The high and rising wealth inequality receives growing attention in modern societies and research on economic inequality. Different surveys (e.g. the Household Finance and Consumption Surveys HFCS, the Survey on Aging and Living Conditions SHARE, or the Luxemburg Wealth Study) have started to collect microdata on wealth. However, most wealth estimates neglect pension wealth that—for most individuals—presents the largest asset.
There are many reasons to consider pension entitlements in wealth analysis. First, the inclusion of pension wealth has profound consequences on the wealth distribution. Second, pension rights are important for the comparison of wealth between countries. According to the standard life-cycle hypothesis, expected pension benefits and private wealth are substitutes (Feldstein, 1974). If pension wealth crowds out private saving, countries with more generous pension schemes should have lower private savings than countries with more limited pension schemes. Third, the omission of pension wealth might result in misleading comparison of wealth between covered and non-covered groups within a country (e.g. wealth differences between employed and self-employed). Fourth, pension wealth has been shown to influence many behaviours and decisions, such as private savings, composition of wealth portfolios or retirement timing. When ignoring pension wealth, studies on these behaviours could yield biased results. Finally, a better knowledge on the inequality of pension wealth can be policy relevant.
At the same time, pension entitlements differ from other assets in several respects: they cannot alleviate income poverty, be used to purchase consumer goods, or be passed on to others. Accordingly, pension entitlements do not confer political power, social influence, or social status. Furthermore, individuals have little control over how their pension assets are invested and restrictions often apply for bequests. Finally, the tax system and welfare state treat pension wealth differently from other assets. A further difference occurs for pay-as-you go systems, where entitlements are not funded.
There is no agreement on how to deal with pension wealth in analysis on wealth inequality. Some scholars argue for ignoring pension wealth, as it does not give “direct personal control over resources” (Alvaredo, Atkinson and Morelli, 2018). The most widely used wealth definition based on microdata is net worth, which refers to the “value of all the assets owned by a household less the value of all its liabilities at a particular point in time” (OECD, 2013, p. 54). Net worth includes private pensions, but not occupational and statutory ones. The term “augmented wealth” refers to the broader definition of wealth that includes entitlements to future pension streams (Davies and Shorrocks, 2000). Organisations such as the OECD suggest using augmented wealth as a measure complementary to net worth (OECD, 2013, p. 67ff). While these definitions are standards in the micro-approach to wealth analyses (see e.g. Wolff, 2015; Cowell, Nolan, Olivera and Van Kerm, 2017, p. 177), they differ from the macro-economic definition of wealth in the core tables in the System of National Accounts (SNA), which includes only funded pension wealth. However, unfunded pension wealth is included in supplementary data of the SNA to facilitate country comparisons (System of National Accounts, 2008, p. 369ff.). Attempts at distributional national accounts, such as the World Inequality Database (WID), usually follow the wealth definition in the core tables of the SNA, which include only funded pension wealth. Apart from theoretical considerations, data availability and data comparability often dictate the wealth definition used in empirical studies.
This study documents the relevance and the distribution of pension wealth in Switzerland and shows how it affects wealth inequality. The analysis is based on a new linkage of administrative and survey data with a matching rate of over 99% of the sample. Survey data come from the CH-SILC in 2015, which contains separate questions on various real and financial assets including private pension accounts. Administrative records are used to estimate entitlements for social security and occupational pensions for non-retired individuals at the basis of yearly incomes since 1981. While the linked data allows a rather precise estimation of statutory pension wealth, the simulation for occupational pension wealth relies on many assumptions. Following the OECD guidelines for Micro Statistics on Household Wealth (OECD, 2013), I apply the accrual method to estimate the present values of pension wealth under the current pension system. For social security pensions and ongoing second-pillar pensions, I estimate the future pension stream on cohort- and gender-specific survival rates assuming a real discount rate of 2%. For the occupational pension, I estimate the accumulated capital, which is influenced neither by life expectancy nor by retirement age.
Several important findings come out of my analysis. In line with experiences from other countries, pension wealth is highly relevant and has an equalising effect on wealth inequality. Twenty-eight percent of augmented wealth consists of entitlements for social security pensions, and another 23% consists of entitlements for occupational pensions. Although the Gini coefficient of net worth amounts to 0.76, it decreases to 0.65 when adding occupational pensions and to 0.55 for augmented wealth. Moreover, there are important wealth differences between men and women and between age groups. In contrast to findings of other countries, wealth accumulation over the life span continues beyond retirement age.
The methodological approach followed has the advantage of making results comparable to estimates for Germany and the USA by Bönke, Grabka, Schröder and Wolff (2019) but comes with some limitations. Firstly, it ignores the heterogeneity in life expectancy by socio-economic groups, which applies also in Switzerland (e.g. Mackenbach et al., 2019; Wanner and Lerch, 2012). As high-income groups live longer compared to low-income groups, they will receive benefits for a longer time and limit the equalising effect of pensions (Haan, Kemptner and Lüthen, 2019). Secondly, the accrual method does not take account of future pension reforms but assesses the situation in 2015. Due to both effects, the redistributive effect of pension wealth will be weaker in reality. The impact of increasing pension age on wealth inequality is simulated as a sensitivity analysis, which shows a small desequalising effect of such reforms.
While analysis of augmented wealth is common in the USA and UK, there have been only a few attempts to measure the distribution of augmented wealth in Europe (Roine and Waldenström, 2009 for Sweden; Maunu, 2010 for Finland; Bönke, Grabka, Schröder, Wolff and Zyska, 2019 for Germany). All studies reveal a mitigating effect of pension assets on wealth inequality. To my knowledge, there have been no attempts so far to empirically measure augmented wealth in Switzerland. This country presents an interesting case, as it has a very high wealth concentration at the top but a rather average wealth inequality when measured with the Gini index. Moreover, the share of private occupational pensions is (with 31% of post-retirement disposable income) among the highest in OECD countries (OECD, 2019).
A simulation for top wealth shares was conducted by Föllmi and Martínez (2017) based on two simplified scenarios for occupational and private pension wealth but not social security pensions. Assuming an equal distribution of pension assets, the simulation showed that the wealth share of the top 10% of the population would decrease by 27 percentage points. Assuming that shares in pension wealth would correspond to share in labour income, the top 10% wealth share was lowered by 20 percentage points. The individual data used in my study show a slightly weaker reduction in inequality because occupational pensions are more unequally distributed than labour income and because pension wealth and other assets are positively correlated.
A related body of literature addresses the effect of pension entitlements on savings. According to the life cycle hypothesis, individuals save during their working lives and commence with dissaving after retirement (Modigliani, 1986). Accordingly, a change in the expected pension benefits should alter private wealth by the same amount (Feldstein, 1974). Although empirical studies find evidence for such crowding out, the rate is considerably lower than one (e.g. Attanasio and Brugiavini, 2003; Bottazzi, Jappelli and Padula, 2006; Chetty, Friedman, Leth-Petersen, Nielsen and Olsen, 2014; Feng, He and Sato, 2011). Explanations for the limited substitution include bequest motives, short-sightedness, liquidity constraints, risk associated with future reforms, individual information and non-marketable future benefits (Bottazzi et al., 2006, p. 2188).
This paper is structured as follows. In Section 2, I briefly present the pension system in Switzerland and review literature on wealth inequality in Switzerland. Section 3 describes the data and methodological approaches used. Section 4 documents the estimation of pension wealth. Results are discussed in Section 5; Section 6 discusses pension reforms and provides sensitivity analysis for the impact of widowhood pensions, different discount rates and increasing retirement age; and Section 7 concludes.