Financial literacy, stock market participation, and hybrid pension systems. A quantitative analysis of the relationship in the context of current pension reform plans in Germany

Thèse de Master , 2024 , 110 Pages , Note: 1,3

Economie politique - Finances

Demographic change will place an increasing burden on pure pay-as-you-go pension systems, such as the one in Germany, which are not designed to accommodate the changing needs of an aging population. As the baby boomers approach retirement, the ratio of workers to retirees will deteriorate, resulting in a growing need for pension payments to be met by a shrinking number of people in the labor force. The retirement age in Germany is 67, and those aged 67 and over currently represent about 20% of the total population, while those aged 20 to 66 make up about 61% (Statistisches Bundesamt, 2024). This means that there are approximately three people in the workforce for every retiree in 2024. After the retirement of the baby boomers, this number will drop to about 2.2 by 2044 and as low as 2.0 by 2064. Furthermore, life expectancy is increasing continuously, necessitating the disbursement of pensions to retirees for a more extended duration (Statistisches Bundesamt, 2023). As a result, in Germany, those born after 1987 are among the losers of the welfare state, as they will contribute more to financing it over their remaining lifetimes in today's values than they will receive in social benefits, including old-age pensions (Stiftung Marktwirtschaft, 2024). The state already contributes more than 100 billion euros per year from tax revenues to the German pension system (Bundesministerium der Finanzen, 2023), while social security contributions still average just under 40% of gross income (Deutsche Rentenversicherung, 2024). Consequently, political reforms are necessary to mitigate the need for ever more tax revenue and higher contributions to finance the social security system and maintain pension levels.

Table of contents

Table of figures

Table of tables

Table of abbreviations

1 Introduction
1.1 Background and motivation
1.2 Previous literature and research contribution
1.3 Methodology and structure

2 Data and variable construction
2.1 A scoring model for selected countries
2.1.1 Degree of funding and DoF score
2.1.2 Financial literacy and FinLit score
2.1.3 Stock market participation and SMP score
2.1.4 DoF score versus SMP score
2.1.5 DoF score versus FinLit score
2.2 Primary variables
2.2.1 Independent DoF variables
2.2.2 Dependent FinLit variables
2.2.3 Dependent SMP variables
2.3 Control variables
2.3.1 Individual characteristics
2.3.2 Risk aversion
2.3.3 Investor protection
2.3.4 Familiarity
2.3.5 Trust
2.3.6 Economic indicators
2.3.7 Political uncertainty
2.3.8 Tax rates

3 In-depth analysis
3.1 Difference-in-differences analysis
3.1.1 FinLit difference-in-differences
3.1.2 SMP difference-in-differences
3.2 Hypothesis testing.
3.2.1 H1: The DoF has a positive impact on FinLit across countries.
3.2.1.1 Bivariate regressions on FinLit and its components.
3.2.1.2 Multiple-variable regressions on FinLit and its components.
3.2.2 H2.1: The DoF has a positive impact on SMP across countries.
3.2.2.1 Bivariate regressions on SMP.
3.2.2.2 Multiple-variable regressions on SMP.
3.2.3 H2.2: A CoO within a funded pension component results in a more pronounced positive impact of the DoF on SMP.

4 Conclusion.
4.1 Summary and magnitude of the results.
4.2 Implications for pension policy and reform plans in Germany.
4.3 Recommendations for future research.

Appendix I: Figures

Appendix II: Tables

Bibliography.

Sources

Table of figures

Figure 1: Statutory DoF for all 19 relevant OECD countries. 7

Figure 2: Combined DoF for all 38 OECD countries. 8

Figure 3: DoF score for all 38 OECD countries. 8

Figure 4: FinLit score and its composition for all 20 relevant OECD countries. 11

Figure 5: Big Three score for all 31 relevant OECD countries. 13

Figure 6: Direct SMP rate for 29 OECD countries. 16

Figure 7: Indirect SMP rate for 24 OECD countries. 16

Figure 8: Total SMP rate (SMP score) for 26 OECD countries. 17

Figure 9: Scatter plot of DoF score versus SMP score. 18

Figure 10: Scatter plot of DoF score versus FinLit score. 19

Figure 11: Predicted direct SMP rates for countries with and without a CoO (1st) 49

Figure 12: Population pyramid for Germany in 2024. 55

Figure 13: Population pyramid for Germany in 2044. 55

Figure 14: Population pyramid for Germany in 2064. 56

Figure 15: Workplace DoF for all 13 relevant OECD countries. 56

Figure 16: Financial knowledge score for all 20 relevant OECD countries. 57

Figure 17: Financial behavior score for all 20 relevant OECD countries. 57

Figure 18: Financial attitudes score for all 20 relevant OECD countries. 57

Figure 19: Financial literacy score for all 20 relevant OECD countries. 58

Figure 20: Knowledge of the time value of money in all 31 relevant OECD countries. 58

Figure 21: Knowledge of simple and compound interest in all 31 relevant OECD countries 59

Figure 22: Knowledge of risk diversification in all 31 relevant OECD countries. 59

Figure 23: Big Three score for all 31 relevant OECD countries. 60

Figure 24: Predicted direct SMP rates for countries with and without a CoO (3rd) 61

Figure 25: Predicted total SMP rates for countries with and without a CoO (1st) 62

Figure 26: Predicted total SMP rates for countries with and without a CoO (3rd) 63

Table of tables

Table 1: FinLit DiD for Poland vs. Germany. 35

Table 2: SMP DiD for relevant OECD countries vs. Germany. 36

Table 3: Model 1: Bivariate regressions of indep. DoF variables on dep. FinLit variables (1) 37

Table 4: Model 1: Bivariate regressions of indep. DoF variables on dep. FinLit variables (2) 38

Table 5: Model 2: Bivariate regressions of indep. DoF variables on dep. FinLit variables (1) 38

Table 6: Model 2: Bivariate regressions of indep. DoF variables on dep. FinLit variables (2) 39

Table 7: Bivariate regressions of indep. DoF variables on dep. SMP variables (1) 41

Table 8: Bivariate regressions of indep. DoF variables on dep. SMP variables (2) 42

Table 9: SMP regression models: Statutory DoF and Sex. 45

Table 10: Magnitude of the potential effect of indep. DoF variables on dep. SMP variables 51

Table 11: Direct SMP regression models: Magnitude of the combined DoF and statutory DoF 51

Table 12: Dataset summary statistics. 64

Table 13: Risk aversion proxy VIF analysis. 65

Table 14: Risk aversion proxy R-squared analysis. 65

Table 15: Risk aversion proxy t-test analysis. 66

Table 16: Familiarity proxy VIF analysis. 66

Table 17: Familiarity proxy R-squared analysis. 66

Table 18: Familiarity proxy t-test analysis. 67

Table 19: The impact of general trust in others on direct SMP. 67

Table 20: Trust proxy VIF analysis. 67

Table 21: Trust proxy R-squared analysis. 68

Table 22: Trust proxy t-test analysis. 68

Table 23: FinLit DiD details for Poland vs. Germany. 69

Table 24: SMP DiD details for Israel vs. Germany. 69

Table 25: SMP DiD details for Poland vs. Germany. 70

Table 26: SMP DiD details for Norway vs. Germany. 70

Table 27: Model 1: Significant coefficients of indep. DoF variables on dep. FinLit variables (1) 71

Table 28: Model 1: Significant coefficients of indep. DoF variables on dep. FinLit variables (2) 72

Table 29: Model 2: Significant coefficients of indep. DoF variables on dep. FinLit variables (1) 72

Table 30: Model 2: Significant coefficients of indep. DoF variables on dep. FinLit variables (2) 73

Table 31: Significant coefficients of indep. DoF variables on dep. SMP variables (1) 73

Table 32: Significant coefficients of indep. DoF variables on dep. SMP variables (2) 74

Table 33: Direct SMP regression model: Combined DoF. 75

Table 34: Direct SMP regression model: Statutory DoF (1) 76

Table 35: Direct SMP regression model: Statutory DoF (2) 77

Table 36: Direct SMP regression model: Statutory DoF (3) 78

Table 37: Direct SMP regression model: Statutory DoF and financial literacy. 79

Table 38: Indirect SMP regression model: Statutory DoF. 80

Table 39: Total SMP regression model: Statutory DoF. 81

Table 40: Direct SMP regression model: Statutory DoF (4) 82

Table 41: Direct SMP regression model: Statutory DoF excl. collective benefits. 83

Table 42: Direct SMP regression model: Workplace DoF excl. collective benefits. 84

Table 43: Direct SMP regression model: Interaction of combined DoF and CoO.. 85

Table 44: Total SMP regression model: Interaction of combined DoF and CoO.. 86

Table 45: Regression models on indirect (I) and total (T) SMP: Magnitude of the statutory DoF 87

Table of abbreviations

CoO = Choice of options

DiD = Difference-in-differences

DoF = Degree of funding of a pension system

EPU = Economic policy uncertainty

FinLit = Financial literacy

GDP = Gross domestic product

GPS = Global Preference Survey

HFCS = Household Finance and Consumption Survey

INFE = International Network on Financial Education

OECD = Organization for Economic Cooperation and Development

SHARE = Survey of Health, Ageing and Retirement in Europe

SMP = Stock market participation

VIF = Variance inflation factor

1 Introduction

1.1 Background and motivation

Demographic change will place an increasing burden on pure pay-as-you-go pension systems, such as the one in Germany, which are not designed to accommodate the changing needs of an aging population. As the baby boomers approach retirement, the ratio of workers to retirees will deteriorate, resulting in a growing need for pension payments to be met by a shrinking number of people in the labor force, as illustrated by the population pyramids for Germany for the years 2024, 2044, and 2064 in Figures 12, 13, and 14, respectively (cf. Appendix I). The retirement age in Germany is 67, and those aged 67 and over currently represent about 20% of the total population, while those aged 20 to 66 make up about 61% (Statistisches Bundesamt, 2024). This means that there are approximately three people in the workforce for every retiree in 2024. After the retirement of the baby boomers, this number will drop to about 2.2 by 2044 and as low as 2.0 by 2064. Furthermore, life expectancy is increasing continuously, necessitating the disbursement of pensions to retirees for a more extended duration (Statistisches Bundesamt, 2023). As a result, in Germany, those born after 1987 are among the losers of the welfare state, as they will contribute more to financing it over their remaining lifetimes in today's values than they will receive in social benefits, including old-age pensions (Stiftung Marktwirtschaft, 2024). The state already contributes more than 100 billion euros per year from tax revenues to the German pension system (Bundesministerium der Finanzen, 2023), while social security contributions still average just under 40% of gross income (Deutsche Rentenversicherung, 2024). Consequently, political reforms are necessary to mitigate the need for ever more tax revenue and higher contributions to finance the social security system and maintain pension levels.

In Germany, the so-called generational capital, a supplementary funded pension component to support the statutory pension insurance, is earmarked. It is essentially a debt-financed equity pension whose returns are expected to help stabilize the pension system from the mid-2030s (Bundesministerium der Finanzen, 2024). Funded pension components have also been implemented in several other countries in response to demographic shifts. One illustrative case is that of Sweden, where a partially funded system was introduced in 1999 (Haupt & Kluth, 2013, p. 3) that initially served as a role model for pension reform plans in Germany (Vogel & Dürr, 2021, pp. 3–5). In Sweden, as in many other countries with a funded pension component, individuals have the option to choose between different pension funds and therefore need to become familiar with the capital markets in order to make an informed decision. If citizens are introduced to the stock market and its benefits as part of this process, this education may promote both financial literacy and stock market participation (SMP) outside the pension system. A comparison of current SMP rates of the Swedish and German populations shows that the former is almost twice as high (SHARE-ERIC, 2024d; Börsch-Supan et al., 2013). In the United States and Denmark, where funded pension components have been in place for even longer than in Sweden, SMP is also much higher than in Germany, Austria, or Italy, where there is no mandatory funding in the pension system (SHARE-ERIC, 2024d; Aladangady et al., 2023; Börsch-Supan et al., 2013). These differences suggest that funded pension components may be an important factor in strengthening financial education and SMP. Funded pension components where there is no option to choose specific pension funds can also promote financial literacy and participation in the stock market, provided that the benefits of equities are explained by policymakers, for example, in the course of introducing a funded pension component, as is currently the case in Germany (Bundesministerium der Finanzen, 2024). If the funding of pension systems has such a positive effect and leads to long-term participation in the stock market, as would be expected from an understanding of basic financial concepts such as compounding through increased financial literacy, funded pension components could even support private old-age provision and provide additional relief to the pension system. Indeed, there are several studies on the importance of financial literacy for SMP and voluntary private retirement savings, as discussed in the following chapter. However, whether funded pension components play a critical role in this relationship is a gap in the existing literature. Consequently, the purpose of this thesis is to examine the impact of the level of funding of a pension system on both financial literacy and SMP.

1.2 Previous literature and research contribution

There is ample evidence of the significance of financial literacy, not only for participation in the stock market but also for private old-age provision. For example, Van Rooij et al. (2011) show that individuals with low levels of financial literacy are significantly less likely to participate in the stock market. Thomas and Spataro (2018) also find that there is a positive and significant correlation between financial literacy and SMP. Furthermore, a lack of financial literacy contributes to a general aversion to financial matters, which in turn may encourage the postponement of pension actions (Leinert, 2017, pp. 89–90). Leinert (2017) thus posits that it is highly probable that financial literacy exerts a positive influence on private old-age provision (p. 98). In fact, the willingness of citizens to save for retirement in private voluntary pension schemes is found to be positively correlated with financial literacy (Cupák et al., 2019). Moreover, Prast and Van Soest (2016) demonstrate that financial literacy has a positive impact on pension-related decisions (p. 118). In summary, the prevailing view in the academic literature is that enhancing individuals' financial literacy may encourage them to participate in the stock market and engage in voluntary retirement savings, thereby improving retirement preparedness and providing relief to the pension system. However, none of these studies examine the reverse relationship, i.e., that the pension system itself, i.e., its structure and funding, may have a positive effect on financial literacy. The lack of research in this field provides an avenue for initial investigation. Consequently, one objective of this thesis is to ascertain whether the level of funding of a pension system has a positive effect on financial literacy.

With respect to fluctuations in SMP, Meister and Schulze (2022) show that the main causes appear to be changes in the transaction and information costs of participation as well as changes in risk exposure (p. 11). The funding of pension systems can play a key role in this respect. First, funded pension components can help stabilize capital markets because pension investments tend to be long-term and less susceptible to short-term fluctuations (Vittas, 1996). This stability can increase confidence in the stock market and reduce information costs, as investors spend less time and resources assessing uncertainty and risk. Additionally, a stronger presence of institutional investors, such as pension funds, often increases the range of financial products and services tailored to individuals' needs (FIAP, 2016, pp. 1, 3). This can lead to better market infrastructure and easier market access, which in turn can reduce the cost of participation. However, the extent to which the level of funding of a pension system actually affects SMP has not yet been fully explored in the existing literature. Using Sweden as an example, Massa et al. (2006) analyze whether the introduction of a funded pension component increases the population's participation in the stock market outside the pension system. They find that the introduction of a funded pension component is associated with a greater likelihood of SMP, especially for those individuals who actively select a particular pension fund. Furthermore, choosing a specific product within the funded pension component serves to diminish the likelihood of exiting the stock market. Consequently, the provision of funded pension components may facilitate the dissemination of information regarding the advantages of SMP, thereby increasing the number of long-term investors and, in turn, the level of voluntary savings (Massa et al., 2006, pp. 2-3, 8, 12, 15-16). Nevertheless, the analysis is country-specific, and therefore, no general statement can be made about the positive impact of the level of funding of a pension system on SMP. Consequently, the second objective of this thesis is to examine whether such an effect exists across several countries, leading to two working hypotheses:

- Hypothesis 1 (H1): The level of funding of a pension system has a positive effect on the financial literacy of a country's population.
- Hypothesis 2 (H2): The level of funding of a pension system has a positive effect on the SMP of a country's population.

Should one or both of these hypotheses prove to be accurate, an additional investigation will be conducted to ascertain whether, as previously postulated, the choice of options (CoO) between different fund managers or pension funds, as is the case in Sweden, for instance, enhances the impact of the level of funding on the corresponding dependent variable. This assumption is based on the premise that individuals are more likely to engage in financial decision-making in countries where a CoO is available and where familiarity with the stock market is required to make informed decisions. In order to test the primary hypotheses and ascertain a pronounced impact of the level of funding through a CoO, the following methodology is employed.

1.3 Methodology and structure

The 38 member countries of the Organization for Economic Cooperation and Development (OECD) are selected as the basis for the dataset due to the high availability and comparability of data within this group. This selection ensures a high level of data quality and consistency, particularly concerning pension system funding and financial literacy, for which data are provided directly by the OECD. Each country is represented by a single observation in the dataset. The following chapter employs a scoring model to categorize the 38 countries in question according to their respective pension system funding, financial literacy, and SMP. This model provides the foundation for subsequent analysis. Moreover, the second part also presents the remaining dataset and all variables considered for the empirical analysis. The third part delves deeply into the relationships between pension system funding and financial literacy, as well as between pension system funding and SMP. First, a difference-in-differences analysis is conducted prior to testing the developed hypotheses. The latter is accomplished through bivariate and multiple-variable regression models. To ensure the robustness of the results, various variables will be controlled for in the multiple-variable regression models.

By following this methodology and structure, this thesis aims to provide a thorough and methodologically sound analysis of the correlations between pension system funding and financial literacy, and between pension system funding and SMP across OECD countries.

2 Data and variable construction

2.1 A scoring model for selected countries

2.1.1 Degree of funding and DoF score

The degree of funding of a pension system (DoF) represents the primary parameter to be analyzed in terms of its influence. The DoF is defined as the proportion of mandatory and quasi-mandatory pension contributions allocated to funded pension components. Quasi-mandatory contributions exist, for example, where there is automatic enrolment into a pension scheme with the possibility of opting out, or where pension schemes are negotiated as part of collective agreements so that the average full-time employee is expected to be covered. The focus is on (quasi-)mandatory pension components, as this guarantees a high level of comparability between individual countries. In view of the large selection of voluntary financial products for private pension provision, which are not subject to automatic enrolment and usually lack comprehensive coverage, the DoF is constrained to the statutory and workplace pillars of old-age provision. The statutory pension pillar is defined as all pension schemes that are subject to the governance and administration of the state. In contrast, the workplace pillar encompasses all pension schemes that are governed or administered through employers. The DoF is calculated separately for each pension pillar to examine whether a possible effect on financial literacy and SMP might be driven more by one of the two pillars. Additionally, a combined DoF is computed across the statutory and workplace pension pillars. The calculation is based on the proportion of total (quasi-)mandatory pension contributions allocated to funded pension components. The proportion of (quasi-)mandatory employee pension contributions allocated to funded pension components is also computed. However, as the aim is to understand how the overall level of funding of the pension system affects financial literacy and SMP, the DoF based on total contributions, including those of the employer, is used in the empirical analysis. A variety of country-specific data sources, including official government and agency publications, academic studies, international organization reports, and OECD country profiles, are used to calculate the DoF in each country. Data used refer to the average full-time employee in each country in 2024, or the most recent year available.

In the majority of cases where a (quasi-)mandatory funded pension component exists within the statutory pension pillar, a pay-as-you-go component is also present. These partially funded systems are referred to as hybrid pension systems in this thesis. The DoF of hybrid pension systems thus reaches a value between 0 and 1. A value of 0 is assigned when there are no (quasi-)mandatory contributions to a funded pension component. Conversely, a value of 1 is assigned when all (quasi-)mandatory contributions are directed into a funded pension component, indicating that the statutory pension pillar is fully funded. The variable representing the level of funding in the statutory pension pillar is designated as “statutory DoF.” In four OECD countries, the statutory pension pillar is fully funded, whereas in eight member states, the statutory pillar is characterized by a hybrid pension system. For the remaining OECD countries, the statutory DoF is zero. In some of these countries with a zero statutory DoF, however, there is still a funded pension component that operates in the background of the statutory pillar, serving as a kind of stabilization mechanism analogous to the earmarked generational capital in Germany. Notwithstanding the absence of direct contributions to these funded pension components, they may nevertheless be regarded as (quasi-)mandatory, given that the entire population is automatically participating in them. Examples of this include the Norwegian government pension funds and several reserve funds, such as those that exist in Canada, France, and Japan. Although the statutory DoF takes the value 0 here, as there are no (quasi-)mandatory contributions, financial literacy and SMP could nevertheless be influenced by these pension components. For this reason, another binary variable is generated, referred to as “statutory mandatory funding,” which takes the value 1 if a funded pension component can be considered (quasi-)mandatory and 0 otherwise, regardless of whether there are direct contributions to this component or not. There are a total of seven OECD countries where there is a (quasi-)mandatory funded pension component without direct contributions, as illustrated in Figure 1 about statutory DoF.

Figure 1 Statutory DoF for all 19 relevant OECD countries

Illustrations are not included in the reading sample

Illustrations are not included in the reading sample

Note. Presentation of all OECD countries with a (quasi-)mandatory funded pension component in the statutory pension pillar and their respective statutory DoF, including those countries where such a component exists but no direct contributions are made to it, i.e., Canada, France, Japan, Korea, Luxembourg, Norway, and the United States.

In contrast, the workplace pension pillar is characterized by a single funded pension component that is either (quasi-)mandatory, in which case contributions must be made, or is not (quasi-)mandatory. Consequently, if there are (quasi-)mandatory contributions, the DoF takes the value 1. For the analysis of this pension pillar, it is therefore sufficient to consider whether contributions must be made to the funded pension component or not. This can be achieved by generating a binary variable that takes the value 1 in cases where (quasi-)mandatory contributions exist and 0 otherwise. This variable is designated as “workplace DoF,” and the 13 OECD countries with a value of 1 are shown in Figure 15 (cf. Appendix I). An additional variable similar to the one called statutory mandatory funding is thus not required for the workplace pension pillar.

The variable indicating the level of funding across both pension pillars is referred to as “combined DoF,” calculated as the proportion of the total (quasi-) mandatory contribution rate across both pension pillars that flows into funded pension components. Figure 2 presents the combined DoF for all 38 OECD countries. Similar to the statutory mandatory funding variable, another binary variable is created, designated as “combined mandatory funding,” which takes the value 1 if any funded pension component across both pension pillars can be considered (quasi-)mandatory and 0 otherwise, regardless of whether there are direct contributions to this component or not. Thus, the combined mandatory funding variable is essentially a combination of the statutory mandatory funding and workplace DoF variables. In the event that one or both of the two variables assumes the value of 1, the combined mandatory funding variable is also assigned the value of 1.

Figure 2 Combined DoF for all 38 OECD countries

Illustrations are not included in the reading sample

Illustrations are not included in the reading sample

Note. Presentation of all OECD countries and their respective combined DoF across the statutory and workplace pension pillars.

Figure 3 DoF score for all 38 OECD countries

Illustrations are not included in the reading sample

Illustrations are not included in the reading sample

Note. Presentation of all OECD countries and their respective DoF scores.

In addition, a parent DoF variable is specifically designed to combine the level of funding and the existence of (quasi-)mandatory funded pension components across both pension pillars. Therefore, the so-called DoF score is calculated as the sum of the combined DoF and the combined mandatory funding variables. Ultimately, the DoF score combines all DoF variables in one key metric, as illustrated in Figure 3. OECD countries with a (quasi-)mandatory funded pension component without direct contributions in the statutory pillar, i.e., where the statutory DoF is 0, now have a DoF score of 1. Those countries with a combined DoF of 1 reach the highest possible DoF score of 2. A total of ten OECD countries have been identified as having a DoF score of 0, indicating the absence of any form of (quasi-)mandatory funding within their pension systems. This group includes Germany, where the DoF score is set to transition to a value of 1 upon the full implementation of the generational capital.

2.1.2 Financial literacy and FinLit score

Atkinson and Messy (2012) define financial literacy as “a combination of awareness, knowledge, skill, attitude, and behavior necessary to make sound financial decisions and ultimately achieve individual financial wellbeing” (p. 14). Based on this definition, the OECD International Network on Financial Education (INFE) has developed a survey instrument to assess the financial literacy of adults from a diverse range of backgrounds in a multitude of countries. The questionnaire draws upon existing financial literacy surveys and includes core questions pertaining to financial knowledge, behavior, and attitudes, from which a final score is derived (Atkinson & Messy, 2012, p. 6). All OECD INFE reports on adult financial literacy published to date provide data for almost 70 countries, including 31 of the 38 OECD member states. There are various alternative sources available to obtain financial literacy data. One example is the Flash Eurobarometer FL525 Survey from 2023, which provides financial literacy scores for 27 European countries (European Commission, Directorate-General for Communication, 2023a). Other pertinent examples include the 2017 report on money, financial literacy, and risk in the digital age published by Allianz, which provides data for ten European states (Coppola et al., 2017, p. 16), and the 2015 S&P Global Financial Literacy Survey, which provides data for 143 countries (Klapper et al., 2015, pp. 23–25). However, the OECD INFE surveys employ a more comprehensive questionnaire, particularly in the most recent iteration, which suggests the generation of more meaningful data. For those OECD member states not covered by OECD INFE surveys, there are often country-specific studies available. One illustrative example is the case of the United States, for which regular nationwide surveys are conducted with the objective of recording financial literacy. These include the TIAA Institute-GFLEC Personal Finance Index and the FINRA Foundation National Financial Capability Study. However, the methodologies employed in these surveys diverge from those utilized by the OECD INFE. For example, the underlying questionnaires differ, and the final scores are computed using a distinct logic. As a result, there is no reliable comparability between OECD INFE surveys on adult financial literacy and such country-specific studies. Consequently, only selected OECD INFE reports are employed as data sources for financial literacy scores.

The initial report of the OECD INFE pilot study on adult financial literacy was published in 2012. One financial knowledge question from this first study was made optional in later questionnaires and excluded from the score computation from 2015 onwards (OECD, 2016, pp. 20, 86). The data necessary to realign the 2012 scores with those from later publications could not be obtained, and more recent data from later publications are available. Consequently, the results of the pilot study are not utilized. The questionnaires or toolkits, as they are referred to by the OECD INFE, also exhibit slight variations between subsequent years. The 2015 toolkit, which includes one less question on financial knowledge than the one originally used in the pilot study, was employed in all studies underlying the 2016, 2017, and 2018 publications, although the latter does not provide data for OECD countries. In 2018, a new questionnaire was released, which largely corresponds to the previous version, but some changes, additions, and deletions have been made (OECD, 2018, pp. 5, 34–40). This toolkit was employed in the research conducted for all publications from 2020 and 2021 prior to the release of a new version in 2022, which was utilized for the study behind the most recent publication from 2023.¹ The 2022 questionnaire incorporates additional inquiries about digital financial products and services, as well as about sustainable finance products and attitudes toward sustainable finance (OECD, 2022, p. 7). Furthermore, some pre-existing questions on offsetting have been designated as optional. Therefore, to assess whether data from the most recent and earlier publications can be used side by side, the ranking is reviewed for those OECD countries for which data are available for 2016/17, 2020/21, and 2023, i.e., for which three data points can be obtained, which is the case for ten countries, including Germany. A comparison of the data shows that each country's ranking changes 55% of the time, particularly in the 2023 report, where the ranking is different 60% of the time compared to the 2016/17 data. For example, Croatia is ranked 8th in 2016/17, 9th in 2020/21, and suddenly 6th in 2023, while Indonesia is ranked 4th in 2016/17, 3rd in 2020/21, and suddenly 9th in 2023. Similarly, Germany's rescaled and normalized financial literacy score in the 2023 report is 0.77, as opposed to 0.66 in 2020/21 and 2016/17. This illustrates that for some countries, the scores and country rankings in the 2023 report differ significantly from those observed in previous publications. Although the basic structure of the toolkits used remains consistent over the years, the difference in scores and rankings in the examples above demonstrates the loss of full comparability between surveys, particularly between the 2023 report and previous ones. As the most recent study uses the most comprehensive questionnaire and provides data for 20 OECD countries, the highest number of OECD member states covered in any other OECD INFE publication to date, the 2023 report is used as the main data source for financial literacy.

Figure 4 FinLit score and its composition for all 20 relevant OECD countries

Illustrations are not included in the reading sample

Illustrations are not included in the reading sample

Note. Data are taken from the latest 2023 report and scaled according to previous OECD INFE publications (OECD, 2023c, Annex D, Table 2.1).²

Figure 4 shows the corresponding data from the 2023 report. Germany ranks 1st in the 2023 report in both financial knowledge and financial behavior, and thus 1st in the overall financial literacy score (FinLit score). With respect to Germany’s and Ireland’s placement, there are two countries at the top where there is no funding in the pension system, i.e., where the DoF score is zero. However, Ireland is immediately followed by six OECD countries with a positive DoF score. To illustrate the country rankings for each pillar of financial literacy, Figures 16 through 19 show the individual scores for financial knowledge, financial behavior, and financial attitudes, as well as the overall financial literacy score, all adjusted to a standardized scale (cf. Appendix I). In order to assess the impact of the DoF not only on financial literacy but also on each of its pillars, the normalized individual scores are also used as dependent variables in the final regression analyses.

Also used in the final regression analyses are the results of the so-called “Big Three” financial literacy questions (Global Financial Literacy Excellence Center, n.d.) developed by Lusardi and Mitchell (2011). These are questions on the basic financial concepts of compound interest, inflation, and risk diversification that in identical or slightly modified form, are part of the standard questionnaire of many well-known and widely used financial literacy surveys, including those from the OECD INFE. Since understanding these three basic financial concepts is thought to be a driver of SMP, especially long-term participation, a positive effect of the DoF on the outcomes of the Big Three may also indicate a positive effect on long-term SMP and thus on private old-age provision. In the OECD INFE toolkits, these three basic financial concepts are assessed in a slightly modified form in the financial knowledge section, namely in the questions on the time value of money, the calculation of simple and compound interest,³ and the diversification of risk. As these questions do not change over the years and are not made optional, they allow full comparability between different OECD INFE publications. Consequently, comparable data on the Big Three can be obtained for 31 OECD countries. Once more, the 2023 report provides data for 20 OECD member states, while the 2020/21 and 2016/17 OECD INFE publications provide data for a further eleven OECD countries. However, as with the overall financial literacy scores above, the ranking is first reviewed for those OECD countries for which data is available for 2016/17, 2020/21, and 2023, i.e., for which three data points can be obtained, which is the case for seven countries, including Germany. A comparison of the data reveals that the ranking of each country remains unchanged in approximately 57% of cases. In the remaining cases, however, it does undergo a slight alteration. To ensure the integrity of the analysis given the time lag between data collection, the additional eleven observations on the time value of money, simple and compound interest, and risk diversification knowledge from the 2020/21 and 2016/17 OECD INFE publications are incorporated into a separate regression model. Figure 5 shows the corresponding data for all 31 OECD countries.

Figure 5 Big Three score for all 31 relevant OECD countries

Illustrations are not included in the reading sample

Illustrations are not included in the reading sample

Note. The Big Three scores are calculated as the sum of the individual results of the relevant questions, each indicating the percentage of respondents who answered correctly, resulting in a total score ranging from 0 to 300. Data for 20 OECD countries are from the latest publication (OECD, 2023c, Annex D, Table 2.7). Data for the remaining eleven countries (indicated in gray) are from previous publications (OECD, 2021, p. 14; OECD, 2020, p. 20; OECD, 2017, p. 19; OECD, 2016, pp. 23–24).

The scores of the eleven OECD countries for which data are drawn from the 2020/21 or 2016/17 publications are presented in gray to indicate that the reliability of the country's position in the overall ranking may be compromised by the time lag in data collection, as discussed in the country ranking analysis above. As with the overall financial literacy score, Germany ranks 1st in the Big Three, although it has no funding in the pension system. To illustrate the country rankings for each of the Big Three, Figures 20 through 23 show the individual results for the time value of money, simple and compound interest, and risk diversification questions, as well as the overall Big Three score, all adjusted to a standardized scale (cf. Appendix I). Again, the results for the eleven OECD countries with data from the 2020/21 and 2016/17 publications are presented in gray.

2.1.3 Stock market participation and SMP score

In line with previous literature, this thesis considers indirect, direct, and total participation in the stock market, while the focus is on the latter two (Guiso et al., 2008; Arts, 2018; Kaustia et al., 2023). Direct participation occurs when an individual directly holds shares. Indirect participation, on the other hand, exists when shares are held through other financial products such as mutual funds, exchange-traded funds, or individual retirement accounts in the voluntary private pension pillar. To ensure a clear distinction from the DoF variables, indirect SMP through (quasi-)mandatory pension schemes in the statutory and workplace pension pillars is explicitly excluded from the calculation. The computed total SMP rates per country indicate what proportion of the population in each country participates either directly or indirectly in the stock market, or both. The most recent Wave 9 of the Survey of Health, Ageing and Retirement in Europe (SHARE), published in 2024, is used as the main SMP data source for all 23 participating OECD countries.⁴ The asset section of the SHARE questionnaire explicitly asks about individual retirement accounts, mutual funds, and shares held by households. Consequently, this section was utilized as the foundation for the calculation of SMP rates in respective countries. The latest wave of the Household Finance and Consumption Survey (HFCS) from 2021 was also considered as a potential source of SMP data. However, the HFCS merely provides data for 19 OECD member states. The only OECD country that is included in the HFCS but not in the SHARE is Ireland. As the questionnaires differ between the SHARE and the HFCS, only the direct SMP rate for Ireland can be obtained, which allows comparability with the SHARE data. Consequently, this is the only data point used from the HFCS. To check for potential data conformity issues, the direct SMP rates for the remaining 18 OECD countries in the HFCS are compared with the corresponding rates obtained for these countries from the SHARE data to check for differences in the country rankings between the two surveys. Apart from a few outliers such as Estonia, Italy, the Netherlands, and Spain, the rankings are very similar. For example, Greece, Hungary, Lithuania, Latvia, and Slovakia are the five countries with the lowest rates of direct SMP in both surveys. Additionally, Finland ranks 1st and Germany ranks 3rd in both surveys. As Ireland would take 7th place in the country ranking in the HFCS and 6th place in the SHARE, it is included as an additional data point alongside the SHARE data (SHARE-ERIC, 2024d; European Central Bank, 2023; Börsch-Supan et al., 2013). For the remaining countries, several country-specific data sources must be consulted to obtain the required information:

- Australia: The 2014 Australian Share Ownership Study provides a total SMP rate that is comparable to the SHARE data (Australian Securities Exchange, 2015, p. 11). However, due to the age of the data point, it is only included in a separate regression specification.
- Canada: The 2019 Survey of Financial Security provides a direct SMP rate that is comparable to the SHARE data (Statistics Canada, 2020). However, as the data point was collected prior to the global pandemic, it is included in a distinct regression specification solely to reflect the most current trends in the initial specification.
- Japan: The latest publication on individual stockholder trends in Japan by the Japan Securities Dealers Association provides a direct SMP rate for 2022 that is comparable to the SHARE data (JASDEC et al., 2023).
- Norway: The latest working paper on the dynamics of SMP in Norway published by Norges Bank provides SMP data from 1993 to 2016 that is comparable to the SHARE data (Galaasen & Raja, 2024, Table 1, fig. 1). However, due to the age of the most recent data point, the corresponding direct, indirect, and total SMP rates are only included in a separate regression specification.
- United Kingdom: The FCA’s May 2022 Financial Lives Survey published in 2023 provides a direct SMP rate that is comparable to the SHARE data (Financial Conduct Authority, 2023, p. 23).
- United States: The 2022 Survey of Consumer Finances provides SMP data that is comparable to the SHARE data (Aladangady et al., 2023, pp. 15–16, 19).

Figure 6 Direct SMP rate for 29 OECD countries

Illustrations are not included in the reading sample

Illustrations are not included in the reading sample

Note. The percentage of population in each country directly participating in the stock market is indicated (SHARE-ERIC, 2024d; Galaasen & Raja, 2024, fig. 1; European Central Bank, 2023; Financial Conduct Authority, 2023, p. 23; Aladangady et al., 2023, pp. 15–16, 19; JASDEC et al., 2023; Statistics Canada, 2020; Australian Securities Exchange, 2015, p. 11; Börsch-Supan et al., 2013).

Figure 7 Indirect SMP rate for 24 OECD countries

Illustrations are not included in the reading sample

Illustrations are not included in the reading sample

Note. The percentage of population in each country indirectly participating in the stock market is indicated (SHARE-ERIC, 2024d; Galaasen & Raja, 2024, fig. 1; European Central Bank, 2023; Financial Conduct Authority, 2023, p. 23; Aladangady et al., 2023, pp. 15–16, 19; JASDEC et al., 2023; Statistics Canada, 2020; Australian Securities Exchange, 2015, p. 11; Börsch-Supan et al., 2013).

Figure 8 Total SMP rate (SMP score) for 26 OECD countries

Illustrations are not included in the reading sample

Illustrations are not included in the reading sample

Note. The percentage of population in each country that participates in the stock market either directly or indirectly or both is indicated (SHARE-ERIC, 2024d; Galaasen & Raja, 2024, fig. 1; European Central Bank, 2023; Financial Conduct Authority, 2023, p. 23; Aladangady et al., 2023, pp. 15–16, 19; JASDEC et al., 2023; Statistics Canada, 2020; Australian Securities Exchange, 2015, p. 11; Börsch-Supan et al., 2013).

In total, suitable SMP data can be identified for 30 of the 38 OECD member states, including direct SMP rates for 29 OECD countries, indirect SMP rates for 24 OECD countries, and total SMP rates for 26 OECD countries. The individual rates are shown in Figures 6 through 8, which also visualize the country rankings. The SMP rates for Australia, Canada, and Norway are presented in gray to indicate that the reliability of the country's position in the overall ranking may be compromised by the time lag in data collection. Since the total SMP rate includes both direct and indirect SMP, it represents the parent SMP variable for the empirical analysis. In the context of the scoring model in this thesis, it is also referred to as the SMP score.

2.1.4 DoF score versus SMP score

In this section, a scatterplot analysis is used to examine whether fluctuations in the DoF correspond to changes in SMP rates, thereby providing insight into the potential impact of pension system funding on SMP. To visualize the relationship, the data for the parent DoF variable (DoF score) is plotted against the data for the parent SMP variable (SMP score) in Figure 9. The scatter plot shows a wide range of SMP scores for different DoF scores. There is a noticeable cluster of countries without pension system funding, with SMP scores predominantly between 0.05 and 0.15. The dashed trend line indicates a positive correlation between the DoF score and the SMP score, i.e., as the DoF score increases, there is a general upward trend in the SMP score, suggesting that higher levels of pension system funding are associated with increased SMP.

Figure 9 Scatter plot of DoF score versus SMP score

Illustrations are not included in the reading sample

Illustrations are not included in the reading sample

Note. Each dot represents an OECD country, with individual DoF scores plotted against corresponding SMP scores. A dashed trend line is included to indicate the general direction of the relationship between the two variables.

At a DoF score between 1.2 and 1.4, there is a significant spread in SMP scores, ranging from close to 0 to over 0.6. This suggests that while some countries with moderate funding in the pension system are active in the stock market, others are not. There are fewer observations for DoF scores above 1.5, but they still show a positive trend, further supporting the correlation. The presence of outliers, particularly at higher SMP scores for mid-range DoF scores, indicates variability in SMP despite similar levels of pension system funding. This variability could be influenced by other determinants of SMP not captured in this graph. Consequently, while this initial visualization provides a basic understanding of the relationship between pension system funding and SMP, further analysis incorporating additional variables is needed to provide deeper insights into the underlying dynamics.

2.1.5 DoF score versus FinLit score

As above, a scatterplot is used to examine whether fluctuations in the DoF correlate with changes in financial literacy levels to gain insight into how pension system funding may affect financial literacy and its components. To visualize the relationship between the DoF and financial literacy, the data for the parent DoF variable (DoF score) is plotted against the data for the parent financial literacy variable (FinLit score) in Figure 10.

Figure 10 Scatter plot of DoF score versus FinLit score

Illustrations are not included in the reading sample

Note. Each dot represents an OECD country, with individual DoF scores plotted against corresponding FinLit scores. A dashed trend line is included to indicate the general direction of the relationship between the two variables.

There is a concentration of countries where the DoF score is zero while the full range of FinLit scores is covered. Also, the dashed trend line suggests a slight negative correlation between the DoF score and the FinLit score, i.e., as the DoF score increases, there is a slight decrease in the FinLit score, suggesting that higher levels of pension funding may be associated with a slight decrease in financial literacy. Note, however, that the regression line is almost flat. For a DoF score between 1.2 and 1.4, there is a notable spread in FinLit scores, ranging from about 0.55 to 0.7. This suggests variability in financial literacy levels among countries with similar levels of pension system funding. There are few observations for DoF scores above 1.5, but they continue to show a slight downward trend in FinLit scores. In addition, there are outliers, particularly at DoF scores of 0, with relatively high FinLit scores (between 0.7 and 0.8), suggesting that some countries maintain high levels of financial literacy despite lower degrees of pension system funding. This, together with the variability in FinLit scores for similar DoF scores, suggests that while pension funding may influence financial literacy, it is not the sole determinant. As a result, while this initial illustration provides a basic understanding of the relationship between the DoF and financial literacy, further analyses that include additional variables are needed to gain a deeper understanding of the underlying dynamics.

2.2 Primary variables

2.2.1 Independent DoF variables

This section provides an overview of the independent DoF variables used in the final regression analyses of the relationship between pension system funding and its impact on financial literacy and SMP. Each variable is summarized below, and the corresponding summary statistics are presented in Table 12 to provide a comprehensive understanding of the dataset used in the analysis (cf. Appendix II).

- Statutory DoF: This variable represents the degree of funding of the statutory pension pillar in each country. It measures the extent to which total pillar-specific (quasi-)mandatory contributions are allocated to a funded pension component.
- Statutory mandatory funding: This is a binary variable that indicates the existence of a (quasi-)mandatory funded pension component in the statutory pension pillar, regardless of whether direct contributions are made to it or not.
- Workplace DoF: This is a binary variable that indicates the existence of a (quasi-)mandatory funded pension component in the workplace pension pillar.
- Combined DoF: This variable represents the combined degree of funding of the statutory and workplace pension pillars in each country. It measures the extent to which total (quasi-)mandatory contributions to both pension pillars are allocated to funded pension components.
- Combined mandatory funding: This is a binary variable that indicates the existence of (quasi-)mandatory funded pension components across the statutory and workplace pension pillars, regardless of whether direct contributions are made to them or not.
- DoF score: The sum of the combined DoF and combined mandatory funding variables to aggregate the level of funding and existence of (quasi-)mandatory funded pension components across both pension pillars.

2.2.2 Dependent FinLit variables

This section provides an overview of the dependent FinLit variables used in the final regression analyses of the relationship between pension system funding and its impact on financial literacy. Each variable is summarized below, and the corresponding summary statistics are presented in Table 12 (cf. Appendix II).

- Financial knowledge score: Average adult financial knowledge score in each country, normalized as a percentage of the maximum possible score.
- Financial behavior score: Average adult financial behavior score in each country, normalized as a percentage of the maximum possible score.
- Financial attitudes score: Average adult financial attitudes score in each country, rescaled and normalized as a percentage of the maximum possible score.
- Financial literacy score: Average adult financial literacy score in each country, rescaled and normalized as a percentage of the maximum possible score.
- Time value of money knowledge: Percentage of adults in each country who answered the time value of money question correctly.
- Simple and compound interest knowledge: Percentage of adults in each country who answered the simple and compound interest questions correctly.
- Risk diversification knowledge: Percentage of adults in each country who answered the risk diversification question correctly.
- Big Three score: The sum of the individual results of the time value of money, simple and compound interest, and risk diversification questions.

2.2.3 Dependent SMP variables

This section provides an overview of the dependent SMP variables used in the final regression analyses of the relationship between pension system funding and its impact on SMP. Each variable is summarized below, and the corresponding summary statistics are presented in Table 12 (cf. Appendix II).

- Direct SMP: Percentage of the population in each country that directly owns stocks.
- Indirect SMP: Percentage of the population in each country that owns stocks indirectly through other financial products such as mutual funds, exchange-traded funds, or individual retirement accounts.
- Total SMP: Percentage of the population in each country that participates directly or indirectly in the stock market, or both.

2.3 Control variables

2.3.1 Individual characteristics

SMP rates and financial literacy levels vary significantly with individual characteristics, as has been extensively documented in the literature. For example, stock ownership increases with income, age, and wealth (Poterba & Samwick, 1995), potentially due to reduced liquidity constraints and a greater willingness to invest in risky assets. Because income, age, and wealth create different levels of exposure to financial products, they are also correlated with financial literacy (Bucher-Koenen, 2009). Moreover, SMP rates are higher for men than for women. This gender participation gap is attributed to men's overconfidence in finance (Barber & Odean, 2001) and women's higher risk aversion (Jianakoplos & Bernasek, 1998). Additionally, women’s lower financial literacy levels contribute to the gender disparity in SMP (Almenberg & Dreber, 2012; Bucher-Koenen et al., 2021), though the gender literacy gap is narrowing with increased educational efforts to build confidence (Blaschke, 2022). Better education correlates with both higher financial literacy and increased SMP (Bertaut, 1998; Bucher-Koenen, 2009; Cole & Shastry, 2009). This is because higher education levels, especially financial education, typically enhance financial literacy, thereby reducing the information costs associated with SMP (Bucher-Koenen, 2009; Van Rooij et al., 2011). Furthermore, Vu et al. (2021) highlight that self-perceived health status can influence the ownership of risky assets. Employment status also influences SMP; households tend to shift from riskier to more stable assets in advance of unemployment and sell stock holdings after job loss due to declining labor income and depleted savings (Basten et al., 2016). Moreover, individuals with left-leaning political orientations are less likely to participate in the stock market, demonstrating that personal values and ideological factors, such as a more negative view of capitalism, drive stock market aversion (Kaustia & Torstila, 2011). All these individual characteristics are considered potential control variables for the final regression analyses, as detailed below, and the corresponding summary statistics are presented in Table 12 (cf. Appendix II).

- Sex: The sex ratio is used, i.e., the ratio of men to women in each country. Data are from the OECD database for the most recent year available, 2022 (OECD, 2023b).
- Age: The old-age dependency ratio, which is the ratio of elderly dependents (aged 65 and over) to the working-age population (aged 20-64), provides an accurate measure of the economic burden and demographic structure in each country. Data are from the OECD database for the latest available year, 2022 (OECD, 2024a).
- Employment: The employment rate is used, which is the ratio of the employed to the working-age population (aged 15-64) in each country. Data are from the OECD database for the most recent year available, 2023 (OECD, 2024g).
- Income: Gross domestic product (GDP) per capita is used because this measure standardizes economic output by population and reflects average economic well-being and living standards, thereby facilitating meaningful cross-country comparisons. Data are from the OECD database for the latest available year, 2023, in USD and converted to current purchasing power parities (OECD, 2024f).
- Wealth: The median net wealth per household in each country is used. Data are from the OECD database for 2019 or the latest available year, converted from local currency to US dollars on May 6th, 2024 (OECD, 2023d).
- Education: Educational attainment in each country is used, expressed as the share of the working-age population (aged 25-64) with a tertiary education. Data are from the OECD database for the most recent year available, 2023 (OECD, 2024e).
- Political orientation: The average of self-reported political attitudes of individuals in each country is used. Data are from the Standard Eurobarometer STD99 spring 2023 (European Commission, Directorate-General for Communication, 2023b).
- Health: The percentage of the population (aged 16 or over) who perceive their health as positive is used. Data are from the OECD database for 2021 or the latest available year (OECD, 2024b).

Furthermore, FinLit variables are considered as control variables in the regression analyses on SMP, and vice versa, as an additional robustness check. To further reduce the risk of omitted variable bias, other country-specific characteristics are also considered potential control variables, especially since SMP and financial literacy have other well-established determinants, as detailed in the following sections.

2.3.2 Risk aversion

Several academic papers discuss the negative association between risk aversion and SMP. For instance, Dohmen et al. (2011) show that a lower willingness to take risks significantly reduces the likelihood of SMP (pp. 539-540). Additionally, Guiso et al. (2018) show that investors' risk aversion increases significantly after the 2008 financial crisis, leading them to sell more stocks. Furthermore, Sias et al. (2020) illustrate that genetic predispositions related to risk aversion significantly impact individuals' likelihood of participating in the stock market. As a result, less risk-averse individuals are likely to be more engaged in financial markets, thereby gaining financial knowledge, which may also lead to increased financial literacy. Guiso et al. (2008) suggest that since risk aversion is a preference parameter that reflects innate characteristics, the distribution should be similar across different populations (p. 2590). However, using the Global Preference Survey (GPS), Falk et al. (2018) support the existence of systematic differences in risk aversion across countries, which can be attributed to both cultural and economic factors. Accordingly, different proxies for risk aversion are considered potential control variables in the final regression analyses.

The OECD database does not provide direct measures of risk aversion. However, it does offer data on related economic and social factors that researchers often use as indirect proxies for risk attitudes or factors that influence attitudes toward risk. These include education, employment, health, and trust, all of which are considered separate control variables in this thesis. The above-mentioned GPS, on the other hand, provides data on the willingness to take risks in 28 OECD countries, which is considered the first direct proxy for risk aversion. The corresponding data were collected as part of the 2012 Gallup World Poll (IZA, 2018; Falk et al., 2016). In addition, to use a more recent data source, the 2021 World Worry Index may serve as another useful proxy for attitudes towards risks, as there is a probable correlation between concern about risks and risk-taking behavior. In general, individuals who are more concerned about potential risks may be less inclined to engage in activities associated with those risks. It is, however, important to note that worry alone may not fully capture individuals' attitudes toward risk. Risk preferences are complex and influenced by a multitude of factors such as cultural norms, personal experiences, perceptions of control, and socioeconomic factors (Falk et al., 2018). Although worry is a component of risk perception, it does not necessarily indicate an individual's willingness to take risks in different contexts. For instance, an individual may express concern about economic instability yet still engage in risky financial investments due to optimism or perceived opportunity. Similarly, individuals might engage in activities with health risks due to perceived benefits or social pressures. Consequently, while the World Worry Index offers valuable insights into the prevalence of concern about specific risks, it may not fully capture the complexities of individuals' attitudes toward risk-taking behavior. Nevertheless, given that the World Worry Index provides data for all 38 OECD member states (Lloyd’s Register Foundation, 2021), the index is considered a second indirect proxy for risk aversion. Furthermore, a question from the 2019 World Risk Poll is considered another direct proxy for risk aversion, as it captures perceptions of risk. The survey question posed to respondents is whether they tend to think of opportunity or danger when they hear the term “risk.” As the World Worry Index is already a measure of concern, this third proxy focuses on a positive attitude toward risk by indicating the proportion of the population that responds “opportunity.” The 2019 World Risk Poll provides data for 36 of the 38 OECD member states (Lloyd’s Register Foundation, 2019). Summary statistics for the three risk aversion proxies are presented in Table 12 (cf. Appendix II).

To test whether these proxies can be used together in a regression analysis without causing multicollinearity problems, their correlation coefficients and the variance inflation factors (VIF) after joint regression on the parent FinLit and SMP variables are analyzed. The low VIFs in Table 13 indicate that the proxies are not highly correlated with each other (cf. Appendix II), which is confirmed by the highest absolute correlation value between the three proxies of 0.286. All proxies could therefore be used side by side in a regression. However, due to the large number of potential control variables, certain proxies are prioritized. The prioritization is primarily based on data availability, as the dataset is already limited by the fact that there are only 38 observations in total, i.e., one observation per OECD country. The focus is therefore on those proxies for which data can be obtained for the largest number of OECD countries, so that as few observations as possible are lost in the final regression analyses. In addition, the explanatory power of each proxy is tested and compared, although this plays a less important role in the prioritization decision than data availability. To test the explanatory power of the proxies, each is first regressed individually on the parent FinLit and SMP variables to compare R-squared values, and then jointly to perform t-tests. In the individual regressions on the FinLit score, the “positive risk perception” variable has the highest R-squared value, while in the individual regressions on total SMP, the “worry” variable has the highest R-squared value, indicating that these two risk aversion proxies have the highest explanatory power, as shown in Table 14 (cf. Appendix II). The significant t-test results of the positive risk perception and worry variables in Table 15 further confirm that these two variables have the highest explanatory power (cf. Appendix II). As these are also the two risk aversion proxies with the largest number of observations, they are the optimal candidates for inclusion in the final regression models.

2.3.3 Investor protection

The level of investor protection can also be identified as a contributing factor to the observed differences in SMP rates across countries. Strong legal protection of outside investors mitigates the risk of expropriation by corporate insiders (La Porta et al., 2013, pp. 426), which makes investing more attractive to a broader range of individuals. Furthermore, wealthy individuals are more inclined to assume control of companies when investor protection is less robust. Therefore, as indicated by Giannetti and Koskinen (2010), if investor protection is indeed weak, the demand of affluent individuals for voting shares drives prices, leading to lower expected returns and thus lower SMP rates of retail investors (p. 160). The same applies if the situation is reversed. As a result, retail investors from countries with less robust investor protection policies hold a greater proportion of their portfolios in foreign equities, given that domestic stocks are not as attractive in comparison (Giannetti & Koskinen, 2010, p. 161). However, when individuals participate in the stock market, they often tend to prefer securities of domestic issuers (French & Poterba, 1991), a phenomenon known as “home equity bias.” Combined with a low level of investor protection, this tendency may result in individuals not participating in the stock market if they are unwilling to invest in foreign companies with which they are not familiar. Since a higher level of investor protection is expected to have a positive effect on SMP, specific proxies, as derived below, are considered potential control variables in the final regression analyses.

La Porta et al. (2013) demonstrate that there is a significant influence of legal origins on laws and regulations, including investor protection, with common law systems offering stronger protections compared to civil law systems (pp. 426–429). Further, they argue that although specific laws and regulations have evolved and adapted locally over time, the core principles, and frameworks of each legal system have remained intact. These enduring legal foundations continue to shape the regulatory environment, including investor protection, in different countries, which in turn influences the observed levels of SMP. To control for the effect of investor protection on SMP, variables are constructed that correspond to the historical origin of OECD countries' legal systems. Specifically, following La Porta et al. (2013, pp. 429-432, 474-475), countries are classified as belonging to one of four legal families, i.e., English common law, French civil law, German civil law, or Scandinavian civil law, which is then translated into four corresponding dummy variables, which take the value 1 if the relevant legal origin applies and 0 otherwise. Summary statistics for these dummies are presented in Table 12 (cf. Appendix II).

2.3.4 Familiarity

As mentioned in the previous section, familiarity may also be a factor in SMP. Individuals frequently demonstrate a preference for investing in companies or industries with which they are more familiar, as they tend to possess a more comprehensive understanding of the operational dynamics and potential for growth of these entities. This familiarity may be derived from personal experiences, such as the use of a company's products or services, or from following news and developments in a particular industry. Furthermore, investors may be more confident in their investment decisions when they possess knowledge about a company's management, products, or competitive position. For example, 401(k) plan participants in the United States show a strong propensity to overweight their own company's stock in their portfolios (Benartzi, 2001; Liang & Weisbenner, 2002; Mitchell & Utkus, 2002). Coval and Moskowitz (1999, 2000) show that even professional asset managers tend to favor local stocks and that these investments outperform those in more distant companies. Weisbenner and Ivkovich (2003) demonstrate the same for retail investors. Brown et al. (2004) were the first to show that the presence of local firms is positively correlated with the likelihood of SMP (pp. 3-6). They also show that if a higher proportion of individuals in the local community own stocks, an individual is more likely to buy stocks as well (Brown et al., 2004, p. 2). Several other studies also suggest that information sharing with peers plays an important role in investment decisions, i.e., initial SMP and subsequent portfolio selection (Duflo & Saez, 2003; Hong et al., 2004; Bursztyn et al., 2014). In addition, Haliassos et al. (2017) and Ouimet and Tate (2017) show that social interactions also improve financial literacy. Early and ongoing exposure to financial information via the community and leveraging knowledge from peers contribute to building this financial literacy. Consequently, by increasing familiarity in the communities, individuals can achieve better financial understanding and make more informed financial decisions. Using data on Facebook friendships, Cannon et al. (2024) even find that economic connectedness, defined as the proportion of one's social network with high socioeconomic status, is the most important factor in household SMP relative to cohesiveness and civic engagement. Overall, familiarity, whether through local firms or peers, can lead to higher levels of financial literacy and SMP.

In the literature, the degree of home bias is often used as a proxy for familiarity (Huberman, 2001; Grinblatt & Keloharju, 2001). As mentioned earlier, home equity bias refers to the tendency of individuals to invest disproportionately in domestic assets relative to foreign assets. The reasons for this bias often include factors such as familiarity with domestic markets, perceived lower risk, and a preference for local investments. However, since home bias is a portfolio weighting parameter for individuals already participating in the stock market, it is not an ideal control variable for the final regression analyses in this thesis. Rather, a parameter is sought that influences the initial decision to participate in the stock market, regardless of the subsequent portfolio weighting, e.g., by capturing the effect of local firms and peers.

For the local firm effect, the number of listed firms per capita is considered as a potential control variable. This measure effectively captures the density of publicly traded companies within a country, reflecting how accessible these firms are to the domestic population. Thus, residents of countries with a higher number of listed firms per capita are more likely to encounter these firms, and they are offered a more diverse range of investment opportunities, consistent with the idea of a local firm effect, i.e., that proximity increases the likelihood of investment. The working-age population (aged 20-64) is used as the denominator to offer a more targeted measure of the availability of publicly traded firms relative to the segment of the population most likely to engage in stock market activities. Population data are from the OECD database for the latest available year, 2022 (OECD, 2023b). The number of listed firms per country is obtained from the World Bank Group for 2022 or the most recent year available (2024a). This first proxy is called “familiarity firms.”

For the peer effect, the OECD community ranking score based on the perceived quality of the social support network in each country, provided by the OECD Better Life Index, is considered a potential control variable (OECD, n.d.). High-quality social support networks indicate strong community involvement and interaction, which is consistent with the concept of the peer effect, i.e., that individuals are influenced by the behavior of their community members. Consequently, these networks also reflect how individuals interact and influence each other's financial behavior, such as SMP, by facilitating the flow of information and increasing trust through shared knowledge and peer recommendations. This second proxy is called “familiarity peers.” Summary statistics for the two familiarity proxies are presented in Table 12 (cf. Appendix II).

As with the risk aversion proxies, the familiarity variables are also examined in more detail and prioritized accordingly. The correlation coefficient between the two proxies is 0.298. Together with the low VIFs in Table 16 (cf. Appendix II), this indicates that the proxies are not highly correlated. Therefore, both proxies could be used side by side in a regression. However, due to the larger number of observations, the second proxy, which captures the peer effect, is preferred. In the individual regressions on the FinLit score and on total SMP, both proxies have a rather low R-squared value, indicating that neither proxy has very high explanatory power, as shown in Table 17 (cf. Appendix II). The insignificant t-test results of the two familiarity proxies in Table 18 further confirm that neither has very high explanatory power (cf. Appendix II). Consequently, the prioritization is based solely on data availability.

2.3.5 Trust

Trust also plays an important role in SMP. Hagman (2015) shows that higher levels of trust, especially general trust in others and trust in government institutions, can increase SMP. Trust helps explain the variation in SMP across individuals and countries and may explain why even some wealthy individuals choose not to participate in the stock market (Hagman, 2015, p. 28). Sapienza et al. (2007) distinguish between belief-based trust, which pertains to the general trust in others, and preference-based trust, derived from past trusting behaviors. While belief-based trust may affect perceptions of market mechanisms, participants, and institutions, preference-based trust likely builds upon individual experiences with financial transactions. Consequently, both components may impact SMP, particularly through different levels of confidence in investor protection and market integrity. Guiso et al. (2008) further posit that trust affects investment decisions due to the perceived risk of being cheated. They demonstrate that less-trusting individuals are less likely to invest in stocks and, if they do, they invest smaller amounts. Moreover, Georgarakos and Inderst (2014, p. 37) find that, especially for households with higher financial capability, the perception of legal protection in financial markets is a determining factor in their decision to participate in the stock market. Specifically, there is a positive correlation between households' willingness to hold risky assets and both their financial capability and the extent to which they trust their legal protection (Georgarakos & Inderst, 2014, p. 38). This suggests that building confidence in the legal and governmental institutions that oversee financial markets is a critical factor in promoting SMP. Such institutions include, for example, national governments, the civil service, the judiciary, and the legislature. As the trust level in those institutions is expected to have a positive effect on SMP, and Guiso et al. (2008, pp. 2559, 2587) suggest that trust is not just a proxy for other determinants of SMP, dedicated control variables are considered for the final regression analyses, as detailed below. Data are from the OECD database for the latest available year, 2021 (OECD, 2023a).

- Trust in national government: This variable represents the aggregate trust that individuals place in the national government's capacity to oversee and regulate financial markets. It encompasses a more comprehensive understanding of the government's effectiveness, stability, and reliability (OpenAI, 2024).
- Trust in civil service: The civil service encompasses the bureaucratic apparatus responsible for implementing government policies, including financial market regulations. This variable is thus indicative of the perceived administrative efficiency, professionalism, and integrity of the civil service in financial market oversight (OpenAI, 2024).
- Trust in judiciary: This variable is indicative of the public's confidence in the judiciary and the legal framework that governs financial markets. It encompasses perceptions of the fairness, impartiality, and effectiveness of legal institutions in resolving disputes, enforcing contracts, and upholding investor rights (OpenAI, 2024).
- Trust in legislature: This variable is an indicator of trust in the legislative branch of government, which plays a role in setting financial market regulations and overseeing regulatory agencies. It reflects perceptions of the effectiveness of the legislative branch, responsiveness to public concerns, and accountability (OpenAI, 2024).

The variables indicate the percentage of respondents who have a high and a moderately high amount of trust in each institution. In line with previous literature (Guiso et al., 2008; Hagman, 2015), the general level of trust in others is also considered a control variable, indicating the percentage of respondents who say that most people can be trusted. The data for this parent trust proxy come from Wave 7 of the World Values Survey (Haerpfer et al., 2022). With these data, the cross-country results of Guiso et al. (2008, p. 2590) on the impact of trust on direct SMP can be replicated with similar significance levels across OECD member states using the English common law dummy as a control variable, as shown in Table 19 (cf. Appendix II). Moreover, Guiso et al. (2008) show that there is a strong correlation between trust and familiarity and that the level of trust plays a crucial role in explaining the effect of familiarity on SMP, because knowledge, e.g., of local firms, reduces mistrust, and because trust facilitates the flow of information, e.g., between the members of a community (p. 2562). Consequently, the trust variables are given a higher priority than the familiarity proxies in the final regression analyses. Summary statistics for the five trust proxies are presented in Table 12 (cf. Appendix II).

As with previous proxies, the trust variables are also examined in more detail and prioritized accordingly. The highest correlation coefficient among the five proxies is 0.936 (between trust in legislature and trust in national government), with the remaining correlation coefficients also ranging between 0.630 and 0.850. Together with the VIFs in Table 20 (cf. Appendix II), this indicates that the proxies are highly correlated with each other. Therefore, using multiple trust proxies side by side in a regression is likely to cause multicollinearity problems. In the individual regressions on total SMP, which is the main dependent variable of interest in this case, the general level of trust in others has the highest R-squared value, as shown in Table 21 (cf. Appendix II). The t-test results in Table 22 further confirm that general trust in others has the highest explanatory power for SMP (cf. Appendix II). Because it also has the largest number of observations, this parent trust variable is prioritized in the final regression analyses.

2.3.6 Economic indicators

The final regression analyses should also control for economic indicators such as historical economic development and stock market performance, as these factors may have a significant impact on SMP as well. For example, Meister and Schulze (2022) show that in the wake of economic crises such as the dotcom crisis, the financial crisis, and the euro crisis, investing in risky assets declined significantly in Germany (pp. 1-2). In contrast, during the recent COVID-19 crisis, there was a sharp increase in SMP, driven mainly by the younger generation. Moreover, Kaustia et al. (2023) show that changes in SMP rates tend to track stock market performance (pp. 5-6). Consequently, controlling for economic indicators can help to isolate the effect of the DoF on SMP from the general economic conditions and ensures that the observed relationship is not spuriously driven by overall economic trends.

When selecting a period for the economic indicators to control for in the final regression analyses, it is critical to consider a timeframe that captures relevant economic trends and market conditions across all OECD member countries. Given the significant economic disruption caused by the COVID-19 pandemic, and the need to capture both immediate and longer-term effects, the period from 2015 to the most recent year available, i.e., 2022, is selected. This timeframe encompasses several distinct economic phases, including the pre-pandemic expansion, the sharp contraction during the COVID-19 pandemic, and the subsequent recovery. This ensures that the results are not unduly influenced by a single economic event or cycle. Moreover, the selected period allows for the observation of different market adjustments that occurred in response to the pandemic and its aftermath. This allows for control of the effects of changes in market conditions over time. In the context of economic development, GDP growth is considered a potential control variable, as it is a fundamental indicator of economic health. To control for stock market performance, on the other hand, the development of the ratio of stock market capitalization to GDP is considered as per La Porta et al. (2013, p. 435). This measure normalizes stock market size as a percentage of each country’s economic output, thereby facilitating meaningful comparisons across OECD countries with disparate levels of stock market development, different economic sizes, and dissimilar economic structures. The geometric average change per year is used as a measure for both variables rather than the absolute change over the entire period because it accurately reflects compound growth, smooths out short-term volatility, and provides a clearer long-term trend, making it a more reliable indicator over time. GDP growth data are from the OECD database (OECD, 2024d). Stock market capitalization to GDP growth data are from the World Bank Group (2024b). Summary statistics for the two economic indicators are presented in Table 12 (cf. Appendix II).

2.3.7 Political uncertainty

Political uncertainty can also be identified as a determinant of SMP. For example, empirical models suggest that political uncertainty requires a higher risk premium for investors (Pástor & Veronesi, 2012), and studies show that political uncertainty increases stock market volatility (Baker et al., 2016; Boutchkova et al., 2012). Consequently, uncertainty may discourage SMP, especially among risk-averse individuals. Indeed, Agarwal et al. (2022) show that during periods of heightened political uncertainty, households reallocate funds from equities to safer assets in response to increased risk. Using data for OECD countries from 1995 to 2016, Gholipour and Dunkley (2019) also show that increased economic policy uncertainty reduces household SMP and leads to a shift to safer assets. Increased labor income risk and increased asset risk, which are particularly binding for less wealthy and lower-income households, appear to drive the negative effect of uncertainty on SMP. Baker et al. (2016) provide an index of economic policy uncertainty (EPU) based on the frequency of newspaper coverage of related keywords.

To capture the overall trend, in line with the economic indicators, the geometric average annual change in the EPU index per country between 2015 and 2022 is considered an additional potential control variable in the final regression analyses. Data are taken from Baker et al. (2024). Summary statistics for the political uncertainty control variable are presented in Table 12 (cf. Appendix II).

2.3.8 Tax rates

Personal tax rates, especially those on capital gains and dividends, can also influence SMP. Higher rates reduce the after-tax return on investment, making equity investment less attractive relative to other forms of saving or consumption (Poterba, 2001). Consequently, individuals may be less inclined to invest in the stock market if the tax burden on returns is perceived as high. In addition, Chetty and Saez (2005) show that favorable tax treatment of individual dividend income encourages corporate dividend payments. This is because dividend taxes represent a double tax on corporate earnings (Poterba & Summers, 1984). Consequently, low taxes on dividends may increase the attractiveness of dividend-paying stocks and thus be a positive signal for SMP.

Overall, taxes on investment income, be it capital gains or dividends, can be interpreted as a form of transactional participation cost that can negatively affect SMP. Therefore, both the personal tax rate on capital gains and the personal tax rate on dividends are considered control variables in the final regression analyses. Net personal tax rates on dividend income are taken from the OECD database (OECD, 2024c). Capital gains tax rates are taken from PwC (2024). The corresponding summary statistics are presented in Table 12 (cf. Appendix II).

3 In-depth analysis

3.1 Difference-in-differences analysis

In order to see whether a difference-in-differences (DiD) analysis is possible, it is first necessary to check when the (quasi-)mandatory funded pension components were introduced in each OECD member state and whether sufficient FinLit and SMP data are available. Overall, there is only one country in the dataset for which a FinLit DiD analysis is possible, i.e., for which data on financial literacy are available before and after the introduction of a (quasi-)mandatory funded pension component.

For an SMP DiD analysis, on the other hand, there are a total of three suitable countries. Consequently, a comprehensive DiD analysis is not possible due to limited data availability. However, DiD estimators can still be computed for the relevant OECD countries to discuss suggestive evidence, as detailed below. For this purpose, Germany is used as the benchmark country without pension funding.

3.1.1 FinLit difference-in-differences

The only suitable country for FinLit DiD analysis is Poland, which introduced the relevant (quasi-)mandatory funded pension component in 2019. As a reminder, FinLit data are available for OECD countries from 2012 to 2023. However, except for the results of individual questions like the Big Three, overall scores for financial literacy, knowledge, behavior, and attitudes only allow sufficient comparability between the 2016/17 and 2020/21⁵ OECD INFE reports. Consequently, the DiD estimators are calculated using data points from 2020 and 2016/17 to analyze the differences in overall scores and Big Three knowledge before and after the introduction of the funded pension component. Since it is also possible to consider the 2023 results for the Big Three, additional corresponding DiD estimators are also calculated using data points from 2023 and 2016/17. Each DiD estimator measures the difference in the change in outcomes between Poland and Germany during the respective periods. The corresponding data are presented in Table 1. Positive DiD estimators suggest that after the introduction of the (quasi-)mandatory funded pension component, the respective FinLit variable increases relatively more in Poland than in Germany, which has no such component, and vice versa, assuming parallel trends.

Table 1 FinLit DiD for Poland vs. Germany

Illustrations are not included in the reading sample

Note. The difference-in-differences for each FinLit variable between Poland and Germany during the respective periods is indicated. Further details are provided in Table 23 (cf. Appendix II).

For the period between 2016/17 and 2020, most DiD estimators are positive. However, Poland and Germany each correspond to only one observation in the dataset, so the results may be due to chance. Moreover, 2020 is only one year after the introduction of the (quasi-)mandatory funded pension component in Poland in 2019, so the DiD estimators for the Big Three between 2023 and 2016/17 probably better reflect the longer-term effects of the introduction of the (quasi-)mandatory funded pension component. The corresponding DiD estimators are all negative, which is consistent with the scatter plot presented in Section 2.1.5. But again, without a larger number of relevant observations, the DiD results are not very reliable.

3.1.2 SMP difference-in-differences

As the first wave of the SHARE was conducted in 2004 and SMP data for Norway are available from 1993 onwards, a total of three countries are eligible for an SMP DiD analysis, i.e., Norway, Israel, and Poland, for which the relevant (quasi-)mandatory funded pension components were introduced in 2006, 2008, and 2019, respectively. Consequently, for Israel and Poland, the DiD estimators are computed using the SHARE data points from the most recent Wave 9 and the last available SHARE data points around or shortly before the introduction of the respective (quasi-)mandatory funded pension component. As for the FinLit DiD, Germany is used as the benchmark country without pension funding. For Israel, the DiD estimators are calculated using the SMP data points from Wave 9 and the SMP data points from Wave 2 of the SHARE, which was conducted in 2006/07. Similarly, for Poland, the DiD estimators are calculated using the SMP data points from Wave 9 and the SMP data points from Wave 7 of the SHARE, which was conducted in 2017. For Norway, except for 2016, only total SMP rates could be obtained. The corresponding DiD estimator is computed using the data point from 2006, which matches the data point from Wave 2 of the SHARE for Germany, and the data point from 2015, which matches the data point from Wave 6 of the SHARE for Germany. The corresponding data are presented in Table 2.

Table 2 SMP DiD for relevant OECD countries vs. Germany

Illustrations are not included in the reading sample

Note. The difference-in-differences for each SMP variable between relevant OECD countries and Germany during the respective periods is indicated. Further details are provided in Tables 24 through 26 (cf. Appendix II).

Since the signs of all SMP DiD estimators are negative, this suggests that the introduction of a (quasi-)mandatory funded pension component may have a negative effect on SMP, which contrasts with the scatter plot presented in Section 2.1.4. However, as with the FinLit DiD estimators, the results are not very reliable due to the limited number of relevant observations. Therefore, to test the hypotheses developed, the next step is to dive into the final regression analyses.

3.2 Hypothesis testing

3.2.1 H1: The DoF has a positive impact on FinLit across countries

3.2.1.1 Bivariate regressions on FinLit and its components

In order to test the first hypothesis, which posits a positive impact of the DoF on financial literacy across countries, a series of bivariate regressions are first conducted, wherein the independent DoF variables are regressed on the dependent FinLit variables. These bivariate scenarios will provide an initial indication of whether there is a significant relationship between the DoF and FinLit. The results of the bivariate regressions across all scenarios are presented in Tables 3 and 4, where the number of observations corresponds to the 20 OECD countries included in the 2023 OECD INFE report on adult financial literacy. This is referred to as the first regression model, designated as Model 1.

Table 3 Model 1: Bivariate regressions of indep. DoF variables on dep. FinLit variables (1)

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses after the coefficients. ** indicates that the coefficient is different from zero at the 5% level, and * at the 10% level.

It is notable that, except for the independent workplace DoF and combined mandatory funding variables, the vast majority of the coefficients exhibit a negative sign, which is consistent with the scatter plot presented in Section 2.1.5. However, only a few results are significant, especially for scenarios with the financial attitudes score as the dependent variable and the statutory DoF as the independent variable. Overall, the bivariate regression of the independent statutory DoF variable on the dependent simple and compound interest knowledge variable demonstrates the highest t-statistic of all scenarios. The workplace DoF is the sole independent DoF variable exhibiting a positive coefficient on all dependent FinLit variables (cf. Table 4). Nevertheless, none of the coefficients of the workplace DoF exhibit a statistically significant difference from zero.

Table 4 Model 1: Bivariate regressions of indep. DoF variables on dep. FinLit variables (2)

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses after the coefficients. * indicates that the coefficient is different from zero at the 10% level.

Table 5 Model 2: Bivariate regressions of indep. DoF variables on dep. FinLit variables (1)

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses after the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level.

Given the presumed positive relationship between knowledge of the Big Three and SMP, particularly in the context of long-term participation, and the availability of comparable Big Three data from older OECD INFE publications, the regression scenarios on the dependent Big Three variables are extended accordingly. Tables 5 and 6 illustrate the results of the analysis, which includes the Big Three data for the additional eleven OECD countries drawn from older OECD INFE publications. The bivariate regressions presented thus include a total of 31 observations. This is referred to as the second regression model, designated as Model 2. Two scenarios in the second regression model yield a positive coefficient, with one of the two exhibiting a statistically significant difference from zero (cf. Table 6). Notwithstanding, the overwhelming majority of scenarios, irrespective of whether they deliver significant or insignificant results, yield negative coefficients. In total, only 25% of the scenarios in the second regression model provide significant outcomes, with most of them indicating a negative relationship between the DoF and financial literacy.

Table 6 Model 2: Bivariate regressions of indep. DoF variables on dep. FinLit variables (2)

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses after the coefficients. ** indicates that the coefficient is different from zero at the 5% level.

3.2.1.2 Multiple-variable regressions on FinLit and its components

The subsequent phase of the analysis entails examining which scenarios also yield significant results in more comprehensive regression specifications that include pertinent control variables. The results of Chapter 2.3 suggest that the most important predictors of financial literacy are individual socio-demographic and socio-economic characteristics, as well as risk aversion, familiarity, and, hence, trust. Consequently, the multiple-variable regression models are constructed in such a way that these control variables are incorporated gradually for each specification. The bivariate regression corresponds to the 1st regression specification. The 2nd specification also includes sociodemographic control variables (i.e., sex, age, and education), while the 3rd adds socioeconomic control variables (i.e., employment and income/wealth). The remaining control variables are added subsequently, if applicable. In the context of multiple-variable regressions, it is essential to monitor the correlation coefficients between predictors and the VIFs in order to identify potential issues of multicollinearity. The threshold for the VIFs is set at 10, which is considered a rule of thumb in the literature not to exceed (O’Brien, 2007). Additionally, an absolute value of 0.8 is set as the maximum limit for the correlation coefficients. The VIF and correlation limits are not exceeded in any of the following multiple-variable regressions presented.

In the first regression model, again only the 20 OECD countries for which FinLit data are available in the 2023 OECD INFE report are considered. The first three regression specifications are applied to every scenario between independent DoF and dependent FinLit variables. Tables 27 and 28 illustrate, for each scenario, which regression specifications yield coefficients for the respective independent DoF variables that are significantly different from zero (cf. Appendix II). In total, only five scenarios in the first regression model yield significant results in higher regression specifications. Three of these scenarios deliver significant results in the 3rd regression specification, in which wealth is used as a control variable instead of income. However, the use of wealth as a control variable also results in the loss of observations, as wealth data could not be obtained for nine OECD countries. Consequently, the regression analysis is rendered less reliable by the replacement of income with wealth, as some observations are lost in the process. Given that almost all coefficients of the DoF variables in the scenarios with significant results exhibit a negative sign, it is not possible to establish a positive relationship between the DoF and FinLit in the first regression model.

In order to increase the number of observations in the scenarios with the dependent Big Three variables, again data for the additional eleven OECD countries drawn from older OECD INFE publications are incorporated into the second regression model. Once more, Tables 29 and 30 illustrate, for each scenario, which regression specifications yield coefficients for the respective independent DoF variables that are significantly different from zero (cf. Appendix II). In the second regression model, all the scenarios that yield statistically significant coefficients in the bivariate regressions lose significance as soon as higher regression specifications are entered. There is only one scenario that yields a significant coefficient in a higher regression specification, namely the 3rd regression specification of the independent combined DoF variable on the dependent risk diversification knowledge variable. In this specification, there are a total of 20 observations in the second model, as opposed to 15 in the same scenario in the first model. In both regression models, the coefficient in this scenario is negative and significantly different from zero at the 10% level.

Given the lack of significant results in the higher regression specifications and the limited number of observations, it is not prudent to include additional control variables at this point. This is because adding further control variables is unlikely to yield significant outcomes and may result in a further reduction of the number of observations, thereby compromising the credibility of the results. In total, only about 7% of all presented scenarios across all three regression specifications in both regression models yield statistically significant results, and the significance of the bivariate regressions is diminished in most cases once control variables are included. Consequently, the presented results do not permit the assertion that there is a significant relationship between the DoF and financial literacy. Because the overwhelming majority of scenarios across all three regression specifications in both regression models yield coefficients with a negative sign, irrespective of whether they are significantly different from zero or not, it can be concluded that the first hypothesis, namely that the DoF exerts a positive influence on financial literacy across countries, is rejected.

3.2.2 H2.1: The DoF has a positive impact on SMP across countries

3.2.2.1Bivariate regressions on SMP

In order to test the second hypothesis, namely that there is a positive impact of the DoF on SMP across countries, again bivariate regressions of the independent DoF variables on the dependent SMP variables are first performed. The regression results for all scenarios are presented in Tables 7 and 8.

Table 7 Bivariate regressions of indep. DoF variables on dep. SMP variables (1)

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses after the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level.

It should be noted that data for Australia, Canada, and Norway are not included in these regressions due to the time lag in data collection. These data points will be incorporated into a separate regression specification. It is striking that the coefficient exhibits a positive sign in all scenarios, indicating a positive correlation between the DoF and SMP. This finding aligns with the scatter plot presented in Section 2.1.4. Of all scenarios, 44% yield significant results, while the primary DoF variables (i.e., DoF score, combined DoF, statutory DoF, and workplace DoF) deliver coefficients significantly different from zero in two-thirds of the respective scenarios. Moreover, it is noteworthy that the observed positive relationship appears to be primarily driven by the workplace pension pillar, as evidenced by the lack of significant coefficients for the two DoF variables pertaining solely to the statutory pillar, i.e., the statutory DoF and statutory mandatory funding variables.

Table 8 Bivariate regressions of indep. DoF variables on dep. SMP variables (2)

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses after the coefficients. *** indicates that the coefficient is different from zero at the 1% level, and ** at the 5% level.

3.2.2.2 Multiple-variable regressions on SMP

In order to gain further insight and examine which scenarios also yield significant results in more comprehensive regression specifications, it is again necessary to add pertinent control variables. The findings of Chapter 2.3 indicate that the most significant predictors of SMP are individual socio-demographic and socio-economic characteristics, and several other factors, in particular risk aversion and trust. Once more, the multiple-variable regression models are constructed in such a way that these control variables are incorporated gradually for each specification. The bivariate regression corresponds to the 1st regression specification. The 2nd one also includes sociodemographic control variables (i.e., sex, age, and education), while the third one adds socioeconomic control variables (i.e., employment and income/wealth). The remaining control variables are added subsequently, if applicable. First, the initial three regression specifications are applied to every scenario between independent DoF and dependent SMP variables. Tables 31 and 32 illustrate, for each scenario, which regression specifications yield coefficients for the respective independent DoF variables that are significantly different from zero (cf. Appendix II). In all scenarios, regardless of the regression specification, the independent DoF variables exhibit a positive coefficient. Regarding the primary DoF variables (i.e., DoF score, combined DoF, statutory DoF, and workplace DoF), the majority of scenarios deliver significant results, indicating a positive relationship between the DoF and SMP. Given that the statutory DoF yields significant coefficients in the higher regression specifications across all dependent SMP variables, it is possible that a potential positive effect of the DoF on SMP is not driven more by the workplace pension pillar as previously thought. In fact, in the higher regression specifications, the workplace DoF only yields a coefficient that is significantly different from zero in the regressions on direct SMP. Overall, the most significant results in the higher regression specifications are achieved in the scenarios with the dependent direct SMP variable for which the most data are available.

In the 3rd regression specification, the coefficients for the statutory DoF and the combined DoF are, at a minimum, significantly different from zero at the 5% level, irrespective of whether income or wealth is employed as a control variable. Consequently, these two scenarios will be subjected to further analysis by incorporating additional control variables. In order to circumvent multicollinearity issues that arise when income and trust in others that have a correlation coefficient of above 0.8 are included in a regression model, wealth is employed as a substitute for income. This approach allows the VIF and correlation limits to be respected even in higher regression specifications. Table 33 shows that the coefficient of the combined DoF loses its significance in the 4th regression specification including positive risk perception and trust in others as additional control variables (cf. Appendix II). However, the incorporation of Canada’s direct SMP data point for 2019 and Norway's for 2016 as supplementary observations in the 5th regression specification reinstates the significance of the combined DoF's coefficient. In contrast, Table 34 shows that the statutory DoF remains a highly significant factor in the 4th regression specification, even in the absence of the supplementary direct SMP data points from Canada and Norway (cf. Appendix II). Upon the incorporation of the additional data points in the 5th regression specification, the t-statistic exhibits a further increase (cf. Table 34 in Appendix II). Overall, the 4th regression specification in the model with the statutory DoF as independent variable provides the most optimal model fit and the highest explanatory power, as indicated by the F-statistics and adjusted R-squared values in Tables 33 and 34 (cf. Appendix II). Nevertheless, the 4th regression specification is also the one with the lowest number of observations at 17. This is because some observations are lost in the process of adding additional control variables because, for some countries, not all data points could be obtained. It is possible to design further specifications with additional control variables in which the number of observations remains the same or even increases slightly (cf. Tables 35 and 36 in Appendix II). However, adding additional control variables likely overfits the regression model due to the still very limited number of observations.

To further ensure the robustness of the results, the control variable of education is replaced with the financial literacy score in the corresponding regression specifications. Both variables are related – with financial literacy exerting a more direct impact on SMP, as it provides the specific knowledge and skills needed to invest. As previously discussed, financial literacy is a crucial factor in SMP. This substitution is thus made to ascertain whether the significance of the coefficient of the statutory DoF on direct SMP observed in the original regression model would be maintained. Substitution is preferred over adding the financial literacy score to the model to avoid overfitting. Table 37 shows that the significance of the statutory DoF is diminished when education is replaced with the financial literacy score as control variable (cf. Appendix II). This suggests that the initial significance of the statuary DoF may be partially attributed to variations in educational attainment. It can thus be posited that education represents an important explanatory variable concerning the effects of the statuary DoF. The reduction in the adjusted R-squared value in the higher regression specifications compared to the initial model with education also demonstrates that the new model incorporating the financial literacy score is capable of explaining a lesser proportion of the variance in the direct SMP (cf. Tables 34 and 37 in Appendix II). This suggests that education may be a superior control for the variance in SMP relative to the financial literacy score. Furthermore, the number of observations also decreases markedly when education is replaced with the financial literacy score. A reduction in the number of observations may result in less robust findings and may be a contributing factor to the loss of significance of the coefficient of the statutory DoF. Given that only 13 observations remain left in the 4th regression specification with the financial literacy score, the control variable wealth yields a VIF of over 15, which demonstrates the existence of problematic multicollinearity (cf. Table 37 in Appendix II). Overall, it can be stated that education may be a superior control variable, which makes the robustness check with the financial literacy score less meaningful.

Using education instead of the financial literacy score, all performed regression analyses with the statutory DoF as independent variable deliver significant results, including those with indirect SMP and total SMP as dependent variables (cf. Tables 38 and 39 in Appendix II). It is therefore reasonable to posit that the statutory DoF could have a positive influence on SMP, given the evidence which suggests a positive relationship between the two even after controlling for the most important predictors of SMP. However, as not all identified influencing factors of SMP can be used in conjunction as control variables due to the limited number of observations, there is a potential risk of omitted variable bias. Furthermore, the bivariate regressions on all SMP variables with the statutory DoF as independent variable yield insignificant results (cf. Table 7). The regression models including control variables are more meaningful than those without, as they permit the exclusion of alternative explanations. In the most favorable scenario, however, the bivariate regression would also be significant. The addition of the control variable sex to the bivariate regression models on all three SMP variables results in the coefficient of the statutory DoF becoming significant (cf. Table 9). This indicates that gender may be an important factor influencing the relationship between the statutory DoF and direct SMP and that this effect is superimposed in the bivariate regression specification.

Table 9 SMP regression models: Statutory DoF and Sex

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level.

In addition, when the supplementary direct SMP data points for Canada and Norway are added, the coefficient of the statutory DoF already becomes significant in the bivariate regression. Moreover, the t-statistic increases in each regression specification once the additional data points are included (cf. Table 40 in Appendix II). This could indicate that the original insignificance in the bivariate regression without the additional data points is due to the sample size being too small or not fully representative. However, it is important to remember that the data points for Canada and Norway are already somewhat outdated, and no statement can be made about how the SMP in both countries has developed in recent years. To put this in context, the total SMP in Norway for 2016 is reported to be around 25%. Between 1998 and 2016, the total SMP fluctuated slightly within a range of 24% to 32%, with a slight downward trend after the turn of the millennium (Galaasen & Raja, 2024, Table 1). Thus, over almost two decades, there have been no drastic changes in SMP in Norway, which supports the validity of the data point. The same applies to Canada, where the direct SMP rate fluctuated only slightly within a range of 7% to 10% between 1999 and 2019 (Statistics Canada, 2020). The increased significance following the inclusion of the two additional data points thus provides further support for the finding of a significant positive correlation between the statutory DoF and SMP.

In certain countries, mandatory pension contributions are used to finance both individual and collective benefits. Collective benefits encompass a range of provisions, including survivors’ pensions and disability benefits, for which there exists some form of risk sharing between contributors. As the proportion of contributions allocated to these collective benefits cannot be distinguished from the remainder in most cases, and therefore not all DoF values in the dataset can be considered fully comparable, relevant countries were marked for a robustness check in case of significant results. In the statutory pillar, a total of 24 countries are identified where contributions are also used to finance collective benefits such as survivor’s pensions or disability benefits, i.e., where risks are shared collectively among contributors. 36 out of 38 OECD countries are relevant for the analysis because Australia and New Zealand do not have (quasi-)mandatory contributions in the statutory pension pillar. The sample size for this pillar is therefore only twelve, leading to nine observations in the bivariate direct SMP regression model using the statutory DoF as independent variable. In contrast, there are only three countries in the workplace pension pillar where contributions also finance collective benefits. However, of the 38 OECD countries, only 13 are relevant, as the rest do not have (quasi-)mandatory contributions in the workplace pension pillar. As a result, the sample size for this pillar is only ten, leading to eight observations in the bivariate direct SMP regression model using the workplace DoF as independent variable. Consequently, in order to conduct the robustness checks, it is necessary to exclude the majority of countries, which makes the outcomes highly unreliable. In addition, most regression specifications are infeasible due to the presence of multicollinearity issues (cf. Tables 41 and 42 in Appendix II). Therefore, the performed robustness checks concerning collective benefits are not very meaningful.

Regardless of whether the results are statistically significant or not, all scenarios across all regression specifications yield positive coefficients for the independent DoF variables. In addition, the four primary DoF variables (i.e., DoF score, combined DoF, statutory DoF, and workplace DoF) yield predominantly significant coefficients in both the bivariate and higher regression specifications, the latter controlling for the most important predictors of SMP. Furthermore, the incorporation of supplementary data points, which, despite being somewhat outdated, can still be regarded as valid, serves to enhance the significance of the findings, as evidenced by the regressions of the independent statutory DoF variable on the dependent direct SMP variable. As this yields a reliable positive correlation between the DoF and SMP, thereby supporting the hypothesis that the DoF has a positive impact on SMP across countries, it is accepted. Nevertheless, the observed positive relationship and statistical significance do not establish causality.

3.2.3 H2.2: A CoO within a funded pension component results in a more pronounced positive impact of the DoF on SMP

As stated in the introduction, it is reasonable to hypothesize that funded pension components in countries where individuals have the option to choose between different funds or fund managers may exert a more pronounced influence on SMP outside the pension system. This is because individuals may be more inclined to engage in financial decision-making and require familiarity with the capital markets to make an informed decision. Given the significant correlation between the DoF and SMP variables, this section examines whether a CoO has the potential to reinforce a positive impact of the DoF on SMP. To test this hypothesis, the data set is divided into two groups, i.e., countries in which there is a CoO in one or both pension pillars in the case of a (quasi-)mandatory funded pension component and countries in which there is none. This division is carried out via an additional variable in the data set, designated as “combined CoO,” which assumes the value 0 if there is no CoO available and 1 otherwise. In order to incorporate all relevant observations into a single regression model, an interaction term is generated between the combined DoF and the combined CoO variables. Moreover, to circumvent the issue of multicollinearity, the combined DoF variable is subjected to a centering process, whereby the mean value of the combined DoF variable is subtracted from each observation's combined DoF value. As a result, the highest VIF can be reduced from 7.60 in the direct SMP regression model with the combined DoF, the combined CoO, and their interaction term to 1.20 when utilizing the centered combined DoF and the corresponding interaction term with the combined CoO. The regression results employing the latter are presented in Table 43 (cf. Appendix II). In all regression specifications, the centered combined DoF demonstrates a significant positive effect on direct SMP. This indicates that a higher DoF is associated with greater SMP, controlling for the CoO and other variables, which is consistent with previous findings. The combined CoO does not show a significant effect in any regression specification. This suggests that the presence of a CoO does not significantly impact SMP when the DoF is accounted for. The interaction term is also not significant in any of the regression specifications, indicating that the effect of the DoF on SMP does not significantly differ between countries with and without a CoO. To visualize the relationship, the predicted direct SMP rates at varying levels of funding for countries with and without a CoO are calculated and plotted. The application of higher regression specifications results in the generation of predicted SMP rates that exceed the observed range within the data set (cf. Figure 24 in Appendix I). Accordingly, for the sake of interpretability, the chart in Figure 11 also visualizes the predicted direct SMP rates based on the 1st regression specification presented in Table 43 (cf. Appendix II). Looking at the latter, direct SMP increases with the degree of funding for both groups (with and without a CoO), supporting the general finding of a positive correlation between the DoF and SMP. Notably, the slope of the red line (countries with a CoO) is slightly steeper, indicating a potentially stronger effect of the combined DoF on direct SMP. However, this difference is modest, and the interaction term is not statistically significant in any of the regression specifications presented in Table 43 (cf. Appendix II). Additionally, the blue line consistently lies above the red line, suggesting higher baseline SMP in countries without a CoO. Nevertheless, the vertical bars around each point represent the 95% confidence intervals for the predicted values. As the confidence intervals of the two lines exhibit considerable overlap, the difference in predicted direct SMP rates between countries with and without a CoO is not statistically significant at the given confidence level.

Figure 11 Predicted direct SMP rates for countries with and without a CoO (1st)

Illustrations are not included in the reading sample

Note. For the 1st regression specification in Table 43 (cf. Appendix II), this chart illustrates the predicted direct SMP rates at different levels of the centered combined DoF for countries with (red line) and without (blue line) a CoO with 95% confidence intervals.

For the higher regression specifications, the slope of the red line turns negative, as illustrated in Figure 24 for the 3rd regression specification (cf. Appendix I). This suggests that, for countries with a CoO, increasing the DoF is associated with a decrease in predicted direct SMP. However, the 95% confidence intervals entirely overlap with those of the blue line, indicating once more that the differences in predictions compared to the countries without a CoO are not statistically significant. Similar outcomes are observed when the analysis is conducted in terms of total SMP, rather than direct SMP (cf. Table 44 in Appendix II and Figures 25 and 26 in Appendix I).

Overall, the results indicate that the presence of a CoO does not significantly alter the relationship between the DoF and SMP. Consequently, the hypothesis that a CoO within a funded pension component results in a more pronounced positive impact of the DoF on SMP is rejected.

4 Conclusion

4.1 Summary and magnitude of the results

As previously stated, a positive impact of the DoF on financial literacy, and in particular on the Big Three, may suggest a beneficial influence on long-term SMP and, consequently, on private old-age provision. Nevertheless, no positive correlation was found between the DoF and financial literacy. On the contrary, in the regression analyses, the signs of the coefficients of the DoF variables are predominantly negative, which, however, were also insignificant in most of the scenarios presented. Therefore, no reliable relationship, either positive or negative, could be established between the DoF and financial literacy. In light of the absence of statistical significance in the vast majority of scenarios, an assessment of the magnitude of the potential effect of the DoF on financial literacy and its components is not warranted. Overall, as there was insufficient evidence to suggest a significant relationship between the DoF and financial literacy, it is not possible to make any statements regarding long-term SMP, which is thought to require a certain degree of financial knowledge. Whether the DoF may also support private old-age provision is thus subject to further analysis.

In contrast, there appears to be a positive relationship between the DoF and SMP. Not only do all scenarios deliver positive coefficients for the DoF variables across all regression specifications, but also the four primary DoF variables (i.e., DoF score, combined DoF, statutory DoF, and workplace DoF) yield predominantly significant results, even after controlling for the most important predictors of SMP. In addition, the inclusion of supplementary data points further increases the size and the significance of the observed coefficients. As a result, a reliable positive correlation between the DoF and SMP can be established. To analyze the magnitude of the potential effect of the DoF on SMP, the significant coefficients of the respective DoF variables are divided by the corresponding standard deviations of each SMP variable. The effect sizes for the significant coefficients of the DoF variables in the bivariate regressions without the supplementary data points from Australia, Canada, and Norway are presented in Table 10. The observed magnitudes range from about 58% to 220% of the respective standard deviation, indicating a substantial potential positive effect of the DoF on SMP. For example, for the highest significance scenario, direct SMP has the potential to increase by more than two times its standard deviation for a one-unit increase in the combined DoF (cf. Table 10).

Table 10 Magnitude of the potential effect of indep. DoF variables on dep. SMP variables

Illustrations are not included in the reading sample

Note. The effect sizes are shown for each bivariate scenario that yields a significant coefficient for the respective DoF variable in the regressions presented in Tables 7 and 8. *** indicates that the result is statistically significant at the 1% level, ** at the 5% level, and * at the 10% level.

Table 11 Direct SMP regression models: Magnitude of the combined DoF and statutory DoF

Illustrations are not included in the reading sample

Note. The effect sizes are shown for each specification that yields a significant coefficient for the respective DoF variable in the regressions presented in Tables 33 and 34 (cf. Appendix II). *** indicates that the result is statistically significant at the 1% level, ** at the 5% level, and * at the 10% level. Five regression specifications are defined (1st: bivariate regression; 2nd: controlling for sex, age, and education; 3rd: additionally controlling for employment and wealth; 4th: additionally controlling for positive risk perception and trust in others; 5th: additionally incorporating data from Canada (2019) and Norway (2016) as supplementary observations).

In order to analyze the magnitude even after accounting for the most important predictors of SMP, the effect sizes are also calculated for the regression specifications that include selected control variables. Table 11 shows the effect sizes for the significant coefficients of the combined DoF and the statutory DoF variables in their regression models on direct SMP presented in Tables 33 and 34 (cf. Appendix II). For the higher regression specifications, the magnitude stays in a high range of about 153% to 221%. For the statutory DoF, the effect size even increases continuously with the regression specifications. Similar outcomes can be observed in Table 45, which shows the magnitude of the significant coefficients of the statutory DoF in the regressions on indirect and total SMP presented in Tables 38 and 39 (cf. Appendix II).

Despite the strong potential positive effect of the DoF on SMP, it is important to remember that the relationship and significance between the two parameters do not prove causality. As a matter of fact, OLS regression cannot distinguish whether the independent variable causes the dependent variable or vice versa, and situations in which the independent variable and the dependent variable simultaneously affect each other can further complicate interpretation. Whether the effect is transmitted from the DoF to SMP thus requires further analysis. The DoF can likely have a positive effect on SMP, as a funded pension component can directly force individuals to make financial decisions, or at least raise their awareness of the benefits of equities. However, the reverse relationship, where high SMP leads to a higher DoF, is also possible, e.g., through increased political pressure due to increased demand for financial products, or easier introduction of funded pension components due to a pre-existing equity culture (OpenAI, 2024). Although the latter is likely to be less direct and more dependent on long-term cultural and infrastructural developments, which argues for an effect of the DoF on SMP, further methods, such as instrumental variable approaches, would first have to be carried out in order to be able to draw causal conclusions. Furthermore, given the lack of statistical significance, an assessment of the magnitude of any altered potential effect of the DoF on SMP due to the presence of a CoO is not warranted.

4.2 Implications for pension policy and reform plans in Germany

The earmarked generational capital in Germany is allocated in the statutory pension pillar, for which the statutory DoF provides the most significant results according to the regression analyses, especially in higher regression specifications. Consequently, the generational capital has the potential to boost SMP outside the pension system, provided that the population is educated about the benefits of equities, which in turn could also increase financial literacy and private old-age provision. Moreover, a CoO, which is not available for the generational capital in Germany, does not significantly alter the relationship between the DoF and SMP. As a result, even without the option to choose between different pension funds or fund managers, the generational capital may exert a positive influence on overall SMP. Nevertheless, the statutory mandatory funding and combined mandatory funding variables do not yield significant results in the regressions on SMP. This suggests that a potential positive effect is driven more by (quasi-)mandatory funded pension components to which direct contributions are made, which will not be the case for the generational capital in Germany (Bundesministerium der Finanzen, 2024). Therefore, in order to have a positive impact on SMP and also on private old-age provision, the government must enforce financial education in the process of introducing the generational capital and, if necessary, consider direct contributions from individuals to the generational capital in the future. Such contributions would help grow the fund faster, accelerate compounding, and thus increase individual benefits at retirement, thereby mitigating demographic shifts. However, a double burden on current contributors would have to be avoided through gradual introduction, government subsidies, transitional financing, and, for example, a flexible contribution option so that individuals can control their own financial burden (OpenAI, 2024).

In addition, the German government is planning to introduce a tax-favored retirement savings account in the voluntary private pension pillar, thereby creating an incentive for long-term investment in securities (AssCompact, 2024). However, the amount of potential government subsidies and the maximum limits for tax-free investments are still open. The government could also go one step further and provide children with such a savings account already at birth, and then inform parents, and later the child, of its progress every year. At a certain age or on certain occasions, such as the purchase or construction of a home, the savings could then be made available. Even small, tax-advantaged amounts invested in a fund designated at birth or chosen by the parents would provide education about how compounding works, thereby increasing financial literacy and knowledge of the Big Three for both children and their parents. In addition, the generation of children who receive such a savings account could be observed and studied to see how they behave in the capital markets as adults, thus ensuring the efficacy of such a program.

4.3 Recommendations for future research

A significant correlation between the DoF and SMP could be established. However, whether a potential positive effect of the DoF leads to long-term SMP, which requires a certain degree of financial literacy, cannot be conclusively clarified, as no positive correlation between the DoF and financial literacy was found. As a result, no clear conclusions can be drawn regarding the strengthening of private old-age provision through funded pension components, which would require long-term SMP outside the pension system. Future research may therefore focus on determining whether the DoF influences long-term SMP.

Moreover, the dataset in this thesis is very limited, i.e., there is a limited number of observations, and the observations are only available at the country level. Therefore, in order to further validate the findings presented and potentially establish a significant relationship between the DoF and financial literacy, the analysis may be extended not only by including additional observations at the country level but also by performing a multi-level regression analysis by incorporating additional observations at the individual level. Furthermore, as previously stated, additional econometric methods, such as instrumental variable approaches, would be needed to establish causality.

Appendix I: Figures

Figure 12 Population pyramid for Germany in 2024

Illustrations are not included in the reading sample

Note. Age distribution by sex in Germany in 2024 (Statistisches Bundesamt, 2024).

Figure 13 Population pyramid for Germany in 2044

Illustrations are not included in the reading sample

Note. Age distribution by sex in Germany in 2044 (Statistisches Bundesamt, 2024).

Figure 14 Population pyramid for Germany in 2064

Illustrations are not included in the reading sample

Note. Age distribution by sex in Germany in 2064 (Statistisches Bundesamt, 2024).

Figure 15 Workplace DoF for all 13 relevant OECD countries

Illustrations are not included in the reading sample

Note. Presentation of all OECD countries with a (quasi-)mandatory funded pension component in the workplace pension pillar and their respective workplace DoF.

Figure 16 Financial knowledge score for all 20 relevant OECD countries

Illustrations are not included in the reading sample

Note. Financial knowledge scores range from 0 to 7 and are normalized as a percentage of the maximum possible score (OECD, 2023c, Annex D, Table 2.5).

Figure 17 Financial behavior score for all 20 relevant OECD countries

Illustrations are not included in the reading sample

Note. Financial behavior scores range from 0 to 9 and are normalized as a percentage of the maximum possible score (OECD, 2023c, Annex D, Table 2.11).

Figure 18 Financial attitudes score for all 20 relevant OECD countries

Illustrations are not included in the reading sample

Note. Financial attitudes scores are rescaled to a range of 1 to 5 and normalized as a percentage of the maximum possible score (OECD, 2023c, Annex D, Table 2.16).

Figure 19 Financial literacy score for all 20 relevant OECD countries

Illustrations are not included in the reading sample

Note. Financial literacy scores are rescaled to a range of 1 to 21 and normalized as a percentage of the maximum possible score (OECD, 2023c, Annex D, Table 2.1).

Figure 20 Knowledge of the time value of money in all 31 relevant OECD countries

Illustrations are not included in the reading sample

Note. The percentage of adults who answered the time value of money question correctly is indicated, adjusted to a standardized scale. Data for 20 OECD countries are from the latest publication (OECD, 2023c, Annex D, Table 2.7). Data for the remaining eleven countries (indicated in gray) are from previous publications (OECD, 2021, p. 14; OECD, 2020, p. 20; OECD, 2017, p. 19; OECD, 2016, pp. 23–24).

Figure 21 Knowledge of simple and compound interest in all 31 relevant OECD countries

Illustrations are not included in the reading sample

Note. The percentage of adults who answered the simple and compound interest questions correctly is indicated, adjusted to a standardized scale. Data for 20 OECD countries are from the latest publication (OECD, 2023c, Annex D, Table 2.7). Data for the remaining eleven countries (indicated in gray) are from previous publications (OECD, 2021, p. 14; OECD, 2020, p. 20; OECD, 2017, p. 19; OECD, 2016, pp. 23–24).

Figure 22 Knowledge of risk diversification in all 31 relevant OECD countries

Illustrations are not included in the reading sample

Note. The percentage of adults who answered the risk diversification question correctly is indicated, adjusted to a standardized scale. Data for 20 OECD countries are from the latest publication (OECD, 2023c, Annex D, Table 2.7). Data for the remaining eleven countries (indicated in gray) are from previous publications (OECD, 2021, p. 14; OECD, 2020, p. 20; OECD, 2017, p. 19; OECD, 2016, pp. 23–24).

Figure 23 Big Three score for all 31 relevant OECD countries

Illustrations are not included in the reading sample

Note. The Big Three scores are calculated as the sum of the individual results of the relevant questions, each indicating the percentage of respondents who answered correctly, resulting in a total adjusted score ranging from 0 to 3. Data for 20 OECD countries are from the latest publication (OECD, 2023c, Annex D, Table 2.7). Data for the remaining eleven countries (indicated in gray) are from previous publications (OECD, 2021, p. 14; OECD, 2020, p. 20; OECD, 2017, p. 19; OECD, 2016, pp. 23–24).

Figure 24 Predicted direct SMP rates for countries with and without a CoO (3rd)

Illustrations are not included in the reading sample

Note. For the 3rd regression specification in Table 43 (cf. Appendix II), this chart illustrates the predicted direct SMP rates at different levels of the centered combined DoF for countries with (red line) and without (blue line) a CoO with 95% confidence intervals.

Figure 25 Predicted total SMP rates for countries with and without a CoO (1st)

Illustrations are not included in the reading sample

Note. For the 1st regression specification in Table 44 (cf. Appendix II), this chart illustrates the predicted total SMP rates at different levels of the centered combined DoF for countries with (red line) and without (blue line) a CoO with 95% confidence intervals.

Figure 26 Predicted total SMP rates for countries with and without a CoO (3rd)

Illustrations are not included in the reading sample

Note. For the 3rd regression specification in Table 44 (cf. Appendix II), this chart illustrates the predicted total SMP rates at different levels of the centered combined DoF for countries with (red line) and without (blue line) a CoO with 95% confidence intervals.

Appendix II: Tables

Table 12 Dataset summary statistics

Illustrations are not included in the reading sample

Note. Summary statistics include measures of dispersion (i.e., standard deviation, minimum, and maximum) as well as the arithmetic mean as a measure of central tendency.

Table 13 Risk aversion proxy VIF analysis

Illustrations are not included in the reading sample

Note. VIF analysis after joint regression of all listed proxies on the parent FinLit and SMP variables, i.e., on the FinLit score and on total SMP.

Table 14 Risk aversion proxy R-squared analysis

Illustrations are not included in the reading sample

Note. Analysis of R-squared values from individual regressions of each proxy on the parent FinLit and SMP variables, i.e., on the FinLit score and on total SMP.

Table 15 Risk aversion proxy t-test analysis

Illustrations are not included in the reading sample

Note. Analysis of t-test results from joint regressions of all listed proxies on the parent FinLit and SMP variables, i.e., on the FinLit score and on total SMP. The t-statistics are shown in parentheses after the coefficients. ** indicates that the coefficient is different from zero at the 5% level, and * at the 10% level.

Table 16 Familiarity proxy VIF analysis

Illustrations are not included in the reading sample

Note. VIF analysis after joint regression of all listed proxies on the parent FinLit and SMP variables, i.e., on the FinLit score and on total SMP.

Table 17 Familiarity proxy R-squared analysis

Illustrations are not included in the reading sample

Note. Analysis of R-squared values from individual regressions of each proxy on the parent FinLit and SMP variables, i.e., on the FinLit score and on total SMP.

Table 18 Familiarity proxy t-test analysis

Illustrations are not included in the reading sample

Note. Analysis of t-test results from joint regressions of all listed proxies on the parent FinLit and SMP variables, i.e., on the FinLit score and on total SMP. The t-statistics are shown in parentheses after the coefficients.

Table 19 The impact of general trust in others on direct SMP

Illustrations are not included in the reading sample

Note. Analysis of the regression results on the direct SMP variable. The t-statistics are shown in parentheses after the coefficients. *** indicates that the coefficient is different from zero at the 1% level.

Table 20 Trust proxy VIF analysis

Illustrations are not included in the reading sample

Note. VIF analysis after joint regression of all listed proxies on the parent FinLit and SMP variables, i.e., on the FinLit score and on total SMP.

Table 21 Trust proxy R-squared analysis

Illustrations are not included in the reading sample

Note. Analysis of R-squared values from individual regressions of each proxy on the parent FinLit and SMP variables, i.e., on the FinLit score and on total SMP.

Table 22 Trust proxy t-test analysis

Illustrations are not included in the reading sample

Note. Analysis of t-test results from joint regressions of all listed proxies on the parent FinLit and SMP variables, i.e., on the FinLit score and on total SMP. The t-statistics are shown in parentheses after the coefficients. * indicates that the coefficient is different from zero at the 10% level.

Table 23 FinLit DiD details for Poland vs. Germany

Illustrations are not included in the reading sample

Note. Values for each FinLit variable for Poland (vs. Germany) and the respective DiD estimators for the indicated periods (OECD, 2016, pp. 24, 80; OECD, 2017, pp. 19, 67; OECD, 2020, pp. 17, 20; OECD, 2023c, Annex D, Table 2.7).

Table 24 SMP DiD details for Israel vs. Germany

Illustrations are not included in the reading sample

Note. SMP rates for Israel (vs. Germany) and the respective DiD estimators for the indicated period (SHARE-ERIC, 2024a; SHARE-ERIC, 2024d; Bergmann et al., 2019; Börsch-Supan et al., 2013).

Table 25 SMP DiD details for Poland vs. Germany

Illustrations are not included in the reading sample

Note. SMP rates for Poland (vs. Germany) and the respective DiD estimators for the indicated period (SHARE-ERIC, 2024c; SHARE-ERIC, 2024d; Bergmann et al., 2019; Börsch-Supan et al., 2013).

Table 26 SMP DiD details for Norway vs. Germany

Illustrations are not included in the reading sample

Note. Total SMP rate for Poland (vs. Germany) and the respective DiD estimator for the indicated period (Galaasen & Raja, 2024, fig. 1; SHARE-ERIC, 2024b; SHARE-ERIC, 2024d; Bergmann et al., 2019; Börsch-Supan et al., 2013).

Table 27 Model 1: Significant coefficients of indep. DoF variables on dep. FinLit variables (1)

Illustrations are not included in the reading sample

Note. Indication of whether the coefficient of the respective DoF variable is significant in any of the first three defined regression specifications (1st: bivariate regression; 2nd: controlling for sex, age, and education; 3rd: controlling for sex, age, education, employment, and income/wealth). ** indicates that the coefficient is different from zero at the 5% level, and * at the 10% level. (i) indicates that the coefficient is only significant when income is used as control variable instead of wealth in the 3rd regression specification. (w) indicates that the coefficient is only significant when wealth is used as control variable instead of income in the 3rd regression specification.

Table 28 Model 1: Significant coefficients of indep. DoF variables on dep. FinLit variables (2)

Illustrations are not included in the reading sample

Note. Indication of whether the coefficient of the respective DoF variable is significant in any of the first three defined regression specifications (1st: bivariate regression; 2nd: controlling for sex, age, and education; 3rd: controlling for sex, age, education, employment, and income/wealth). * indicates that the coefficient is different from zero at the 10% level. (w) indicates that the coefficient is only significant when wealth is used as control variable instead of income in the 3rd regression specification.

Table 29 Model 2: Significant coefficients of indep. DoF variables on dep. FinLit variables (1)

Illustrations are not included in the reading sample

Note. Indication of whether the coefficient of the respective DoF variable is significant in any of the first three defined regression specifications (1st: bivariate regression; 2nd: controlling for sex, age, and education; 3rd: controlling for sex, age, education, employment, and income/wealth). *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level. (w) indicates that the coefficient is only significant when wealth is used as control variable instead of income in the 3rd regression specification.

Table 30 Model 2: Significant coefficients of indep. DoF variables on dep. FinLit variables (2)

Illustrations are not included in the reading sample

Note. Indication of whether the coefficient of the respective DoF variable is significant in any of the first three defined regression specifications (1st: bivariate regression; 2nd: controlling for sex, age, and education; 3rd: controlling for sex, age, education, employment, and income/wealth). ** indicates that the coefficient is different from zero at the 5% level.

Table 31 Significant coefficients of indep. DoF variables on dep. SMP variables (1)

Illustrations are not included in the reading sample

Note. Indication of whether the coefficient of the respective DoF variable is significant in any of the first three defined regression specifications (1st: bivariate regression; 2nd: controlling for sex, age, and education; 3rd: controlling for sex, age, education, employment, and income/wealth). *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level. (i) indicates that the coefficient is only significant at the indicated level when income is used as control variable instead of wealth in the 3rd regression specification. (w) indicates that the coefficient is only significant at the indicated level when wealth is used as control variable instead of income in the 3rd regression specification.

Table 32 Significant coefficients of indep. DoF variables on dep. SMP variables (2)

Illustrations are not included in the reading sample

Note. Indication of whether the coefficient of the respective DoF variable is significant in any of the first three defined regression specifications (1st: bivariate regression; 2nd: controlling for sex, age, and education; 3rd: controlling for sex, age, education, employment, and income/wealth). *** indicates that the coefficient is different from zero at the 1% level, and ** at the 5% level. (i) indicates that the coefficient is only significant when income is used as control variable instead of wealth in the 3rd regression specification.

Table 33 Direct SMP regression model: Combined DoF

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. *** indicates that the coefficient is different from zero at the 1% level, and ** at the 5% level. The 5th regression specification incorporates data from Canada (2019) and Norway (2016) as supplementary observations.

Table 34 Direct SMP regression model: Statutory DoF (1)

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level. The 5th regression specification incorporates data from Canada (2019) and Norway (2016) as supplementary observations.

Table 35 Direct SMP regression model: Statutory DoF (2)

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level.

Table 36 Direct SMP regression model: Statutory DoF (3)

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level. The 5th regression specification incorporates data from Canada (2019) and Norway (2016) as supplementary observations.

Table 37 Direct SMP regression model: Statutory DoF and financial literacy

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level.

Table 38 Indirect SMP regression model: Statutory DoF

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level. The 5th regression specification incorporates Norway's 2016 SMP data point as an additional observation.

Table 39 Total SMP regression model: Statutory DoF

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level. The 5th regression specification incorporates Norway's 2016 SMP data point as an additional observation.

Table 40 Direct SMP regression model: Statutory DoF (4)

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level. All regression specifications incorporate data from Canada (2019) and Norway (2016) as supplementary observations.

Table 41 Direct SMP regression model: Statutory DoF excl. collective benefits

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. Variables omitted by STATA due to multicollinearity are marked accordingly.

Table 42 Direct SMP regression model: Workplace DoF excl. collective benefits

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. Variables omitted by STATA due to multicollinearity are marked accordingly.

Table 43 Direct SMP regression model: Interaction of combined DoF and CoO

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level. The 4th regression specification incorporates data from Canada (2019) and Norway (2016) as supplementary observations. The regression specification incorporating risk and trust as additional control variables is not included in the analysis due to the presence of multicollinearity issues (VIF > 10).

Table 44 Total SMP regression model: Interaction of combined DoF and CoO

Illustrations are not included in the reading sample

Note. The t-statistics are shown in parentheses below the coefficients. *** indicates that the coefficient is different from zero at the 1% level, ** at the 5% level, and * at the 10% level. The 4th regression specification incorporates Norway's 2016 SMP data point as an additional observation. The regression specification incorporating risk and trust as additional control variables is not included in the analysis due to the presence of multicollinearity issues (VIF > 10).

Table 45 Regression models on indirect (I) and total (T) SMP: Magnitude of the statutory DoF

Illustrations are not included in the reading sample

Note. The effect sizes are shown for each specification that yields a significant coefficient for the statutory DoF in the regression on indirect (I) and total (T) SMP presented in Tables 38 and 39. *** indicates that the result is statistically significant at the 1% level, ** at the 5% level, and * at the 10% level. Five regression specifications are defined (1st: bivariate regression; 2nd: controlling for sex, age, and education; 3rd: additionally controlling for employment and wealth; 4th: additionally controlling for positive risk perception and trust in others; 5th: additionally incorporating data from Norway (2016) as supplementary observation).

Bibliography

Agarwal, V., Aslan, H., Huang, L., & Ren, H. (2022). Political uncertainty and household stock market participation. Journal of Financial and Quantitative Analysis, 57 (8), 2899–2928. https://doi.org/10.1017/s0022109022000114

Aladangady, Aditya, Bricker, J., Chang, A. C., Goodman, S., Krimmel, J., Moore, K. B., Reber, S., Henriques Volz, A., & Windle, R. A. (2023). Changes in U.S. Family Finances from 2019 to 2022: Evidence from the Survey of Consumer Finances [Report]. Board of Governors of the Federal Reserve System. https://doi.org/10.17016/8799

Almenberg, J., & Dreber, A. (2012). Gender, stock market participation and financial literacy (SSE/EFI Working Paper Series in Economics and Finance, No 737). Stockholm School of Economics. https://EconPapers.repec.org/RePEc:hhs:hastef:0737

Arts, L. (2018). Financial literacy and stock market participation: The moderating effect of country-specific social connectedness [Thesis, University of Groningen, Uppsala University]. https://www.diva-portal.org/smash/get/diva2:1193563/FULLTEXT01.pdf

Atkinson, A., & Messy, F. (2012). Measuring financial literacy: Results of the OECD/International Network on Financial Education (INFE) pilot study (OECD Working Papers on Finance, Insurance and Private Pensions, No. 15). OECD Publishing. https://doi.org/10.1787/5k9csfs90fr4-en

Australian Securities Exchange. (2015). The Australian Share Ownership Study [Report]. https://www.asx.com.au/documents/resources/australian-share-ownership-study-2014.pdf

Baker, S. R., Bloom, N., & Davis, S. J. (2016). Measuring economic policy uncertainty. The Quarterly Journal of Economics, 131 (4), 1593–1636. https://doi.org/10.1093/qje/qjw024

Barber, B. M., & Odean, T. (2001). Boys will be boys: Gender, overconfidence, and common stock investment. The Quarterly Journal of Economics, 116 (1), 261–292. https://doi.org/10.1162/003355301556400

Basten, C., Fagereng, A., & Telle, K. (2016). Saving and portfolio allocation before and after job loss. Journal of Money, Credit and Banking, 48 (2–3), 293–324. https://doi.org/10.1111/jmcb.12301

Benartzi, S. (2001). Excessive extrapolation and the allocation of 401(k) accounts to company stock. The Journal of Finance, 56 (5), 1747–1764. https://doi.org/10.1111/0022-1082.00388

Bergmann, M., Kneip, T., De Luca, G., & Scherpenzeel, A. (2019). Survey participation in the Survey of Health, Ageing and Retirement in Europe (SHARE), Wave 1-7 (based on Release 7.0.0; SHARE Working Paper Series, No. 41-2019). MEA, Max Planck Institute for Social Law and Social Policy.

Bertaut, C. C. (1998). Stockholding behavior of U.S. households: Evidence from the 1983-1989 Survey of Consumer Finances. The Review of Economics and Statistics, 80 (2), 263–275. https://doi.org/10.1162/003465398557500

Blaschke, J. (2022). Gender differences in financial literacy among teenagers - Can confidence bridge the gap? Cogent Economics & Finance, 10 (1). https://doi.org/10.1080/23322039.2022.2144328

Börsch-Supan, A., Brandt, M., Hunkler, C., Kneip, T., Korbmacher, J., Malter, F., Schaan, B., Stuck, S., & Zuber, S. (2013). Data resource profile: The Survey of Health, Ageing and Retirement in Europe (SHARE). International Journal of Epidemiology, 42 (4), 992–1001. https://doi.org/10.1093/ije/dyt088

Boutchkova, M., Doshi, H., Durnev, A., & Molchanov, A. (2012). Precarious politics and return volatility. The Review of Financial Studies, 25 (4), 1111–1154. https://doi.org/10.1093/rfs/hhr100

Brown, J. R., Ivković, Z., Paul A. Smith, & Weisbenner, S. (2004). The geography of stock market participation: The influence of communities and local firms [Working paper]. Board of Governors of the Federal Reserve System. https://www.federalreserve.gov/pubs/feds/2004/200422/200422pap.pdf

Bucher-Koenen, T. (2009). Financial literacy and private old-age provision in Germany – Evidence from SAVE 2008 (MEA Discussion Paper, No. 192-2009). SSRN. https://doi.org/10.2139/ssrn.1534664

Bucher-Koenen, T., Alessie, R., Lusardi, A., & Van Rooij, M. (2021). Fearless woman: Financial literacy and stock market participation (NBER Working Paper Series, No. 28723). National Bureau of Economic Research. https://doi.org/10.3386/w28723

Bursztyn, L., Ederer, F., Ferman, B., & Yuchtman, N. (2014). Understanding mechanisms underlying peer effects: Evidence from a field experiment on financial decisions. Econometrica, 82 (4), 1273–1301. https://doi.org/10.3982/ecta11991

Cannon, B., Hirshleifer, D., & Thornton, J. (2024). Friends with benefits: Social capital and household financial behavior (NBER Working Paper Series, No. 32186). National Bureau of Economic Research. https://doi.org/10.3386/w32186

Chetty, R., & Saez, E. (2005). Dividend taxes and corporate behavior: Evidence from the 2003 dividend tax cut. The Quarterly Journal of Economics, 120 (3), 791–833. https://doi.org/10.1093/qje/120.3.791

Cole, S., & Shastry, G. K. (2009). Smart money: The effect of education, cognitive ability, and financial literacy on financial market participation (HBS Working Paper, No. 09–071). Harvard Business School. https://hbswk.hbs.edu/item/smart-money-the-effect-of-education-cognitive-ability-and-financial-literacy-on-financial-market-participation

Coppola, M., Langley, G., Sabatini, M., & Wolf, R. (Eds.). (2017). When will the penny drop? Money, financial literacy and risk in the digital age (International Pension Papers, No. 1/2017). Allianz. https://gflec.org/wp-content/uploads/2017/01/AGI-IPP_1-17_Financial-Literacy_FINAL.pdf

Coval, J. D., & Moskowitz, T. J. (1999). Home bias at home: Local equity preference in domestic portfolios. The Journal of Finance, 54 (6), 2045–2073. https://doi.org/10.1111/0022-1082.00181

Coval, J. D., & Moskowitz, T. J. (2000). The geography of investment: Informed trading and asset prices (CRSP Working Papers, No. 502). SSRN. https://dx.doi.org/10.2139/ssrn.214138

Cupák, A., Kolev, G. I., & Brokešová, Z. (2019). Financial literacy and voluntary savings for retirement: Novel causal evidence. European Journal of Finance, 25 (16), 1606–1625. https://doi.org/10.1080/1351847x.2019.1641123

Dohmen, T., Falk, A., Huffman, D., Sunde, U., Schupp, J., & Wagner, G. G. (2011). Individual risk attitudes: Measurement, determinants, and behavioral consequences. Journal of the European Economic Association, 9 (3), 522–550. https://doi.org/10.1111/j.1542-4774.2011.01015.x

Duflo, E., & Saez, E. (2003). The role of information and social interactions in retirement plan decisions: Evidence from a randomized experiment. The Quarterly Journal of Economics, 118 (3), 815–842. https://doi.org/10.1162/00335530360698432

Falk, A., Becker, A., Dohmen, T., Enke, B., Huffman, D., & Sunde, U. (2018). Global evidence on economic preferences. The Quarterly Journal of Economics, 133 (4), 1645–1692. https://doi.org/10.1093/qje/qjy013

Falk, A., Becker, A., Dohmen, T. J., Huffman, D., & Sunde, U. (2016). The preference survey module: A validated instrument for measuring risk, time, and social preferences (IZA Discussion Paper, No. 9674). IZA. https://docs.iza.org/dp9674.pdf

FIAP. (2016). Contributions of the individually-funded pension systems and their advantages over the public PAYGO programs (Pensions Note, No. 11). https://www.fiapinternacional.org/wp-content/uploads/2016/01/Pensions-Note-No-11_v12.12.2016.pdf

Financial Conduct Authority. (2023). Financial lives 2022: Key findings from the FCA’s Financial Lives May 2022 survey [Report]. https://www.fca.org.uk/publication/financial-lives/financial-lives-survey-2022-key-findings.pdf

French, K. R., & Poterba, J. M. (1991). Investor diversification and international equity markets (NBER Working Paper Series, No. 3609). National Bureau of Economic Research. https://doi.org/10.3386/w3609

Galaasen, S. M., & Raja, A. (2024). The dynamics of stock market participation (Working Paper, No. 5). Norges Bank. https://www.norges-bank.no/contentassets/a6b861424aa6474a9f481ff10f1de517/wp-5-24-stock_dynamics.pdf?v=21022024151956

Georgarakos, D., & Inderst, R. (2014). Financial advice and stock market participation. SSRN. https://doi.org/10.2139/ssrn.1641302

Gholipour, H. F., & Dunkley, M. E. (2019). Economic policy uncertainty and household financial assets. Applied Economics Quarterly, 65 (2), 101–114. https://doi.org/10.3790/aeq.65.2.101

Giannetti, M., & Koskinen, Y. (2010). Investor protection, equity returns, and financial globalization. The Journal of Financial and Quantitative Analysis, 45 (1), 135–168. https://www.jstor.org/stable/27801477

Grinblatt, M., & Keloharju, M. (2001). How distance, language, and culture influence stockholdings and trades. The Journal of Finance, 56 (3), 1053–1073. https://econpapers.repec.org/RePEc:bla:jfinan:v:56:y:2001:i:3:p:1053-1073

Guiso, L., Sapienza, P., & Zingales, L. (2008). Trusting the stock market. The Journal of Finance, 63 (6), 2557–2600. https://doi.org/10.1111/j.1540-6261.2008.01408.x

Guiso, L., Sapienza, P., & Zingales, L. (2018). Time varying risk aversion. Journal of Financial Economics, 128 (3), 403–421. https://doi.org/10.1016/j.jfineco.2018.02.007

Hagman, M. (2015). The effects of trust on stock market participation: A cross-sectional study based on 15 countries [Master thesis, Umeå University]. https://www.diva-portal.org/smash/get/diva2:840123/FULLTEXT01.pdf

Haliassos, M., Jansson, T., & Karabulut, Y. (2017). Financial literacy externalities [Working paper]. CEPR. https://cepr.org/system/files/2023-02/Financial%20Literacy%20Externalities.pdf

Haupt, M., & Kluth, S. (2013). Take a chance on me - Can the Swedish premium pension serve as a role model for Germany’s Riester scheme? (MEA Discussion Paper, No. 266-2013). SSRN. https://doi.org/10.2139/ssrn.2265950

Hong, H., Kubik, J. D., & Stein, J. C. (2004). Social interaction and stock-market participation. The Journal of Finance, 59 (1), 137–163. https://scholar.harvard.edu/files/stein/files/social_interactionjoff.pdf

Huberman, G. (2001). Familiarity breeds investment. The Review of Financial Studies, 14 (3), 659–680. https://doi.org/10.1093/rfs/14.3.659

Jianakoplos, N. A., & Bernasek, A. (1998). Are women more risk averse? Economic Inquiry, 36 (4), 620–630. https://doi.org/10.1111/j.1465-7295.1998.tb01740.x

Kaustia, M., Conlin, A., & Luotonen, N. (2023). What drives stock market participation? The role of institutional, traditional, and behavioral factors. Journal of Banking & Finance, 148, 106743. https://doi.org/10.1016/j.jbankfin.2022.106743

Kaustia, M., & Torstila, S. (2011). Stock market aversion? Political preferences and stock market participation. Journal of Financial Economics, 100 (1), 98–112. https://doi.org/10.1016/j.jfineco.2010.10.017

Klapper, L., Lusardi, A., & Van Oudheusden, P. (2015). Financial literacy around the world: Insights from the Standard & Poor’s Ratings Services Global Financial Literacy Survey [Report]. Global Financial Literacy Excellence Center. https://gflec.org/wp-content/uploads/2015/11/3313-Finlit_Report_FINAL-5.11.16.pdf

La Porta, R., Lopez-De-Silanes, F., & Shleifer, A. (2013). Law and finance after a decade of research. In G. M. Constantinides, M. Harris, & R. M. Stulz (Eds.), Handbook of the economics of finance (Vol. 2, pp. 425–491). Elsevier. https://doi.org/10.1016/b978-0-44-453594-8.00006-9

Leinert, J. (2017). Welchen Einfluss hat Financial Literacy auf die Altersvorsorge? Vierteljahrshefte Zur Wirtschaftsforschung, 86 (4), 83–101. https://doi.org/10.3790/vjh.86.4.83

Liang, N., & Weisbenner, S. (2002). Investor behavior and the purchase of company stock in 401(k) plans - The importance of plan design (NBER Working Paper Series, No. 9131). National Bureau of Economic Research. https://doi.org/10.3386/w9131

Lusardi, A., & Mitchell, O. S. (2011). Financial literacy around the world: An overview. Journal of Pension Economics & Finance, 10 (4), 497–508. https://doi.org/10.1017/s1474747211000448

Massa, M., Karlsson, A., & Simonov, A. (2006). Stock market participation and pension reform. SSRN. https://doi.org/10.2139/ssrn.927389

Meister, L., & Schulze, K. (2022). How shocks affect stock market participation (DIW Roundup, No. 142). Deutsches Institut für Wirtschaftsforschung. https://www.diw.de/de/diw_01.c.860745.de/publikationen/roundup/2022_0142/how_shocks_affect_stock_market_participation.html

Mitchell, O. S., & Utkus, S. P. (2002). Company stock and retirement plan diversification (Pension Research Council Working Paper, No. 2002-4). SSRN. https://doi.org/10.2139/ssrn.304461

O’brien, R. M. (2007). A caution regarding rules of thumb for variance inflation factors. Quality and Quantity, 41 (5), 673–690. https://doi.org/10.1007/s11135-006-9018-6

OECD. (2016). OECD/INFE international survey of adult financial literacy competencies [Report]. https://web-archive.oecd.org/2018-12-10/417183-OECD-INFE-International-Survey-of-Adult-Financial-Literacy-Competencies.pdf

OECD. (2017). G20/OECD INFE report on adult financial literacy in G20 countries [Report]. https://www.oecd-ilibrary.org/docserver/04fb6571-en.pdf?expires=1722793674&id=id&accname=guest&checksum=8E4776CCC9F2742BD3AC1F363E66A272

OECD. (2018). OECD/INFE toolkit for measuring financial literacy and financial inclusion [Report]. Retrieved June 26, 2024, from https://www.oecd.org/financial/education/2018-INFE-FinLit-Measurement-Toolkit.pdf

OECD. (2020). OECD/INFE 2020 international survey of adult financial literacy [Report]. https://web-archive.oecd.org/temp/2022-08-09/555847-launchoftheoecdinfeglobalfinancialliteracysurveyreport.htm

OECD. (2021). OECD/INFE report on financial literacy and resilience in APEC economies [Report]. https://web-archive.oecd.org/2021-10-22/614030-Financial-Literacy-and-Resilience-in-APEC-Economies.pdf

OECD. (2022). OECD/INFE toolkit for measuring financial literacy and financial inclusion 2022 [Report]. Retrieved June 26, 2024, from https://www.oecd.org/financial/education/2022-INFE-Toolkit-Measuring-Finlit-Financial-Inclusion.pdf

OECD. (2023c). OECD/INFE 2023 International Survey of Adult Financial Literacy (OECD Business and Finance Policy Papers, No. 39). OECD Publishing. https://doi.org/10.1787/56003a32-en

Ouimet, P., & Tate, G. A. (2017). Learning from coworkers: Peer effects on individual investment decisions. SSRN. https://doi.org/10.2139/ssrn.3072577

Pástor, L., & Veronesi, P. (2012). Uncertainty about Government Policy and Stock Prices. The Journal of Finance, 67 (4), 1219–1264. https://doi.org/10.1111/j.1540-6261.2012.01746.x

Poterba, J. M. (2001). Taxation and portfolio structure: Issues and implications (NBER Working Paper Series, No. 8223 ). National Bureau of Economic Research. https://www.nber.org/papers/w8223

Poterba, J. M., & Samwick, A. A. (1995). Stock ownership patterns, stock market fluctuations, and consumption (Brookings Papers on Economic Activity, No. 2). The Brookings Institution. https://www.brookings.edu/wp-content/uploads/1995/06/1995b_bpea_poterba_samwick_shleifer_shiller.pdf

Poterba, J. M., & Summers, L. (1984). The economic effects of dividend taxation (NBER Working Paper Series, No. 1353). National Bureau of Economic Research. https://doi.org/10.3386/w1353

Prast, H., & Van Soest, A. (2016). Financial literacy and preparation for retirement. Intereconomics, 51 (3), 113–118. https://doi.org/10.1007/s10272-016-0588-9

Sapienza, P., Toldrà-Simats, A., & Zingales, L. (2007). Understanding trust (CRSP Working Paper, No. 621). SSRN. https://doi.org/10.2139/ssrn.1010223

Sias, R., Starks, L., & Turtle, H. (2020). Molecular genetics, risk aversion, return perceptions, and stock market participation (NBER Working Paper Series, No. 27638 ). National Bureau of Economic Research. https://doi.org/10.3386/w27638

Thomas, A., & Spataro, L. (2018). Financial literacy, human capital and stock market participation in Europe. Journal of Family and Economic Issues, 39 (4), 532–550. https://doi.org/10.1007/s10834-018-9576-5

Van Rooij, M., Lusardi, A., & Alessie, R. (2011). Financial literacy and stock market participation. Journal of Financial Economics, 101 (2), 449–472. https://doi.org/10.1016/j.jfineco.2011.03.006

Vittas, D. (1996). Pension funds and capital markets: Investment regulation, financial innovation, and governance (Viewpoint Note, No. 71). The World Bank. https://documents1.worldbank.org/curated/pt/981591468769542423/pdf/multi-page.pdf

Vogel, J., & Dürr, C. (2021). Für eine erfolgreiche Bewältigung des demographischen Übergangs – durch eine gesetzliche Aktienrente und Fachkräfteeinwanderung [Concept paper]. FDP-Bundestagsfraktion. https://www.fdpbt.de/sites/default/files/2021-02/210215_Duerr_Vogel_Gesetzliche_Aktienrente.pdf

Vu, T., Li, C., & Liu, C. (2021). Effects of the financial crisis on household financial risky assets holdings: Empirical evidence from Europe. International Review of Economics & Finance, 71, 342–358. https://doi.org/10.1016/j.iref.2020.09.009

Weisbenner, S., & Ivkovich, Z. (2003). Local does as local is: Information content of the geography of individual investors’ common stock investments (NBER Working Paper Series, No. 9685). National Bureau of Economic Research. https://doi.org/10.3386/w9685

Sources

AssCompact. (2024, July 23). Bundesregierung bestätigt: Altersvorsorgedepot kommt. AssCompact: Fachmagazin Für Risiko- und Kapitalmanagement. https://www.asscompact.de/nachrichten/bundesregierung-bestätigt-altersvorsorgedepot-kommt

Baker, S. R., Bloom, N., & Davis, S. J. (2024). Economic Policy Uncertainty Index [Dataset]. https://www.policyuncertainty.com/index.html

Bundesministerium der Finanzen. (2023, January 13). Christian Lindner gibt Startschuss für das Generationenkapital. https://www.bundesfinanzministerium.de/Content/DE/Standardartikel/Themen/Internationales_Finanzmarkt/Altersvorsorge/startschuss-generationenkapital.html

Bundesministerium der Finanzen. (2024, March 5). Das Generationenkapital. https://www.bundesfinanzministerium.de/Web/DE/Themen/Internationales_Finanzmarkt/Altersvorsorge/Generationenkapital/generationenkapital.html

Deutsche Rentenversicherung. (2024, July 1). Werte der Rentenversicherung: Aktuelle Zahlen aus dem Bereich der Sozialversicherung. Retrieved August 4, 2024, from https://www.deutsche-rentenversicherung.de/DRV/DE/Experten/Zahlen-und-Fakten/Werte-der-Rentenversicherung/werte-der-rentenversicherung.html

European Central Bank. (2023, July 19). Household Finance and Consumption Survey (HFCS): 2021 wave [Survey]. https://www.ecb.europa.eu/stats/ecb_surveys/hfcs/html/index.en.html

European Commission, Directorate-General for Communication (2023a). Flash Eurobarometer FL525: Monitoring the level of financial literacy in the EU (v1.00) [Dataset]. http://data.europa.eu/88u/dataset/s2953_fl525_eng

European Commission, Directorate-General for Communication (2023b). Standard Eurobarometer STD99: Standard Eurobarometer 99 - Spring 2023 (v1.00) [Dataset]. http://data.europa.eu/88u/dataset/s3052_99_4_std99_eng

Global Financial Literacy Excellence Center. (n.d.). Three questions to measure financial literacy. Retrieved August 4, 2024, from https://gflec.org/wp-content/uploads/2015/04/3-Questions-Article2.pdf

Haerpfer, C., Inglehart, R., Moreno, A., Welzel, C., Kizilova, K., Diez-Medrano, J., Lagos, M., Norris, P., Ponarin, E., & Puranen, B. (2022). World Values Survey Wave 7 (2017-2022): Cross-national data-set (v4.0.0) [Dataset]. In WVS Database. World Values Survey Association. https://doi.org/10.14281/18241.18

IZA. (2018). Global preferences survey: Elicit risk, time, and social preferences [Dataset]. In 2012 Gallup World Poll. https://gps.iza.org/home

JASDEC, Japan Securities Dealers Association, & Ministry of Internal Affairs and Communications (Japan). (2023, September 20). Share of the population owning stocks in Japan from fiscal year 2014 to 2022 [Dataset]. In Statista. Retrieved June 22, 2024, from https://www.statista.com/statistics/1445395/japan-share-of-individual-stockholders-in-the-population/

Lloyd’s Register Foundation. (2019). World Risk Poll Data [Dataset]. In World Risk Poll. https://wrp.lrfoundation.org.uk/world-risk-poll-data

Lloyd’s Register Foundation. (2021). Risk indexes: The World Worry Index [Dataset]. In World Risk Poll. https://wrp.lrfoundation.org.uk/risk-indexes

OECD. (n.d.). OECD Better Life Index [Dataset]. In Gallup World Poll. https://www.oecdbetterlifeindex.org/topics/community/

OECD. (2023a). Trust and democratic governance - Government at a glance indicators, 2023 edition [Dataset]. In OECD Data Explorer. https://data-explorer.oecd.org/vis?fs[0]=Topic%2C1%7CGovernment%23GOV%23%7CTrust%20and%20democracy%23GOV_TD%23&pg=0&fc=Topic&bp=true&snb=1&vw=ov&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_GOV_TDG_SPS_GPC%40DF_GOV_TDG_2023&df[ag]=OECD.GOV.GIP&df[vs]=1.0&pd=2021%2C2021&dq=A..TRUST_LE%2BTRUST_CL%2BTRUST_CS%2BTRUST_NG. . ...&ly[cl]=MEASURE&ly[rw]=REF_AREA&to[TIME_PERIOD]=false

OECD. (2023b). Historical population data [Dataset]. In OECD Data Explorer. https://data-explorer.oecd.org/vis?tm=sex%20ratio&pg=0&snb=20&vw=tb&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_POPULATION%40DF_POP_HIST&df[ag]=OECD.ELS.SAE&df[vs]=1.0&pd=2022%2C2022&dq=.POP.PS.M%2BF%2B_T.Y20T64%2B_T.&to[TIME_PERIOD]=false

OECD. (2023d). Wealth distribution database [Dataset]. In OECD Data Explorer. https://data-explorer.oecd.org/vis?tm=wealth&pg=0&snb=25&vw=ov&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_WEALTH%40DF_WEALTH&df[ag]=OECD.WISE.INE&df[vs]=1.0&pd=2016%2C&dq=.A.NW..MEDIAN.&ly[cl]=TIME_PERIOD&ly[rw]=REF_AREA%2CCOMBINED_UNIT_MEASURE&to[TIME_PERIOD]=false

OECD. (2024a). Pensions at a glance: Old-age to working-age ratio [Dataset]. In OECD Data Explorer. https://data-explorer.oecd.org/vis?tm=working-age&pg=0&fs[0]=Measure%2C0%7COld-age%20to%20working-age%20ratio%23OAWAR%23&fc=Measure&snb=1&vw=ov&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_PAG%40DF_PAG&df[ag]=OECD.ELS.SPD&df[vs]=1.0&pd=2022%2C2052&dq=.A.OAWAR. . ..&ly[cl]=TIME_PERIOD&ly[rw]=REF_AREA&to[TIME_PERIOD]=true

OECD. (2024b). Current well-being [Dataset]. In OECD Data Explorer. https://data-explorer.oecd.org/vis?pg=0&snb=2&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_HSL%40DF_HSL_CWB&df[ag]=OECD.WISE.WDP&df[vs]=1.0&pd=2020%2C2021&dq=.5_2_DEP%2B5_2.._T._T._T.&to[TIME_PERIOD]=false&vw=tb&tm=net%20wealth%20per%20capita

OECD. (2024c). Combined (corporate and shareholder) statutory tax rates on dividend income: Net personal tax [Dataset]. In OECD Data Explorer. https://data-explorer.oecd.org/vis?tm=net%20personal%20tax&pg=0&snb=38&vw=ov&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_TAX_CIT%40DF_CIT_DIVD_INCOME&df[ag]=OECD.CTP.TPS&df[vs]=1.0&dq=.A. . .. . .&lom=LASTNPERIODS&lo=1&to[TIME_PERIOD]=false&ly[cl]=MEASURE&ly[rs]=SECTOR&ly[rw]=REF_AREA

OECD. (2024d). Economic Outlook 115 [Dataset]. In OECD Data Explorer. https://data-explorer.oecd.org/vis?fs[0]=Topic%2C1%7CEconomy%23ECO%23%7CEconomic%20outlook%23ECO_OUT%23&pg=0&fc=Topic&bp=true&snb=2&vw=tb&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_EO%40DF_EO&df[ag]=OECD.ECO.MAD&df[vs]=1.1&pd=2015%2C2022&dq=.GDPV_ANNPCT.A&to[TIME_PERIOD]=false

OECD. (2024e). Educational attainment - Regions [Dataset]. In OECD Data Explorer. https://data-explorer.oecd.org/vis?tm=educational%20attainment&pg=0&snb=5&vw=tb&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_REG_EDU%40DF_ATTAIN&df[ag]=OECD.CFE.EDS&df[vs]=1.0&dq=A.CTRY. . ..Y25T64._T..MEAN.&ly[cl]=COMBINED_MEASURE&ly[rw]=COMBINED_REF_AREA&to[TIME_PERIOD]=false&lo=1&lom=LASTNPERIODS

OECD. (2024f). Annual GDP and consumption per capita, US $, current prices, current PPPs [Dataset]. In OECD Data Explorer. https://data-explorer.oecd.org/vis?pg=0&fc=Measure&snb=37&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_NAMAIN10%40DF_TABLE1_EXPENDITURE_HCPC&df[ag]=OECD.SDD.NAD&df[vs]=1.0&dq=A.AUS%2BAUT%2BBEL%2BCAN%2BCHL%2BCOL%2BCRI%2BCZE%2BDNK%2BEST%2BFIN%2BFRA%2BDEU%2BGRC%2BHUN%2BISL%2BIRL%2BISR%2BITA%2BJPN%2BKOR%2BLVA%2BLTU%2BLUX%2BMEX%2BNLD%2BNZL%2BNOR%2BPOL%2BPRT%2BSVK%2BSVN%2BESP%2BSWE%2BCHE%2BTUR%2BGBR%2BUSA. . .B1GQ_POP. . .. . ..&ly[rw]=COMBINED_UNIT_MEASURE%2CREF_AREA&to[TIME_PERIOD]=false&lo=1&lom=LASTNPERIODS&vw=ov&tm=GDP%20per%20capita

OECD. (2024g). Employment rate [Dataset]. In OECD Data Explorer. https://data-explorer.oecd.org/vis?tm=employment%20rate&pg=0&fs[0]=Measure%2C0%7CEmployment%20rate%23EMP_WAP%23&fc=Measure&hc[Topic]=Jobs%20%3E%20Employment&hc[Measure]=Employment%20rate&snb=3&vw=tb&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_LFS%40DF_IALFS_EMP_WAP_Q&df[ag]=OECD.SDD.TPS&df[vs]=1.0&dq=.EMP_WAP. . .Y._T.Y15T64..A&ly[rw]=REF_AREA&to[TIME_PERIOD]=false&lo=1&lom=LASTNPERIODS

OpenAI. (2024). ChatGPT-4o [Software]. https://openai.com/chatgpt/

PwC. (2024). Capital gains tax (CGT) rates [Dataset]. In Worldwide Tax Summaries. https://taxsummaries.pwc.com/quick-charts/capital-gains-tax-cgt-rates#qc-31760ebe-6770a061-445aba74-9a9d0e84-d4a1c12a-2aa29b76-49ea25dd-2cc3f309-a5f408f7-e4ad428c-4485ee21-08fc2c25-8335f40c-102eb896-d89d9935-d93d1c31-7cab83e5-a2e129a2-561fb3a9-32edea04-db06599b-fa07c45c-7b9bc1c3-c4860299-32ef6d0f-bf687ec6-d0d615e9-9abe89e8-2af61a0a-e0ac4430-011e0058-bb3ce427-7a194092-b12ecd5d-f8285483-08d05275-3c42b2a9-5d196309

SHARE-ERIC. (2024a). Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 2 (Release version: 9.0.0) [Dataset]. https://doi.org/10.6103/SHARE.w2.900

SHARE-ERIC. (2024b). Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 6 (Release version: 9.0.0) [Dataset]. https://doi.org/10.6103/SHARE.w6.900

SHARE-ERIC. (2024c). Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 7 (Release version: 9.0.0) [Dataset]. https://doi.org/10.6103/SHARE.w7.900

SHARE-ERIC. (2024d). Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 9 (Release version: 9.0.0.) [Dataset]. https://doi.org/10.6103/SHARE.w9.900

Statistics Canada. (2020). Assets and debts held by economic family type, by age group, Canada, provinces and selected census metropolitan areas, Survey of Financial Security [Dataset]. In Table 11-10-0016-01. https://doi.org/10.25318/1110001601-eng

Statistisches Bundesamt. (2023). Sterbefälle und Lebenserwartung: Entwicklung der Lebenserwartung in Deutschland seit 1871/1881. Retrieved July 11, 2024, from https://www.destatis.de/DE/Themen/Gesellschaft-Umwelt/Bevoelkerung/Sterbefaelle-Lebenserwartung/sterbetafel.html

Statistisches Bundesamt. (2024). 15. koordinierte Bevölkerungsvorausberechnung für Deutschland [Dataset]. Retrieved June 8, 2024, from https://service.destatis.de/bevoelkerungspyramide/index.html

Stiftung Marktwirtschaft. (2024, April 4). Sozialabgabengedenktage 2024: Wie lange müssen wir für die soziale Sicherung arbeiten? https://www.stiftung-marktwirtschaft.de/inhalte/presse-und-aktuelles/pressedetails/sozialabgabengedenktage-2024-wie-lange-muessen-wir-fuer-die-soziale-sicherung-arbeiten/show/News/

The World Bank Group. (2024a). Listed domestic companies, total [Dataset]. In World Federation of Exchanges database. Retrieved May 29, 2024, from https://data.worldbank.org/indicator/CM.MKT.LDOM.NO

The World Bank Group. (2024b). Market capitalization of listed domestic companies (% of GDP) - OECD members [Dataset]. In World Federation of Exchanges database. Retrieved May 29, 2024, from https://data.worldbank.org/indicator/CM.MKT.LCAP.GD.ZS?locations=OE

This paper uses data from SHARE Waves 2, 6, 7, and 9 (DOIs: 10.6103/SHARE.w2.900, 10.6103/SHARE.w6.900, 10.6103/SHARE.w7.900, 10.6103/SHARE.w8.900, 10.6103/SHARE.w9.900) see Börsch-Supan et al. (2013) for methodological details. The SHARE data collection has been funded by the European Commission, DG RTD through FP5 (QLK6-CT-2001-00360), FP6 (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, SHARELIFE: CIT4-CT-2006-028812), FP7 (SHARE-PREP: GA N°211909, SHARE-LEAP: GA N°227822, SHARE M4: GA N°261982, DASISH: GA N°283646) and Horizon 2020 (SHARE-DEV3: GA N°676536, SHARE-COHESION: GA N°870628, SERISS: GA N°654221, SSHOC: GA N°823782, SHARE-COVID19: GA N°101015924) and by DG Employment, Social Affairs & Inclusion through VS 2015/0195, VS 2016/0135, VS 2018/0285, VS 2019/0332, VS 2020/0313 and SHARE-EUCOV: GA N°101052589 and EUCOVII: GA N°101102412. Additional funding from the German Ministry of Education and Research, the Max Planck Society for the Advancement of Science, the U.S. National Institute on Aging (U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, OGHA_04-064, BSR12-04, R01_AG052527-02, HHSN271201300071C, RAG052527A) and from various national funding sources is gratefully acknowledged (see www.share-eric.eu).

[...]

---

¹ The results for Malaysia and Spain in the 2023 report were drawn from samples taken in 2021 using the 2018 toolkit (OECD, 2023c, Annex D, Table 2.1).

² The financial attitudes scores were rescaled from an initial range of 1-5 in previous publications to a new range of 0-4 in the 2023 report (OECD, 2023c, p. 68). In this thesis, however, this rescaling is reversed in order to restore comparability of the 2023 scores with those from previous OECD INFE publications.

³ Knowledge of simple and compound interest is assessed in two separate questions that are combined into one result, which is available in all OECD INFE publications.

⁴ The latest Wave 9 of the SHARE provides SMP data for the following 23 OECD countries: Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Israel, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Poland, Portugal, Slovakia, Slovenia, Spain, Sweden, and Switzerland (SHARE-ERIC, 2024d; Börsch-Supan et al., 2013).

⁵ The 2020 OECD INFE report includes FinLit data for Poland and Germany, while the 2021 report does not.

• Université: University of Mannheim (Business School)
• Note: 1,3
• Auteur: Luis Martin Scherer (Auteur)
• Année de publication: 2024
• Pages: 110
• N° de catalogue: V1661839
• ISBN (PDF): 9783389160060
• ISBN (Livre): 9783389160077
• Langue: anglais
• mots-clé: Financial literacy, stock market participation, hybrid pension systems, pension reform plans, Germany