Trends in Mediterranean Inequalities 1950-2015

This article is aimed at analysing the trends of economic, social and institutional inequality among the Mediterranean countries in the period 19502015. After the examination of the inequalities in GDP per capita among and within nations, we present a Human Development Index (HDI) that includes a measure of democratic achievements. Main result is that inequalities in income, after the rise from the 1950s onwards, declined from the start of the twenty-first century. Inequalities in HDI, instead, constantly diminished in the period under examination, while a process of democratization occurred. On the whole, despite the convergence among Mediterranean countries, economic inequalities are much deeper than those in social indicators.

Divided among the three continents of Europe, Asia and Africa, for a long time the Mediterranean formed an interrelated set or a "world economy": an area of the globe where economies, cultures and societies interacted with one another (Braudel, 1979). Even today, within the global economy, the Mediterranean countries are joined together both by flows of production factors and trade of goods (Tirkides and Theophanous, 2011). Massive legal and illegal flows of people daily move from the less developed southern and eastern economies towards the North of the Mediterranean. The trade of goods between the EU and the MEDA (Mediterranean countries in the Euro-Mediterranean Partnership) represents in 2015 around 9 percent of the total EU external trade (European Commission, 2015). Energy sources play a central role in these exchanges. The EU nations are the main investors in the countries of southern Mediterranean (Ferragina, 2010). This paper's aim is to contribute to the debate in progress on global inequalities through the analysis of economic, social and political disparities in the Mediterranean and their changes from 1950 onward. In section 1 we focus on income inequalities (both between and within the Mediterranean countries); in section 2, we discuss the method for the calculation of a Mediterranean Human Development Index (HDI) and present our results. We will see that, although declining from about 2000, Mediterranean economic disparities are still remarkable, while those in social and political indicators are much less profound.

Related literature
On the World scale, increasing international divergence among nations in living standards was a main feature of economic development from the start of modern growth up to the 1950s or 1960s (Bourguignon and Morrisson, 2002). From then on, divergence and inequality stabilised on a high level or slowed down until about 2000 (Schultz, 1998;Xavier Sala-i-Martin, 2006;Milanovic, 2006Milanovic, , 2011Milanovic, , 2012Berry and Serieux, 2006;Ricolfi et al. 2015). However, only from the beginning of the 21 st century, the trend of inequality in income levels across countries seems to have clearly inverted (Milanovic 2013a). The decline in inequality from 2000 onwards has been largely caused by the sustained growth of populous countries, mainly China and India and other emerging economies such as Indonesia and Brazil (Clark, 2011;Milanovic, 2013a). Yet, rising inequality within states in income distribution from the 1970s appears to limit, at least in part, the effects of between states increasing equality in average incomes. Some economists share the opinion that, in the last two centuries, international inequality described an inverted U curve, similar to that suggested by Simon Kuznets (1995) in personal income distribution during the first wave of modern growth, and by Jeffrey Williamson (1965) in regional inequality among regions within the modernising nations. Not all scholars, however, subscribe to the downward global decline of inequality among nations (Heshmati, 2006, 84;Korzeniewicz and Moran, 1997;Anand and Segal, 2008).
The trend of inequality among the Mediterranean countries has been investigated by relatively few studies. The main results indicate how disparities in per capita GDP levels among the Mediterranean economies increased until 1970, stabilised until the mid-80s, and then increased (Daniele and Malanima, 2013). Only in the first decade of 2000s, the slowing down of growth rates of more developed economies fostered a sort of convergence among the Mediterranean nations (Daniele and Ansani, 2014). The partial failure of the Euro Mediterranean Partnership, that is the absence of a process of trade and economic integration in the Mediterranean region, was one of the causes of the divergence between the northern Mediterranean countries and the Middle East and North African nations (Amendola, 2011). A review of the changes in inequality in income distribution has been offered by Capasso and Astarita (2011), that assembled Gini index data, from the mid 1960s to mid 2000s, for some Mediterranean nations. Results show how, from 1960s onward, inequality slightly declined in the high-income Mediterranean countries and, also, in Turkey and Morocco. After the break-up of Yugoslavia, in the nations of the Adriatic region, inequality in income distribution sharply increased during the 1990s; then, in the 2000s, inequality declined, albeit with remarkable national differences. The nexus between income distribution and economic growth has been analysed, in a panel of 18 Mediterranean countries for the period 1995−2012, by Amendola and Dell'Anno (2014), which found a statistically significant non-linear relationship (an inverted Ushaped curve) between inequality and growth. Mediterranean countries characterised by medium income inequality (a pre-redistribution Gini index of approximately 40-45) had the highest growth rates in the considered period.

Inequalities among nations
In 2015, with 510 million inhabitants on the whole, the Mediterranean countries represented 7 percent of the World population and produced 10 percent of World product. In the first decade of the third millennium, 37 percent of the entire Mediterranean population lived in the four richest countries of the Latin region: Portugal, Spain, France and Italy (Figure 1). This 37 percent produced 70 percent of the whole gross product and consumed 60 percent of the total commercial energy (Bartoletto and Malanima, 2014).
Although remarkable, the economic disparities on the Mediterranean scale are, however, much lower than on the World scale. The World's richest countries enjoy a per capita GDP sixty times as great as that of the poorest. In the Mediterranean, the average GDP in France is six times that of Algeria.  and the other southern and eastern countries. Although the economies of these countries are far from homogeneous, the average level of development is sensibly lower than that of the northern Mediterranean. Among these two groups of countries, disparities grew until 2000, since the rates of per capita GDP growth were in the North higher than in the South-East, while the increase in population was lower (Table 1).    Despite the convergence, inequalities in living standardsthose regarding material conditionare still profound and continue to represent, together with conflicts, the fundamental cause of the massive northwards migratory movements from the South-East. The relative economic progress in the South and East, does not depend, however, on the industrial success or innovation, but on energy exports (Algeria, Libya and, to a lesser degree, Tunisia, Egypt and Syria) and exports of other raw materials, such as cotton or phosphate (Syria, Egypt, Tunisia, Morocco), together with the rise of tourism (Corm, 2011: 129).
In order to set this change in Mediterranean disparities in a long-term perspective, we lack data on GDP before 1950 for most Mediterranean countries. We can avail, however, of information on real wages for ten countries, dating back as far as the middle of the nineteenth century for any fifty years (Caruana Galizia, 2015). The results we reach are synthesized through the values of the Theil index in Figure 5. We see that, until about 1950, inequalities among Mediterranean countries were relatively low. They increased fast after the Second World War, when the countries of the Latin area or North Mediterranean experienced high rates of growth.
After that period, Mediterranean inequalities diminished. On the whole, the Mediterranean economies described a clear turned U curve in terms of per capita GDP.

Inequalities within nations
So far we calculated Mediterranean inequality as if, in each country, everybody enjoyed the same average income. This concept of inequality does not take into account inequality in personal income distribution (Toniolo and Walker, 2000;Milanovic, 2005). While World inequality among nations is diminishing, the fast increase of inequality within some populous countries, such as China (Lin and Xu, 2008), might neutralise, at least in part, the process of convergence (Atkinson, Piketty and Saez, 2011;Liberati, 2015;Morrisson and Murtin, 2011;IMF 2013;Stiglitz, 2012). On the World scale, convergence seems to be much weaker or inexistent at all whenever personal income distribution is included (Clark, 2011;Milanovic, 2013a). A widespread opinion is that a "sustained increase in income inequality started in the late 1970s in practically all developed nations" (Milanovic, 2016); although the trend and chronology is still controversial (Atkinson, 2015;Dabla-Norris et al., 2015;Ricolfi et al., 2015).
Data on personal income distribution within the Mediterranean nations is, in some cases, lacking and is not always homogenous (Atkinson and Brandolini, 2001). We can draw, however, the overall trend of inequality among individuals. In Table 2  Firstly, the number of data notably varies according to the available surveys. Typically, the number of observations is higher for developed countries. Data are reported for five-years periods: in the case of countries with more available Gini coefficients the average is taken. Secondly, the sources of data are different. The main source we used is the dataset compiled by Milanovic (2013b), which presents "standardized" Gini values for a large sample of countries on the basis of eight primary household surveys. For some years and countries (as specified in the sources of Table 2), we supplemented these data with Gini coefficients of disposable income from other sources (World Bank, Eurostat and OECD datasets). Finally, we used Gini coefficients based on net income, with the exceptions of Egypt and Morocco, for whom they refer to gross income. The heterogeneity of data sources and their low reliability for some countries are an obvious limit to the analysis.
For the Mediterranean, we can also avail of data on the income of the top 10 percent of the population; sometimes used as an indicator of inequality in personal income distribution (as done by Atkinson, Piketty and Saez, 2011;Piketty, 2014).
Historical data on income accruing to the top 10 percent for three Mediterranean countries such as France, Spain and Italy, suggests diverse trends. While in Italy inequality in income rose from 1984, it was diminishing in France from 1964 and was almost stable in Spain (Atkinson, Piketty and Saez, 2011). Data provided by the World Bank (World Development Indicators, 2015) for the income held by the richest 10 percent concerns only some countries in the period 1985-2014. A comparison between this data and Gini indices reported in Table 2 shows a relatively high correlation when data becomes more plentiful, that is for the periods 2000-05 and 2005-10 (r=0.90 and r=0.85 respectively).
Following Milanovic (2005Milanovic ( , 2013, we computed the Gini index of total inequality GT for the Mediterranean region, combining between countries and within countries inequality according  The level of inequality, measured by Gini indices, increased by 15-20 percent over the thirty years. Whenever we include the available figures for the previous decades 1940-79 (whose reliability is, however, dubious and therefore are not presented in Table 2 and Figure 6), inequality within countries adds 20-25 percent to the Gini index of inequality among nations.
The decomposition of the Gini index of total inequality into inequality among nations, confirms how Mediterranean inequality depends primarily on the different economic conditions among the nations and especially between North and South-East.

Inequality in human development
Disparities in average income, both between and within countries, measure only one element of the international differences in well-being. Embodying the Amartya Sen's capabilities approach to individual well-being (Sen, 1999) Table 3.
Although our sources differ from those used by UNDP (2015), the result we reach is highly correlated to that by UNDP (for the year 1980, r=0.97; for 2000 r=0.96; for 2014 r=0.93). We report also the HDI for the Mediterranean as a whole by weighting, for each index, national data by the respective population share.
The trends in the three dimensions and in the HDI for the Mediterranean are reported in   Figure 8 outlines the trends in the HDI for five Mediterranean regions. Regional estimates have been calculated by weighting national data by the share of respective population on regional total and, then, computing the dimension indices according to the methodology described above.
Between 1960 and 2014 the HDI improved in all regions, even though remarkable differences in levels continue to exist. According to our estimates, North Africa is the region with lower HDI.
The inequality in each dimension is estimated through the Atkinson inequality measure A (Atkinson, 1970). These inequality measures have been worked out by UNDP for 2013. For life expectancy, the coefficient is computed through the current inequality in mortality patterns.
Inequality in the education dimension is proxied by inequality in years of schooling of the adult population, while inequality in income is provided by Gini indices (UNDP, 2013). We can then estimate the loss due to inequalities in each dimension as: We see in Table 4 that, in the Mediterranean, the I-HDI is 13.5 percent lower than the HDI.
In Morocco and Egypt, we register the main losses due to inequality. On the whole, countries with less human development also have more dimensional inequality and thus larger losses in human development due to inequality, while people in developed countries experiences the least inequality in human development.

Inequality in political rights
The role of political institutions in economic development has been widely recognized by scholars, although the results on their consequences in terms of standard of living are not conclusive. Some countries' experience indicates, in fact, how economic growth can flourish under different political institutions; non-democratic regimes included (Barro, 1996). There is, however, a growing evidence that democracy fulfils a positive and sizeable effect on growth (Papaioannou and Siourounis, 2008;Acemoglu et al., 2014). The nexus between democracy and human development is, at least theoretically, less uncertain. Since democratic systems empower people, including the poor, the prevailing view is that elected governments are more concerned with people interests than the autocratic ones. Empirical studies, mainly based on cross-countries regressions, reach, however, mixed conclusions. While some of them confirm this view (Gerring et al., 2012), others find that the correlation between democracy and human development, differently measured, is weak or totally absent (Ross, 2006). In agreement with the fundamental approach by Sen (1999), we can argue, however, that democratic achievements express human development by themselves, regardless of their possible effects on specific human indicators, such as life expectancy or education (UNDP, 2002). This concept has been clearly stated by Inglehart and Welzel (2005: 152): "human development advances with the growth of three components: (1) objective capabilities, based on socioeconomic resources, that enable people to act according to their own choices; (2) subjective motivations, based on self-expression values, that emphasize acting according to one's autonomous choices; (3) and legal entitlements, based on civil and political liberties, that allow people to act on the basis of their autonomous choices".
In order to take democracy achievements in the Mediterranean countries into ac-count, we introduce a modified HDI that includes a fourth dimension index concerning political institutions.
To measure institutions, we used the Polity 2 variable drawn from the Polity IV dataset ( Marshall, Gurr and Jaggers, 2014; The estimates for twenty countries and the Mediterranean as a whole are presented in Table   5. Comparing this data with the HDI, we can appreciate the gains (losses) due to different political regimes. During the period under examination, however, a process of democratization occurred in the Mediterranean. This process interested, for example, the Balkan region following the dissolution of Yugoslavia and, as regards Polity IV classifications, also some countries such as Tunisia and Jordan which progressed from autocracy to democracy. Different political regimes continue to characterise the Mediterranean area: in 2014, Syria was classified as an autocracy (-9), Morocco and Egypt (-4) as "closed anocracies", Algeria (2) as "open anocracy", according to the definition of "anocratic systems" as neither fully democratic nor fully autocratic. On the whole, however, from 1960 to 2014, the HDI-P for the Mediterranean notably improved, increasing from 0.46 to 0.73 (+59 percent).
These results show a decreasing trend and then a kind of "institutional convergence" among the Mediterranean nations. The process of democratisation is summarized by Figure 10 that reports the trend of the unweighted average of the Polity2 index for the Mediterranean countries.
This trend is similar to that for the whole World, where the number of democratic countries notably increased during the second half of twentieth century: actually, it doubled since the 1980s until now (Wejnert, 2014 The causes of the diffusion of democratic institutions are diverse. According to the modernization theory (Lipset, 1959), it is fundamentally due to socioeconomic factors: increasing income and education, formation of a middle class, transformation of social values associated with cultural and institutional changes. Independently of the proximate causes, in the Mediterranean, democratization and economic development went hand in hand. In the 1970s, democracy progressed in southern European states (Portugal, Spain, Greece); in the 1990s, a democratic transition occurred in the Balkan; more recently, on the wave of the Arab Spring, a request of more representative political systems emerged in North Africa and Middle East (Robbins, 2015). While in some countries changes toward more representative institutions have occurred, in others revolutionary waves engendered uncertain outcomes or were followed by instability and conflicts.

Economic and socio-political convergence
A comparison of our results for economic and socio-political inequality in the Mediterranean reveals that inequality is in 2014-15 deeper from an economic than socio-political viewpoint. Per capita GDP of the richest economy in the hierarchy (France) is 5.8 times that of the weakest (Algeria), while HDI in the first country of the hierarchy (Israel) is 1.71 that of the lowest (Morocco) and HDI-P in the most advanced nation (again Israel) is 2.38 times that of the lowest (Syria).
In order to specify the trend of inequality we use the ordinary procedure to test absolute βconvergence. On the basis of the previous analysis, we estimated the following equation: where g is the yearly rate of per capita GDP growth and y is the initial level of per capita While in equation (8), that is in the period 1950-2000, the coefficient β is not significant and has a positive sign that indicates divergence, equation (9)  The results confirm a significant convergence process in both variables. Figure 11 summarises the relationship between inequalities in income and in socio-political indicators. The relationship GDP per capita and HDI-HDI-P is not linear and can be proxied by a log regression curve. We see on the horizontal axis that while economic inequality is relatively remarkable, socio-political inequalities are still wide in countries with per capita GDP lower than 10,000 international 1990 dollars PPP. As soon as per capita GDP exceeds 10,000 dollars, socio-political inequality shrinks in the range 0.80-0.90.

Fig. 11. Relationship between per capita GDP and socio-political indicators 2014
Note: GDP per capita 1990 $ PPP. Sources: see text.

Conclusion
On a global scale, inequality peaked immediately after the Second World War.