A. Poverty

Poverty has declined significantly world-wide during the last two decades and its reduction has been especially remarkable in Asia. In the East Asia and Pacific region, the poverty headcount ratio at $1.9/day in 2011 PPP went from 53.7 percent in 1993 to 3.6 percent in 2013 (Table 1). As stated in Zhuang, Kanbur, and Rhee (2014) and in Aoyagi and Ganelli (2015), the progress in poverty reduction in developing Asia was faster than in any other region, at any other time in recorded history.

^{Table 1.}

Extreme poverty headcount ratio ($1.9/day poverty line).

Source: World Bank, PovcalNet. Note: The headcount ratio is the percentage of the of population living in households with consumption or income per person below the poverty line and is based on the average monthly household per capita income or consumption expenditure from the survey in 2011 PPP.

^{Table 1.}

Extreme poverty headcount ratio ($1.9/day poverty line).

Region	Headcount, in percent
	1993	2013
East Asia and Pacific	53.7	3.6
South Asia	44.9	16.1
Sub-Saharan Africa	59.7	42.4
Latin America and the Caribbean	14.0	4.6
Middle East and North Africa	7.0	2.6
Europe and Central Asia	5.2	1.6
World Total	34.0	11.2

Source: World Bank, PovcalNet. Note: The headcount ratio is the percentage of the of population living in households with consumption or income per person below the poverty line and is based on the average monthly household per capita income or consumption expenditure from the survey in 2011 PPP.

View Table

^{Table 1.}

Extreme poverty headcount ratio ($1.9/day poverty line).

Region	Headcount, in percent
	1993	2013
East Asia and Pacific	53.7	3.6
South Asia	44.9	16.1
Sub-Saharan Africa	59.7	42.4
Latin America and the Caribbean	14.0	4.6
Middle East and North Africa	7.0	2.6
Europe and Central Asia	5.2	1.6
World Total	34.0	11.2

Source: World Bank, PovcalNet. Note: The headcount ratio is the percentage of the of population living in households with consumption or income per person below the poverty line and is based on the average monthly household per capita income or consumption expenditure from the survey in 2011 PPP.

China has contributed the most to the poverty decline in East Asia and Pacific by lifting about 642 million people out of extreme poverty, with the poverty headcount ratio at $1.9/day declining from 56.6 percent to 1.9 percent between 1993 and 2013 (compared to a total decline of about 826 million of people in East Asia and Pacific between 1993 and 2013). Notable progress has been also achieved in ASEAN countries, where poverty has been sharply declining and, as a result, the percentage of population living on $1.9/day has fallen below the low- and middle-income countries’ average (Figure 1).

View Full Size

Poverty headcount ratios⁴ in selected ASEAN countries.

Citation: IMF Working Papers 2020, 118; 10.5089/9781513549194.001.A001

Source: World Bank, PovcalNet.Note: Based on data availability and contains missing data points.

• Download Figure
• Download figure as PowerPoint slide

Poverty has declined faster in countries that have started with higher poverty headcount ratios. For example, Vietnam’s extreme poverty headcount ratio declined from 53 percent in 1992 to 2 percent, while in Indonesia the decline was from 58.5 percent in 1993 to 6.5 percent in 2016. On the other hand, the Philippines saw a less marked poverty declining speed, from 13.9 percent in 2000 to 6.1 percent in 2015.

Despite the comparatively small percentage of people living below the extreme poverty line of $1.90 a day, less extreme poverty headcount ratios at $3.2 and $5.5 levels remain relatively high in some ASEAN countries. For instance, Indonesia and Philippines have poverty headcount ratios at the $3.2 and $5.5 per day lines higher than the average of low-and middle-income countries. Nearly 33 percent of the population were living below $3.20 per day and more than 60 percent were living below $5.50 per day respectively in Indonesia and Philippines in 2015. It should also be noted that, despite overall progress in reducing poverty, around 33 million of people were still living below the $1.90 per day extreme poverty line in ASEAN⁵ in 2015.

B. Income Inequality

Accompanying the dramatic decline of poverty headcount ratios, GDP per capita growth in ASEAN has been sustained and positive in the last two decades, except for two episodes in the aftermath of the Asian crisis and global financial crisis in the early 2000s and in 2009 respectively. In most recent years, GDP per capita growth in some ASEAN was exceeding the low- and middle-income countries average (Figure 2).

View Full Size

GDP per capita growth in selected ASEAN countries.

Citation: IMF Working Papers 2020, 118; 10.5089/9781513549194.001.A001

Data Source: World Bank.

• Download Figure
• Download figure as PowerPoint slide

Despite the steady income per capita growth and dramatic poverty reduction enjoyed by ASEAN countries, they have also experienced a significant increase in income inequality in recent decades. Balakrishnan et al. (2013) and Aoyagi and Ganelli (2015) showed that the increase in inequality as measured by the Gini index in Asia in the last two decades was worse than in other emerging markets. Aoyagi and Ganelli (2015) and Huang and Wan (2019) observed that, although the Gini coefficients in developing Asia are on average lower (i.e. they indicate a more equal income distribution) than those observed in sub-Saharan Africa, Latin America, and the Caribbean, developing Asia has experienced a higher growth rate of inequality compared to those other regions.⁶ Huang and Wan (2019) highlighted that about half of the Asian economies, covering around 80 percent of the region’s population, have experienced worsening income distribution as measured by the Gini index in the past two decades.

Figure 3 presents the change in Gini indexes for selected developing Asian countries from the 1990s to the most recently available data point, compared to a regional aggregate population-weighted measure. Notable progress in achieving more equal income distribution is observed in Malaysia, Philippines, and Thailand, while in China, Indonesia, and Lao PDR equality has worsened. Although in Indonesia Gini coefficient increased between early 1990s and 2013, it declined through 2017 and this trend is expected to continue (Doumbia and Kinda 2019). In addition, income inequality in Indonesia remains in line with the levels in other major Asian emerging economies.⁷ For Mongolia and Vietnam the Gini index shows only a slight improvement. As a result, the population-weighted average of the Gini index in developing East Asia and Pacific has increased from 33.5 percent in the 1990s to 38.4 percent in the 2010s.

View Full Size

Change in Gini Index in developing East Asia and Pacific.

Citation: IMF Working Papers 2020, 118; 10.5089/9781513549194.001.A001

Data Source: World Bank, authors’ calculation (regional aggregate) based on available data for Asian countries.

• Download Figure
• Download figure as PowerPoint slide

Figure 4 shows changes in income distributions in ASEAN, comparing the income share held by the richest and poorest 20 percent. Clearly, the income share held by the top 20 percent has not changed much over past decades and is exceeding 40 percent in Indonesia, Malaysia, Philippines, Thailand and Vietnam. While there was a notable decline in this share in Malaysia, Philippines, and Thailand; in Indonesia the share of the top 20 percent group has been higher in recent years than in the 1990s. Similarly, while in Malaysia, Philippines, and Thailand the bottom 20 percent group has recently received a higher income share than in the 1990s, in Indonesia and Vietnam the income share held by the bottom 20 percent has declined since 1990s.

View Full Size

Income distribution in ASEAN.

Citation: IMF Working Papers 2020, 118; 10.5089/9781513549194.001.A001

Data Source: World Bank, PovcalNet. Note: The percentage share of income is the share that accrues to subgroups of population indicated by quintiles. Total consumption expenditure is used to estimate the income share for Indonesia, Philippines, Thailand, Vietnam. Disposable income is used to calculate the income share for Malaysia.

• Download Figure
• Download figure as PowerPoint slide

Nevertheless, the developments reported show the importance of looking at inequality and growth issues in a unified way in ASEAN. This is consistent with the results by Ostry et al. (2014), who found that inequality is a robust determinant of the pace and duration of growth, and the reduction in inequality is associated with higher and more durable growth. In addition, IMF (2015) found an inverse relationship between the income share held by the top 20 percent group and economic growth. Specifically, the study found that an increase in the income share of the top 20 percent by 1 percentage point is associated with a 0.08 percentage point decline in GDP growth in the following 5 years. On the other hand, an increase in the income share of the bottom 20 percent by 1 percentage point is associated with a 0.38 percentage point increase in GDP growth in the following 5 years.

At the same time, social spending is relatively low in Asia, partly reflecting lower revenue collection, and, among other factors, insufficient and poorly targeted social policies are highlighted as bottlenecks to reduce inequality in Asia (see more details on inequality in Asia in Chandra et al. 2016).

A. Inclusive Growth Determinants

Despite the recent significant attention to the concept of inclusive growth, the literature on its determinants and possible policies to foster it remains limited. A number of studies has specifically focused on fiscal redistribution as a tool to equalize income through taxes and transfers. Earlier research has emphasized the tradeoff between “equality” and “efficiency” (Okun 1975) and, in this light, policies to improve equality via taxes have been long seen as dampening incentives to work and to invest, leading to declining economic activity. On the other hand, recent studies have suggested that well-designed fiscal redistribution policies could benefit the poor and foster growth through a number of channels, including improving human capital, health, infrastructure, and education (Benabou 2000; Garcia-Penalosa and Turnovsky 2007; Ostry et al. 2014; Aoyagi and Ganelli 2015).

Previous research suggest that monetary policy might have an impact on inclusive growth. Albanesi (2007) suggests that high inflation widens the gap between rich and poor, since the poor have little choice other than reducing their consumption during period of high inflation and/or unemployment. In this regard, monetary policy aiming at macro-stability can reduce income inequality by reducing inflation and minimizing output volatility, thus leading to improved conditions for the poor (Romer and Romer 1998; Aoyagi and Ganelli 2015).

A few studies have also focused on the impact of macro-structural reforms on growth inclusiveness. Zhuang et al. (2014) have highlighted that technological progress, globalization and market-oriented reforms have increased inequality in developing Asia while driving economic growth. Balakrishnan et al. (2013) suggest that recent economic growth in Asia has been less inclusive and less pro-poor despite decline falling poverty rate. According to them, fiscal policies targeting health, education, as well as labor market reforms and a more inclusive financial system can indeed broaden benefits of growth in Asia. Jain-Chandra et al. (2016) found that promoting structural reforms in Asia—such as broader access to education, health, and financial services—could have a substantial effect on reducing inequality.

In the remainder of this section, we report our contribution to the empirical analysis of the determinants of inclusive growth. We draw on the existing literature and extend it by considering structural inclusive growth drivers in addition to monetary and fiscal drivers. Two novel structural drivers of inclusive growth that we include in our analysis are female labor force participation (FLFP) and financial inclusion led by digitalization (proxied by the number of mobile cellular subscriptions per 100 people).

IMF (2019) has stressed that if women’s employment equaled men’s, economies would be more resilient and economic growth will be higher. However, female labor force participation in some ASEAN countries is much lower than the average for East Asia and Pacific. According to ILO estimates, in 2017 the ratio of female to male labor force participation rate stood at 63.5 percent in Indonesia, 65.5 percent in Malaysia, 61.5 percent in the Philippines, 78 percent in Thailand, and 88 percent in Vietnam, while the average ratio for East Asia and Pacific stood at 77.7 percent in 2017.⁹

The intuition for including digitalization is in line with IMF (2018a), which highlighted a positive association between GDP per capita and an index of usage of digitalization that includes six different indicators. Although Asia is experiencing rapid digitalization (IMF 2018), bank account ownership remains comparatively low in some Asian countries.¹⁰ In this context, mobile devices can serve as a useful mean of money transfer (Figure 6), hence, we use this variable as a proxy for financial inclusion.¹¹

View Full Size

Digital payments and mobile subscriptions in developing countries.

Citation: IMF Working Papers 2020, 118; 10.5089/9781513549194.001.A001

Note: Solid line is a linear trendline.Data Source: Global Findex Database 2017 by World Bank and International Telecommunication Union.

• Download Figure
• Download figure as PowerPoint slide

Another contribution of this research is that, we decompose our inclusive growth indicator into two components – income growth and equity growth – to analyze transmission channels of inclusive growth drivers. Finally, we apply our empirical results to scenario analysis of ASEAN countries’ reform plans.

B. Cross-Country Panel Model and Variables

To analyze the impact of monetary, fiscal and macro-structural reforms on inclusive growth in ASEAN we employ a standard empirical panel cross-country regression model. Our regression sample includes eleven Asian countries, of which six are ASEAN members: Indonesia, Malaysia, Philippines, Lao PDR, Thailand, and Vietnam. Other countries in the panel are Bangladesh, China, India, Mongolia, and Sri Lanka.¹² The period covered is 1992 to 2017.

Our dependent variable is the measure of inclusive growth used by Anand et al. (2013), which integrates average income per capita growth and the equity index growth. In other words, we define inclusive growth as income growth adjusted for an equity component.¹³ The equity index is based on estimates of income distribution based on the PovcalNet database (the official online repository of World Bank poverty data).¹⁴ The available income measure is based on net disposable income only for Malaysia, for other countries in our sample total per capita consumption expenditure data was used to approximate income, according to PovcalNet. Both measures are “net” in their nature and include public and private transfers.

As in Anand et al. (2013), our measure of inclusive growth $\left(\Delta {\overline{y}}^{*}\right)$ can be interpreted as a real income per capita growth $\left(\Delta \overline{y}\right)$ adjusted for the change in the equity index (Δω).¹⁵ This implies that inclusive growth $\mathrm{\Delta }{\overline{y}}^{*}$ is equivalent to per capita income growth $\mathrm{\Delta }\overline{y}$ if the income distribution is unchanged. When the distribution becomes more equal (unequal) $\mathrm{\Delta }{\overline{y}}^{*}$ shifts upwards (downwards). Under this definition, an increase in inclusive growth requires i) an increase in per capita income growth and ii) an increase in equity growth or a combination of both.¹⁶

Our proxy of inclusive growth is regressed on a set of monetary, fiscal and macro-structural variables which may potentially impact inclusive growth as discussed above. Our regressors include: the CPI inflation rate; fiscal redistribution, defined as the difference between market Gini and net (after tax) Gini¹⁷; female labor force participation (FLFP), defined as a percentage share of females who participate in the labor force as a share of total female population (age 15 and older); productivity growth (GDP per person employed); net FDI inflows as percentage of GDP; credit to private sector by banks in percent of GDP ratio, which is an indicator of financial deepening; the savings ratio, which is equal to net national savings as a percentage of GNI; an indicator aimed at capturing the positive impact of digitalization on financial inclusion, defined as the growth rate of mobile cellular subscriptions per 100 people.¹⁸ In our regression analysis we control for country and time fixed effects.

Our empirical cross-country model is specified as follows:

Where the dependent variable is the proxy of inclusive growth¹⁹ $$\Delta {\overline{y}}^{*}{}_{i,t}$$ in country i at time t. $${\overline{y}}_{i,t}$$ is the initial level of per capita PPP-adjusted income (log), x_i,t is the set of monetary, fiscal and structural determinants of inclusive growth described above over period t. u_i is a time-invariant unobserved country-specific effect, γ_t is an unobserved period effect that is common across countries. ν_i,t are errors, i. i. d. , that vary across countries i and time t. In the next subsection we empirically analyze the model and discuss the results.

In addition to assessing the impact of the regressors on the inclusive growth measure $$\mathrm{\Delta }{\overline{y}}^{*}$$, we also decompose our dependent variable into the income per capita component $$\mathrm{\Delta }\overline{y}$$ and the equity index component Δω. This allows us to analyze the impact of the regressors on both components of inclusive growth to clarify whether they affect inclusive growth through the growth of income per capita or by improvements in equality.

Important limitations of our research should be noted. First, as we already discussed above, the availability of income distribution data for developing Asia is limited and, for most of the countries in our sample, income distribution data reflects the distribution of consumption expenditure and not income itself. As explained in World Bank (2016), income-based measures of welfare indicate the potential set of goods and services that could be purchased by a household or an individual. On the other hand, consumption-based measures indicate the purchased amount of goods and services, meaning that consumption-based measures do not capture opportunities but rather realized outcomes.

According to World Bank (2016), there is also incomplete coverage of household surveys among top earners because of inadequately reported entrepreneurial and capital incomes. This might be problematic in inequality analysis as it underestimates the living conditions of the rich. In short, inequality of consumption might not be perfectly correlated with income inequality (consumption-based measure might deviate from that of net disposable income depending on households’ savings). Moreover, income is more volatile while consumption is much smoother (there are cases in which households declare zero income but positive consumption due to dissaving or to incorrectly measured consumption of home-produced goods²⁰) which, according to the World Bank (2016), makes consumption-based welfare a preferred aggregate in measuring poverty (where agriculture and informal sector play an important role) .

It is important to note that the inequality captured by consumption data might be less wide than that captured by income data. For example, consumption-based measures might suggest a decline in inequality if low-income groups consume more, but inequality might still be widening if measured based on income, in cases in which the income of higher-income groups increases as well and they save it rather than consuming it. Although the use of different variables to measure inequality may complicate the interpretation of the empirical results, since our inequality measure is predominantly based on consumption distribution²¹, it is likely that the degree of inequality remains underestimated, given that consumption tends to be smoother than income.

A second limitation of our research is that we are unable to include several potentially significant inclusive growth drivers due to limited data availability in terms of time span. Those include measures of accessibility to and quality of education, quality of overall infrastructure, accessibility to healthcare, financial inclusion, fintech, and other indicators. Investigating the possible impacts of such factors on inclusive growth if longer time series become available is a promising avenue for future research.

Our empirical analysis is potentially subject to endogeneity, which could be caused by reverse causality between our dependent variable (inclusive growth) and our regressors, as well as by omitted variables which might be correlated with variables included in the model. For the first case, potential variables which may cause reverse causality in the model are inclusive growth and the inflation rate. Although we expect that relatively lower inflation will positively contribute to inclusive growth, these relationships need to be empirically tested for endogeneity, otherwise endogeneity in fixed-effect estimation may produce inconsistent results. A possible way to detect reverse causality²² is by using Arellano–Bond estimator regression with instrumental variables. The results of the Arellano-Bond dynamic panel-data estimation and the panel regression excluding the inflation rate are both consistent with our main empirical results (Table 2) and are presented in Appendix, Table 1. Another way to test for endogeneity bias is to exclude suspect variables. In our case, even when we exclude the inflation rate, the results are consistent with our benchmark model (Appendix, Table 2). For the second case, which is endogeneity caused by omitted variable, the lagged dependent variable was included into the regression to capture unobserved effect (Appendix, Table 3), and the results are in accordance with our benchmark model (presented in Table 2).

^{Table 2.}

Determinants of Inclusive Growth, Panel Regression Results.

Source: Authors’ calculation. * denotes statistical significance at 10 percent level, ** denotes statistical significance at 5 percent level, *** denotes statistical significance at 1 percent level. Standard errors are in parentheses.

^{Table 2.}

Determinants of Inclusive Growth, Panel Regression Results.

	Dependent Variable
	$$\Delta {\overline{y}}^{*}$$ – inclusive growth	$$\mathrm{\Delta }\overline{y}$$ – income per capita growth	∆ω – equity index growth
Lagged GDP per capita	1.925** (0.92)	-0.899 (0.70)	2.824*** (0.64)
CPI inflation rate	-0.030 (0.03)	-0.016 (0.03)	-0.013 (0.02)
Redistribution	0.687*** (0.19)	0.402* (0.15)	0.285** (0.13)
Female labor force participation	0.268*** (0.06)	0.203*** (0.044)	0.066* (0.04)
Productivity growth	0.850*** (0.04)	0.886*** (0.03)	-0.036 (0.01)
FDI	0.073*** (0.02)	0.086*** (0.02)	-0.013 (0.01)
Credit-to-GDP ratio	0.003 (0.01)	0.009 (0.01)	-0.006 (0.00)
Savings ratio	0.063*** (0.02)	0.034*** (0.01)	0.029** (0.01)
Digitization	0.480* (0.27)	0.295 (0.21)	0.185 (0.18)
Number of observations	215	215	215
Number of countries	11	11	11
R-squared	0.81	0.89	0.20
Prob > F	0.000	0.000	0.000

Source: Authors’ calculation. * denotes statistical significance at 10 percent level, ** denotes statistical significance at 5 percent level, *** denotes statistical significance at 1 percent level. Standard errors are in parentheses.

View Table

^{Table 2.}

Determinants of Inclusive Growth, Panel Regression Results.

	Dependent Variable
	$$\Delta {\overline{y}}^{*}$$ – inclusive growth	$$\mathrm{\Delta }\overline{y}$$ – income per capita growth	∆ω – equity index growth
Lagged GDP per capita	1.925** (0.92)	-0.899 (0.70)	2.824*** (0.64)
CPI inflation rate	-0.030 (0.03)	-0.016 (0.03)	-0.013 (0.02)
Redistribution	0.687*** (0.19)	0.402* (0.15)	0.285** (0.13)
Female labor force participation	0.268*** (0.06)	0.203*** (0.044)	0.066* (0.04)
Productivity growth	0.850*** (0.04)	0.886*** (0.03)	-0.036 (0.01)
FDI	0.073*** (0.02)	0.086*** (0.02)	-0.013 (0.01)
Credit-to-GDP ratio	0.003 (0.01)	0.009 (0.01)	-0.006 (0.00)
Savings ratio	0.063*** (0.02)	0.034*** (0.01)	0.029** (0.01)
Digitization	0.480* (0.27)	0.295 (0.21)	0.185 (0.18)
Number of observations	215	215	215
Number of countries	11	11	11
R-squared	0.81	0.89	0.20
Prob > F	0.000	0.000	0.000

Source: Authors’ calculation. * denotes statistical significance at 10 percent level, ** denotes statistical significance at 5 percent level, *** denotes statistical significance at 1 percent level. Standard errors are in parentheses.

C. Empirical Results

Table 2 present the results of our regressions for our preferred model. It shows that fiscal redistribution has a significant impact on inclusive growth, both through per capita income growth and equity growth. This result is consistent with the findings of Ostry et al. (2013), who found that the combined direct and indirect effects of redistribution are on average pro-growth, and with those of Aoyagi and Ganelli (2015), who estimated that the redistribution impact on inclusive growth could be sizable in emerging Asia. Table 2 indicates that redistributive policies (implying an increase of 1 point in the difference between market and net Gini) in developing Asia can boost per capita growth by about 0.4 percent and improve equity by almost 0.3 percent at the same time.

Although positively contributing to inclusive growth, the Gini-based redistribution measure remains low in ASEAN (Figure 6). As mentioned in Aoyagi and Ganelli (2015), policymakers in Asia have traditionally used macroeconomic policy to support growth, rather than to redistribute income. Aoyagi and Ganelli (2015) also highlight that although fiscal redistribution is associated with inclusive growth, there could be cases in which some redistributive policies might actually hurt growth. Thus, it is important for policy makers to focus on those fiscal policy instruments which will have the least negative efficiency impact. Our results confirm their finding. In addition, we show that redistribution is beneficial for both income per capita growth and equality growth. We, however, do not analyze possible effects of any specific redistributive policy in this research, we believe those policies should be tailored according to country-specific needs.

View Full Size

Fiscal redistributions in ASEAN.

Citation: IMF Working Papers 2020, 118; 10.5089/9781513549194.001.A001

Data source: Standardized World Income Inequality Database (SWIID).

• Download Figure
• Download figure as PowerPoint slide

For instance, IMF (2014) highlights that fiscal policies need to be carefully designed to balance distributional and efficiency objectives and suggests a mix of instruments for developing countries. Those include expanding conditional cash transfer programs as administrative capacity improves; expanding noncontributory means-tested social pensions; and improving access of low-income families to education and health services and others. In addition, Claus et al. (2012) have assessed the effectiveness of redistributive fiscal policies and have quantified the impact of taxation and government expenditures on income distributions in Asia. Clements et al. (2015) emphasized that distributional impact of fiscal policy is determined by the composition of latter. Although our fiscal redistribution variable does not illustrate relative importance of tax related redistribution compared to spending related one, it would be interesting to analyze alternative indicators of fiscal policy such as ratio of direct to indirect taxes and components of spending in future research.

Our empirical results also show that FLFP is a positive and statistically significant contributor to inclusive growth through both channels. The coefficients reported in Table 2 suggest that a one percentage point increase in FLFP can approximately raise GDP per capita growth by 0.2 percentage point and the equity index growth by 0.07 percentage point. Regarding the impact of FLFP on inclusive growth through the GDP per capita channel, our results are in line with Elborgh-Woytek et al. (2013) and Gonzales et al. (2015), who suggests that higher FLFP is associated with higher economic growth. Cuberes and Teignier (2015) found that gender gaps in entrepreneurship and workforce participation lead to income losses and negatively affect productivity. In addition, UNESCAP (2017) finds that increasing women’s participation in the economy could add an additional US$12 trillion to annual global output by 2025. Based on the UNESCAP report, across the ten ASEAN members, an estimated 61.3 million women entrepreneurs (about 9.8 percent of the total ASEAN population) own and operate businesses.

On average, higher gaps in labor force participation between men and women may result in inequality of earning between sexes creating income inequality (Gonzales et al. 2015). According to our results, bringing more women into the workforce will have a positive effect on equality. It was recently widely discussed that gender and income inequalities are linked²³ and that existing inequality of opportunities between males and females, including unequal access to education, healthcare and finance (i.e. inequality of opportunities which affect the ability of females to participate in the labor force), is strongly associated with income inequality.

Our results suggest that labor productivity growth is likely to affect inclusive growth through an increase in per capita income, while not contributing to equity improvement. According to our results, net FDI inflows have a positive and significant impact on GDP per capita growth and therefore on inclusive growth, as also highlighted in Anand et al. (2013). While in our analysis financial deepening, measured as the credit-to-GDP ratio, does not have a significant impact on inclusive growth measure (in line with Anand et al. 2013), the net savings ratio significantly and positively affects inclusive growth through both channels. Historically, Asia has had comparatively higher savings ratios attributed to insufficient availability of social services such as healthcare and education, limited public pensions, and higher rates of natural disasters. Those higher rates of savings can help smooth consumption and enhance investment, thus improving equality and supporting per capita growth. Finally, our analysis shows that the proxy of digitalization has a significantly positive impact on inclusive growth through both channels.

Interestingly, lagged GDP per capita is insignificant determinant of GDP per capita growth. This result is in line with Aoyagi et al. (2015), and is consistent with the observation by Solow (1956) that the rate of growth tends to fall as the income level rises. On the other hand, higher income per capita has positive and significant effect on the equity index, which suggests that ASEAN countries have entered the phase where economic growth produces more equal income distribution (following Kuznets 1955).

D. Robustness Checks

In addition to our benchmark model we have also estimated a number of alternative specifications to confirm the robustness of our empirical results. Appendix Table 3 presents the results for the specifications in which we have included alternative regressors, such as the lagged dependent variable (to further control for endogeneity bias), total labor force participation instead of FLFP, and trade openness. The results are broadly consistent with those for the benchmark estimation presented in Table 2. Finally, we have performed additional robustness checks by excluding two economies (China and India) from our benchmark model. In this alternative specification, the signs and significance of the regressors are broadly consistent with our benchmark estimation (Table 4).