Equity Holdings of Australian Baby Boomers-Comparing Life Cycle Phases

As a disproportionally large subgroup of the population, funding the impending retirement of the baby boomer generation (born between 1946 and 1966) is a challenge to Governments worldwide. Consequently, an increased onus on being self-funded in retirement mean that unlike any generation before them, baby boomers need to have well-developed financial literacy skills, in addition to access to financial education and advice. While Australia introduced compulsory retirement savings in 1992 via the Superannuation Guarantee Charge, the baby boomer population generally lack sufficient Superannuation savings to fully fund retirement (median superannuation balance is $47,000), and the scheme has some limitations in its coverage (those on welfare payments like disability support are not covered) (Keane, 2015). Thus, many baby boomers will draw on other assets such as equities and property to help fund consumption.

This paper focuses on the role of direct equity investment in satisfying this need. As an asset class, equities can play an important role in facilitating several key savings functions: they aid in providing diversification benefits to household portfolios that are often heavily weighted to property ownership; they have a higher risk/return profile than other assets commonly held by households (i.e. cash, property); and, they are highly liquid which assists in satisfying savings motives including storing wealth for a home deposit, precautionary savings and bequests. It is essential for investors to be aware that equities are an asset class that experiences a relatively high degree of price volatility compared to other assets, and thus investors may experience relatively high returns over long time horizons but risk losing money in any given year. This risk is a significant exposure for the baby boomer cohort who require as much capital as possible upon retirement when working incomes are foregone.

Many Australian households invest in equities both directly and indirectly through their superannuation fund holdings, self-managed superannuation funds (SMSFs), and other investment vehicles. Direct equity ownership is well represented among Australian households. With privatisation of many government organisations in the 1980s, favourable taxation treatment of dividends, and continued growth in returns for the ten years prior to 2014, household participation in equities in Australia of 27 per cent is comparable to the UK (also 27%, see Banks and Tanner, 1996) but lower than the US (with 36% participation, see Bertaut and Starr-McCluer, 2000). However, falls in equity prices since the start of the global financial crisis (GFC) saw the market capitalisation of the Australian stock market decline by 14 per cent between mid-2007 and March 2010. During this period, households sold equities to increase their cash deposits in response to the uncertain environment (West, 2015). Consequently, the GFC shook the confidence and retirement plans of ‘baby boomers’ to the extent of ‘many (older) workers deciding to delay retirement and save longer following losses in superannuation and direct equity investment’ (Humpel, et al. 2010, p. 130).

Thus, there is much interest in quantifying how vulnerable households may be to equity market fluctuations, particularly the portion of the population that are in the pre-retirement, rapid wealth accumulation phase. After briefly exploring the underlying theories of the ‘life cycle hypothesis’ and ‘portfolio theory’, this paper surveys the literature to get a sense of how well households implement investment concepts when managing their asset portfolios. Given the empirical evidence on households generally, it is expected that the equity holdings of baby boomers in the consolidation phase will be higher than they were in the accumulation phase, both in dollar values and portfolio share.

Accordingly, the wealth module of the Household Income and Labour Dynamics in Australia (HILDA) survey is employed, and the 2002 cross-section is used to represent the accumulation phase (baby boomers aged 38 to 56) and the 2014 cross-section represents the consolidation phase (baby boomers aged 50 to 68). Difference in means, logit and tobit analysis methodologies assist with assessing the difference between the equity holdings of the baby boomer generation in the accumulation and consolidation phases. In addition, the characteristics that determine equity ownership are examined in both phases. This particular cohort is of interest given the impending retirement of this disproportionally large subgroup of the population, and potential exposure to financial risk.

The outcomes of this study aim to inform policy makers, financial advisers and financial educators as to the exposure of Australian households to direct equity investment and therefore equity market volatility. As a consequence of this vulnerability, households may not be well prepared for retirement and defer to Government safety nets. On the other hand, as highlighted by portfolio theory, households can benefit from the addition of equities in the form of increased diversification and potential higher returns that are needed to fund a long retirement.

The extensive literature on household portfolios shows consistency across a number of jurisdictions on asset ownership, with much research concentrated in the United States (US). First, households are found to hold very simple asset portfolios that meet their needs (Bertaut and Starr-McCluer, 2000; McCarthy, 2004; Bucks, Kennickell and Moore, 2006). Second, direct equity investment is confined to those households that have higher levels of wealth, income, education, financial literacy, are older, and have a positive attitude to financial risk-taking. Where studies have investigated equity holdings with more detailed data, studies show that diversification within equity portfolios is very low (Goetzman and Kumar, 2008), households own few individual stocks (Kelly, 1995; Polkovnichenko, 2005), and there is evidence of local bias (Kelly, 1995; Huberman, 2001; Benartzi and Thaler, 2001).

Third, as the LCH model predicts, age is a significant determinant of asset ownership. In the early career building life cycle phase, households tend to overinvest in housing, and dependent children inhibit wealth accumulation and diversification (Bruekner, 1997; Flavin and Yamashita, 2002; Cocco, 2005; Pelizzon and Weber, 2008). As households progress to the consolidation and retirement phases, the portfolio share of equities rises, driven by the need to fund consumption in retirement and for bequest purposes (Cocco, 2005; Yao and Zhang, 2005; Kim, et al. 2012). However, equity holdings may be impacted by ageing and health issues, as these households divest of equities to fund retirement and medical expenses (Yao, Hanna and Lindamood, 2004; Edwards, 2010).

Fourth, macroeconomic conditions and events, such as the Great Depression, World Wars, the GFC, or even periods of high economic growth may impact asset holdings (Jianakoplos and Bernasek, 2006; Fukuda, 2009; Malmendier and Nagel, 2011). For example, many people who experienced the Great Depression are reluctant to invest in equities (Malmendier and Nagel, 2011).

Fifth, attitudes to financial risk-taking are a characteristic that can greatly influence asset holdings, particularly higher risk assets like equities. Research shows that risk attitudes are not stable over time and people become more risk averse as they age, that is, that people may change how they respond to questions in between survey periods (Halek and Eisenhauer, 2001; Olivares, Diaz and Besser, 2008; Yao and Curl, 2011; Bucciol and Miniaci, 2011). Evidence for changing risk preferences over time is found in Australia (West and Worthington, 2013a, 2013b). Studies also show that financial risk tolerance tends to increase when equity returns increase and decrease when stock returns decrease (Bilias, Georgarakos and Haliassos, 2010; Yao and Curl, 2011; Malmendier and Nagel, 2011).

Sixth, studies consistently find females are significantly more risk averse than their male counterparts and thus have low participation rates in direct equity investment (Olivares, Diaz and Besser, 2008; Yilmazer and Lyons, 2010), another contention supported by Australian data (West and Worthington, 2013a). Thus, females bypass opportunities to earn higher returns on higher risk assets like equities, which particularly compounds a wealth accumulation disadvantage that women experience in taking time out of the workforce to care for children and part-time work (Parr, Ferris and Mahuteau, 2007; Jefferson and Ong, 2010; Austen, Jefferson and Ong, 2010).

Finally, it is not surprising that many of the characteristics discussed thus far feed into levels of financial literacy. Individuals who demonstrate higher levels of financial literacy tend to have higher levels of portfolio diversification and thus invest in equities (Guiso and Jappelli, 2008; Abreu and Mendes, 2010). Those with low financial literacy levels are more likely to avoid equity investments (Van Rooij, Lusardi and Alessie, 2011) which is evidenced by research in Australia (Worthington, 2005, 2006, 2009; ASIC, 2011). Much of the financial literacy literature identifies people who are young (aged under 24), elderly (aged over 70), women, those on low incomes, people with low levels of education, and people from non-English speaking backgrounds or of Aboriginal or Torres Strait Islander descent as being characteristics associated with low levels of financial literacy (Dvorak and Hanley, 2010; Hung, Yoong and Brown, 2012). Low levels of financial literacy and consequential financial decisions mean that returns earned fall short of the theoretical returns that could be earned by well-informed, disciplined investors (Burtless, 2010).

From this general examination of household portfolio composition it is evident that households have difficulty implementing a diversification strategy, and although they generally affirm the LCH, there are many factors that complicate their financial decisions. While there is extensive literature on household characteristics and asset investment, there is a distinct lack of literature that compares the investment patterns of the same households during the accumulation phase or consolidation phase, thus this paper is a novel contribution to the literature. This study also investigates household characteristics that are significant determinants of holding direct equities and having a higher portfolio share of equities for the baby boomer cohort in both the accumulation and consolidation phases, as determinants may change over time.

The consequence of uneducated financial decisions may have wider reach than impacting the household alone. In a report on national financial education strategies, the OECD’s International Network on Financial Education noted that households were partially responsible for widening the GFC because of their lack of understanding financial issues, as they ‘accepted (sometimes unknowingly) to support more financial risk than what they could afford’ (2009, p. 4). The OECD notes that the lack of awareness of households of the risks they face is a major problem, given the increased sophistication of financial products and the transfer of risk to households that are taking on more financial risk and responsibility for credit and retirement decisions. Households do not seek protection from risks either because they do not understand the need to be protected, or they overestimate their protection and their understanding of the risks they face (Laboul, 2011).

The purpose of this study is to understand the direct equity investment patterns of the baby boomer cohort during the accumulation phase and the consolidation phase. Quantifying the equity investment of baby boomers in the consolidation phase is of particular interest to assess their exposure to volatility and potential capital loss in this very important pre-retirement life cycle phase. Accordingly, this study employs data from the Australian HILDA survey, which is an annual survey of households on various topics including income, labour market and family dynamics, with additional modules added in various years. The survey aims to follow the 6,542 households and 13,969 individuals interviewed in wave 1 in 2001. As this study aims to compare the equity ownership of the accumulation phase with the consolidation phase of the baby boomers, the 2002 cross section will be used to proxy for the baby boomer cohort in the accumulation phase, as baby boomers are aged 38 to 56 (i.e. born between 1946 and 1966) and the 2014 sample will be used to proxy the consolidation phase, as baby boomers are aged 50 to 68 in 2014. The baby boomer cohort consists of 4,725 people in the 2002 cross-section of the HILDA survey and 4,914 people in the 2014 cross-section.

The wealth module is implemented every four years, and questions relate to the householder’s assessment of their investment in assets such as bank accounts, cash investments, equities, superannuation, cash-in values of life insurance policies, trust funds, the family home and other property, business assets, vehicles and collectibles. The level of equity investment is used for direct investment in equities, but unfortunately, the HILDA survey does not collect detailed information on equity ownership other than an overall estimate of value. As this study is aiming to get an understanding of exposure to equity market risk in general terms, this level of data suffices and detailed analysis is left to future research.

The two dependent variables are defined as follows. The first dependent variable is a dummy variable for direct equity ownership (EQT_owner), where positive values (>$0) are identified as 1 and non-holders (≤$0) are 0. The second is the portfolio share of equities (EQT_{portfolioshare}), which is the dollar investment in direct equities as a proportion of the total assets. Table 1 shows the descriptive statistics for the two dependent variables and the asset classes for both the sample population and the baby boomer population, for 2002 and 2014. As can be seen in Table 1, the baby boomer population had a slightly higher participation rate for equity ownership than the sample population (48.4% vs 43.7%) for 2002 and for 2014 (39.5% vs 31.8%). Indeed, comparison of the two samples show that baby boomers have higher participation rates in most of the asset classes. By inspection alone, it is significant to note that the participation of baby boomers in direct equities reduced by approximately nine per cent from 2002 to 2014 (0.484 to 0.395). This reduction was mostly offset by increases in participation in Other Property (0.233 to 0.304).

Regarding portfolio share, Table 1 shows that there are several assets that dominate the portfolio for the samples, such as the Family Home (32% to 44%), Superannuation (20% to 26%) and Vehicles (7% to 12%). For the sample population compared to (baby boomers), direct investment in equities comprised 3.9 (3.2) per cent of the total asset portfolio in 2002 and 2.3 (2.4) per cent in 2014. In 2014, the baby boomers reduced the portfolio share of equities by 0.8 per cent (0.032 to 0.024). This reduction could be due to either or both a reduction in the dollar value of equities and a rise in property prices (considering portfolios are heavily weighted towards property).

As outlined in the literature review, factors that other studies highlight as significant determinants of asset selection include levels of wealth, income, education, age, gender, health, and household structure, as well as financial literacy and financial risk-taking factors. The descriptive statistics, expected signs and names for these variables are provided in Tables 2 and 3 for 2002 and 2014. Unfortunately, the HILDA survey does not ask questions that can test respondent’s knowledge of financial concepts directly. Financial literacy literature indicates that goal setting and planning is an important characteristic of a financially literate person (Dvorak and Hanley, 2010). Hence, the questions regarding ‘Savings Habits’ (asks respondents to indicate how regularly they save) and ‘Investment Horizon’ (asks respondents to indicate how far ahead they plan when investing) are employed as financial literacy proxies. Responses that are towards the ‘Save regularly by putting money aside’ and ‘More than 10 years ahead’ spectrum are considered to be financially literate, while the opposite end of the spectrum as shown in Table 3 is considered less financially literate.

The reference categories are designated with a hashtag (^#) and are chosen based on being important for policy implications or that they comprise a relatively large proportion of the population. Results are interpreted with respect to the constant being equal to the mean of the reference category, and then the coefficients of each of the dummy variables in the regression is equal to the difference between the mean of the group coded 1 and the mean of the reference category. The p-value associated with that coefficient is the test of the category versus the reference category.

Variance inflation factors (VIFs) were used to test for multicollinearity between some groups of the independent variables. Where the variable has been transformed into a dummy variable, the original variable has been used. For tests of age, education, income and net assets, VIFs were 1.088 for age, 1.048 for education, 1.307 for income and 1.231 for net assets. Tests of gender, household structure, health and income yielded VIFs of 1.001 for gender, 1.048 for household structure, 1.034 for health and 1.082 for income. The financial literacy and financial risk variables resulted in VIFs of 1.153 for saving habits, 1.188 for investment horizon, and 1.179 for financial risk-taking. As a rule of thumb, VIFs greater than ten warrant further examination, but there is no issue here as the VIFs are universally very small (less than 1.4) (O’Brien, 2007).

The analysis consists of two parts. To compare the accumulation and consolidation phases, a difference in means test will be used to examine significant differences in means between equity ownership and the portfolio share of equities in 2002 and 2014. Then, a regression of the household factors on equity ownership and the portfolio share of equities is applied to 2002 and 2014 data sets. The two models are defined as follows.

Given that this dependent variable is specified as a binary response (0, 1), a logit model is appropriate and it models the probability of a positive outcome (1) given a set of regressors by maximum likelihood. The log likelihood function for logit is:

where S is the set of all observations j, such that y_j≠0, F(z) = e^z/(1+e^z), and w_j denotes the optional weights. lnL is maximised where 0 corresponds to the ‘constant only’ model and 1 corresponds to perfect prediction for a discrete model (in which case the overall log likelihood is 0).

As EQT_{portfolioshare} is naturally censored between 0 and 1, a tobit model is appropriate. The expected value of y is given as:

where Ø is the standard normal density and Φ is the standard normal cumulative distribution function, y_i is the observed variable, x_i denotes the vector of exogenous and fully observed regressors, β is the coefficient, σ is the standard deviation. As two-tailed censoring applies, estimates are observed in the interval [L, U], where L (lower limit or zero) and U (upper limit or 1) are the known censoring points.

This study seeks to measure the extent to which Australian baby boomer households ascribe to financial theory by increasing direct equity investment in the consolidation life cycle phase and garner insight into vulnerability to equity market volatility as they approach retirement. The 2002 and 2014 cross sections of the HILDA survey were employed to compare equity ownership and equity portfolio share of the baby boomer population in the accumulation phase and the consolidation phase. A number of conclusions can be drawn from the findings. First, the average level of equity investment and portfolio share are significantly different in 2002 to 2014. However, the descriptive statistics show that the direction of the change is not as predicted as participation in equity investments was reduced in 2014. Furthermore, the logit and tobit models highlighted differences in the significance of characteristics between the two periods, such as variances in household structure, income level and gender. Thus it can be argued that transitioning to the consolidation phase induces a change in financial decision-making, but that in this study the impact of the GFC and ongoing changed macroeconomic circumstances could not be disentangled.

Second, an elevated level of wealth is highly significant for equity participation and portfolio share. The results of the 2014 model show an increase in magnitude of high wealth level, and it can be inferred that the GFC may have exacerbated the wealth divide, and thus the wealthier portion of the population were able to continue holding equities while those in lower wealth categories liquidated their holdings. Concomitantly, higher levels of financial literacy and willingness to take financial risks were consistent with equity investment, and in combination with exacerbating the wealth divide, the market volatility derived from the GFC may have led people with lower levels of financial literacy to divest of equity ownership while those with higher levels persisted.

Overall, the results did not confirm our hypothesis that baby boomer households have increased direct equity ownership in the consolidation phase. While an avenue for future research, it is likely that the GFC reduced household investment in equities, particularly those household members who may have lower levels of wealth, income, financial literacy and willingness to take financial risk. Accordingly, these households are missing out on important diversification benefits that equities can provide as well as potential for higher portfolio returns. However, it may be of comfort to policy makers that baby boomer households are not at peak levels of equity market volatility exposure in this important pre-retirement consolidation phase.

These results have important implications for public policy and financial educators and advisers. Clearly, higher levels of financial literacy have positive impacts on minimising the risk exposure of a household’s asset portfolio by the inclusion of direct equity investment as an important component of a diversification strategy. More financial literacy education may improve financial decisions made by those at the lower levels of income and wealth. Future research may benefit from more detailed decomposition of equity ownership and controlling for the macroeconomic environment.