The Redistributive Preferences of the Well-Off

Demand for redistribution varies across income group, but the income-related variation appears to be heterogeneous across countries. Beyond the sole absolute level of inequalities, we argue that socio-economic attitudes toward redistribution also depend on the structure of inequalities and more precisely on the relative position of each group in a specific distribution of income. More precisely we hypothesize that the structure of inequality immediately below a group is indicative of the economic cost faced in case of downward mobility; the concentration of incomes above a group (and the ability to tax them) determines the potential gain of this group from raising tax and benefit provision. We test this argument empirically by examining the attitudes of the upper income quintile, relatively to other income groups. In line with our hypothesis we find that the well-off, who are generally less supportive of redistribution than lower income group, (i) tend to be more in favour of redistribution when income inequalities between P90 and P50 rise; (ii) decrease their support for redistributive policy when national income becomes more concentrated (and more subject to tax avoidance) at the very top of the distribution (Top 0.1%).

1. Argument

The present paper focuses on the redistributive preferences of the well-off. We statistically define the well-off as the fifth quintile in the income distribution, meaning that individuals belonging to this group are clearly wealthier than the median income (to be found in the third quintile) but also distinct from the Top 0.1% income who belong to the super rich (to be sure, very rich people are usually underrepresented -if not inexistent- in survey data). We consider the political preferences of the well-off to be of key interest for two reasons: (i) their political weight might definitely prevail over their electoral weight; (ii) they are in a pivotal position between households that have a reduced capacity to contribute to taxation, and households with greater facility to achieve fiscal optimization. Indeed, the “upper middle class” often appears in the media or in the political discourse as the primary taxpayer for the redistribution system.

Our analysis starts with the following puzzle: there is a strong cross-country variation in the distribution of preferences for redistribution across income groups. To explain this stylized fact, we refine the standard political economy argument that highlights the impact of average level of inequalities on preferences for redistribution in two ways. First, we argue that preference formation is determined by the structure of income inequalities, not only by its level. Indeed, while we acknowledge that inequalities provide information about the distribution of taxable income (which might explain demand for redistribution), we posit that the relative position of individuals given the inequality structure of the economy is informative of the economic cost of moving downward in the social ladder. Second, we argue that preference formation is also determined by individuals’ ability to benefit from the tax-benefit system given the (exogenously) constraint taxation of top-incomes. Piketty and Saez (2007) indeed show that the tax rate of the Top 0.1% income is not correlated to the general level of taxation: when the general tax rate increases, the tax rate for top incomes is not necessarily raised.

1. Preview of results

The econometric analysis uses newly available data on inequalities (World Top Incomes Database), mixed with OECD macro data (OECD earnings distribution database and OECD income distribution and poverty database) and micro survey data on preferences for redistribution (ISSP data, several modules) for a maximum of 44 countries over the time span 1980-2013. The Gini index being an aggregate measure of the level of inequality, we preferably use the P90-P50 index and the income share of the Top 0.1% incomes to gather information on the level of inequality immediately below and above the category of interest. Doing so, we directly test the impact of inequality structure on individual support for redistribution.

Our findings highlight two key determinants of the preferences of the well-off. First, a higher dispersion in the upper half of the income distribution (high P90-P50 index) increases the demand for redistribution for everyone but this impact is even stronger for the well-off. This presumably reflects the risk of downgrading: Q5 individuals who consider the size of the step turned out to be greater ask for more social insurance. Second, the share of income received by the Top 0.1% negatively influences the support for redistribution; this is statistically significant for the Q5 individuals, presumably because of the inability to force highly concentrated revenues to contribute to the welfare state. Indeed, households can expect a redistributive gain only to the extent that they are able to tax incomes above them. The well-off are aware that any tax increase will probably be passed on to themselves, and react negatively to a distancing of the Top 0.1%. This might be especially true at a time when corporate income tax is constrained by international tax competition.

To sum up, our findings show that a higher dispersion of earnings in the upper-half of the distribution increases the support for redistribution of the well-off, while a higher compression of top incomes decreases it. We assume that the first mechanism is related to insurance reasons, while the second mechanism is linked to the renowned incapacity of the state to tax the super-rich.