Who makes the most of matching contributions? An analysis of member choices in DC pension plans

Who makes the most of matching contributions? An analysis of member choices in DC pension plans

Introduction About the research We analysed the contribution decisions of almost 80,000 employees, working within 18 companies, who were members of their employer s defined contribution (DC) pension plan between 2010 and 2013. 1 We examined the influence of three factors on what contribution rate employees choose when faced with a matching contribution structure. 1. The role of the default contribution rate prior research has shown that defaults can have a profound effect on member behaviour; both via member inertia and employees potentially inferring that the default is the appropriate rate. 2. The role of the contribution structure we examine how different contribution structures infl uence employee choices. 3. The role of personal characteristics how are choices correlated with an employee s age, gender and salary. 2 towerswatson.com

Matching structures are a very popular way of providing DC pensions for employees in the UK, with over half of FTSE 100 employers using this approach. 2 Despite their popularity, until now, there has been little evidence on employee behaviour which would enable us to answer the question how much will a typical employee contribute?. Key findings The results suggest that: 1. Members most commonly choose to contribute at either the minimum or the maximum contribution rate. 2. The default contribution rate is pivotal. 3. Increasing additional matched contributions does increase the average contribution rate, but affordability limits employee contributions (with diminishing numbers at the maximum). 4. The level of matching matters. 5. Contributions are higher from older members and from those on higher salaries. This information can be used to predict the contributions and outcomes from different DC designs. 1 Companies with non-matching structures or with matching structures based on age, service, and so on, are excluded from the analysis. The average employee member was in their mid-30s, with an average salary of around 32,000. Some 47% were female. 2 Towers Watson FTSE100 Defi ned Contribution Pension Scheme Survey (2013). Who makes the most of matching contributions? 3

The role of defaults Prior research has shown that default options can have a profound effect on member behaviour. Our research found that members behaviour was relatively polarised. Most members contribute either the minimum or the maximum contribution rate (to get the maximum match from the employer) with a minority selecting various points in between. Where the default is set is found to have a profound effect on the numbers contributing at the maximum rate. 4 towerswatson.com

But for those workers who would otherwise have opted in, some are likely to end up with lower contributions and pensions than might otherwise have been the case. Figure 01 shows that in cases where there was no default contribution rate, almost two thirds of members choose to contribute at the maximum, with only 20% at the minimum. Without the potential influence of a default, most members in our data choose to contribute at the maximum rate. Once an employer has started automatically enrolling employees, it has to set a default contribution rate. These data suggests that by automatically enrolling employees and forcing a default employee contribution rate, we end up with lower average contributions (unless the default is the maximum). This illustrates that auto-enrolment comes at a cost. Auto-enrolment raises take-up rates and helps to ensure greater numbers of employees contribute something. But for those workers who would otherwise have opted in, some are likely to end up with lower contributions and pensions than might otherwise have been the case. In the subsequent graphs in Figure 01 we can see the default rate matters and that where the default is set strongly influences member choices. Where the default rate was set at the minimum matched contribution rate, only around 25% of members were contributing at the maximum but nearly 40% at the minimum rate. By contrast, where the maximum is the default, 77% of members remain at the maximum. Figure 01. Contribution choices according to default contribution rate Percentage 80 70 60 64 77 50 40 30 37 37 37 38 20 10 20 16 26 25 13 10 0 No default At minimum Intermediate At maximum Members contributing at the... Minimum rate Intermediate rates Maximum rate Default contribution rate Who makes the most of matching contributions? 5

The role of the contribution structure Not only do employees respond to the default contribution rate, but also the contribution design. Figure 02 shows the role of the contribution structure on member choices. We start from an assumption that there is no default contribution rate and examine how different contribution structures infl uence member choices using a statistical model for the infl uences on contributions choices (see Appendix on page 11). Figure 02 describes the different contribution structures we examine. The base case assumes a minimum employee contribution rate of 3%, a minimum employer contribution rate of 3% and that additional employee contributions are matched 1:1 up to a further 3% (that is, the maximum contribution would entail 6% from employee and 6% from the employer: 12% in total). In the columns that follow we examine the following scenarios: A higher match rate in a typical matching design if an employee contributes 1 more so does this employer. We examine the expected effect of moving to a 2:1 matching rate. Allowing the employee to contribute more matched contributions instead of contributing an additional 3% they can contribute a further 6%. Lowering the core employee contribution rate from 3% to 0%. Increasing the core employer contribution rate so if the employee contributes the minimum (3%) they instead receive 6% rather than 3%. A design tells employees what they should regard as normal. 6 towerswatson.com

Figure 02. Average contribution rates according to contribution structure Employee contributions Base case Higher matching Employer contributions Higher additional contributions Lower core employee contributions Higher employer core 0% - - - 3% - 1% - - - 4% - 2% - - - 5% - 3% 3% 3% 3% 6% 6% 4% 4% 5% 4% 7% 5% 5% 7% 5% 8% 6% 6% 9% 6% 9% 7% 7% 8% 8% 9% 9% Average employee contribution Average cost to employer 4.5% 5.0% 5.0% 2.5% 4.7% 4.5% 7.0% 5.0% 5.5% 7.7% In Figure 02 we can see: The higher the matching ratio the more likely the member is to pay at the maximum the incentive effects of matching do indeed work (with the average contribution rate 5% compared to a maximum 6%). Where the design aim is to enable good pensions, matching is therefore an effective tool. Where minimum employee contributions are lower, members are more likely to pay the maximum contribution rate. When employees can choose contribution rates between 3% and 6% the average contribution is around the mid-point (4.5%). When employees can choose contribution rates between 0% and 3% the average contribution is close to the maximum (2.5%). Conversely, the higher the employee contribution rate required, the lower the percentage of employees we would expect to contribute at the maximum rate. This illustrates the framing effect of a contribution design. A design tells employees what they should regard as normal. By framing employee contributions as being between 3 and 6%, employees choose to contribute more. Increasing employer minimum contributions does little to alter member contributions employees do not seem to decide to reduce some of their own contributions as a result and the average employee contribution rate is only marginally higher (4.7%) relative to the base case (4.5%). This is as we might hope. Increasing employer contributions will give rise to higher overall contributions and therefore pensions. Increasing the maximum (matched) employee contribution rate produces an increase in member contributions suggesting some members take advantage of the opportunity to contribute more. That the effect is relatively small (the average contribution rate is now 5% compared to a maximum of 9%) suggests some members have an upper tolerance level of contributions, above which they are reluctant to go. In turn, this means offering a matching design that requires high employee contributions at the upper level may appear an eye catching and generous design, but in fact large numbers of employees may not contribute at this level due to affordability concerns. The contribution design therefore will change the outcomes. To look at the two extremes, if you want better outcomes for engaged employees, increasing the matching ratio will increase employee contributions. At the other extreme, offering wide bands of employee contributions, but defaulting contributions to the minimum level may minimise employer costs, while still offering a competitive design. To look at the two extremes, if you want better outcomes for engaged employees, increasing the matching ratio will increase employee contributions. Who makes the most of matching contributions? 7

The role of demographic influences Our research finds there is a strong correlation between the contribution rates chosen and the members age and salary. Matching designs give us age- (and salary) related contributions by proxy. Figure 03 presents the average contribution position according to age, salary and gender. We can see men contribute marginally more than women. Older members contribute signifi cantly more than younger members and those on higher salaries contribute more than those on lower incomes. Figure 03. Average contribution rates according to member demographics Salary Age 60 to 64 55 to 59 50 to 54 45 to 49 40 to 44 35 to 39 30 to 34 25 to 29 <25 96k and above 62k to 96k 48k to 62k 40k to 48k 32k to 40k 24k to 32k 16k to 24k Less than 16k 6 Minimum 7 8 9 10 11 12 Maximum Contribution rate (% of salary) Female Male 8 towerswatson.com

To illustrate how these graphs might translate into very different outcomes, we have shown potential contributions for five single members of the baseline case. Example members Age 35 Age 24 Age 52 Age 35 Age 35 Salary 30,000 Salary 30,000 Salary 30,000 Salary 15,000 Salary 65,000 Average employee contribution rate (%) Average total contribution rate (%) Cost to the average member (per annum) 4.7% 3.8% 4.9% 4.1% 5.2% 9.4% 7.6% 9.8% 8.2% 10.4% 1,410 1,140 1,470 615 3,380 Older members contribute significantly more than younger members and those on higher salaries contribute more than those on lower incomes. Who makes the most of matching contributions? 9

Suggested actions Where DC contribution designs are under review, these fi ndings can inform the design process permitting employers to more accurately estimate both costs and outcomes at retirement. Where costs are higher than desired, or projected outcomes too low, understanding how alternative designs can work might help bring outcomes or costs to desired levels perhaps with limited change to total employer cost. Knowing the expected level and cost of contributions, given a design and a workforce, provides an excellent measure for engagement. Where employees are contributing more than we might expect, they are likely to be highly engaged. Benchmarking a workforce against this data will help employers identify those populations that are more or less engaged than expected, and then act accordingly. Benchmarking a workforce against this data will help employers identify those populations that are more or less engaged than expected, and then act accordingly. What else can make a difference? The preceding analysis has examined a large number of employees to see how they typically choose contribution rates according to their age, gender, salary and the plan s contribution design. However, other factors (not accounted for here), can also make a material difference to the contribution choices made by workers. Communication can also strongly infl uence member choices. A plan which strongly suggests that the maximum is a good contribution rate and presents the maximum as a default, is likely to have different outcomes to a plan that simply presents a set of options. Other external infl uences can also affect members opinion of how much to contribute. For example, events such as a stock market crash, may lead to negative perceptions of investing, which affect employees willingness to save. Alternatively, the economic environment at the time of joining (and how much members think they can afford) may also determine the contribution choices of employees over the long-term. 10 towerswatson.com

Appendix: The statistical analysis Typically each employer has a different contribution design. Therefore, simply comparing the contribution rate observed is not informative, unless we account for the plan design. An employee contributing 2% could be contributing the maximum allowed by the plan. For others, this could be less than half their potential contribution. To allow for such scenarios, this analysis defi nes a new variable based on employee contributions relative to the contribution structure they are faced with. This variable takes a value of zero if an employee contributes the minimum possible, and one if they contribute the maximum matched contribution. This denotes where on the range of feasible contributions (between the minimum and maximum) an employee is. For example: If the employee can contribute between 1% and 3% and decides to contribute 1% this implies a value of zero. If the employee can contribute between 1% and 3% and decides to contribute 2% this implies a value of half (that is, they are contributing half way up the possible range) Such a variable is consistent across all companies and the role of different contribution designs and demographic factors can easily be accounted for (the average employee contribution position, between zero and one, can be modelled as a function of scheme and employee characteristics). A statistical model estimating contribution choices on this new scale is estimated using a fractional logit GLM model. This model estimates the infl uence of age, gender, salary and the contribution structure on employees contribution choices holding other factors constant (that is, the model accounts for the fact that females may be less well paid than males and any effect of gender is identifi ed separately from salary). The model is then used to estimate employee contribution rates adjusting for these other factors. Standard errors were clustered at the company level. Further information For further information please contact your Towers Watson consultant or: Jonathan Gardner +44 1737 274097 jonathan.gardner@towerswatson.com Will Aitken +44 20 7227 2006 will.aitken@towerswatson.com Who makes the most of matching contributions? 11

About Towers Watson Towers Watson is a leading global professional services company that helps organisations improve performance through effective people, risk and financial management. With 14,000 associates around the world, we offer solutions in the areas of benefits, talent management, rewards, and risk and capital management. Towers Watson 21 Tothill Street Westminster London SW1H 9LL Towers Watson is represented in the UK by Towers Watson Limited. The information in this publication is of general interest and guidance. Action should not be taken on the basis of any article without seeking specific advice. To unsubscribe, email eu.unsubscribe@towerswatson.com with the publication name as the subject and include your name, title and company address. Copyright 2013 Towers Watson. All rights reserved. TW-EU-2013-35021. December 2013. towerswatson.com