The Trade-Off Between Economic Performance and Environmental Quality: Does Financial Inclusion Asymmetrically Matter for Emerging Asian Economies?

This study examines the role of nancial inclusion on the environment-economic performance in the top ve Asian emerging economies. The data used for empirical investigation covers the time period from 1995 to 2019. Financial inclusion is measured through bank branches, bank credit, and insurance premiums. To check long-run associations, the panel-ARDL approach has been employed for empirical analysis. The empirical evidence conrms the signicant associations between nancial inclusion-GDP nexus and nancial inclusion-CO2 nexus. The ndings show that bank branches and bank credit have a signicantly positive impact on economic growth and CO2 emissions in the long-run. However, insurance premium has no impact on economic growth but it exerts a signicant negative impact on carbon emissions in the long-run. Furthermore, energy consumption is highly sensitive to economic growth and carbon emissions. The study delivers imperative points for pollution eradication and attaining sustained economic growth. There is a need for government-level efforts to align the targets of nancial inclusion with economic growth and environmental policies.


Introduction
It was the early 1990s when the term nancial exclusion came to the fore and pointed out the inadequate amount of bank branches and the limited accessibility to these branches as a big hurdle in a more liberal, dynamic, and vibrant nancial sector (European Commission, 2008). Previously, the term nancial exclusion was used to de ne obstacles to access the primary nancial services and goods from the point of view of both users of these services (demand side) and producers of these services (supply side) (Rahim et al., 2009). To promote nancial inclusion demand-side works side by side the supply side. Poverty is the main factor that hinders nancial inclusion because if a majority of people are living below the poverty line they don't have enough savings to deposit in the bank accounts. Likewise, if the pace of the economy is slow then the level of investment in the economy is also sluggish resulting in low demand for loans and other nancial services. The tendency to save more can shift the poor people from low-income brackets to higher ones thus increasing their role in banks and nancial institutions which can cause the nancial services to push upwards (Reserve Bank of India, 2013). Reserve bank of India (RBI) de nes nancial inclusion as, "the process of ensuring access to appropriate nancial products and services needed by vulnerable groups such as weaker sections and low-income groups at an affordable cost fairly and transparently by mainstream institutional players." the period 1995-2019. The structure of the study is as follows. Section two describes the data and methods followed by the results and discussions in section three. In section four, we conclude the study.

Methods And Data
To capture the impact of nancial inclusion on economic growth and CO2 emissions in Asian emerging economies, we have borrowed the following long-run model from Van et al. (2021) and Zaidi et al. (2021).
In the above models (1) and (2), GDP per capita (GDP) and carbon emissions (CO 2 ) are taken as the dependent variables and among the independent variables, nancial inclusion (FI) is included as the main variable, while energy consumption (EC), Trade, and population are control variables in our analysis.
The model discussed above is a long-run model and to get short-run estimates, as well, we describe this model in the form of error correction format. In doing so, we rely upon a method that gives estimates of long-run effects with short-run effects in a single step as follows: considered genuine only if they are co-integrated and the co-integration among the variables is con rmed through the negative and signi cant estimate attached to ECM t-1 . To get the estimate of ECM t-1 rst, we generate a series of residuals labelled as ECM by using equations (3 & 4). We then replace the lagged value of this series (ECM t-1 ) in equations (3 & 4) in place of the lagged-level variables and estimate the new equation with the same number of lags as used originally. The size of the estimate attached to ECM t-1 describes the speed of adjustment towards long-run equilibrium. This method has the advantage that it can estimate e ciently for the small number of observations. Moreover, this technique can take care of the integrating properties of the variables i.e. we should not worry about whether the variable is stationary at level or rst difference because it can accommodate the mixture of variables with I(0) and I(1).
Both tests have used their own tabulate new critical values for testing. The Hausman test is used to con rm the ARDL-PMG or ARDL-PM models are su cient for this empirical analysis. In the end, we check causality in a non-linear framework by conducting the panel causality test of Hatemi-J (2012).

Data
For empirical investigation, data has been taken for the period ranging from 1995 to 2019 for the top ve emerging economies of Asia including China, India, Japan, Indonesia, and Turkey. Table 1 provides a discussion on complete de nitions of variables, their abbreviations, and descriptive analysis of data. Data on GDP per capita, carbon emissions, energy use, trade, and population growth is sought from the World Bank. However, data on bank branches, bank credit, and insurance premiums are sourced from IMF. GDP per capita is measured at constant 2010 US$. Data on carbon emissions are measured in kilotons. Bank branches are taken as bank branches per 100,000 adults. Bank credit is measured as bank deposits in percentage. Data on life and non-life insurance premium volume to GDP (%) is taken to measure insurance premium. Data series on energy use is measured as kg of oil equivalent per capita. Trade is measured in percentage of GDP. The annual percentage of population growth is used to measure the population growth variable.

Empirical Results And Discussion
Panel-ARDL requires that none of the variables in the model is I(2) and the panel unit root tests tell us about the stationarity of our variables To that end, we have used panel unit root tests namely Levin, Lin and Chu (LLC), I'm, Pesaran and Shin (IPS) and ADF-Fisher. The results of the LLC show that most of the variables are stationary at rst difference except BC, Insurance, and POP. However, when we apply IPS and ADF tests all the variables are stationary at rst difference except the variable of POP. Table 2 ndings con rm that we can apply the panel-ARDL technique. As the frequency of our data is annual we have imposed a maximum of three lags and optimal lag selection is based on Akaike Information Criterion (AIC). Note: ***p<0.01; **p<0.05; and *p<0.1 After con rming the preliminary condition of Panel-ARDL we are now in a position to start the discussion on the estimates of our variables. Our dependent variables are GDP and CO2 emissions and we have used three different proxies of nancial inclusion; bank branches, bank credit, and insurance. For both GDP and CO2 models we have included all the proxies of nancial inclusion one by one. Table 3, shows the results of both in the short and long run. Moreover, cointegration tests and other diagnostics are also reported in table 3. First of all, we want to con rm whether our long-run results are cointegrated or not. Two tests of cointegration i.e. ECM t-1 and Kao con rm that our long-run estimates of GDP and CO2 are cointegrated meaning they are genuine or valid. Hausman test results have supported the panel ARDL-PMG model. First, we discuss the long-run results of the GDP and CO2 models in detail, and then the short-run results in brief.
The long-run estimates of BB and BC, in the GDP model, are positively signi cant and in the case of Insurance, the estimate is insigni cant. As the variables are taken in the log form we can explain them by saying that a 1% increase in the bank branches and bank credits facilities improve the GDP by 0.021% and 0.271%. The estimate of bank credit is large as compared to the estimate of bank branches suggesting that instead of the number of branches, improved credit facilities are more helpful in increasing the GDP of the economy. As the number of branches and credit facilities in an economy increases the production activities also increase due to the easy availability of loans and other nancial services for investment in large projects that can help the economy to grow at a great pace. Moreover, nancial inclusion connects a large number of people to the nancial system of the country that brings them into the mainstream economy which also helps in the development of the economy (Sharma, 2016).  The control variables EC and Trade helps the economy to grow as well -a 1% rise in EC improves the economic growth of the country by 0.853%, 1.739%, and 1.628% and a 1% rise in Trade improves the economic growth of the economy by 0.032%, 0.039%, and 0.012%. However, a 1% rise in the POP only improves the economic growth in the rst model by 0.458% whereas it is not statistically noticeable in the second and third models. Now we will discuss the long-run estimates of CO2 models. The estimated coe cients of BB and BC are positively signi cant, whereas the estimated coe cient of Insurance is signi cantly negative. In the elasticity form, we can elaborate these results by saying that a 1% rise in the bank branches and credit facilities by banks increases the CO2 emissions by 0.015% and 0.1417%. However, a 1% rise in Insurance decreases the CO2 emissions by 0.0181%. Theory suggests that nancial inclusion can affect the environment positively or negatively. Our results are suggesting that improved nancial inclusion due to the increased number of bank branches and credits facilities help the nancial sector to develop and grow which is considered as a driver in nurturing the economy due to the surge in the availability of production and consumption loans that also increase the energy demand and thus give rise to CO2 emissions (Frankel and Romer, 1999). On the other side, as the economy grows due to better nancial inclusion of the society more sophisticated and advanced technologies developed in the production process can help to reduce CO2 emissions. Similarly, the availability of credit facilities also speeds up the investment in renewable energy projects that also exert less burden on the environment. Banks provide individual loans for energy-e cient products such as LEDs, DC inverters, fuel-e cient cars, etc., besides banks also provide easy credit to the house owners for installing solar energy. The positive impact of nancial inclusion on CO2 emissions is supported by le et al. (2020), however, Renzhi and Baek (2020) found an inverted U-shape relationship between CO2 emissions and nancial inclusion.
The variable of energy consumption exerted a positive impact on the CO2 emissions in all the models by the amount of 1.259%, 1.992%, and 1.635%.
Conversely, a 1% rise in Trade reduces the CO2 emissions by 0.004% only in model six, whereas in models four & ve the impact of Trade is insigni cant. Finally, the estimated coe cient of POP (0.557%) is signi cant and negative in model four, while insigni cant in models ve & six. In the short run, the estimates in the GDP models are providing us with an inconclusive picture as most of them are insigni cant and appeared with mixed signs at most lags.
Similarly, the short-run estimates in all CO2 models are mostly insigni cant and provide inconclusive results. Table 4, provide the results of the Granger causality which con rm one-way causality running from GDP→BC, GDP→Insurance, CO2→BC, Insurance→CO2. However, bi-directional causality is found between GDP↔BB. For detailed results see Table 4.

Conclusion And Policy Implications
The objective of the study is to investigate the role of nancial inclusion on environmental quality and economic performance in the top ve emerging economies of Asia including China, India, Japan, Indonesia, and Turkey for the period 1995 to 2019. Bank branches, bank credit, and insurance premiums are used to measure nancial inclusion. The panel-ARDL method is employed for empirical investigation. It is found that long-run panel cointegration exists between the focused variables of the study. In the long-run, bank branches impact on economic and environmental performance is positive inferring that as the bank branches increase it leads to increase economic performance and pollution emissions. Bank credit also results in increasing economic growth and pollution emissions in the long-run. The impact of insurance premium on economic performance is statistically insigni cant revealing that there is no association between the insurance premium and economic growth in the long run. However, insurance premiums exert a signi cant negative impact on carbon emissions con rming that the increase in insurance premiums results in reducing carbon emissions in the long-run. It is also evident that energy consumption is positively associated with economic growth and pollution emissions in the long-run. Trade impact is positive on economic growth in all three models but this effect is negative on pollution emissions only in the insurance premium regression model in the long-run. Population growth has a signi cant impact on economic-environmental performance only in bank branches regression in the long-run. In the short-run, bank branches have a positive impact on carbon emission revealing that an increase in bank branches results in rising carbon emissions. Bank credit has no association with the economic-environmental performance nexus in the short-run. Insurance premium impact is negative on economic-environment performance in the short-run concluding that due to an increase in insurance premium economic growth and carbon emissions will decrease.
Authorities and policymakers of these economies should follow and embrace mitigation methods, including the adoption and installation of digital nancial inclusion in the future. Asian emerging markets should maintain sustainable development via nancial inclusion. Through nancial inclusion, the funds from the nancial institutions can be directed towards the projects of green and clean energy. Moreover, the funds should be transferred to those rms, businesses, and individuals who are involved in green innovations. The government should also articulate strict rules for the nancial institutions to lend nance for renewable energy and environmental-friendly projects, and this can be more fruitful through digital nancial inclusion. The governments should also remove barriers from digital nancial inclusion such as affordability, documentation, and trust.
The study has a limitation of the availability of data. The data on digital nancial inclusion is not available earlier than 2004. Therefore, we have not included digital nancial inclusion in the analysis. This study used only three variables of nancial inclusion based on usage and access to the formal nancial services factors, while many other factors were not considered in the analysis due to the unavailability of relevant data for the Asian emerging economies.
Future studies should also use other proxies of digital nancial inclusion. Future studies can be conducted on the same topic by covering more updated models and data. Future researchers may also analyze at the micro-level in the high-polluted economy.

Declarations
Ethical Approval: Not applicable Consent to Publish: Not applicable Authors Contributions: This idea was given by Liu Dong and Yuantao Xie. Muhammad Hafeez, Liu Dong, Yuantao Xie, and Ahmed Usman collected the data, computed data analysis and wrote the complete paper. While Liu Dong and Yuantao Xie read and approved the nal version.
Consent to Participate: I am free to contact any of the people involved in the research to seek further clari cation and information Funding: Not applicable.