Using this labor-capital decomposition it is, in principle, possible to attribute shifts in top income shares to shifts in top earnings shares, top capital income shares, and factor shares. A practical empirical problem, however, is that in most cases we lack data on the cross distributions over long periods of time. Roine and Waldenstr€om (2008), studying Sweden, is an exception. Thanks to a particular form of combined income and wealth tax it is possible to calculate the distribution of wealth ranked both by wealth and total income.128 Figure 7.22 shows that the share of total wealth when ranked by total income is somewhat lower than when ranked by wealth, but the two series are highly correlated, suggesting that there is significant overlap between the two distributions.