Wealth inequality has characterised societies over many centuries. Two surprising aspects of the phenomenon are its extent and its persistence: a tiny fraction of the population owns a substantial share of a country’s wealth, and the extent of wealth inequality rarely changes from one year to the next. To give an extreme example of wealth concentration, suppose one could expropriate the wealthiest American resident (Elon Musk) and re-distribute his wealth to the poorest two million American households. This ‘Robin Hood’ policy would be enough to lift these households to middle class.1
A quick look at the literature
A substantial amount of research has been devoted to understanding what causes the huge concentration of wealth at the top (see Benhabib and Bisin 2019 for a recent review) and what leads this inequality to persist over time – even across several centuries (Barone and Mocetti 2016). One conclusion is that to rationalise persistent wealth inequality, one needs at least three ingredients:
- systematic heterogeneity in returns to wealth, whereby people differ persistently in how much capital income they are able to generate from their wealth;
- some intergenerational correlation in wealth and possibly in its returns; and
- some sorting of the wealthy at the time of family formation.
In previous work with Andreas Fagereng (Fagereng et al. 2020), we examined population data from Norway and found that the first two features receive large empirical support. Our findings have been confirmed for several other countries.
In new work (Fagereng et al. 2022), we focus on the role of ‘assortative mating’. Assortative mating has been neglected in the literature, only being used to argue that children of wealthy parents tend to marry the offspring of similarly wealthy parents. There are two important issues that the literature has left unanswered. First, do people really match just based on their parents’ wealth at the time of marriage, or is personal wealth relevant as well (or even more)? Second, do people match not only based on the level but also the growth rate of their wealth (i.e. on wealth returns)? Since, in all societies, marriage (and in a broader sense, cohabitation) is the predominant institution of household formation, the two questions help us understand inequality. What we typically measure is inequality across households, and that is mediated by the assortative mating process. If people assort on their wealth at marriage, wealthy women will end up marrying wealthy men, amplifying wealth concentration.
If people with a higher ability to generate returns marry spouses with a similar ability, wealth over the lifespan of these couples may also potentially grow faster than the wealth of those with lower returns, leading to increasingly diverging levels of wealth – and thus greater inequality – as couples move over their life course.
Our contribution: Sorting on whose wealth?
In our recent work, we highlight several novel facts regarding the assortative mating process along the two dimensions of wealth and returns to wealth and shed new light on how marital sorting affects wealth inequality at the time of marriage and also over the life cycle of the couples. We use population data from Norway, which are distinct (and unique) in that one can observe the wealth of the whole population before and after marriage and the returns to wealth of each individual before marriage and those of the couple after marriage.
Our data confirm the findings from previous literature that marriages are assorted on wealth. However, since we can match individuals with their parents and observe parental wealth, we can also study whether people sort on their own wealth or on their parents’ wealth. Differently from previous literature (e.g. Charles et al. 2013), we find that what matters is not parents’ wealth but the spouses’ own pre-marriage wealth: once the latter is accounted for, parents’ wealth plays no significant role in explaining assortative mating. Sorting on own wealth is significant, and its strength is documented in Figure 1, which shows a ‘heat map’ plotting the number of marriages observed for each combination of ventiles of the bride’s pre-marriage wealth and ventiles of the groom’s pre-marriage wealth.
Figure 1 Heat map of assortative mating on wealth
On the map, darker areas correspond to lower frequencies. Clearly, more marriages are observed along the left-to-right diagonal, consistent with assortative mating. Sorting on own wealth is distinct from sorting on income or education. The heat map has exactly the same property if we look at spouses that have the same level of education or couples that come from similar parts of the income distribution.
Sorting on returns to wealth
The implication of sorting on own wealth is that wealth inequality at marriage is amplified by the matching process in the marriage market. Individuals appear to sort also on their personal returns to wealth: men who earn higher pre-marriage returns on their wealth are more likely to match with women who also earn higher returns. This is again documented with a heat map (Figure 2). Furthermore, sorting on returns is quantitatively as strong as sorting on wealth. And quite importantly, it is not a reflection of sorting on wealth.
Figure 2 Heat map of assortative mating on returns to wealth
This is important to stress because one important feature of the data is that the wealthy tend to earn higher returns – a so-called scale dependence effect (Gabaix et al. 2016) – implying that wealthier people earn higher returns even if they have the same financial sophistication or risk tolerance as less wealthy people. As Figure 3 shows, the shape of the heat map is unaffected if we focus on sorting on returns among spouses belonging to the top quartile of pre-marriage wealth.
Figure 3 Heat map of sorting on returns conditional on wealth
Who manages household wealth?
How does sorting on returns affect the evolution of the couple’s wealth after marriage? Unlike sorting on wealth – the effect of which is easy to predict since marriage simply adds up the wealth that each spouse brings to the family – the effect of matching on returns depends on who, after marriage, ends up managing the household’s assets. For instance, one potential scenario is where the two spouses equally share the responsibility of managing their pooled wealth, so that the return on household wealth approximately equals a weighted average of the return of each spouse. This is the rate at which household wealth would evolve after marriage.
An alternative scenario is one of full specialisation, in which financial management of the household’s assets is assigned to the spouse with the highest ability to generate returns. In this case, household wealth will grow faster than when spouses share asset management responsibilities (except when there is perfect sorting on returns, in which case the ‘household finances management rule’ is irrelevant).
Comparing individual returns before marriage and household returns after marriage allows us to infer the ‘management rule’. We find that the spouse with the highest pre-marriage return on wealth carries 80% of the decision power, while the other spouse accounts for the remaining 20%. However, this result masks interesting forms of heterogeneity. First, men command a slightly higher weight than would be warranted by their pre-marriage returns alone, possibly a reflection of gender norms.
Moreover, among wealthy families – those classified in the top decile of the wealth distribution – the ‘management rule’ is one of full specialisation: the spouse with the highest pre-marriage return appears to be fully in charge of managing the household’s assets. This implies that, by effectively eliminating reversion to the mean by sharing financial responsibilities, the assets of wealthy households grow even faster over time, boosting wealth concentration. This in turn suggests one mechanism that generates persistence in wealth concentration is the allocation of decision power between spouses: when stakes are large – i.e. among the very wealthy – the incentive is also stronger to grant more discretion to the spouse with greater capabilities to manage the family assets. All else equal, this full specialisation reinforces the concentration of wealth at the top of the distribution.
Matching dynamics and inequality
Our evidence on assortative mating is relevant for the debate around the causes of both the extreme wealth concentration as well as its evolution over time – an issue that has attracted considerable attention in recent years. It is easy to show analytically that societies where marriages are assorted on wealth exhibit more wealth concentration at the point of marriage; with return heterogeneity, they also grow more unequal during the life cycle of a marriage, and the dynamic effects are amplified by assortative mating on returns.
Regarding the dynamics of wealth concentration, because mating patterns and wealth management arrangements are very likely to evolve over time (e.g. because social norms and relative gender skills change), time variation in assortative mating and wealth management allocation rules can be independent and, so far, unnoticed causes of changes in wealth inequality over time. Figure 4, Panel A, documents a significant decline in Norway of assortative mating on wealth over time (from around 0.23 to around 0.14). Panel B shows the time evolution of the Gini coefficient for wealth at marriage and documents that it has declined too. The association between the indexes in Panels A and B, albeit crude, is suggestive of the potential importance of assortative mating for the time evolution of wealth inequality.
Figure 4 The dynamics of assortative mating and wealth inequality
Barone, G, and S Mocetti (2016), “Intergenerational mobility in the very long run: Florence 1427–2011”, Bank of Italy Working Paper No. 1060.
Benhabib, J, and A Bisin (2019), “Skewed wealth distributions: Theory and empirics”, Journal of Economic Literature 56(4): 1261–91.
Charles, K K, E Hurst and A Killewald (2013), “Marital sorting and parental wealth”, Demography 50(1): 51–70.
Fagereng, A, L Guiso, D Malacrino and L Pistaferri (2020), “Heterogeneity and persistence in returns to wealth”, Econometrica 88(1): 115–70.
Fagereng, A, L Guiso and L Pistaferri (2022), “Assortative mating and wealth inequality”, CEPR Discussion Paper 17148.
Gabaix, X, J-M Lasry, P-L Lions and B Moll (2016), “The dynamics of inequality”, Econometrica 84(6): 2071–111.
1 Elon Musk’s wealth is estimated at $237 billion (“The world’s real-time billionaries”, Forbes). According to the 2019 Survey of Consumer Finances, the median wealth of the American families is $122,000. If the poorest families have zero wealth, redistributing Elon Musk’s assets evenly among them could bring 1.94 million families to the median.