It is fundamental to quantify the size of the developing world’s middle class, as well as how it has changed and varied among nations. In terms of development, the middle class benefits society as a whole by producing the consumer spending that drives the economy, acting as a reliable source of labor for businesses, and contributing significantly to tax revenue, which is used to fund public services. It supports policy reforms and the execution of growth-enhancing institutional changes and public investments. In addition, the middle class serves as a significant factor in determining the overall stability of a society, providing citizens with a sense of security and providing the foundation for economic growth. It is also important to study the middle class because it can provide insights into the fight against poverty. Nations with a higher share of middle class have better chances at fighting poverty than those with a lower share.
There is, however, long-standing disagreement over the definition of the middle class due to its subjective and relative nature. It is generally accepted that the middle class includes those who possess more wealth and income than the average person, but it is difficult to say exactly where the line is drawn. There is also disagreement over whether the middle class should be determined by income, wealth, or if its definition should change depending on the economy. It is unclear if the middle class should be defined differently in emerging nations as opposed to the developed ones. Nevertheless, researchers would ideally like to have information on a wide range of factors to fully assess the middle class. Among these factors are the person’s family background, education, employment, and wage income, as well as whether he lives in a decent house, owns a car, has a mobile phone, or considers himself “middle-class”.
To simplify matters and because of the frequent lack of quality data, especially in developing countries, many studies reduce the middle-class analysis to a monetary indicator (income) that approximates a person’s social status. We define the middle class from this economic perspective in two ways: in absolute terms, referring to a specific income level that is the same for every country, or in relative terms, referring to the middle-income segment of each country. With the absolute definition, the emphasis is on a particular degree of financial security or affluence in accordance with a range of income or assets. This range typically varies across countries. In contrast, the relative definition looks at a comparison between one’s income to that of other people in the same nation. This definition assesses the differences between a person’s income and the incomes of those around them and is usually used to measure economic mobility and whether middle-class households can move up or down the income ladder.
Various definitions of the middle class have emerged over the years, whether in absolute or relative terms. Focusing on the growth of the middle classes in developing countries, this article, therefore, makes use of the following definitions:
They define the middle class in 13 developing countries in terms of daily consumption as between $2 and $4 or $6 and $10 daily. Duflo and Banerjee, in their article, consider that the difference between the poor and the middle class is whether a person has a regular, well-paid wage job. A job that gives him the mental space he needs to do all those things that the “middle-class” also does, such as having fewer children and taking better care of their education and health. In this sense, they make an essential link between an economical approach of the middle-class and its sociological dimension.
He uses a similar description of the middle-class to Banerjee and Duflo’s while explicitly distinguishing between the “Western middle class” and the “developing world’s middle class”. He argues that someone identifies as being in the developing world’s middle-class if that person is not poor by developing country standards, though still poor by advanced country standards. More precisely, a person is “middle-class” if he lives in a household with consumption per capita between the median poverty line of developing countries, namely $2 a day at 2005 purchasing power parity and the US poverty line of $13 a day.
In their paper entitled “Decomposing World Income Distribution: Does the world have a middle class?” they define the middle-class using the mean income of Brazil and Italy. The upper bound is $50 per day, which equals the average income in Italy, the least wealthy among G7 members, and the lower bound is equal to the mean earnings in Brazil ($12 per day). However, the problem with Milanovic and Yithzaki’s approach is that it does not distinguish between population groups who earn less than $12 a day but are “middle-class”.
He focused on the global middle-class by defining the middle class as people in the group earning between $11 and $110 a day (2011 purchasing power parity). The upper boundary ($110) is double Luxembourg’s median income, and the lower threshold ($11 per day) equals the average between the poverty lines of Italy and Portugal. According to Kharas, in this range, the income elasticity of consumption appears to be greater than one, and the demand for a range of new goods and services increases. It then stimulates growth through product differentiation, branding, and marketing.
Their paper “Stuck in The Tunnel: Is Globalization Muddling the Middle Class?” suggested a measure of the middle class based on the size and income share of the households around the median and their income status relative to their wealthier counterparts. The middle class represents households in the middle of the income distribution in each country, i.e., families with per capita income in the range of 75 and 125 percent of the median household per capita income. They have shown that while the poorest have the greatest needs and deserve attention, the political support and economic participation of those in the middle will be critical to market-driven economic growth and, thus, long-term poverty reduction.
As the future of Africa’s economy lies in the middle class, it is difficult to define who falls into this key group due to a lack of high-quality data sources across the continent. In this report, we have used the Gallup World Poll database to measure the middle-class in Africa based on the different definitions listed above to compare them, as each definition raises another matter concerning the definition of the middle-class. The Gallup database depicts the most comprehensive and farthest-reaching survey in the world. Through yearly, nationally representative polls with comparable measures across nations, the survey reaches more than 99% of the world’s adult population. The database largely contains surveys on either income or household consumption expenditures, which we used to calculate a person’s daily spending for our research.
This study includes data from 46 of the 54 countries that compose Africa. In some sections, we aggregated the continent into its five mutually exclusive geographical and economic regions, as shown in the table below:
Based on the various household income surveys provided by the Gallup database, we obtained individuals’ daily income, which we classified according to whether or not they were “middle class.” It allowed us to represent the annual middle class share at the national level that we aggregated at the regional level. The following figures present the evolution of the African middle class according to the different definitions used, aggregated at the regional level, from 2009 to 2020. The estimates presented in the figures below are partly based on proprietary data that is very costly. We are using here the latest data we have access to.
Following Duflo and Banerjee’s definition, we see that the northern and southern parts of Africa have the largest share of middle-class individuals. However, the percentage of the middle-class share has declined in those regions in the last few years. It fell from 30% in 2018 to 24% in 2020 in the northern region, and from 25% in 2017 to 14% in the south. The East, Central, and West areas, on the other hand, experienced a general increase in the middle class from 2010 to 2020. The proportion of the middle class has risen from 5%, 3%, and 5% in 2009 to 11%, 10%, and 10% in 2020, respectively.
Ravallion, Milanovic, Yitzhaki, and Kharas’ definitions follow the same trends as Duflo and Banerjee’s, as illustrated in the images below. As for the definition of Birdsall et al., while the regions with the largest middle class are the same as in previous definitions of the middle class, it is worth noting that the evolution of the middle-class share does not exhibit any patterns here in comparison to the others.
These results indicate that the African middle class has experienced a general increase over the past decade, although there are regional variations.. This is important information for policy makers, as it can help them assess the effectiveness of their strategies to promote economic development in the African continent.
You can zoom the coordinate systems using the scroll bar just below the graph, or switch between the line and bar chart, by selecting the corresponding options at the top right of the chart.
You can zoom the coordinate systems using the scroll bar just below the graph, or switch between the line and bar chart, by selecting the corresponding options at the top right of the chart.
You can zoom the coordinate systems using the scroll bar just below the graph, or switch between the line and bar chart, by selecting the corresponding options at the top right of the chart.
You can zoom the coordinate systems using the scroll bar just below the graph, or switch between the line and bar chart, by selecting the corresponding options at the top right of the chart.
You can zoom the coordinate systems using the scroll bar just below the graph, or switch between the line and bar chart, by selecting the corresponding options at the top right of the chart.
The evolution of the middle-class share in Africa over time was depicted in the previous graph, but this does not provide information on the size of the middle class in each nation. In order to comprehend this, it is necessary to consider the middle class in terms of population size in each nation and year.
The maps below illustrate how, based on various definitions, the population of the African middle class has changed geographically from 2009 to 2020. Use the “play” button and timeline in the bottom-left corner of the chart to see how this has changed since 2009. You can find out information about a country’s middle class size at a given time by clicking on any one of its names.
Focusing on population size rather than the percentage of the middle class in Africa, we find that the top three most populous nations in terms of “middle-class” people are Nigeria (about 9.5 million in 2009 and 22.9 million in 2020), South Africa (10.5 million in 2009 and 8.3 million in 2020), and Egypt (9.5 million in 2009 and 19.5 million in 2020). It’s also important to note that the middle-class population in Ethiopia and Algeria has grown significantly over the past two years (18.4 million for Ethiopia and 15.3 million for Algeria in 2020).
The map above gives also our yearly data coverage: the grey areas on the maps represent the missing data.
The map above gives also our yearly data coverage: the grey areas on the maps represent the missing data.
The map above gives also our yearly data coverage: the grey areas on the maps represent the missing data.
The map above gives also our yearly data coverage: the grey areas on the maps represent the missing data.
The map above gives also our yearly data coverage: the grey areas on the maps represent the missing data.
We discussed the annual distribution and share of the middle class in Africa in the section above; now, let us take it a step further by identifying the top African nations with the highest middle class population, and share according to different definitions. The table below contains aggregated data for 46 African countries over the period of 2009 to 2020. It ranks African countries according to the highest proportion of their middle-class population (in the third column) or the highest population size belonging to the middle-class (in the last column).
From the analysis of the table, it is evident that Mauritius, Libya, Algeria, Tunisia, and South Africa are the African countries with the highest proportion of middle-class population, with an average of more than 20% over the period of 2009 to 2020. In contrast, in terms of the middle class size, Nigeria is the most populous country with middle-class individuals followed by Egypt and South Africa. It shows that if Mauritius, Libya, Algeria and Tunisia have higher levels of trust from an investment perspective, Nigeria and South Africa have the highest potential from a business perspective. Sierra Leone and Liberia, on the other hand, have a middle-class share of less than 1%. With regards to population size of the middle-class, Comoros, Gambia Central African Republic, and Liberia are at the bottom, with an average population size of less than 500,000. The analysis of this table provides useful insights into the proportions of middle-class populations in different African countries.
As previously stated, the middle class represents an opportunity for economic development, and it stands to reason that the developing nations with the highest proportions of the middle class are more likely to have higher levels of development and vice versa. This section will concentrate on the first direction of this correlation. In the figures bellow, we show a yearly scatter plot between the middle-class population and the GDP per capita for African economies after a log transformation. We used the GDP per capita to capture the economic output and the purchasing power of a citizen in a specific nation. The countries are represented by the points in the charts by reporting their GDP per capita on the vertical axis and the corresponding number of middle-class residents on the horizontal axis each year. Additionally, to get a better understanding of the relationship between two variables, we added a regression line to the plot. This line captures the linear relationship between the two and makes sure the vertical distances between the data points (here the countries) and the line are minimized. Also shown is the confidence interval surrounding the line.
Our regression model for each year has the following linear functional form: \[ log(GDP\ per\ capita) = \beta \ log(Middle\ Class \ population) + \epsilon\]
With the exception of Birdsall et al.’s definition, which suggested a negative relationship between the middle class and the GDP per capita, all the absolute definitions in the plot below show a noisy positive relationship between the middle class and GDP per capita.
According to what we discovered above, a country’s prosperity (measured by GDP per capita) and its middle-class population are positively correlated. However, in our prior analysis, we did not account for any country-specific characteristics that may be correlated with the GDP per capita and do not change over time. In this section, we attempt to statistically remove these effects by including year and country fixed-effect in the following fixed-effect regression: \[log(GDP\ per\ capita)_{it} = \beta \ log(Middle\ Class \ population)_{it} + \gamma \ X_{it} + \epsilon_{it} \]
We also included in the vector X, several control variables such as income inequality and regime type, to account for other factors that could influence our dependent variable. Our data on income inequality comes from the Standardized World Income Inequality Database (SWIID), which has provided a wide data coverage by maximizing the comparability of income inequality among nations and years. We focused on the market Gini index before taxes, hypothesizing that a higher GDP per capita is associated with a more equitable distribution of income and wealth, thus indicating lower Gini coefficients. For the type of regime, we used the Polity2 Index, developed by the Center for Systemic Peace. This index evaluates and measures the degree of democracy and autocracy in a country, providing a score ranging from -10 (autocratic) to +10 (democratic). To control for the potential effect of the performance of each regime type on our dependent variable, we constructed and added to the regression a democracy and authoritarian score, both ranging from 0 to 10.
The outcome supports our earlier findings by demonstrating a positive relationship between the GDP per capita and the middle-class population. More precisely, for every 1% increase in the size of the middle class in a country, the GDP per capita increases by about 0.1%. The result remains generally the same after controlling for income inequality and regime type (democracy or authoritarianism).
These findings support the hypothesis that a growing middle class is associated with higher average income of a nation’s citizens. However, as previously stated, this relationship is reciprocal. Starting from here, we will now focus on the second direction of the correction between the middle class population and the Gdp per capita i.e we will investigate how the size of the middle class population changes in relation to Gross Domestic Product (GDP) per capita in African economies.
| (1) | (2) | (3) | (4) | (5) | |
|---|---|---|---|---|---|
| log(Middle-Class size) | 0.083 *** | 0.091 *** | 0.080 *** | 0.076 *** | 0.076 *** |
| (0.009) | (0.009) | (0.009) | (0.009) | (0.009) | |
| Gini index | 0.016 | 0.012 | 0.012 | 0.010 | |
| (0.008) | (0.008) | (0.008) | (0.008) | ||
| Regime type | 0.073 ** | 0.061 | 0.028 | ||
| (0.022) | (0.032) | (0.039) | |||
| Democracy Score | 0.003 | ||||
| (0.005) | |||||
| Authoritarian Score | -0.017 | ||||
| (0.012) | |||||
| N. obs. | 376 | 366 | 303 | 297 | 297 |
| R squared | 0.206 | 0.253 | 0.260 | 0.247 | 0.253 |
| F statistic | 85.592 | 54.343 | 30.145 | 20.630 | 21.213 |
| P value | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| All models are estimated by a panel fixed effect regression with the standard errors in parentheses. In each estimation, the GDP per capita(log) is the dependent variable. statistically significant coefficients are followed with stars. *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level | |||||
| (1) | (2) | (3) | (4) | (5) | |
|---|---|---|---|---|---|
| log(Middle-Class size) | 0.083 *** | 0.092 *** | 0.080 *** | 0.077 *** | 0.077 *** |
| (0.009) | (0.009) | (0.009) | (0.009) | (0.009) | |
| Gini index | 0.017 * | 0.013 | 0.012 | 0.010 | |
| (0.008) | (0.008) | (0.008) | (0.008) | ||
| Regime type | 0.073 ** | 0.061 | 0.028 | ||
| (0.022) | (0.032) | (0.039) | |||
| Democracy Score | 0.003 | ||||
| (0.005) | |||||
| Authoritarian Score | -0.017 | ||||
| (0.012) | |||||
| N. obs. | 376 | 366 | 303 | 297 | 297 |
| R squared | 0.211 | 0.262 | 0.264 | 0.252 | 0.258 |
| F statistic | 88.465 | 56.934 | 30.887 | 21.182 | 21.779 |
| P value | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| All models are estimated by a panel fixed effect regression with the standard errors in parentheses. In each estimation, the GDP per capita(log) is the dependent variable. statistically significant coefficients are followed with stars. *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level | |||||
| (1) | (2) | (3) | (4) | (5) | |
|---|---|---|---|---|---|
| log(Middle-Class size) | 0.055 *** | 0.060 *** | 0.052 *** | 0.051 *** | 0.051 *** |
| (0.007) | (0.007) | (0.007) | (0.007) | (0.007) | |
| Gini index | 0.006 | 0.004 | 0.002 | -0.001 | |
| (0.008) | (0.008) | (0.008) | (0.008) | ||
| Regime type | 0.073 ** | 0.042 | 0.019 | ||
| (0.023) | (0.033) | (0.041) | |||
| Democracy Score | 0.008 | ||||
| (0.005) | |||||
| Authoritarian Score | -0.020 | ||||
| (0.012) | |||||
| N. obs. | 375 | 365 | 302 | 296 | 296 |
| R squared | 0.160 | 0.204 | 0.207 | 0.211 | 0.213 |
| F statistic | 62.626 | 40.920 | 22.368 | 16.728 | 16.953 |
| P value | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| All models are estimated by a panel fixed effect regression with the standard errors in parentheses. In each estimation, the GDP per capita(log) is the dependent variable. statistically significant coefficients are followed with stars. *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level | |||||
| (1) | (2) | (3) | (4) | (5) | |
|---|---|---|---|---|---|
| log(Middle-Class size) | 0.062 *** | 0.068 *** | 0.060 *** | 0.058 *** | 0.059 *** |
| (0.007) | (0.007) | (0.007) | (0.007) | (0.007) | |
| Gini index | 0.010 | 0.009 | 0.007 | 0.004 | |
| (0.008) | (0.008) | (0.008) | (0.008) | ||
| Regime type | 0.070 ** | 0.044 | 0.009 | ||
| (0.023) | (0.033) | (0.040) | |||
| Democracy Score | 0.006 | ||||
| (0.005) | |||||
| Authoritarian Score | -0.023 | ||||
| (0.012) | |||||
| N. obs. | 375 | 365 | 302 | 296 | 296 |
| R squared | 0.188 | 0.232 | 0.236 | 0.236 | 0.242 |
| F statistic | 76.128 | 48.280 | 26.521 | 19.265 | 19.972 |
| P value | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| All models are estimated by a panel fixed effect regression with the standard errors in parentheses. In each estimation, the GDP per capita(log) is the dependent variable. statistically significant coefficients are followed with stars. *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level | |||||
| (1) | (2) | (3) | (4) | (5) | |
|---|---|---|---|---|---|
| log(Middle-Class size) | 0.044 | 0.028 | 0.014 | 0.018 | 0.019 |
| (0.025) | (0.025) | (0.025) | (0.025) | (0.025) | |
| Gini index | -0.010 | -0.011 | -0.012 | -0.015 | |
| (0.009) | (0.008) | (0.009) | (0.009) | ||
| Regime type | 0.066 ** | 0.030 | 0.008 | ||
| (0.025) | (0.036) | (0.045) | |||
| Democracy Score | 0.009 | ||||
| (0.006) | |||||
| Authoritarian Score | -0.022 | ||||
| (0.013) | |||||
| N. obs. | 376 | 366 | 303 | 297 | 297 |
| R squared | 0.009 | 0.009 | 0.035 | 0.049 | 0.051 |
| F statistic | 3.010 | 1.439 | 3.101 | 3.254 | 3.383 |
| P value | 0.084 | 0.239 | 0.027 | 0.013 | 0.010 |
| All models are estimated by a panel fixed-effect regression with the standard errors in parentheses. In each estimation, the GDP per capita(log) is the dependent variable. statistically significant coefficients are followed with stars. *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level | |||||
The United Nations estimates that Africa will experience more than half of the world’s population growth between now and 2050. Therefore, it is important to understand how this rapid population growth will impact both the African middle class and the working-age population in Africa.
The chart below shows the change in the population split by region of the African working-age population. It provides historical data going back to 1950 and a population projection to 2100 based on scenarios developed by the Wittgenstein Center.
In each scenario, the potential direction of population evolution is presented depending on the population’s growth scenario. For example, the second (medium development) scenario is the middle-of-the-road scenario and is the most likely path each country will take in the future. It combines the average fertility across all nations with average mortality, average migration, and the global education trend.
As we see in the charts below, there has been a significant increase in the working-age population over the last few years in all the African regions. It seems to continue growing, especially in Africa’s western and eastern areas. However, under the scenario of rapid development, it is anticipated that the working-age population in those regions will almost reach today’s levels by 2060.
From this point on, we will be curious about how the future growth in the working-age population will affect the middle class in Africa. But let us first consider the historical dynamics that existed between the African middle class and its working-age population
The working-age population of Africa is growing exponentially, at least in the western and eastern parts of the continent, as evidenced by the figures above. Therefore, it is crucial to understand the historical relationship between Africa’s middle class and its working-age population as well as how this relationship will change in the future.
The scatter plot below shows the middle class size (on the vertical axis) against the working-age population (on the horizontal axis) after a log transformation. The countries are represented by the points in the charts. To get a better understanding of the relationship between the middle class and the working-age population, we added a regression line to the scatter plot to capture the linear relationship between the two. We also added to the plot the confidence interval surrounding the line, which gives us an indication of the accuracy of the regression line. Furthermore, the confidence interval also helps to identify outliers which could affect the precision of the linear relationship.
The analysis of the graph reveals a positive correlation between the working-age population and the size of the middle class. This suggests that countries with a larger working-age population are likely to have a larger middle class. To confirm this hypothesis, we will perform in the following section a fixed-effect regression on our panel dataset, which consists of 46 African countries over the period of 2009 to 2020.
The results of the fixed-effect regression presented in the tables below show a positive relationship between the size of the middle class and the working-age population in Africa. We used a panel fixed effect regression to account for any unobserved heterogeneity across countries, as well as for time-invariant effects. This method provides more precise estimates of the impact of the independent variables and enhances the explanatory power of the model. Following Duflo and Banerjee definition of the middle class for example, the model show a positive and significant coefficient of 3.5. It shows that a 1% increase in the number of people who are working-age is associated with in a 3.5% increase in the number of people who are middle class. These results are robust given that they only slightly change when we control for income inequality and the GDP per capita. Income inequality and GDP per capita also seem to have an impact on the African middle class. Results show that while GDP per capita is positively correlated with the size of the middle class, income inequality plays the opposite role. If we use Duflo and Banerjee’s definition of the middle class as an example, we can see that a percent point (1%) increase in GDP per capita is correlated with a 0.3% increase in the size of the middle class, but that when income inequality is increased by one unit, the middle class size decreases by about 18%.
The future composition of the African middle class can be predicted using the strong correlation between the working-age population, income inequality, and GDP per capita.
| (1) | (2) | (3) | |
|---|---|---|---|
| log(Working-age population) | 3.558 *** | 3.162 *** | 2.755 *** |
| (0.217) | (0.251) | (0.261) | |
| log(GDP per capita) | 0.694 ** | 1.017 *** | |
| (0.264) | (0.282) | ||
| Gini index | -0.162 *** | ||
| (0.038) | |||
| N. obs. | 380 | 376 | 366 |
| R squared | 0.447 | 0.465 | 0.501 |
| F statistic | 268.816 | 142.939 | 107.053 |
| P value | 0.000 | 0.000 | 0.000 |
| All models are estimated by a panel fixed effect regression with the standard errors in parentheses. In each estimation, the log(middle class population size) is the dependent variable. statistically significant coefficients are followed with stars. Data on GDP per capita comes from the Worl Bank (WDI) and income inequality data comes from Standardized World Income Inequality Database (SWIID). *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level | |||
| (1) | (2) | (3) | |
|---|---|---|---|
| log(Working-age population) | 3.610 *** | 3.194 *** | 2.761 *** |
| (0.218) | (0.252) | (0.262) | |
| log(GDP per capita) | 0.732 ** | 1.084 *** | |
| (0.265) | (0.283) | ||
| Gini index | -0.169 *** | ||
| (0.038) | |||
| N. obs. | 380 | 376 | 366 |
| R squared | 0.451 | 0.471 | 0.509 |
| F statistic | 273.287 | 146.340 | 110.448 |
| P value | 0.000 | 0.000 | 0.000 |
| All models are estimated by a panel fixed effect regression with the standard errors in parentheses. In each estimation, the log(middle class population size) is the dependent variable. statistically significant coefficients are followed with stars. Data on GDP per capita comes from the Worl Bank (WDI) and income inequality data comes from Standardized World Income Inequality Database (SWIID). *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level | |||
| (1) | (2) | (3) | |
|---|---|---|---|
| log(Working-age population) | 3.683 *** | 3.011 *** | 2.437 *** |
| (0.330) | (0.384) | (0.405) | |
| log(GDP per capita) | 1.206 ** | 1.813 *** | |
| (0.404) | (0.437) | ||
| Gini index | -0.165 ** | ||
| (0.058) | |||
| N. obs. | 379 | 375 | 365 |
| R squared | 0.273 | 0.292 | 0.323 |
| F statistic | 124.655 | 67.762 | 50.798 |
| P value | 0.000 | 0.000 | 0.000 |
| All models are estimated by a panel fixed effect regression with the standard errors in parentheses. In each estimation, the log(middle class population size) is the dependent variable. statistically significant coefficients are followed with stars. Data on GDP per capita comes from the Worl Bank (WDI) and income inequality data comes from Standardized World Income Inequality Database (SWIID). *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level | |||
| (1) | (2) | (3) | |
|---|---|---|---|
| log(Working-age population) | 3.707 *** | 2.972 *** | 2.390 *** |
| (0.309) | (0.356) | (0.373) | |
| log(GDP per capita) | 1.319 *** | 1.874 *** | |
| (0.374) | (0.403) | ||
| Gini index | -0.193 *** | ||
| (0.054) | |||
| N. obs. | 379 | 375 | 365 |
| R squared | 0.303 | 0.331 | 0.368 |
| F statistic | 144.417 | 80.971 | 61.881 |
| P value | 0.000 | 0.000 | 0.000 |
| All models are estimated by a panel fixed effect regression with the standard errors in parentheses. In each estimation, the log(middle class population size) is the dependent variable. statistically significant coefficients are followed with stars. Data on GDP per capita comes from the Worl Bank (WDI) and income inequality data comes from Standardized World Income Inequality Database (SWIID). *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level | |||
| (1) | (2) | (3) | |
|---|---|---|---|
| log(Working-age population) | 0.329 ** | 0.287 * | 0.326 * |
| (0.114) | (0.133) | (0.142) | |
| log(GDP per capita) | 0.041 | -0.064 | |
| (0.140) | (0.153) | ||
| Gini index | -0.019 | ||
| (0.020) | |||
| N. obs. | 380 | 376 | 366 |
| R squared | 0.024 | 0.023 | 0.028 |
| F statistic | 8.314 | 3.862 | 3.105 |
| P value | 0.004 | 0.022 | 0.027 |
| All models are estimated by a panel fixed effect regression with the standard errors in parentheses. In each estimation, the log(middle class population size) is the dependent variable. statistically significant coefficients are followed with stars. Data on GDP per capita comes from the Worl Bank (WDI) and income inequality data comes from Standardized World Income Inequality Database (SWIID). *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level | |||
As we show in the previous regression, there is a strong link between the African-working class and its middle class. With this link in mind, we were able the get a projection of the African middle class per region, per middle-class definition, and under different scenarios. The breakdown per definition and scenario is fascinating: it gives an idea about the African middle-class’s future in various settings.