et al. (2020) used quantile regression to estimate convergence, however, Islam (1995), Peng et al. (2005), Long et al. (2017) posits that the use of fixed effect for absolute β-convergence estimation is most preferable especially for the analysis of differing times and places or the amalgamation of the two thence. Moreover, the fixed and random effects have been widely used and the differences between these two have been clearly defined. The random effect seeks to establish a relationship between the unobservable individual effects and the exogenous variables. Following suit, the pooled OLS regression is used in the estimation of the conditional β-convergence. The econometric model for the representation of the conditional convergence is written below as formula (2):

$$\begin{gathered} \Delta EE_{{it}} = \alpha _{ + } \beta _{1} EE_{{i(t - 1)}} + + \beta _{2} Trade{\text{ }}Competitiveness_{{it}} + \beta _{3} GreenTechnology_{{it}} \hfill \\ \quad \quad \quad + \beta _{4} Human{\text{ }}Capital_{{it}} + \beta _{5} Industry_{{it}} + \beta _{6} Urbanizationi_{{it}} + ~\mu _{{it}} \hfill \\ \end{gathered}$$
(2)

Here, the differential environmental efficiency is the regressor and conditional β-convergence is achieved if the coefficient of the lagged environmental efficiency is negative. Trade competitiveness represents the Asia Pacific’s (APAC) strengths in competition and ability to trade in goods on the global market under free and fair conditions, green technology represents APAC’s ability to apply technology in energy production and consumption to enhance energy performance for achieving environmental efficiency. Human capital is a representation of APAC’s education and skills towards economic growth and green behavior. Industrialization is the embodiment of APAC’s economic activities and urbanization represents APAC’s compact city structure and consumption.

3 Methodology and Data

3.1 Metafrontier Super Epsilon-Based Measure

DMUs tend to have different technology level, institution. If DMUs are assessed w.r.t the same production frontier, environmental efficiency will has estimate biases. Conventional DEA might neglect environmental technology heterogeneity. Metafrontier can overcome the demerits of Conventional DEA considering environmental technology heterogeneity. When DMUs have the same efficiency value, we cannot compare them. Super efficiency can address the shortcomings of CCR model, discriminateing different DMUs with the same value (Andersen and Petersen 1993). Tone and Tsutsui (2010) proposed an Epsilon-based measure to address the shortcomings of previous radial and non-radial DEA methods. Thus we use a meta-frontier super EBM measure to analyze the environmental efficiency, considering the heterogeneity of environmental technology. This approach allows for greater flexibility in assessing environmental efficiency and estimating technological gaps for countries using a variety of technologies in comparison to the world's total technology. Broadstock et al. (2016), Long et al. (2018), Zhang et al. (2021) have all applied the Meta-Frontier in estimating environmental efficiency in different empirical scenarios. Following Tone and Tsutsui (2010) and Long et al. (2018) the linear programming of EE with respect to (w.r.t.) non-oriented metafrontier super EBM can be represented in formula (3):

$${\text{Min}}\frac{{\theta - \varepsilon _{{\text{x}}} \mathop \sum \nolimits_{{{\text{i}} = 1}}^{{\text{M}}} \frac{{{\text{w}}_{{\text{i}}}^{{\text{x}}} {\text{s}}_{{\text{i}}}^{ - } }}{{{\text{x}}_{{{\text{i}}0}} }}}}{{\varphi + \left( {\varepsilon _{{\text{y}}} \mathop \sum \nolimits_{{{\text{r}} = 1}}^{{\text{S}}} \frac{{{\text{w}}_{{\text{i}}}^{{\text{y}}} {\text{s}}_{{\text{r}}}^{ + } }}{{{\text{y}}_{{{\text{r}}0}} }} + \varepsilon _{{\text{b}}} \mathop \sum \nolimits_{{{\text{t}} = 1}}^{{\text{K}}} \frac{{{\text{w}}_{{\text{i}}}^{{\text{b}}} {\text{s}}_{{\text{t}}}^{ - } }}{{{\text{b}}_{{{\text{t}}0}} }}} \right)}}$$

subject to

$$\begin{gathered} \mathop \sum \limits_{{{\text{j}} = 1}}^{{\text{J}}} \mathop \sum \limits_{{{\text{n}} = 1,{\text{n}} \ne {\text{k}}}}^{{{\text{N}}^{{\text{j}}} }} \lambda _{{\text{x}}} = \theta {\text{x}}_{{\text{i}}} - {\text{s}}_{{\text{i}}}^{ - } {\text{~}}, \hfill \\ \mathop \sum \limits_{{j = 1}}^{J} \mathop \sum \limits_{{n = 1,n \ne k}}^{{N^{j} }} \lambda y = \varphi y_{0} + s_{r}^{ + } ~, \hfill \\ \mathop \sum \limits_{{{\text{j}} = 1}}^{{\text{J}}} \mathop \sum \limits_{{{\text{n}} = 1,{\text{n}} \ne {\text{k}}}}^{{{\text{N}}^{{\text{j}}} }} \lambda {\text{b}} = \varphi {\text{b}}_{0} - {\text{s}}_{{\text{r}}}^{ - } {\text{~}}, \hfill \\ \mathop \sum \limits_{{{\text{j}} = 1}}^{{\text{J}}} \mathop \sum \limits_{{{\text{n}} = 1,{\text{n}} \ne {\text{k}}}}^{{{\text{N}}^{{\text{j}}} }} {\text{u}}_{{\text{n}}}^{{\text{j}}} = 1{\text{~}}, \hfill \\ {\text{u}}_{{\text{n}}}^{{\text{j}}} \ge 0, \hfill \\ {\text{s}}_{{\text{r}}}^{ + } \ge 0,{\text{~s}}_{{\text{i}}}^{ - } \ge 0,{\text{~s}}_{{\text{t}}}^{ - } \ge 0 \hfill \\ \end{gathered}$$
(3)

where \(\lambda\) and \(u\) are the respective weights assigned to DMUs under group-frontier and meta-frontier, respectively. The technical frontier, according to the group-frontier model, is limited by resource utilization efficiency, economic trend, and technological level (Hayami, 1969; Hayami and Ruttan, 1970). It is possible to speculate on the nature of sub-technologies that suggest a group of firms' production capabilities. Formula (4) denotes linear programming with respect to non-oriented super EBM under group frontier.

$${\text{Min}}\frac{{\theta - \varepsilon_{x} \mathop \sum \nolimits_{i = 1}^{M} \frac{{w_{i}^{x} s_{i}^{ - } }}{{x_{i0} }}}}{{\varphi + \left( {\varepsilon_{y} \mathop \sum \nolimits_{r = 1}^{S} \frac{{w_{i}^{y} s_{r}^{ + } }}{{y_{r0} }} + \varepsilon_{b} \mathop \sum \nolimits_{t = 1}^{K} \frac{{w_{i}^{b} s_{t}^{ - } }}{{b_{t0} }}} \right)}}$$

subject to

$$\begin{gathered} \mathop \sum \limits_{{n = ,{\text{n}} \ne {\text{k}}}}^{N} \lambda y = \varphi y_{0} + s_{r}^{ + } \hfill \\ \mathop \sum \limits_{{n = 1,{\text{n}} \ne {\text{k}}}}^{N} \lambda b = \varphi b_{0} - s_{r}^{ - } \hfill \\ \sum \lambda = 1, s_{r}^{ + } \ge 0, s_{i}^{ - } \ge 0, s_{t}^{ - } \ge 0 \hfill \\ \end{gathered}$$
(4)

where both oriented and non-oriented EBM models combine the merits of radial and non-radial models. \(\sum \lambda =1\) indicates VRS.

3.2 Absolute β-Convergence

This research incorporates the estimation of β-convergence of environmental efficiency with the intent to regress the percentile increase of environmental efficiency on the primary level. A β- convergence is achieved when a negative β, significant and vary from zero is recorded. The concept of β-convergence here indicates that countries that start off with lower efficiency scores are most likely to experience a higher growth and outrun those countries that begun with a higher efficiency score. β-convergence is typified by the two major types namely the conditional and unconditional β-convergence. The unconditional β- convergence suggests that those countries with a low initial growth will be able to catch up with those with higher initial levels at a stable level (Tables 1, 2, 3).

Table 1 Summary of Variables
Full size table
Table 2 Descriptive Statistics of data
Full size table
Table 3 Non-Parametric Test of environmental efficiency w.r.t.meta frontier and group frontier
Full size table

3.3 Conditional β-Convergence

The conditional β-convergence, on the other hand, exhibits the ability of different subgroup countries to converge at a certain level as according to country-specific conditions.

3.4 Stochastic Convergence

LLC (2002) assumes that the individual panel exhibits similar first order auto correlation coefficient, which is one of its disadvantages. Im et al. (2003) role of renewable energy consumption(2003) recommends a robust unit root test that is able to run for different dynamic panel per individual unit root, to correct the flaws with LLC (2002). The IPS posits that only a part of the series can be stable in terms of the alternate hypothesis as against the LLC. It is presented as formula (5):

$$\Delta {y}_{it}={a}_{oi}+{\rho }_{i}{y}_{it}+{\varepsilon }_{it}$$
(5)

Here also \({\rho }_{i}\) represents the coefficient AR (3), which is dynamic across ρi individuals, very distinct with the LLC (2002). The null hypothesis is hence:

\({H}_{0}:{\rho }_{i}=0\) for all \(i\)

The null hypothesis of IPS test is with the assumption that the entire series are not stable, meaning that the countries can diverge. The alternate is hereby written as:

\({H}_{i}:{\rho }_{i}<0\) for \(i=\) 1,2,…,N

The alternative hypothesis of IPS test says that a part of the individual processes can be stable, meaning that there can be convergence. In this study, these unit roots are employed to analyze the panel. The IPS, LLC, Augmented Dickey Fuller (ADF) will be tested to determine whether environmental efficiency converges or diverges.

4 Results and Discussion

4.1 Environmental Efficiency Comparison

There have been some arguments as to whether or not there is any distinction between environmental efficiency (Meta frontier) and environmental efficiency (Group Frontier). For the purpose of this study, it is imperative to determine the difference and so two non-parametric tests have been used. The Wilcoxon rank-sum test is a parametric test that uses the mean of variables to decipher whether the two variables are the same. According to the result from the Wilcoxon rank-sum test, the null hypothesis is rejected whiles the alternate is accepted that the median for the two variables is significantly different. The Kolmogorov–Smirnov tests use the distribution of the two datasets to determine whether they are the same. According to the Kolmogorov–Smirnov test, the null hypothesis is rejected whiles the alternate hypothesis is accepted that the distribution for both groups is different. The results hereby verify that there is a difference between environmental efficiency (Meta frontier) and environmental efficiency (group frontier).

Figure 1 below shows the environmental efficiency in each of the selected countries within the Asia Pacific in years, according to the Meta Frontier. From 1990 to 2000 to 2010 and 2018, showcasing 10 years intervals and the results show that every 10 years, environmental efficiency keeps increasing in some of the countries, probably due to the reduction in the use of fossil fuel and the use of green technologies as well as innovations. This notwithstanding, environmental efficiency also decreased in many of the countries especially, China and India probably due to their emerging economies and their high use of energy for industrial purposes.

Fig. 1
figure 1

Environmental efficiency w.r.t. meta frontier of the Asia Pacific countries

Full size image

In 1990, the most efficient countries were Australia, Bangladesh, Indonesia, Japan, Pakistan, the Philippines, and Thailand. Others like India, Korean Republic, New Zealand, Singapore, and Vietnam were inefficient. In the year 2000, environmental efficiency fluctuated and Malaysia, Korean Republic, and Singapore which hitherto, were inefficient, recorded the most efficient scores of 1.00 whiles environmental efficiency reduced in Bangladesh, Australia, Thailand, Pakistan, and the Philippines. China, however, increased slightly to 0.99 although, still remained inefficient. In 2010, China improves drastically and becomes efficient together with Bangladesh, Indonesia, Japan, Korean Republic, New Zealand, Vietnam, the Philippines, and Thailand. Environmental efficiency is yet again reduced in Singapore and Malaysia. In 2018, as indicated by the turquoise blue, most of the countries were inefficient, save Indonesia, Japan, Thailand, and Vietnam.

The environmental efficiency of the individual countries in the Asia Pacific in years, according to the Group frontier is also shown in Fig. 2 above. The results indicate that as of 1990, China was the most inefficient with an average environmental efficiency value of 0.44. Japan is the most efficient country of the year with an average EE of 1.17, followed by New Zealand with 1.01. The dark grey color represents the efficient countries. In the space of 10 years, in the year 2000, the results indicate that most of the countries were inefficient although there had been an improvement in the values. Only six out of the 14 countries, namely; Australia, Indonesia, Japan, the Korean Republic, and Thailand were efficient with Bangladesh as the most efficient with the highest environmental efficiency value of 6.32. In 2010, there had been an improvement in environmental efficiency as indicated by the dark grey and black colors. Eight of the countries were efficient whiles the remaining six namely; Bangladesh, Indonesia, Malaysia, New Zealand, Singapore, and Thailand were inefficient. In 2018, however, nine of the countries had become efficient as indicated by the black and dark blue colors. The turquoise blue represents the inefficient countries which include China, New Zealand, Pakistan, and Vietnam.

Fig. 2
figure 2

Environmental efficiency w.r.t. group frontier of the Asia Pacific countries

Full size image

4.2 Convergence Analysis

4.2.1 Absolute β-Convergence

The absolute β–convergence analysis seeks to examine the likelihood that, countries with initial low environmental efficiency scores can grow to catch up with countries with higher environmental efficiency scores in due course. We then run absolute β–convergence under the Group frontier environmental efficiency and the Meta Frontier environmental efficiency. The results in Table 4 below show that there is absolute β-convergence in the whole panel as well as in all the regions under the group frontier. It is important to note that the signs preceding the coefficient are a representation of the initial environmental efficiency, hence the results hereby establish that there is a tendency of a “catch-up” between the initial low environmental efficiency and inefficient countries in the whole panel as well as by regions.

Table 4 Absolute β-convergence Analysis w.r.t. group frontier
Full size table

Table 5 below is the result of the Meta Frontier environmental efficiency and it shows a similar trend with that of the Group Frontier. The signs in the panel and in all the regions are negative and statistically significant, which means there is a tendency of absolute β- convergence.

Table 5 Absolute β-convergence Analysis w.r.t. meta frontier
Full size table

4.2.2 Conditional β-Convergence

The purpose of the conditional β-convergence analysis is to determine whether environmental efficiency can finally achieve a state of steadiness in the presence of the aforementioned control variables as well as to examine them as determinants of environmental efficiency growth. The results of the truncated and Tobit regressions showed similar results, exhibiting the robustness of the results (see Table 6). The lagged environmental efficiency is negative and significant in the main panel and the in rest of the regions. This means that environmental efficiency converges. Southeast Asia has the highest convergence tendency at (−0.755), whiles Oceania has the lowest convergence tendency at (−1.225). Green technology improves environmental efficiency in the whole panel and the rest of the regions save South Asia and Oceania where it hampers environmental efficiency. Human capital promotes environmental efficiency in the panel and in all regions. Human capital has a significant role in climate change mitigation because educated people tend to be more responsive to environmental regulations. Therefore, in countries where people have attained a certain level of education, it is likely their actions will directly or indirectly influence environmental efficiency. As it was highlighted in the results, advancing the human capital level has the tendency to promote environmental regulation. This in effect can also increase environmental efficiency in those countries. Industrial structure significantly promotes environmental efficiency in the whole panel and in Oceania but reduces environmental efficiency in the rest of the regions. l The industrial structure of a country promotes economic growth and propels an economy towards social and economic modernization.

Table 6 Conditional β-Convergence of environmental efficiency w.r.t meta frontier
Full size table

Industrialization is a catalyst for high energy consumption which is a possible reason why it hampers environmental efficiency in East, South, and Southeast Asia. However, the type of energy used can help improve environmental efficiency, hence the mixed effects of industrial structure on environmental efficiency. Muhammad et al. (2022) found industrial structure to have a mixed effect on environmental efficiency. Urbanization improves environmental efficiency in South Asia and Oceania but causes a deterioration in environmental efficiency in the panel and in the rest of the regions. Meanwhile, trade competitiveness reduces environmental efficiency in the whole panel and in Oceania and advances environmental efficiency in Southeast, East, and South Asia. Freer and expanded trade tend to have an adverse effect on environmental performance. When countries participate in competitive deregulation to weaken environmental protection, trade openness and competitiveness may result in a global drop in environmental standards. Less stringent environmental regulations in a country skew relative production costs among trading partners, resulting in comparative advantage in the manufacture of polluting commodities, leading to a specialization in the export of those goods. This heightens exporting pollution, hence the result shows that trade competitiveness reduces environmental efficiency in some of the regions, although it also promotes environmental efficiency in most of the regions. This confirms the findings of Hu et al. (2018) that expanding the trade scale could serve as an effective way of reducing carbon emissions, thereby narrowing the convergence gap between sectors and countries. Trade on its own could be a driver for clean production (Forslid et al. 2018).

The results for the group frontier are similar to that of the Meta Frontier. The lagged environmental efficiency is also negative and significant, which is a confirmation of environmental efficiency convergence in the main panel and all the regions. Green technology improves environmental efficiency except in the main panel and in Oceania where it reduces environmental efficiency. Human capital negatively impacts environmental efficiency in South and Southeast Asia but increases environmental efficiency in the rest of the regions. Industrial structure has a mixed impact on environmental efficiency as it positively affects environmental efficiency in all the regions except Oceania and East Asia. Urbanization also has both negative and positive effects on environmental efficiency. Trade competitiveness also reduces environmental efficiency in all the regions save Oceania (see Table 7).

Table 7 Conditional β-Convergence of environmental efficiency w.r.t group frontier
Full size table

4.2.3 Stochastic Convergence

Mathematically, stochastic convergence acts on the theory that a series of largely random or erratic measures can be predicted to settle into a pattern from time to time. When there is stationarity in a trend and β-convergence at the same time, stochastic convergence is achieved. The inference here is that the shocks in relative environmental efficiency would be but momentary, moreover, once a country achieves an initial efficiency score above the mean, the growth rate should eventually be negative in order to achieve convergence. With regards to stochastic convergence, the Levin–Lin–Chu (LLC) unit root test is applied to determine whether the environmental efficiency of the APAC regions will converge or diverge. The Im-Pesaran-Shin unit root test is as well-used as robust. Important to note is that, the presence of unit root in the data will be an indication and confirmation of environmental efficiency divergence. On the other hand, the absence of unit roots indicates environmental efficiency convergence. Stochastic convergence is discussed under the group frontier and the Meta Frontier. The results are presented in Table 8 below:

Table 8 Stochastic Convergence of environmental efficiency
Full size table

The p values are less than 1% in the Whole panel, East Asia and Oceania, and less than 10% in South Asia under both the Meta Frontier and Group Frontier in the LLC and IPS. This means that environmental efficiency converges from 1990 to 2018 in these regions. However, environmental efficiency diverges in Southeast Asia under Meta Frontier but converges under the Group Frontier as the probability values were greater than 10% and less than 1% respectively, in both the LLC and IPS. According to the coefficients, the convergence steadiness tendency is highest in South Asia, followed by Southeast Asia. East Asia recorded the lowest coefficient in both the Meta Frontier and Group Frontier, which means that the convergence tendency is very low in the region. Oceania recorded the highest coefficient under the Meta Frontier which also means that the convergence tendency from 1990 to 2018 is high.

4.3 Robust Analysis of CO2

In order to assess the validity of the results of the convergence of environmental efficiency, the data was re-analyzed the other way around. Below in Fig. 3, is a map showing carbon emissions in the selected Asia Pacific countries. The results indicate that there is an increase in carbon emission in all 14 countries in this decadal analysis. This means that every ten years, CO2 emission keeps rising. The dark color indicates the countries with the highest emission and in all the years thus 1990, 2000, 2010, and 2018, China is recorded as the highest emitter of CO2, followed by Korea. India and Japan then follow with the grey color. CO2 emission, however, reduced in Japan in 2010 and 2018 as the dark grey turns to turquoise blue. The turquoise blue indicates the countries with lower CO2 emissions and all the decades shown, emissions in Australia, New Zealand, Singapore, Indonesia, Pakistan, and Malaysia have remained relatively lower.

Fig. 3
figure 3

Carbon Emission of the Asia Pacific countries in years

Full size image

Placing CO2 beside environmental efficiency as the dependent variable, the results are revealed in Table 8 below. The results show that the lagged CO2 is negative and therefore a confirmation of convergence of CO2 in South Asia, Southeast Asia, and Oceania although they are all not significant. This also confirms the lagged environmental efficiency results. This means that there is absolute β-convergence.

Table 9 above shows the results for the conditional β-convergence and the findings are in tandem with the environmental efficiency convergence results. Whiles carbon emission increases, environmental efficiency decreases. According to the robust analysis, environmental efficiency decreases as carbon emission increases as the result shows the two to be moving in opposite directions. CO2 and trade competitiveness converge at all levels. Green technology and trade competitiveness and human capital all reduce CO2 emission in the whole panel whiles industrialization and urbanization have mixed influence on CO2 emission.

Table 9 Conditional β–Convergence of CO2 emission
Full size table

5 Conclusion and Policy Implications

5.1 Summary

This paper mainly studies the convergence of environmental efficiency across the Asia Pacific region. First, environmental efficiency is explored across the Asia Pacific region. Stochastic convergence via panel unit root test can overcome the demerits of conventional β-convergence with potential invalid inference. Furthermore, we also analyze how green technology and human capital impact environmental efficiency from the perspective of endogenous growth. Third, we also analyze effect of trade competitiveness impact environmental efficiency. Furthermore, environmental efficiency is evaluated via metafrontier super EBM efficiency. Metafrontier has the merits to consider the heterogeneity of environmental technology. Super efficiency has the advantages of discriminating different DMUs with the same efficiency value. EBM can address the demerits of conventional CCR and SBM(slacks-based measure). CCR has the demerits of ignoring input excesses or output shortfalls. SBM has the demerits that cannot allow for the proportionate changes of inputs and outputs. Last, robust analysis on CO2 convergence is also conducted.

Stochastic convergence revealed confirmation of environmental efficiency convergence in all the regions from 1990 to 2018, except for Southeast Asia which diverged under the Meta Frontier but converged under the Group frontier. East Asia recorded the lowest coefficient value indicating a lower convergence tendency in the region whiles Oceania recorded the highest coefficient value, which also is an implication of a higher convergence tendency from 1990 to 2018. Also, countries with low initial environmental efficiency growth established a significant tendency to catch up with those with high initial environmental efficiency growth as the absolute β-convergence was confirmed in all the regions under both the Meta Frontier and Group Frontier. Results for conditional β-convergence revealed that Trade competitiveness has a positive impact on environmental efficiency but decreases environmental efficiency in Oceania and in the main panel. Human capital has a mixed impact on environmental efficiency under the group frontier but improves environmental efficiency under the meta frontier. Urbanization, however, reduces environmental efficiency at all levels whiles industrialization has a positive but insignificant relationship with environmental efficiency at all levels except in East Asia which was significant at 10%.

Under the Group Frontier, the lagged environmental efficiency was negative and significant in all regions and in the panel. Trade competitiveness reduces environmental efficiency in all regions except in Oceania. Green technology has a mixed effect on environmental efficiency.Human capital negatively impacts environmental efficiency only in South Asia. Industrialization has a mixed impact on environmental efficiency. Urbanization also has a mixed impact on environmental efficiency albeit not significant. Trade competitiveness has both negative and positive impacts on environmental efficiency. Using CO2 as the dependent variable to check robustness, the results indicated a continuous rise in carbon emission in all the countries, although the highest in China. The absolute β-convergence was confirmed in all the regions whiles conditional β-convergence indicated that human capital, and green technology reduced CO2 in all regions. Urbanization, Industrialization, and Trade competitiveness, however, were also found to increase CO2.

5.2 Policy Implications

Upon the aforementioned findings, the following recommendations are given as enumerated below.

  1. 1.

    The economies of Asia Pacific, especially the Southeast Asia Pacific region, are encouraged to invest more in human capital because a literate nation will understand the complexities and implications of environmental inefficiencies. More education means more innovative technology developments.

  2. 2.

    Environmental regulations must be strengthened even under international competition. The kind of commodities for production and trade must be green inspired in order to optimize environmental performance.

  3. 3.

    Secondly, with the relentless pursuit of economic growth, production and industrialization play a key role in reaching environmental sustainability goals. It is essential for policymakers to boost industrial structures, combined with strong environmental regulations, nurture and direct them towards green industrialization. This we believe will boost the speed of environmental efficiency convergence.

  4. 4.

    Policymakers should lay down necessary and effective procedures to control the inevitability of urbanization and its resultant effects on the environment. In this study, urbanization had a mixed effect on the convergence of environmental efficiency. This is suggestive that when people move from the rural areas into the urban cities, in as much as resources are excessively exploited, education and skills will also increase, and innovation is more likely to be developed because people will have access to a better life and education, and environmental efficiency can improve.

  5. 5.

    Green technology should be optimized and encouraged. Policymakers need to transition from the use of primary energy, coal and fossil fuel, into green technologies and convergence will be achieved. The results strongly emphasized on green technology as a potent driver of high environmental performance.

5.3 Limitation of the Study

This study has mainly focused on the human capital and trade competitiveness in achieving environmental efficiency convergence in 14 Asia Pacific countries, while establishing the non-linear relationship between the variables. However, due to lack of available data, other Asia Pacific countries could not be explored.

5.4 Future Direction of the Study

A study such as this, is of much relevance in studying the environmental performance and convergence of other areas in the world. Hence, this study could be extended to other regions such as Europe and Africa for the purpose of comparison. Also, the data used was collected from pre-covid period and if data from post-covid-19 period could be factored in future study, there could be fresh and more interesting findings.