Financial Inclusion and Extreme Poverty in the MENA Region: A Gap Analysis Approach

Eradicating extreme poverty remains one of the most significant and challenging Sustainable Development Goals (SDGs) in the Middle East and North African (MENA) region. The latest World Bank statistics from 2018 show that extreme poverty in MENA increased from 2.6% to 5% between 2013 and 2015. MENA ranks third among developing regions for extreme poverty, and fell short of halving extreme poverty by 2015 – the target established by the United Nations’ Millennium Development Goals, the precursor to the SDGs. Using system General Method of Moments dynamic panel estimation methodology on annual data for 11 MENA countries and 23 emerging markets (EMs) over the period 1990 – 2017, this study begins by estimating the impact of financial inclusion – using measures of access – on the eradication of extreme poverty by 2030, the first goal of the SDGs. The results of the study indicate that, on one hand, financial access measures have a positive, statistically significant impact on reducing extreme poverty for the full sample as well as the MENA region. The second part of the study employs a gap analysis against four poverty targets—0%, 1.5%, 3%, and 5%—and shows that no MENA country and few EM countries will be able to close the extreme poverty gap and reach the target of 0% by 2030 by depending solely on improvements in financial access. These targets are based on the two benchmarks set by the World Bank and the UN, with intermediaries to capture error and give a fuller picture of what is possible. However, if improvements in financial inclusion alone can bring every EM and MENA country except Djibouti and Romania to bring the most accessible target of reducing global extreme poverty to no more than 5% by 2030.


I. INTRODUCTION
The world has made remarkable progress reducing extreme poverty in the past 25 years.
Between 1990 and 2015, the number of people living on less than $1.90 per day -the international benchmark for extreme poverty -dropped by one billion, bringing us closer to the United Nations Sustainable Development Goals (SDGs). Still, the benefits of economic growth have reached regions, countries, and individuals unevenly. An unacceptably large number of people, more than 700 million, still live in poverty across the globe, and extreme poverty is becoming more entrenched in some places, especially those fraught by violent conflicts and weak institutions.
In the Middle East and North Africa (MENA), a region particularly vulnerable to fragility, eradicating extreme poverty remains one of the most challenging of the SDGs. 3 MENA ranks third among developing regions for extreme poverty. According to the latest World Bank data, the proportion of the population living under $1.90 a day rose from 2.6% to 5% between 2013 and 2015, while the number of poor nearly doubled from 9.5 to 18.6 million.
Although extreme poverty is much higher in sub-Saharan Africa, the pace at which it is growing in MENA presents a blunt warning that progress cannot be taken for granted. The erosion of past gains risks fueling political, economic, and environmental crises, threatening to exacerbate the circumstances of those already struggling to protect their lives and livelihoods.
While research on poverty reduction in the region tends to focus on financial development and governance, less attention has been paid to the role of financial inclusion. SDG 1 4 -eliminating poverty in all its forms -explicitly highlights the importance of access to financial services. Indeed, evidence from Argentina, India, Kenya, Malawi, Niger, and other countries demonstrates the ways in which financial inclusion can impact poverty (Klapper, El-Zoghbi, and Hess 2016). When people are included in the financial system, they are better able to improve their health, invest in education and business, and make choices that benefit their entire families. Financial inclusion advances governments, too: Introducing vast segments of the population into the financial system by digitizing social transfers, for example, can cut government costs and reduce leakage, with benefits that ripple across society.
Yet, the links between financial inclusion and poverty reduction in MENA are less established. This study aims to analyze the importance of financial inclusion in addressing 3 Formerly Millennium Development Goals, from 2000Goals, from to 2015 More specifically Target number 1.1 of SDG 1. extreme poverty by 2030, the year UN Member States set as a target for achieving the SDGs.
We assess improvements in financial inclusion against four targets of 0%, 1.5%, 3%, and 5% living in poverty. 5 In so doing, this study seeks to answer the following questions: Do different types of financial inclusion indicators (focusing on those for access) affect poverty alleviation directly? Is this effect the same across samples of MENA countries and emerging markets (EMs) more broadly? Are the countries under study able to use financial inclusion tools exclusively to close the poverty gap by 2030? The remainder of the paper is divided as follows: Section II briefly reviews the literature; Section III describes the data used; Section IV highlights the methodology employed and the model specification; Section V presents our results; and Section VI concludes. An appendix appears at the end of the paper.

II. LITERATURE REVIEW
When the World Bank called for an end to extreme poverty by 2030 and the UN set a more ambitious goal to eliminate poverty in all its forms, everywhere, progress from the previous two decades had given the international community reason to be hopeful. More than a third of the world lived in extreme poverty in 1990, by 2015, the ratio had fallen to one tenth (World Bank 2018). However, the pace of poverty reduction has slowed, and for millions of people in sub-Saharan Africa and MENA, poverty is on the rise. The 2015 figure proved the low point.
Over the past several years, researchers have used various dimensions of financial inclusion to point to the causal relationship between financial inclusion and economic growth (Honohan 2004;Demirguc-Kunt and Klapper 2012;Cumming et al. 2014;Klapper, El-Zoghbi, and Hess 2016;El-Zhoghbi, Holle, and Soursourian 2019). A study in India found measures like banking penetration, deposits, and availability and use of banking services boosted growth between 2004 and 2013 (Sharma 2016). Another in Kenya estimated that the expansion of a mobile money service significantly contributed to per-capita income growth (Beck et al. 2018).
And in countries in the MENA region, scholars have demonstrated the impact of financial 5 The World Bank goal of ending extreme poverty would reduce the proportion of people living on less than $1.25 a day (in 2005 constant dollars, which is $1.90 in 2020 dollars) to 3% by 2030. The SDG in which it is embedded calls for that indicator to fall to 0%. The World Bank acknowledges that 0% extreme poverty is an excellent goal, but suggests it is unrealistic, noting that some people whose available money falls below $1.90 a day may only experience this circumstance for a short time. It is also the case that reported rates may not be completely reliable. For example, governments are incentivized to underreport poverty rates and may have no knowledge of black market and other cash income. A 3% rate could be statistically indistinguishable from a 0% rate. development (Hamdi and Hakimi 2015), banking concentration (Abouzayed and Fayoumi 2016), and households' financial access (Emara and El Said 2019) on growth.
When it comes to countries in MENA, the link between financial inclusion and the poor is no clearer. Indeed, many researchers turn elsewhere to understand poverty in the region. Banerji and Humphreys (2003) focus on good governance as a crucial component of poverty relief, while Ncube, Anyanwu, and Hausken (2013) find that domestic investment, trade openness, exchange rates, income per capita, and oil rents are key poverty-reducing variables. Neaime and Gaysset (2017) use General Method of Moments (GMM) and Generalized Least Squares models to conclude that population, inflation, and trade openness have significant effects on poverty, whereas financial inclusion does not appear to alleviate it.
Still, it is increasingly recognizing that lack of access to finance in MENA is a severe restriction on economic growth and poverty alleviation, as the poor struggle to accumulate savings and cover critical health and education expenses (Alvarez de la Campa 2010, Pearce 2011). The region lags others on key indicators of bank deposits and loan accounts, and despite the expansion of bank branches and microfinance institutions in some MENA countries, vast segments of the population are still cut off from financial services (Pearce 2011). The limited availability and quality of data, especially as it relates to financial technology, remains another challenge to poverty-reduction efforts in MENA.
Our paper seeks to build upon this evidence base. By using access measures of financial inclusion, we contribute to the growing literature that investigates the direct link between financial inclusion and poverty reduction, with a focus on MENA countries and emerging markets. We use system GMM dynamic panel estimation methodology on annual data for eleven MENA countries and twenty-three emerging markets. We conclude with a gap analysis, which follows Panda and Kumar's method for calculating the difference between specified targets and projections in the context of achieving the Millennium Development Goals (MDGs; see Section IV). Such an analysis requires projecting global poverty to 2030 under various assumptions.
Researchers at the World Bank Group recently showed, setting specifications such as household welfare growth, economic growth, and inequality to "nowcast" poverty data from 2015 household surveys to the present, and then project rates to 2030 (2018). Crespo Cuaresma et al. present another econometric tool for forecasting poverty rates. Their model combines country-specific historical estimates of income distribution, using Beta-Lorenz Curves with projections for changes in GDP and population demographics to create "poverty paths," by country, up to 2030 (2018).

III. DATA
The data set is constructed as a panel of country observations from the World Development

IV. MODEL SPECIFICATION & METHODOLOGY
Using system GMM dynamic panel estimation methodology on annual data for 11 MENA countries and 23 EMs over the period 1990 -2017, the first part of the study estimates the role of financial inclusion -using measures of access -in eradicating extreme poverty (the first goal of SDG). To perform such an analysis, the following dynamic panel regression methodology is used: …N, t = 1990,…T Where Povit denotes the Poverty headcount ratio at $1.90 a day as a percent of the population of country i, at time t, Povit-1 is the lagged poverty variable, and Xit-1 is the vector of explanatory variables. These include the annual GDP growth rate, inflation rate, trade as a percentage of GDP, mobile cellular subscriptions per 100 people, and the annual population growth rate. The variable FIit-1 represents financial inclusion indicators that cover different financial access areas of the financial system in country i at time t, and εit is the error term.
The system GMM combines Equation (1)  In the second step required growth for the SDG1 for each country is computed using a compound growth rate formula as follows, where r is the required poverty head count ratio growth rate as defined by the percent of the population living under $1.90, Pov2030 is the poverty head count ratio in the year 2030, Povl is the poverty head count ratio in the latest available year, and k is the year of the latest available value of the poverty measure.
Next, the actual growth in the financial access indicator, or FinAcc, as measured by the principal component analysis of ATMs per 100,000 adults, bank accounts per 1,000 adults, and bank branches per 100,000 adults, is calculated using the following semi log trend function, where a is the constant of the regression and b is the growth rate in the access indicator.
The coefficients of Equation (6) are estimated using time series regression for each country in the MENA sample in a turn. The next step entails using the coefficient φ of Equation (4) together with the parameter b of Equation (6) to project extreme poverty percent in the year 2030 as follows, !"# :efe = !"# p (1 + n (δ + φ)) :efe,g Hence, re-writing Equation (7), the growth rate of the financial access indicator that is required to close the poverty gap by the year 2030 is computed as follows, The projected 2030 value of the poverty head count ratio, Pov2030, computed using Equation (7), is assumed to depend solely on the improvement in the financial access services.
The difference between the targeted 2030 value of the projected poverty level and the targeted 2030 poverty level, which can be bridged by other non-financial factors affecting poverty such as economic growth, inflation rate, openness of the economy, population growth or the spillover effects of financial inclusion on other SDGs that are expected to reduce extreme poverty (see Section II).
Accordingly, using the results of Equation (5), an SDG gap analysis is undertaken to compute and analyze the difference between the targeted and the projected values for the poverty head count ratio and a financial access gap analysis to compute the growth in the financial access indicator required to close the extreme poverty gap by the year 2030 if the group of countries in our sample depend solely on improvement in financial access services.  (1) term. The first column shows the results of a regressing poverty on its own lag only.

V. ESTIMATION RESULTS
The results show an AR(1) coefficient of the poverty head count ratio of 0.94% of the population, representing a short-term positive correlation between poverty and its own lag.
Adding GDP growth rate to the model, Column 2 shows results. The inclusion of this variable does not have a large impact on the sign or significance of the AR(1) coefficient. The coefficient for the GDP growth rate is negative as expected, and is statistically significant where a one percent increase in GDP growth rate results in a drop in the poverty head count ratio by about 0.081% of population.
As shown in Column 3, adding inflation rate alters neither the sign nor the statistical significance of the previous two regressors. However, as the results show, the coefficient of inflation rate does not have a statistically significant impact on the poverty head count ratio in any of the six regressions of this table.
Column 4 shows the results of the regression that adds population growth rate. Adding this regressor does not significantly affect the coefficients of the included regressors. The coefficient for the population growth rate is statistically significant and positive as expected where a one percent increase in the population growth rate results in an increase in the poverty head count ratio by about 0.19 % of population.
Next, Column 5 includes the variable trade as a percent of GDP and shows that the To analyze the impact of access to financial services on the poverty head count ratio, shows the results of the full sample for the first access indicator, ATMs machines, which has a statistically significant negative impact on the poverty head count ratio, where a one unit increase in ATMs leads to a fall in the poverty head count ratio by about 0.64% of the population. Similarly, Column 3 shows that a one unit increase in bank accounts per 1000 adults leads to a statistically significant decrease in the poverty head count ratio by about 2.02% of the population. Column 5 shows that a one unit increase in bank branches per 1000 adults leads to a statistically significant decrease in the poverty head count ratio, which is about 0.96% of the population.
Next, to analyze the impact of access to finance in the 11 MENA countries of our sample, an interaction term of the dummy variable MENA is added to the regression. In Columns 2, 4, and 6 the dummy variable for the MENA region is interacted with atm, ba, and bb, respectively. The interaction terms are statically insignificant, with the exception of the interaction term of atm.
Column 7 shows that the variable "acc," a linear combination using the principal component analysis of the three access to finance indicators, ATM machines, bank branches, and accounts, is negative and statistically significant. A one unit increase in acc leads to a statistically significant decrease in the poverty head count ratio, about 2.19% of the population.
Column 8, however, shows that the interaction term of the dummy for the MENA region with the acc indicator has a statistically insignificant impact on the poverty head count ratio.  Using estimated total effect of the acc index for the MENA region computed in Table (6), the SDG gap analysis for the MENA sample is performed and the results are presented in Tables 7, 9, 11, and 13 corresponding to a poverty target of 0%, 1.5%, 3%, and 5%, respectively. In each table, the third column computes the required SDG growth rate using Equation (5), the fourth column computes the 2030 SDG projection using Equation (7), and the fifth column computes the SDG Gap by subtracting the 2030 targeted poverty level from the poverty projected level for 2030, or Column (4). Column (6) computes the required increase in the financial access index using Equation (8) and Column (7) estimates the actual growth in the financial access index using Equation (6). Finally, Column (8) computes the 2030 financial access gap by subtracting Column (7) from Column (6).
If we assess the results based on the 0% target, Table (7) shows that, based on the latest available value for poverty head count ratio, none of the countries in the sample have achieved the targeted level. However, Jordan and Iran are already close to the targeted poverty level with a latest poverty value of 0.1 and 0.2, respectively. Other countries, such as Djibouti and Yemen, are way above the targeted poverty levels with a latest available poverty head count ratio of 22.5% and 18.8% of the population, respectively. The annual poverty growth rate would have to fall by 0.3% and 0.65%, respectively, for them to reach 0% poverty by 2030.
As per the results of Column (5), the estimation of the 2030 gap shows that Yemen, Djibouti, and Iraq will perform the worst out of the entire MENA sample with a poverty head count ratio gap of 7.81%, 3.68%, and 1.65%, of the population respectively. Those three countries would need to achieve an annual increase in the financial inclusion access index of 0.75%, 0.73%, and 0.63%, respectively, in order to reach 0% poverty in 2030. The results of Column (8) show that those three countries will miss the poverty target. Yemen's financial inclusion growth gap is 0.69%, while both Djibouti and Iraq's is 0.60%. On the other hand, Iran and Jordan will be performing the best out of the entire MENA sample with a predicted poverty head count ratio gap of only 0.05% and 0.09%, of the population respectively. Those two countries will be able to close the poverty gap in 2030 if they increase the growth rate of the financial inclusion index by 0.59% and 0.47%, respectively Similarly, for the EMs sample, based on the latest available values for poverty head count ratio in each country, Column (2) of Based on the poverty gap analysis of Column (5), the results show that Romania, Brazil, and Philippines will lag behind in closing the extreme poverty gap in 2030. Their projected poverty head count ratio is 5.23%, 2.09%, and 1.90% of the population, respectively. To close the poverty gap in 2030, Romania will require an increase in the financial inclusion index by about 0.27% while Brazil and Philippines will require 0.26%, and 0.23%, respectively.
China, by contrast, will have almost closed its poverty gap in 2030. Its projected rate poverty head count ratio is only 0.0003% of the population and its projected financial inclusion index growth gap is about 0.06% in the year 2030. The projected poverty ratios in India and Indonesia are projected to be around 0.02% and 0.03% of the population, respectively, with a projected gap in the required growth of the financial inclusion index of 0.11% and 0.14%, respectively.
Table (9) analyzes the 1.5% poverty target. Based on the latest available value for poverty head count ratio, the only MENA countries that will not achieve this target are Djibouti, Iraq, Tunisia, and Yemen. These countries would require a fall in the annual poverty growth rate of 0.15%, 0.03%, 0.01%, and 0.15% to achieve the 1.5% poverty target by 2030. The financial inclusion index would have to grow annually by 0.19%, 0.04%, 0.02%, and 0.19%, respectively.
Table (10) shows that eleven countries out of the EMs sample have already achieved the 1.5% poverty target. Out of the remaining twelve countries, only Brazil, Philippines, and Romania will not achieve the target. For these three countries, the projected poverty gaps are 0.59%, 0.40%, and 3.73%. Thus, to achieve the 1.5% target the poverty head count ratio would need to fall by 0.09%, 0.10%, and 0.09%, respectively, by the year 2030. Brazil and Romania would both require growth of 0.04%, in the financial inclusion index and Philippines would require growth of 0.05%.
Assessing the results based on the 3% poverty target, according to Table (11), the latest available data shows that all but two of the MENA countries can achieve it; Djibouti and Yemen have a projected poverty gap of 1.32% and 2.81%, respectively. These two countries would have to achieve a growth of financial inclusion index of 0.11% and 0.10%, respectively, to close the poverty gap. Similarly, for the EMs sample, the latest available data on Table (12) shows that twelve countries have already achieved the target. Of the remaining eleven countries, only one country, Romania, will miss the target by the year 2030. Its projected poverty gap is 2.23%. An estimated 0.02% annual growth in the financial inclusion index will be required to close this gap.
Much as they will not achieve the 3% target, Djibouti and Yemen are the only MENA countries that are not already achieving the most flexible poverty target, 5%, as yet. The projection analysis shows that Djibouti is on track and will be able to close the poverty gap by the year 2030, as shown on Table (13). However, Yemen will miss the target by 2.81% with a predicted required increase in the financial inclusion index of 0.10% to close the poverty gap.
Most of the countries in the EMs sample are also achieving the 5% target as shown on Table ( 14). Of the six that are not, only Romania will not achieve 5% by 2030. But the predicted gap is only 0.23% and the required growth in the access index is only about 0.004%.

VI. CONCLUSION
Using system GMM dynamic panel estimation methodology on annual data for 11 MENA countries and 23 emerging markets (EMs) over the period 1990 -2017, the study uses several measures of financial inclusion that cover access side of financial services to analyze its impact on eradicating extreme poverty (SDG 1).
The results of the study show that financial access index (or acc index comprising of atms, ba, and bb) has a statistically significant impact on reducing extreme poverty for the full sample as well as the MENA sample. The results confirm that a one-unit increase in the acc index results in a fall in poverty head count ratio by about 2.22% for the full sample and only about 0.79% for the MENA sample.
Using the acc index, we then employed a gap analysis, using it to predict the ability of MENA and EM countries to achieve extreme poverty goals by 2030 if they were to depend only on the improvement in financial services and no other factors. The study assesses the impact of the improvements in financial inclusion on the achievability of the SDG 1 against four poverty targets 0%, 1.5%, 3%, and 5%. Two of these targets correspond to those assigned by the SDG and the World Bank (0% and 3%, respectively). This analysis incorporates additional targets to make up for any statistical errors the others did not and to give a fuller picture of what is possible and what is probable. The World Bank may be correct to suggest that 3% worldwide is more realistic than 0%; it also points out that eliminating poverty is a country-by-country endeavor. Four target points allowed this research to illuminate how that endeavor may unfold (World Bank 2015). The study concludes that if we assess the impact of the improvements in financial inclusion against the most restrictive target of 0% poverty by 2030, and if the current trends of financial access measures continue, then none of the MENA countries and the majority of the EMs countries will not be able to achieve the poverty goal if they depend only on the improvement in financial access services and no other factors.
However, if we assess the impact of the improvements in financial inclusion against the most flexible target of reducing global extreme poverty to no more than 5% by 2030, the study concludes that all countries with the exception of two countries -Djibouti and Romania -will be able to achieve the poverty goal by 2030 if they depend solely on improvements in financial services and no other factors. These results justify dedicating significant resources to such improvements.
Policy considerations can be directed towards developing and promoting the infrastructure Yet, the lack of data availability, including on financial technology, in MENA countries remains a major limitation for analyses of fintech's impact on poverty alleviation. As delivery and usage of financial technology is predicted to magnify the impact of financial inclusion on poverty reduction both directly -as shown in this paper -and indirectly -through channels related to other SDGs. Additionally, governments in MENA must take data availability and quality more seriously if they are to reverse the acceleration of extreme poverty in the digital age.  atm Bank accounts per 1,000 adults

APPENDIX
Number of depositors with commercial banks per 1,000 adults. Depositors with commercial banks are the reported number of deposit account holders at commercial banks and other resident banks functioning as commercial banks that are resident nonfinancial corporations (public and private) and households. For many countries data cover the total number of deposit accounts due to lack of information on account holders. The major types of deposits are checking accounts, savings accounts, and time deposits.
ba Bank branches per 100,000 adults Commercial bank branches are retail locations of resident commercial banks and other resident banks that function as commercial banks that provide financial services to customers and are physically separated from the main office but not organized as legally separated subsidiaries.
bb Access Index.
The principal component of the last three indicators.