Raising pension awareness through PensionTV: Evidence from two Dutch pension funds

(1)

Raising pension awareness through PensionTV: Evidence

from two Dutch Pension Funds

A thesis submitted in partial fulfillment of the requirements for the degree of

Master of Science in Public Administration

Track Economics & Governance

Leiden University

Department of Governance and Global Affairs

Author:

Supervisor:

Amy Sevink

Dr. Jim Been

S1686852

Second reader:

Prof. Dr. Marike Knoef

(2)

(3)

Abstract

This study analyses the effectiveness of the interactive pension communication tool PensionTV on 1) pension knowledge, 2) appreciation, and 3) activation. To identify the effect of PensionTV, we rely on a non-randomized field experiment using poll-questions and a survey administered to a sample of 221 pension plan participants of the pension fund of the Fashion, Interior, Carpet and Textile industry and 673 pension plan participants of the company pension fund for IBM. Our results show positive and statistically significant effects of PensionTV on self-reported pension knowledge (1.94/10) and appreciation (0.37/10). Women start with a lower starting position on average but the increase in knowledge or appreciation is considerably higher than for men. However, the results of the OLS regressions suggest that this difference cannot be attributed to gender, but rather to a difference in prior knowledge. Respondents with one-point higher prior knowledge score on average 0.34 points higher on ex-post subjective knowledge and 0.66 lower for the change in subjective knowledge. The effect is thus smaller for those who start with more knowledge. We find a similar effect for prior appreciation. Furthermore, PensionTV activates pension plan participants to look at their personal pension planner. For pension funds where participants have relatively little knowledge prior to PensionTV, the program appears to be the most effective, both in terms of impact and cost-effectiveness. In conclusion, PensionTV turns out to be an effective means of communication and is highly appreciated by pension plan participants.

JEL-codes: D14, D83, G23, G53, J32

Keywords: Pension Communication, PensionTV, Financial Literacy, Dutch Pension System

Acknowledgements

I would like to express my gratitude to my supervisor Dr. Jim Been for his dedicated support and guidance. Many enjoyable and insightful talks led to the development and improvement of this research. Furthermore, I want to thank my family and friends for encouraging and supporting me during my entire study at Leiden University.

(4)

1. Introduction

The Dutch pension system is considered to be one of the most generous pension systems worldwide (OECD, 2019; Arnold, 2019). The Netherlands is ranked number one in Mercer's Global Pension Index, which is based on three basic elements: adequacy, sustainability and integrity (Mercer, 2019). In the Netherlands, every person accrues compulsory state pension (AOW), regardless of work history. A full state pension lifts households and individuals beyond the poverty level (Central Bureau for Statistics, 2019). In addition to the state pension, approximately 90% of employees accumulate pension through the mandatory occupational pension. However, as a result of the ageing population, pressure on collective pensions has increased in many European countries, including the Netherlands (Wendel et al., 2016; Van Rooij, Lusardi & Alessie, 2012). The Dutch government is trying to make households more responsible for their financial planning, which increases the dependence on private financial decisions.

The high mandatory nature and the limited freedom of choice are characteristic for the Dutch pension system. This may indicate that people are reluctant to invest in pension awareness. Hence, effective pension communication can lead to a relatively large increase in knowledge. Estimates of the effectiveness of PensionTV in the Netherlands might therefore be upper bound estimates. When people have to make more pension choices themselves, they may be more inclined to invest in pension awareness by themselves, given the potentially higher returns.

The life cycle theory indicates that consumption smoothing over the life cycle leads to the greatest benefit. A pension plan participant has to know how the pension system works and how much pension they automatically accumulate in order to know whether this is sufficient to maintain the desired standard of living. Not everyone is well-informed about the amount and conditions of their pension arrangement and therefore insufficiently prepared for retirement decisions (Nell, 2017; Prast & Van Soest, 2016; Wendel et al., 2016; Van Rooij, Lusardi & Alessie, 2012). This makes effective communication about pensions essential. The purpose of pension communication is to enable participants to undertake adequate financial planning for their retirement and to ensure confidence in the pension system and the pension funds (Dutch Authority for the Financial Markets, n.d.; Nell, 2017).

Pension communication is regulated by the Dutch Pension Communication Act. Nevertheless, pension funds are allowed to deploy additional media such as Facebook ads, a website and newsletters. A renown Dutch TV-broadcaster has launched a new pension

(7)

communication tool called PensionTV. This interactive webinar aims to increase general and fund-specific pension knowledge, trust in the pension fund and appreciation for the pension arrangement. Furthermore, it aims to motivate and activate pension plan participants to log in at the personal pension planner to see their personal pension entitlements. Participants can participate interactively by answering poll-questions and asking questions themselves. The first results about the effectiveness are promising and PensionTV is already awarded with the PBM Communication Prize 20181_{. However, the method by which the first macro results are} announced is not sufficiently advanced to allow any definitive conclusions to be drawn.

In this study, extensive research is conducted into the effectiveness of PensionTV on pension plan participants. We take two Dutch pension funds that likely differs in their type of participants: the pension fund of the Fashion, Interior, Carpet and Textile Industry (Mitt) and the company pension fund for information technology company IBM (SPIN). The research question is as follows: What effect does PensionTV has on 1) pension knowledge, 2) appreciation and 3) activation? The first component measures self-perceived pension knowledge prior and after watching PensionTV, while the second component measures the appreciation for the pension arrangement and the level of trust in the pension fund. Thirdly, the actions taken to get informed about one’s personal pension are measured. We will also investigate how participants experience PensionTV as a means of communication and whether they would like to see a broadcast of PensionTV more often.

1.1 Academic relevance

This thesis adds to the literature on pension communication in the following way. There is already a vast amount of research in the field of pension communication, pension awareness and retirement planning. Developments in the field of pension communication are happening rapidly. Pension funds are launching innovative tools that help people understand their pensions and the actions they can take themselves. The effectiveness of many pension communication instruments has already been investigated (e.g. Knoef, Been & Van Putten, 2018; Dinkova et al., 2019; Nell, 2017). They looked into the effectiveness of the Uniform Pension Overview, the use of informative letters, Facebook ads, telephone helpdesks and more. Brüggen, Post and Schmitz recognise that new technological tools such as an interactive online pension planner

1_{PBM's annual pension symposium is a leading event for the pension sector in the Netherlands. The}

organization wants to contribute to better communication in the pension sector and has therefore - well before the Pension Communication Act - set up the PBM Communication Award. In this way, best practices are collected for pension funds and the entire pension sector (PBM, 2019).

(8)

can be a promising solution to promote pension awareness and consequently stimulate retirement planning.

The success of pension communication will depend on the ability to set realistic and measurable targets that can be achieved in a timely, cost-effective and innovative manner to achieve maximum impact. It will also depend on the ability and willingness of pension funds and the government to learn from previous pension communication campaigns (Atkinson et al., 2012; Atkinson et al., 2015). The evaluation of pension communication programmes is therefore of great importance. However, scientific research into the effectiveness of pension communication interventions is still far from complete (Brüggen & Post, 2019). The effectiveness of PensionTV has hardly been researched. Some first macro results have been announced. Although these results are promising, there has not yet been a scientific study into its effectiveness (Van Huijgevoort, 2020). This study therefore contributes to the pension communication literature by studying how this innovative pension communication instrument can help to increase knowledge, appreciation and user experiences. Moreover, this dissertation studies how PensionTV can contribute to motivating and increasing the use of the online pension planner.

1.2 Societal relevance

The importance of analysing the effectiveness of PensionTV on pension awareness is fourfold. First, pension communication can inform employees about their pension arrangement and encourage them to start retirement planning on time (e.g. Dinkova et al., 2019; Knoef, Been & Van Putten, 2018). Research has shown individuals who are better informed about their own pension, are better capable to act in a rational way (Jappelli & Padula, 2013). Moreover, financially illiterate households more often make worse financial decisions compared to households that are considered to be financial literate (Van Rooij, Lusardi & Alessie, 2012). If PensionTV does increase (subjective) pension knowledge, and individuals become more aware of their pension and better informed about the Dutch pension system, more rational financial decisions might be made. However, based on this research, we can indicate whether PensionTV leads to higher self-perceived pension knowledge. Whether it has an effect on saving behaviour or the moment of retirement cannot (yet) be measured. This would be an interesting topic for future research to better understand the effectiveness of PensionTV in the longer term.

Secondly, the better-informed individuals are, the more responsive they are to designed pension incentives because they better understand the argumentation of the government (Chan

(9)

and Stevens, 2008). The results of this study are therefore important for the government. If PensionTV is more effective than other pension communication tools, it might be applied by other pension funds too.

Furthermore, confidence in the Dutch pension system has been poor for years. Research among 1,607 employees who accumulate pension through their employer shows that pension knowledge and trust go hand in hand. After acquiring knowledge, employees indicate that they have more confidence in the pension system. Confidence among the group that received information increases by 15% (Motivaction, 2016). In addition, Böhnke, Brüggen and Post demonstrate that participants with limited subjective knowledge are more likely to feel negative about their future financial well-being (2019). PensionTV aims to increase participants' knowledge and to increase confidence in the sector and the pension fund, which is necessary for the functioning of the system.

Finally, this new pension communication instrument is potentially very cost-effective. The promotion, production and broadcasting of PensionTV costs about €17,000.-. The broadcast can easily be adapted to the target group of other pension funds. The costs for the production of PensionTV are fixed, which means that the cost per viewer decreases as the number of viewers increases. For pension funds where participants have relatively little knowledge prior to PensionTV, the program appears to be the most effective, both in terms of impact and cost effectiveness. The challenge is therefore to get people with a low level of appreciation or knowledge to participate in PensionTV. If the pension fund succeeds in increasing the number of participants in addition, the costs of PensionTV per participant can remain limited.

1.3 Structure of the thesis

This paper continues as follows; the next section illustrates the structure of the Dutch pension system and its characteristics since it is fundamentally different from other countries. Section 2 also explains the Dutch pension communication legislation and (the effectiveness of) additional media used by pension funds. Section 3 of the paper explains the life cycle theory and the effect of pension knowledge on retirement savings. Afterwards, PensionTV will be further explained and information about pension fund Mitt and SPIN will be provided. In section 5 the hypotheses are formulated and the research method will be explained. Data and summary statistics are described in section 6. Descriptive statistics and estimation results are presented in section 7. In section 8 we present recommendations for government policy, pension

(10)

funds and the format of PensionTV. Finally, we conclude in section 9 and provide some additional discussion of the results.

(11)

2. Institutional background

2.1 Dutch Pension System

The Dutch pension system consists of three pillars: state pensions, mandatory fully funded occupational pension arrangement and individual voluntary pension saving. The state income is unrelated to work history. Moreover, the second pillar is mandatory to almost all employees. Mandatory savings for retirement are typical Dutch. If employees assume their first and second pillar pension inadequately to maintain their standard of living, they can make additional pension arrangements in the private third pillar (Debets et al., 2018: 14).

2.1.1 First pillar: state pensions (the so-called AOW)

The first pillar is a state income arrangement, the so-called AOW. The gross monthly income is based on the number of years the individual has lived or worked in the Netherlands. An individual accrues 2% AOW for each insured year. Consequently, 50 years of accrual are required to obtain the full state pension, which means that individuals who came to the Netherlands later in life may not be entitled to full state pension. The state income arrangement provides a gross monthly income, starting at the statutory retirement age. For one individual, this amounts to a gross amount of approximately 1,255 euros per month. For partners, if both are of pensionable age, the gross is 860 euros per month per person2_{. In case a person has not} accrued full AOW benefits, the income may be lower than the minimum social assistance. This happens, for example, if someone has accrued less AOW benefit as a result of a stay abroad. In this case, it is possible to receive supplementary income provision for the elderly (AIO), supplementing income to the minimum of social assistance (Rijksoverheid, n.d.).

The state pension arrangement is partially funded by the pay-as-you-go system. Due to funding issues, the central government makes additional payments from the general resources (taxes). Until a few years ago, the mandatory retirement age was 65. However, due to the ageing of the population, the dependency ratio has increased. As the ratio increases, the productive part of the population will be taxed more heavily in order to maintain the pensions of economically dependent persons. This system was no longer sustainable in the Netherlands. In addition,

(12)

people are getting older and, consequently, they will receive AOW for a longer period of time, which will increase the burden on the system. In order to keep the state pension affordable, it was decided to gradually increase the retirement age from 65 to 67 from 1 January 2012, depending on the date of birth. A recent study by Atav, Jongen and Rabaté shows that raising the retirement age in the Netherlands has been effective (2019). They found an increase in the employment rate of 16 percentage points. The cost savings resulting from the increase are substantial despite an increase in participation in social insurance. From 2022 onwards, the retirement age would be linked to life expectancy (Rutte & Samsom, 2012). Thus, the retirement age in the Netherlands depends on the remaining average life expectancy and therefore depends on someone's date of birth. However, in the 2019 pension agreement, it was decided that the state retirement age will increase at a delayed pace to the age of 67 in 2024 (Rijksoverheid, 2019a). The state pension age is known 5 years in advance.

2.1.2 Second pillar: mandatory fully funded occupational pension

This pillar is most relevant for the scope of this thesis and will, therefore, explained in more depth. Mandatory occupational pension accumulation is agreed upon in collective labour market agreements in the Netherlands. When employers offer a pension scheme, employees are obliged to participate. As a result, approximately 90% of Dutch employers accrue supplementary pension together with their employer. Pension funds can be set up for a sector, or especially for a large company. Both the employer and the employee pay pension contributions, usually to a pension fund. Employers pay about 2/3 of the total pension contributions and employees pay 1/3 part (Rijksoverheid, n.d). These premiums are invested by pension funds to be able to pay out supplementary pensions in the future.

Pension funds must be able to meet their future payments. Whether there is sufficient money in the pension funds is determined by the interest rate curve. This actuarial interest rate has only fallen in recent years because it follows the real interest rate. The lower the notional return, the more money must be in the pension funds to be able to meet the requirements in the future. If the coverage ratio falls below a critical level, pension funds must lower retirement benefits. In the past, retirees could count on a replacement rate of 90% of their net income (Debets et al., 2018; 11). Indexation of pension benefits to cost-of-living increases was the rule. However, in recent years the financial situation of pension funds has not allowed indexation of pensions within the rules that apply in the current pension system. This means that pensions

(13)

cannot adjust with the rising prices, which means the purchasing power of pensions decreases. This applies to both pensioners and employees who are still building up pension (Debets et al., 2018; Opheikens, 2019).

In the new pension agreement was already decided to lower the required coverage ratio from 104 to 100 percent. The next measuring moment was scheduled for 31 December 2019. Pension funds that are below the standard for the sixth consecutive year should reduce benefits. Due to the persistently low interest rates, pension benefits at many pension funds had to be reduced next year (Vogels, 2019). However, under pressure from the labour unions, the opposition parties and the pension funds, Minister Koolmees decided to postpone the pension cuts this year (Rijksoverheid, 2019b). Only for pension funds with less than 90% of the required capital, reductions in pension benefits are still required. To allow employees to also benefit from the intervention, the Minister states that funds should temporarily not increase the pension contribution or reduce the amount accrued by participants. This means that the financial reserves will be used and that current employees may suffer from this in the future. Koolmees his decision to postpone pension cuts is taken to work out the new pension agreement.

The second pillar used to be completely defined benefit (DB) premiums. With DB, the total accrued capital is converted at the retirement date, into a lifelong fixed payment. The amount of the pension benefit is fixed and the amount of the required pension contributions may vary. Both in the Netherlands and in other countries around the world, individual choice-based defined contribution (DC) plans are receiving increasing attention (Böhnke, Brüggen and Post, 2019). In recent years, the number of pension schemes in the Netherlands based on a contribution agreement has increased. In 2015, 15 percent of pension entitlements in the Netherlands were based on a defined contribution agreement. In 2016, the share of DC agreements was 19 percent and in 2017, the percentage of DC agreements increased to 21 percent (Molenaar-Cox & Woestenburg, 2018; Woestenberg, 2020). With Defined Contribution, the level of the pension benefit depends on the premium paid and the investment return achieved. The amount of the pension benefit is determined at retirement. Participants can choose between a fixed (stable) or variable payment. With a variable payment, part of the accrued capital continues to be invested. This may achieve a higher pension outcome, but it also creates uncertainty. When offering a choice between fixed and variable, pension providers must provide participants with relevant information about the consequences and risks of both options. The number of pension participants using flexible pension payments is currently small but is growing rapidly (Dutch Authority for the Financial Markets, 2019; Prast, 2020). Another relevant trend regarding the second pillar pension is the introduction of a lump sum payment

(14)

on the retirement date. Participants can then choose to receive a certain amount (maximum 10% of the saved pension capital) at once when they retire. Future pension benefits will of course then be lower. This offers more flexibility and freedom of choice and is in line with the agreements made in the 2019 pension agreement (Rijksoverheid, 2019c).

At the moment, the Dutch pension system is still generous and there is still relatively little freedom of choice, which is very typical for the Dutch case. Jappelli and Padula demonstrate that in countries with generous social security benefits, there will be fewer incentives to save for old age and build up wealth and therefore less reason to invest in financial knowledge (2013). Mandatory occupational pension benefits have become uncertain due to the threat of pension cuts. In addition, the number of DC pension schemes is increasing and there is an increase in freedom of choice, resulting in a shift of risks from the employer to the employee. It is regularly argued that the shift to defined contribution and allowing more freedom of choice automatically solves the problem of low pension involvement. However, research in Sweden shows that this is not always the case. Freedom of choice is appreciated but not always used (Böhnke, Brüggen and Post, 2019). The shift to defined contribution therefore requires a strong pension communication. Chapter 3 discusses the relationship between financial literacy and pension choices in more detail.

2.1.3 Third pillar: individual voluntary pension saving

The third pillar consists of additional individual voluntary pension saving. Because of the royal mandatory saving, the individual voluntary pension saving, is relatively small in the Netherlands (Debets et al., 2018). For individuals with low occupational pension savings (e.g. self-employed persons) that want to retire before their statutory retirement age or people who do not make ends with their AOW and employer pension, the third pillar can be of importance. A study examined by the Dutch Authority for the Financial Markets found that one third of the population carry the risk of having a net retirement income that is lower than the income they suggest themselves to need after retirement (Wendel et al., 2016; see also Van Rooij, Lusardi & Alessie, 2012). These studies show that the first and second pillar pensions solely are relatively often not sufficient to continue the standard of living.

(15)

2.2 Pension communication

Large-scale pension discounts could be avoided this year, but without indexation for most funds, the situation is still problematic. At the same time, the Dutch cabinet is continuing to review the pension system. A challenging period for the sector, in which the importance of effective communication is unprecedented. At the same time, the shift of risks from the employer to the employee and the increasing freedom of choice are increasing the importance of pension awareness and thus the importance of effective pension communication (Debets et al., 2018).

Before explaining the Dutch legislation in the field of pension communication and supplementary media, the objectives of the pension communication are clarified. Eight goals are obtained based on interviews with 25 pension organizations (Nell, 2017). The main goal is to enable pension participants to make independent decisions, make choices and take action. Moreover, they need to know and understand their pension arrangement. Ten organizations have indicated that they believe it is important for pension participants to have confidence in the pension fund and that this is or should be an explicit goal of pension communication. Confidence in the pension system has been poor for years. The Dutch have poor expectations for their second pillar pension. What is the reason for this? Minister Koolmees pointed out that the financial crisis has weakened the financial position of pension funds. Funding ratios fell by as much as 20 percentage points in 2008 and fell further due to low interest rates. Pensions have not been index-linked for years, reducing purchasing power, and there is a threat of pension cuts (Wolzak, 2018). Effective pension communication might improve trust in the system and in pension funds. Furthermore, plan participants also need to know and understand their own pension situation and the effect certain life events may have on their retirement. Lastly, participants should be aware of the risks that their pension arrangement entails and be able to make a financial planning (Nell, 2017: 33). The Dutch Authority for the Financial Markets (AFM) describes the purpose of pension communication as follows: to enable the participant to arrange adequate financial planning for his old age (Dutch Authority for the Financial Markets, n.d.). Hence, there are plenty of reasons to ensure effective pension communication. The following section explains the Dutch pension communication legislation.

(16)

2.2.1 Dutch legislation on pension communication

In the Netherlands, pension funds must comply with the Pension Communication Act introduced in 2015. According to article 62 of the Pension Communication Act, the purpose of pension communication is: "to enable the pensioner to consult information on his pension rights and entitlements in a clear and comprehensible manner ... The purpose of the pension register is also to provide insight into the amount of the pension to be achieved, the choices relating to retirement and the consequences of these choices and of important events for the pensioner's retirement". In order to achieve this objective, the law prescribes a number of instruments.

With the online Pension Register (in Dutch: mijnpensioenoverzicht.nl (MPO)), the sector contributes to informing participants about accrued and achievable pensions on the retirement date. Both first and second pillar pensions are included in the Pension Register. The MPO also provides insight into the impact of life events on pensions, such as unemployment or early retirement. Partners within a household can log in together to get a more complete overview of the couple's collective situation (Federation of the Dutch Pension Funds, n.d.).

Besides the online pension register, pension funds are obliged to use the following communication channels. First, the Uniform Pension Overview (UPO) periodically provides participants, former participants and retirees with an overview of their pension (Pensioenmodellen, n.d.). Debets et al. estimated the effect of the Uniform Pension Overview on pension knowledge and active pension planning (2018). They followed a large sample of individuals during a long time period and they found that the introduction of the UPO might have a small positive effect on pension knowledge. However, the results are unstable and depend on the identifying assumption.

Second, pension123 is sent to new pension plan participants within three months. The purpose of this communication channel is to provide the member with information about the most important components of the pension scheme. The participant will receive answers to the following important questions: How do I build up pension savings? What choices do I have? How secure is my pension? When do I have to take action? The member decides for himself how detailed he will take the information: in outline (layer 1), with an explanation of the outline (layer 2) or detailed (layer 3). In addition, there is the pension comparison tool, which is part of Pension123. The purpose of this instrument is to make it clear to the participant that pension schemes differ from each other and that it is useful to be aware of these differences, especially when changing jobs (Pensioenmodellen, n.d.).

(17)

2.2.2 Evaluation Pension Communication Act

With the introduction of the Pension Communication Act in 2015, a commitment was made to evaluate the Act after three years. The evaluation shows that about half of the participants indicate that they know when they can retire and how much pension they can expect. One third of the employees know whether their pension is sufficient, what risks are involved and what options are available. Pension knowledge differentiates and appears to be linked to age, income and stage of life. The group that knows a lot is relatively old, has a higher income and has to deal with an approaching pension. Simply offering information is not enough to activate people they take the most essential steps to arrange their pension on time. Pension administrators have indicated that they do not consider the costs of complying with the legal information instruments to be in balance with the attainment of policy objectives. For example, pension administrators consider the personal environments on their websites to be more effective. This is also the reason why administrators invest in other, in their opinion more important, instruments (Van Waveren, Kuin & Duysak, 2019).

Three quarters of the participants use one or more sources of information to gain insight into their own pension. This suggests that the use of pension communication tools is important for participants. Participants consider mijnpensioenoverzicht.nl to be the most useful. The UPO (and the online environments of the pension administrator) also contribute, although to a lesser extent, to achieving the policy objectives. Pension123 hardly contributes to achieving the policy objectives (Van Waveren, Kuin & Duysak, 2019).

Minister Koolmees, in response to the evaluation report, wrote a letter to the House of Representatives (Koolmees, 2020). Based on the evaluation, Koolmees no longer considers it necessary to legally prescribe pension providers to use the standard models for the Pension123 and the UPO. Minister Koolmees expects that pension providers will take the opportunity to make pension information more suitable for their target group. Furthermore, he indicates that the provision of information should be more demand-driven rather than supply-driven, which is more (cost-) effective. Moreover, it is important to focus more on activating participants. Koolmees is thinking of making it compulsory to mention a concrete action. These policy intentions have yet to be discussed with the pension sector and subsequently in the House of Representatives. What is certain is that Koolmees wants to give pension funds more freedom to adapt pension communication to the wishes of their target group.

(18)

2.2.3 Additional pension communication media

Research shows that pension funds are not satisfied with the Pension Communication Act. The law often requires information to be provided in a complete and comprehensible manner. However, information can never be 100 percent complete and 100 percent understandable, resulting in a trade-off between these two goals. In addition, many pension funds state that statutory media such as the UPO and Pension123 are insufficiently read or understood by their clients (Nell, 2017: 34). They also experience that the law is difficult to interpret and find it difficult to meet all requirements.

This negative attitude towards the Pension Communication Act has resulted in three main communication strategies among pension organizations. Most accept the situation as it is and do not try to ameliorate it, some choose to optimize the legally required instruments of communication and others focus on the design of additional means of communication, such as websites and alternative brochures, because they think those media are more effective (Nell, 2017: 173). Pension funds mention three arguments for the use of additional media. Firstly, they want to provide participants with information for which there is no possibility within the statutory means (e.g. information about current developments concerning pensions). That is why pension funds are looking for other ways to reach their participants. Secondly, pension funds indicate that the legally required media are insufficiently understood by their clients. With additional communication, pension funds explain the use of the legally required funds. Finally, pension funds prefer to use a different use of language than is legally required. Difficult and technical terms are often used to comply with the law. In the evaluation study on the Pension Communication Act, participants frequently suggest simplifying information and using understandable language (in Dutch: 'Jip-en-Janneketaal') (Van Waveren, Kuin & Duysak, 2019). In additional communication, funds are free to use language that is understandable to everyone.

(19)

Figure 1. Additional media used by pension funds (Nell, 2017: 35)

In Figure 1, additional communication media used by pension funds are divided into three categories: written, digital and oral. Dozens of additional media are already being used by pension funds, for example videos, webinars, quizzes or games and information meetings. PensionTV is an interactive webinar to which participants actively contribute. Informative videos are also used and questions can be asked during the broadcast. PensionTV is innovative as Figure 1 does not yet show a means of communication such as PensionTV. Chapter 5 explains PensionTV in more detail.

How effective are existing additional communication instruments? The remainder of this section will evaluate the effectiveness of various pension communication instruments using available literature. Knoef, Been and Van Putten analysed the causal effect of an informative letter on pension awareness versus the effect of a Facebook ad and examined whether positive or negative framing is more effective in increasing pension awareness (2018). They conducted a randomised field experiment with a representative sample of 4,000 people in the Netherlands and found the Facebook ad successful in activating respondents to log into the online pension register. The online pension register is a legally required pension communication instrument. Thus, social media can be useful in optimising the impact of legally required means of communication such as the online pension register. PensionTV also encourages participants to go to their pension planner via a link to activate them to learn more about their pension. The letter is less effective in stimulating to log in at the online register. However, the negatively framed letter is successful in raising pension literacy.

In collaboration with a company specialising in technologically advanced pension planners, Brüggen, Post and Schmitz have created three different versions of an online pension planner (2019). The basic pension planner only allowed participants to check their pension

(20)

situation and the composition of their future pension income. With the second pension planner with average interactivity, participants could choose to adjust certain variables that affect their retirement income. The highly interactive pension planner also included an interactive budget tool and showed whether the accrued pension income was adequate to cover the desired expenditures. Brüggen, Post and Schmitz collected data using an online panel consisting of 285 participants. They find a positive effect of interactivity on the number of clicked options. Women prefer average interactivity while men prefer the highest level of interaction. Moreover, they find a positive effect on the participants’ intention to check their personal pension situation in the coming three to six months.

Another field experiment in the Netherlands examined 8563 employees with mandatory occupational pension arrangements. Dinkova et al. tested the effectiveness of digital pension documents by looking at the duration of the online visit and click behaviour (2019). The additional pension documents were customized for three age categories with different pension goals. It is considered to be important for young participants to know how their pension is arranged. Older and middle-aged groups want to be informed about the choices they have within their retirement plan and whether they are on track to an adequate pension. They found customization has the intended effect for older employees: they clicked on relevant information and not on non-relevant information.

Elling and Lentz also conducted research into customized pension communication. They distinguish four forms of customization: segmentation of target groups, personalization of documents, tailoring and, finally, personal interaction with participants. The latter form is seen as the ultimate form of customization. Both the legislator and the pension providers consider tailor-made pension communication as an opportunity to increase the motivation of the pension participants and to make the information more understandable and easier to find (Elling & Lentz, 2019: 4). By means of 21 semi-structured interviews, the authors collected information on how pension organisations deal with tailor-made solutions. Although most organizations indicate that customization is important, Elling and Lentz hardly see it implemented in practice. Many organizations do invest in a pension planner in which participants can make choices and then see results that apply to them. This planner can be seen as a form of customization in which the content is personalized. Technological developments, particularly in the field of digitalization and changing media use, create more opportunities for pension communication. Customization is not only about the content, but also about the channel on which it is offered. The authors see a future for the use of video, animation and chat. However, they argue that despite the digital innovations, real interaction still mainly takes place

(21)

in meetings, for example in the workplace, with (groups of) employees. Innovative opportunities are therefore not yet exploited here (2019: 28). PensionTV responds to this by developing a new interactive communication instrument that enables direct contact with the participants via the chat function.

(22)

3. Theoretical framework

3.1 Life cycle theory

The life cycle theory assumes that it is optimal to spread consumption over the life cycle. This is based on the fact that an extra unit of consumption produces increasingly less additional economic utility. The first 100 euro income per week is very valuable. However, if someone already earns 5000 euros, 100 euros extra is not that much. So, this 100 euros is less beneficial than the first 100 euros. When consumption is saved in good times to be used in times when less money is being made, it yields a consumer more benefit throughout his life than if he consumes a lot today and, consequently, can consume little after retirement (Wendel et al., 2016). The life cycle model assumes that individuals are rational and able to spread consumption over life time and make decisions to maximize life cycle utility. Hence, consumers should consider beforehand what part of their income they consume and what they save for their retirement. However, households often deviate from the optimal plan because of two reasons. First, it is difficult for households to calculate the optimal percentage of savings because this requires a certain degree of financial knowledge. A second reason why consumers do not save enough for their pension is that, even if the correct savings ratio was known, households may not have the self-control to reduce current consumption in favour of future consumption (Thaler and Benartzi, 2004). To protect people from retirement where the standard of living cannot be continued, most employees in the Netherlands are obliged to join a pension fund (Bovenberg, et al., 2007). As has been mentioned before, individuals get increasing opportunities to make their own pension choices. This results in more responsibility on the part of the individual. In order to be able to make sensible choices, information is of great importance. This research focuses on the effect of information provision and pension communication on the pension awareness of individuals.

3.2 Financial literacy

A large part of the world’s population lacks the financial knowledge to make well-informed decisions that are in their own interest (Debets et al., 2018; Lusardi & Mitchell, 2014; Van Rooij, Lusardi & Alessie, 2012). Financial literacy can be measured through three questions that measure whether someone understands the compound interest rate principle, the inflation principle and the principle of risk diversification (Lusardi & Mitchell, 2014). These three questions are used in the Aegon Retirement Readiness survey to measure financial literacy.

(23)

This survey includes information from 14,400 employees and 1,600 retirees in 15 countries across Europe, the Americas, Asia and Australia. Only 30 percent were able to answer the three financial literacy questions correctly.

In the Netherlands, it was only 27 percent who were able to give correct answers (Aegon Retirement Readiness Survey, 2019). Moreover, only 30 percent of the labour force indicate they are well informed about the pension system in the Netherlands (Wijzer in geldzaken, 2018). The majority do not even know what their income will be after retirement. In the past two years, more people have become aware of the importance of being involved in their own retirement. Yet the majority of the labour force mentioned they need to be more involved in their financial situation after retirement (Wijzer in geldzaken, 2018). People seems aware of the importance of preparing for retirement, but they still procrastinate and do not engage in it. Given the shift to DC, it is advisable to make policies to improve the financial literacy of the population, so that sensible choices are more likely to be made.

Moreover, Lusardi and Mitchell find there are differences in financial literacy among socio-economic groups. For example, women are less financially literate than men and older people give themselves a higher score, despite scoring poorly on the financial literacy questions. They also find that women more often indicate that they do not know an answer. This suggests that women are relatively well aware of their lack of financial knowledge which makes them extremely suitable for financial education programs (Lusardi & Mitchell, 2014). In our research we will condition the effectiveness of PensionTV for gender and age. We will also investigate whether women are more likely to indicate that they do not know the answer.

3.3 Role of information on retirement planning

So, financial literacy is low in the world and in the Netherlands. The question is, how does financial literacy (or more specifically pension literacy) matters for economic decision-making, including pension planning? Research by Van Rooij, Lusardi and Alessie shows that financial-literate people are better able to make rational financial decisions than individuals who have little financial knowledge. This also has consequences for retirement planning and savings (2012). The more financially capable are also likely to engage in pension planning (Lusardi & Mitchell ,2014: p. 22). This is supported by several studies in different countries. The studies show that the correlation between financial literacy and different pension planning measures is quite robust. Moreover, Bernheim and Garret found that employer-based retirement education significantly increase retirement savings (1996: p. 34). Furthermore, financial literacy is

(24)

strongly related to retirement savings, choice of investment portfolios, understanding of pension arrangements and subjective financial security (Jappelli & Padula, 2013).

Hence, financial literacy has, on average, a positive effect on pension planning. Critics may, nevertheless, question the relevance of this to the Netherlands. Since the Dutch pension system is relatively generous, and the vast majority of employees save through compulsory defined benefit pension plans with mandatory contributions, it may be discovered that they are currently over-saving and, when financial literacy increases, they will adjust their savings downwards. However, research from the Authority for the Financial Markets (AFM) shows that replacement rates of the Dutch compulsory pension system are often below what is expected and insufficient to provide the desired standard of living. Van Rooij, Lusardi and Alessi analysed the relationship between financial literacy and retirement planning in the Netherlands and found a positive and statistically significant effect of financial literacy on pension planning (2012). Moreover, they found that confidence in economic knowledge is positively related to the chance someone is likely to calculate how much they need to save for retirement purposes (Van Rooij, Lusardi and Alessi, 2012: p. 467).

3.4 Government incentives

Even if individuals rationally choose not to invest in pension knowledge, it is socially desirable to increase financial literacy (Knoef, Been & Van Putten, 2018). Pension literacy makes people better understand government incentives. Policies to retire later are often based on raising the normal retirement age and introducing benefit reductions in the event of early retirement (Giesechke & Yang, 2017). Financial incentives can induce people to actually continuing to work longer. However, retirement decisions require pension knowledge. Individuals who lack financial knowledge do not understand what the financial incentives are for and what they are trying to achieve, which makes them less effective (Bernheim & Garrett, 1996: p. 1). Chan and Stevens found that ill-informed people do not react to actual retirement incentives while individuals that are well-informed respond stronger than previously thought (2008: p. 253). Effective pension communication may therefore benefit the effectiveness of government policies.

(25)

4. A new pension communication instrument: PensionTV

‘’The fragile reputation of the pension sector makes it difficult to communicate effectively with its participants. Tailor-made communication and a good connection to the perceptions of your participants helps in this respect. Inventiveness is needed, as well as the courage to try something different. This year's winners show that this is possible". These words of praise were spoken by the jury when PensionTV was awarded with the PBM Communication Prize 20183_. What appealed to the jury is that this concept can be applied quite simply and with a limited budget by any other pension fund (PBM, 2018).

4.1 The format

PensionTV is a new, interactive pension communication instrument with the aim of informing pension plan participants and increasing pension knowledge. PensionTV can be seen as a 70-minute crash course in pension planning. It provides information about the Dutch pension system as well as fund-specific information. Explanatory videos and animations reinforce the transfer of information. For the production of PensionTV, there has been a collaboration between the pension fund and RTL, one of the largest television producers in the Netherlands. This collaboration delivers high-quality television. PensionTV is an accessible concept in which employees can easily participate. The pension fund invites employees to take part in the interactive webinar by means of e-mails, explanatory videos and flyers. Plan participants can already ask questions in advance so that the show is tailored to the needs of the viewers. Hence, PensionTV offers customization with this innovative concept. There is the possibility for personal interaction during the broadcast via the live chat function. A team is available behind the scenes to answer chat questions during the show and some questions are answered directly by the speakers of the pension fund. Participants have the choice to watch the interactive webinar live or watch the broadcast afterwards.

During the webinar information about the viewers is collected4_{. These are questions of} knowledge and opinion. Viewers can simply answer the poll-questions by clicking on the answer they think is correct. After a few minutes, the answer to the question is discussed in the

(26)

broadcast and respondents are able to see what others have filled in. During the break, viewers have the opportunity to look at their pension fund's pension planner via a link that appears on the screen. The break is quite short to take a detailed look at the pension planner, but those who clicked on the link are interested and motivated to look at their personal pension during or after the broadcast. After the broadcast, viewers will be asked to fill in a survey with questions about demographics and again about appreciation and knowledge. The survey questions are presented in appendix B.

Four broadcasts of PensionTV have been made for the pension funds SPIN, Mitt, PNO Media, and ppi Cappital. In this study, the broadcasts for pension funds Mitt and SPIN are analysed. The exact content of PensionTV is tailored to the wishes of the pension fund, so that the communication matches the needs of the pension participants as much as possible. The questions that were asked to the Mitt sample during and after the broadcast therefore deviate from the questions that were asked to the SPIN sample. However, the set-up of the two webinars is the same, so that the effectiveness of PensionTV for Mitt and SPIN can be compared. Nevertheless, it has to be taken into account that the questions are different and a one to one comparison is therefore not possible.

In the following sections, we provide some background information on the Mitt and SPIN pension funds and briefly compare the two funds. In the institutional framework, we already briefly mentioned sector-specific and company-specific funds. We have an example of each with pension fund Mitt and SPIN.

4.2 Pension fund Mitt

Mitt is the pension fund of the Fashion, Interior, Carpet and Textile industry. At the end of 2018, pension fund Mitt had nearly 20,000 active plan participants. These are employees who are currently building up pension with Mitt (Mitt, 2019). The average age of all employees in the Mitt sector is 43.8 years. A little over 40 percent of employees in the industry are women. The level of education in the sector is comparable to the average level of education in the Netherlands. Mitt's workforce consists of 28% employees with a higher professional education (hbo) degree or academic (wo) degree. Over 40% have an MBO, HAVO or VWO education level, approximately 23% have a VMBO background and 8% primary education (Mitt, 2017). For pension fund Mitt, the subjective pension knowledge of the participants before and after PensionTV is measured, as well as the change in trust in pension fund Mitt. Trust in a pension fund is probably partly dependent on the financial position of the fund. The coverage ratio

(27)

indicates how financially healthy a pension fund is. At the time of the broadcast (November 2019), Mitt‘s coverage ratio was 96.6% (Mitt, 2020). If the coverage ratio is 100%, the pension fund has enough money to pay out all future pensions, but in the long term it has no money left over to deal with risks or to increase pensions. This means that, like many other pension funds in the Netherlands, Mitt's financial position is inadequate. The pension plan participants of pension fund Mitt accrue pension in a DB plan. When pensions have to be reduced, the participants immediately notice this in their financial accounts. The poor financial position can have a negative impact on the confidence in pension fund Mitt.

4.3 Pension fund SPIN

SPIN is the company pension fund for IBM, which is the largest information technology company in the world (IBM, n.d.). They administer the pensions for about 14,000 people who work or have worked at IBM. The average salary per employee at IBM Netherlands is 108,000 euros, which is three times the average Dutch income (Meij, 2019). Because the average salary is so high, IBM employees are relatively more dependent on second pillar pensions than employees in the Mitt industry. The questions that were asked during the IBM broadcast are therefore more focused on the second pillar pension and the specific pension arrangement. The appreciation for the pension scheme is measured before and after PensionTV. The majority of the participants (8503 individuals) have a defined benefit pension plan. The employee pays a contribution of 3% of the pensionable base. IBM will pay the rest (SPIN, n.d.). 2114 persons have a defined contribution scheme and 3281 persons have a combination of DB and DC. The people with a DC scheme can choose from 3 investment profiles: defensive, basic and investing. They can change profiles every month and SPIN invests within the chosen profile according to age cohort. This means that the investment mix shifts from risk-bearing to risk-averse as the retirement age approaches (Ramaer, 2019). Every month, IBM and the employee make a contribution. The personal contribution consists of a maximum of 5% of the DC base. In addition, one can opt for an extra personal contribution.

SPIN's financial position is exceptionally good. Where many pension funds run the risk of having to cut back, the current coverage ratio of IBM's pension fund is far above the required coverage ratio. At the time of PensionTV (September 2019), the coverage ratio was 129.2% (SPIN, 2020). Because of the persistently high coverage ratio, DB participants do not have to fear pension cuts. SPIN is an exception and therefore the valuation of the pension scheme will probably be higher on average than for other pension funds.

(28)

4.4 Comparison

Hence, pension fund Mitt and SPIN are quite different from each other. Mitt is a sector-specific pension fund that administers the pension scheme for many employers in Fashion, Interior, Carpet and Textile industry. As a result, the pension plan participants are probably more heterogeneous than at pension fund SPIN, where only IBM employees are affiliated. Moreover, the average salary and educational level of SPIN participants is very high compared to Mitt. Furthermore, the type of pension plan also differs between these two funds. Mitt only operates a defined benefit scheme, while SPIN operates DB, DC and combination schemes. Finally, the coverage ratio of the pension funds is also very different. SPIN is in an extremely financially healthy position, where pension fund Mitt is doing everything in its power to prevent pension cuts in the coming years. Because of these differences, it is interesting to investigate the effectiveness of PensionTV for these pension funds and to compare the results. It is plausible that the effectiveness of PensionTV differs per pension fund, given the difference in characteristics of participants, the difference in pension scheme and the difference in financial position of the two funds.

(29)

5. Hypotheses and methods

5.1 Hypotheses

To be able to make causal statements about the effectiveness of PensionTV as a means of pension communication, we have defined three hypotheses. The first hypothesis involves the effectiveness of PensionTV. We compare results before PensionTV with results after the interactive webinar and by using descriptive statistics and a paired t-test we can make an initial statement about the effectiveness of PensionTV. To determine the degree of causality we need hypotheses 2 and 3. We correct for gender and age, prior information (subjective pension knowledge and appreciation) and the type of pension scheme. We assume that these variables together indicate a person's heterogeneity in terms of knowledge and preferences to participate in PensionTV. Based on these three hypotheses we will formulate a concluding statement on the effect of PensionTV.

5.1.1 Effectiveness of PensionTV

The purpose of this study is to determine the effectiveness of PensionTV. For this, several independent variables are used to make a statement about the effect on; subjective and objective pension knowledge, appreciation, trust and activation. We expect that PensionTV will have a positive effect on those outcome variables, as the aim of PensionTV is to increase knowledge and trust. The literature in the field of pension communication substantiates this (e.g. Knoef, Been & Van Putten, 2018; Dinkova et al., 2019; Brüggen, Post & Schmitz, 2019).

H10: PensionTV has no effect on 1) knowledge, 2) appreciation and 3) activation H11: PensionTV has a positive effect on 1) knowledge, 2) appreciation and 3) activation

5.1.2 Effectiveness of PTV by demographics

Secondly, we investigate whether demographics do explain the variation in outcome variables. Lusardi and Mitchell found that there are differences in financial literacy among socio-economic groups (2014). For example, women are less financially literate than men and woman are well aware of their lack of knowledge. We expect that demographics can partially explain variation in the effectiveness of PensionTV. Because the shortcoming of nonrandomization in

(30)

our research it is necessary to control for the effect of demographics on the variables of interest. This partly determines the degree of causality of the effects we find for the first hypothesis.

H20: Demographics do not explain variation in the effectiveness of PensionTV H21: Demographics do explain variation in the effectiveness of PensionTV

5.1.3 Effectiveness of PTV by prior information

Unfortunately, apart from age and gender, we have no data available on other characteristics, such as education, position and salary. However, we do have information about the subjective pension knowledge of Mitt participants prior to PensionTV. This variable includes information on unobserved characteristics such as motivation, interest in pensions, knowledge and (time-) preferences. We therefore assume that, along with gender and age, this variable indicates the heterogeneity of individuals to participate in PensionTV.

For pension fund SPIN, we have no information available about the subjective pension knowledge of participants. For SPIN, however, information is available on the type of pension scheme and the appreciation for the SPIN pension scheme. The valuation of the pension scheme prior to PensionTV indirectly provides information about non-observed characteristics such as knowledge of the pension scheme, interest in the subject and time availability to dig into the subject.

H30: Prior information does not predict the effectiveness of PensionTV. H31: Prior information does predict the effectiveness of PensionTV.

In addition, some SPIN participants have a defined contribution scheme, whereby they can choose their own investment profile. As a result, participants with this type of scheme may be more involved in their pension than defined benefit participants. There are studies that argue that a switch from defined benefit to defined contribution schemes would solve the problem of low involvement. However, there are also studies that show that freedom of choice is valued but ultimately little use is made of it (Brüggen, Böhnke & Post, 2019). The freedom of choice could therefore lead to a higher valuation of the pension scheme but, on the other hand, more freedom of choice also entails more risk. In addition to prior appreciation, we therefore control our results for the type of pension scheme.

(31)

5.2 Methods

This research takes the form of a quantitative field experiment. Unfortunately, the sample is not randomized and there is no control group. Participants could self-selected themselves into subscription and participation and there is no information available for the individuals who did not participate, which is a limitation of this research. It would be interesting to know the characteristics of people who do not participate in PensionTV and to compare them with the sample that does participate. We condition for different background characteristics and we compare our sample with the entire population of the pension funds. Nevertheless, these data must be interpreted with caution because the sample we use is not a representative sample of the pension fund due to potential self-selection bias. We use prior information (subjective knowledge or appreciation) in the OLS-regressions as a way to condition for this potential self-selection bias. Despite the absence of a control group, the effect of PensionTV can be determined on the basis of a pre- and measurement. The difference between pre- and post-measurement is caused by the broadcast of PensionTV.

The study was conducted in the form of a survey, with data being gathered during the PensionTV broadcasts and through online surveys afterwards. PensionTV is customized to the wishes of the pension fund. The poll-questions therefore differ per pension fund. While Mitt provides information about subjective and objective pension knowledge, SPIN does provide information about satisfaction with the pension scheme before and after watching PensionTV. We are therefore forced to report the results of Mitt and SPIN separately and the results cannot be compared one-to-one. Moreover, during the pension communication award ceremony5_{, the} chairman of the jury stressed the importance of corporate communication and reputation management for the pension sector. "Your own position as a sender of important pension messages has a major impact on their effect” (PBM, 2018). The effects of PensionTV can therefore differ per pension fund, especially since the financial position of Mitt and SPIN is very different.

We start by directly comparing the subjective knowledge or appreciation prior to PensionTV with the measurements after PensionTV. Based on descriptive statistics, we can already make initial statements about the effect of PensionTV. We also perform a paired t-test, which compares the difference in the means from the two variables measured (before and after)

(32)

on the same set of subjects, while taking into account the fact that the scores are not independent. The t-test excludes individuals who have only completed one measurement moment (Student’s t-test, 2020). The difference is therefore caused by PensionTV and not because the sample from the pre-measurement differs from the post-measurement sample.

The most commonly known linear regression method is the Ordinary Least Squares (OLS) method. By employing several OLS-regressions, we will attempt to estimate the effect of demographics on the effectiveness of PensionTV. The OLS method minimises the sum of the squared deviations of the measured values from the estimated linear function (Sen & Srivastava, 2003). We can easily include prior information (subjective knowledge, appreciation and type of pension scheme) to correct for some unobserved characteristics. The main advantage of this method is that the estimated coefficients can be directly interpreted as the average marginal effect. These results can be used in statements about the impact of PensionTV and can be easily understood by a wide audience.

As with almost every method, there can also be difficulties and pitfalls in applying the least squared error regression in practice. The least squares regression can perform very badly when some points in the data have excessively large or small values for the dependent variable compared to the rest of the data. The reason for this is that since OLS-method is concerned with minimizing the sum of the squared error, any data point that has a dependent value that differs a lot from the rest of the data will have a disproportionately large effect on the results. Outliers can therefore dramatically distort the least squares solution and hence will lead to less accurate predictions (Sen & Srivastava, 2003: 157-158). To mitigate this problem, we plot the data and, if there are obvious outliers in the data, we will do a robustness check by deleting outliers.

(33)

6. Data and summary statistics

6.1 Data

The data in this study is drawn from two main sources: pension fund Mitt and pension fund SPIN. Figure 2a and 2b present the structure of the data and provides insight into the number of individuals who have been invited to participate in PensionTV, the ratio of registrations and the actual number of participants. Over 10,000 employees have been invited by pension fund Mitt to participate in PensionTV. This is about half of the active participants of the pension fund. Only 844 of these people have registered to participate and 221 of them started watching PensionTV. Interestingly, while SPIN invited only 1492 plan participants, more than half of them selected themselves for PensionTV and 673 of them truly watched. Notice that due to missing values, the number of observations used in the regressions will usually be smaller. However, the participation rate for SPIN is much higher than for Mitt, which indicates other unobserved characteristics, such as interest in the topic, time-availability or level of education. These characteristics make SPIN’s plan participants more likely to select themselves to participate in PensionTV than those of Mitt. However, this difference in participation rate can also be caused by a possible difference in promotion of PensionTV among employees. Unfortunately, we do not have this information. Moreover, the questions used for the Mitt broadcast deviate from the questions used for the SPIN webinar. The results will therefore be presented separately, although we will make comparisons.

6.2 Summary statistics

Participation

Participation was measured using two variables: the number of minutes the broadcast was watched (timeonline) and whether the broadcast was finished. The descriptive statistics are presented in Table 1. The broadcast lasts 72 minutes in total. The dummy variable finished is 1 if one watched longer than 70 minutes, since relatively many people watched 71 minutes. PensionTV appears to keep the focus of the participants for a long time. The average time online is 54 and 61 minutes, for Mitt and SPIN participants respectively. Almost sixty percent has completed the broadcast. On average, we find a substantially higher completion rate for the SPIN sample (63.5%) compared to the Mitt sample (40.5%). It makes sense that the

Raising pension awareness through PensionTV: Evidence from two Dutch pension funds