Income inequality has increased significantly in many developed and developing economies over the last decades, with significant variations across countries and regions. At the same time, the rise of emerging countries has contributed to the reduction of inequality between countries. These conflicting trends have attracted considerable interest among academics, policymakers, and the global public.
Unfortunately we face important limitations in our collective ability to measure income inequality. During the past 15 years, following up on Kuznets' (1953) pioneering attempt, a number of authors have used administrative tax records to construct long-run series of top income shares.1 Yet they have two important shortcomings. First, they do not offer for the bottom segments of the distribution the same detailed decomposition as for the top part. Second, there is a large gap between national accounts, which focus on economic aggregates and macroeconomic growth, and inequality studies, which focus on distributions using survey and tax data but usually without trying to be fully consistent with macro aggregates.
Two recent papers – Piketty et al. (2018) for the US and Garbinti et al. (2018) for France – attempt to bridge the gap between national accounts and inequality studies more systematically than has been done in the past. For France, our contribution is both methodological and substantive.
At the methodological level, we combine national accounts, tax, and survey data to build ‘distributional national accounts’ (DINAs), that is, homogenous annual series on the distribution of national income over the 1900-2014 period. In contrast to previous attempts to construct top income series for France, which are based upon fiscal income, our estimates capture 100% of national income and cover the entire distribution, from bottom to top percentiles. These series also offer a detailed breakdown by age, gender, and income categories over the 1970–2014 period. This work focuses upon the distribution of pre-tax income.2
At a more substantive level, our new income series reveal two main facts. First, they deliver higher inequality levels than the usual tax-based series for the recent decades, because the latter miss a rising part of capital income. In particular, regarding how growth has benefited to the population, very different patterns appear between the 1950-1983 period – where growth mainly benefits to most of the population – and the 1983-2014 period – where it only benefits to the very top percentiles. Second, gender inequality in labour income declined in recent decades, albeit fairly slowly among top labour incomes (for example, the female share among top 0.1% earners was only 12% in 2012, compared with 7% in 1994 and 5% in 1970).
Long-run trends in income inequality: The ‘Thirty Glorious Years’ are not over for everyone
Figure 1 depicts the long-run evolution of income inequality in France over the 1900-2014 period. First, it is useful to have in mind the general evolution of average income in France. Per adult national income has increased considerably in the long run. However, the growth has been far from steady and happened mostly during the 1945-1980 – often referred to as the ‘Thirty Glorious Years’ in France. The growth rate of per adult national income has been negative during the 1900-1945 period (-0.1% per year), then jumped to 3.7% per year over the 1945-1980 period, and finally was divided by almost four over the 1980-2014 period (0.9% per year).3 The major long-run transformation is the rise of the share going to the bottom 50% (the ‘lower class’) and the middle 40% (the ‘middle class’) and the decline of the share going to the top 10% (the ‘upper class’).
Figure 1 Income shares in France, 1900-2014
Indeed, during the 1914-1945 period, the top 10% income share fell abruptly from more than 50% of total income on the eve of WWI to slightly more than 30% of total income in 1945. A rise in inequality appears during the reconstruction period and up until 1967-1968, followed by a large reduction of inequality between 1968 and 1983, which is well-known to be due to a large compression of wage inequality4 and a significant reduction of the capital share.
A new increase in inequality starts around 1983, as the newly elected left-wing government puts an end to the very fast rise in wages (substantially faster than output growth, particularly for bottom wages) that had occurred between 1968 and 1983. The top 10% income share declined somewhat after the 2008 financial crises, but it is still significantly higher than in the early 1980s. Most importantly, the top 1% income share (Figure 2) rises significantly between 1983 and 2007, from less than 8% of total income to over 12% over this period, i.e. by more than 50%. This is less massive than in the US,5 but this is still fairly spectacular. Between 2008 and 2013, the top 1% share fluctuates between 10% and 12%, which is still significantly larger than in the low inequality point of the early 1980s. Moreover, the higher we go at the top of the distribution, the higher the rise in top income shares.
Figure 2 Top 1% income share in France, 1900-2014
Given the relative stagnation of average income in France since 1980 (at least as compared to the previous decades), this spectacular rise of very top incomes has not gone unnoticed. As in other countries, the large increase in very top managerial compensation packages in recent decades was largely covered by the media and has shown to the broader public that the Thirty Glorious Years are not over for everyone. Indeed, during the 1950-1983 period, per adult real income rose at almost 4% per year for most of the population, except for very top percentiles, whose incomes grew at about 1.5% per year. Between 1983 and 2014, we observe the opposite pattern: for most of the population real growth rates were about 1% per year or less, except for very top percentiles, who enjoyed real growth rates up to 3% per year (Figure 3). Rising inequality in recent decades has been less massive in France than in the US (where bottom 50% pre-tax incomes did not grow at all), but it has nevertheless been fairly substantial.6
Figure 3 Annual real growth rate by time periods in France
This sharp rise of very top incomes since the 1980s is due both to top capital incomes and top labour incomes. Regarding the rise of top capital incomes, one should distinguish between two effects:
- the rise of the macroeconomic capital share on the one hand (an evolution that is due to a combination of economic and institutional factors, including the decline of labour bargaining power and the lift of rent control, and that was reinforced by corporate privatization policies); and
- the rise of wealth concentration on the other hand.
Regarding the rise of top labour incomes, it is worth stressing that it occurred only at the very top (i.e. above the 95th percentile) and mostly within the top 1% and top 0.1%. It is difficult for standard explanations based upon technical change and changing supply and demand of skills to fully explain this concentration of rising inequality at the very top. It seems more promising to stress the role of institutional factors governing pay setting processes for top managerial compensation, including corporate governance, the decline of unions and collective bargaining processes, and the drop in top income tax rates (see Piketty et al. 2014).
A declining but still high gender gap
Finally, our detailed breakdowns by age and gender allow us to explore new dimensions of inequality dynamics together with the top income dimension. The general conclusion is that although gender gaps have declined significantly in recent decades, they are still extremely high. As of 2012, men earn in average 1.25 times more than women when they are 25and this gender gap continuously increases with age, up to 1.64 when they are 65 (Figure 4). This reflects the fact that women face much lower probabilities of being promoted to higher-wage jobs during their career.
Figure 4 Gender gap by age, France, 1970-2012
Taking a longer temporal perspective, we find that the gender gap used to be much larger. In particular, the French labour model around 1970 appears clearly to be a ‘patriarchal’ model, with men earning 3.5 to 4 times more labour income than women between the ages of 30 and 55. In effect, money income at that time was not something destined for women. While there is a continuous decline in gender inequality in labour income during recent decades (partly due to a dramatic increase of the share of working women), women still do not access higher-paying jobs. In 2012, the female share is only 16% among top 1% labour earners (and 12% among top 0.1%, compared with 7% in 1994 and 5% in 1970), with a very moderate upward trend observed since 1994 (Figure 5).
Figure 5 Share of women in fractiles of top labour incomes in France, 1970-2012
Source: Garbinti et al. (2018).
Bozio A, B Garbinti, J Goupille-Lebret, M Guillot and T. Piketty (218), “Inequality and Redistribution in France: Evidence from Post-Tax Distributional National Accounts (DINA)”, PSE, in progress.
Garbinti B, J Goupille-Lebret J and T Piketty (2018), “Income Inequality Dynamics in France 1900-2014: Evidence from Distributional National Accounts (DINA)”, Journal of Public Economics 162: 63-77
Kuznets, S (1953), Shares of Upper Income Groups in Income and Savings, National Bureau of Economic Research.
Piketty T, E Saez and G Zucman (2018), “Distributional National Accounts: Methods and Estimates for the United States”, Quarterly Journal of Economics 133(2): 553-609
Piketty T, E Saez and S Stantcheva (2014), "Optimal Taxation of Top Labor Incomes: A Tale of Three Elasticities", American Economic Journal: Economic Policy 6(1), 230-271
 See for instance Piketty (2001, 2003), Piketty and Saez (2003), Atkinson and Piketty (2007, 2010), Alvaredo et al. (2011-2017).
 In a companion paper (Bozio et al. 2018), we include taxes and transfers in our series in order to measure the post-tax post-transfers distribution of income.
 We observe similar patterns in most European countries and in Japan, and to a lesser extent in the US and in the UK (where the shocks created by WWI and WWII were less damaging than in Continental Europe and Japan).
 This large compression of wage inequality was driven in particular by very large increases in the minimum wage following the 1968 social protests.
 In the US the top 1% share has reached about 20% of total income.
 In the bottom 20% of the distribution we observe negative or below average growth between 1983 and 2014, reflecting rising unemployment and under-employment. Between the 20th and the 40th percentile, cumulated growth has been somewhat above average (about 40-50%, as opposed to less than 30% between the 40th and the 95th percentiles). This reflects the fact that bottom wages and pensions have increased somewhat more than wages and pensions in the middle and upper middle of the distribution.
 Taking into account all men and women in a given age group, whether they work or not.