Capital gains are the biggest contributor to wealth and inequality – lower taxes

1 Introduction Over the past two decades inequality has attracted increasing attention, with development of inter- national data (Alvaredo et al., 2018a,b), even longer run series (Piketty and Saez, 2003; Moriguchi and Saez, 2008; Bartels, 2017; Garbinti et al., 2018), and the development of Distributional National Accounts (Alvaredo et al., 2016; Piketty et al., 2018). A recent strand of literature has studied measurement of income sources for those at the top (Piketty et al., 2018; Smith et al., 2019). Smith et al. (2019) highlight the role that income-shifting plays in changing the characteristics we associate with those at the top: in particular, top earners largely receive income from human, not financial, capital. This income-shifting is in large part driven by tax incentives to adjust the timing of in- come receipts (le Maire and Schjerning, 2013; Miller et al., 2019). However, such incentives not only encourage shifting between income forms, but from income into capital gains. The difference in tax rates makes these incentives much larger, and the fungibility between income and gains is precisely why the optimal tax literature regularly calls for common marginal rates between income and gains (Slemrod, 1995; Slemrod and Kopczuk, 2002; Mirrlees et al., 2010; Blundell, 2012, 2016). Since capital gains are not part of fiscal income, they are excluded from standard income inequality measures. Including them can therefore not only change the pattern of receipts among those at the top, but also which individuals are at the top, and what share of total remuneration they receive. This paper studies the implications of including capital gains for measured inequality in the UK. We combine confidential administrative data on the universe of UK taxpayers over more than two decades to understand where gains come from, who receives them, and the implications for inequality. We show three main results. First, capital gains are largely shifted income: they largely reflect returns to human capital, rather than simply exogenous changes in the value of assets held. This highlights the limitations of focusing on a narrow fiscal income definition when studying inequality, since the line between income and gains is often blurred. Second, particularly at the narrowest top shares, including gains substantially changes who is at the top. One in ten individuals in the top 1% when gains are included are not in the top 1% based on fiscal income alone. ‘Joiners’ to the top 1% are more likely to be business owners, to be older, and to be female than ‘stayers’ who are at the top on either definition. Third, including gains substantially increases the share of total resources going to those at the top, as well as the observed trend in top shares. Including gains increases the pre-tax share of resources to the top 0.01% by 60% in 2017; their post-tax share almost doubles. Strikingly, including gains also changes the narrative for UK inequality over the past two decades: 2 while top fiscal income shares have remained largely constant since the late 1990s, once gains are included there has been a large and sustained rise. Capital gains are taxed at a lower rate than ordinary income. The top tax rate on labour income in the UK is 47%, for dividends it is 38.1%, while on capital gains it is between 10% and 28% depending on the type of underlying asset. In the US the top federal tax rate is 40.8% for income from labour or ordinary dividends, while on capital gains it peaks at 23.8%, and similar patterns hold elsewhere. The patterns we highlight are therefore likely to be of much broader relevance than to the UK alone. They matter for the time series of top income shares, as incentives to shift remuneration into gains vary with policy; for cross-country comparisons as these incentives vary across locations; and for the type of individuals we see in top shares. To establish where gains come from, we make use of capital gains tax filings. These document not only the amount of gains received by individuals, but the asset classes from which they derive. Around 71% of gains come from unlisted shares and ‘other assets’—a category comprising mainly business assets eligible for ‘entrepreneurs’ relief’, which lowers the tax rate to 10%, compared with a 20% standard rate. Such gains are substitutes for returns that could have been taken as capital income, and may in many cases actually be repackaged labour income (Miller et al., 2019). A further 5% of gains are from ‘carried interest’: payments for fund managers which depend on their investment performance (human capital) rather than equity stake. As in the US, the UK treats these as a return on investment. Three-quarters of all gains therefore come from returns that could just as well have been received as income. Most of this will being missing from national income, as well as from fiscal income, so will not show up in any measure of top share inequality based on income only (Alvaredo et al., 2016; Piketty et al., 2018).1 To understand who receives these gains, we link capital gains and income tax records. We first construct a measure of fiscal income using the deidentified administrative income tax data, to match the commonly used ‘Survey of Personal Incomes’ (Atkinson, 2002, 2007; Jenkins, 2017; Joyce et al., 2019). Combining it with capital gains tax records, we see that gains are extremely concentrated at the top of the existing income distribution.2 Around 50% of all gains by value go to the top 1% of individuals ranked by income, compared to just 13% of all fiscal income. At the indivdual level, nine out of ten people who are in the top 1% when ranking by income are also at the top when gains are included. However, the one in ten ‘joiners’ to the top 1% by total remuneration (income plus gains) 1While national accounts will pick up the part of gains that come from disposal of businesses that previously had retained earnings, they will not pick up intangibles like goodwill. 2It is also important to note that, conditional on having gains, the average amount of gains is similar across most of the income distribution. Although this has little aggregate effect, exclusion of gains is therefore important for understanding inequality further down the income distribution as well. Tables A1 and A2 provide the full distribution. [image omitted] 3 look very different in terms of their characteristics. In line with our first result, and with Smith et al. (2019), owners of businesses are much more likely to be at the top. Those at the top are also substantially older, and more likely to be female. This change in ranking matters because, while recent developments in inequality measurement are able to distribute all parts of national accounts, they do this in proportion to components of fiscal income (Piketty et al., 2018). By construction they miss any components of inequality that change when gains are added to that series. If gains were proportional to some income component, most likely dividends, then that might serve as a proxy and receive higher weight when scaling up to the national accounts (even assuming all real gains are captured in national accounts components). However, we see a prevalence among joiners of individuals with majority pension income, as well as of investors, who would not receive such scaling. While the share of gains going to the top 1% has remained relatively constant over the past twenty years, aggregate gains have risen sharply, tripling in the last decade. The share of total remuneration going to the top has risen in parallel, from 14.1% in 2009 to 16.8% in 2018. By contrast the share of income going to the top has been rising much more slowly: from 12.7% to 13.8% over the same period. Using data on asset holding lengths, we show this pattern holds when gains are considered on an accruals – rather than realisation – basis, while stripping out some of the excess volatility caused by policy changes. Including gains not only changes the nature of inequality trends, but also the apparent effectiveness of policy. Since gains are taxed at lower rates than income, the impact of including gains on post-tax inequality is even larger; the lack of redistribution of gains highlights again why they are a preferred form of remuneration for those that can repackage their income. Top share series including gains have been considered before (Piketty and Saez, 2003; Moriguchi and Saez, 2008; Alvaredo and Saez, 2009; Roine and Waldenstr ̈om, 2012; Alvaredo et al., 2013; Bartels, 2017). Authors have largely rejected these series in favour of an income only series, either because adding gains makes little difference (Bartels, 2017) or because adding gains makes a big difference to the volatility of the series (Piketty and Saez, 2003; Moriguchi and Saez, 2008; Alvaredo et al., 2013). We show that excluding gains is problematic even when interest is primarily in income inequality, because gains can contain a substantial share of shifted income. International differences in the taxation of gains relative to income – as well as differences over time – will influence how large an issue this is in a particular context, but exclusion of gains may bias down top income shares where shifting into gains is occuring.3 While the volatility of realised gains series is understandably seen 3These differences also help explain why in some contexts authors found inclusion of gains to add little, while in others they are substantial. [image omitted] 4 as unappealing, recent work studying estimating these on an accruals basis is likely to be preferred to complete exclusion. Our top share series including accrued gains is no more volatile than the top income series, but does pick out a different trend and level for top share inequality.