## Pension Reform and Stock Market Development

## Source details

**Canonical URL:** [Pension Reform and Stock Market Development](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025049-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025049-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025049-print-pdf.pdf.json)

---

### I. Introduction — motivations, channels, and policy framing
- Demographic and macroeconomic motivations:
  - "one in six people in the world by 2050 will be over age 65, up from one in eleven in 2019."
  - Weak social security systems can raise household savings and hinder rebalancing from investment to consumption (examples: China).
  - In many EMDEs, persistent housing boom/bust cycles and resource misallocations are driven by: weak social security systems and underdeveloped domestic capital markets.
- Policy proposal:
  - Promote fully-funded, individually-owned, collectively-managed, and mandatory/incentivized (FICMI) pension schemes (example: 401(k) in the U.S.).
- Three principal contributions:
  - Empirical and theoretical link between FICMI pension schemes and stock market development, with effects on growth, income distribution, and household savings.
  - Identification of frictions (individual impatience, network externalities, coordination failure) that hinder long-term equity investment and how FICMI funds can mitigate them.
  - Practical policy guidance on enabling conditions, fiscal costs of transition, and design choices (including qualitative guidance on contribution rates).
- Major channels through which FICMI affects stock markets and households:
  - Long-term equity capital channel:
    - FICMI funds are patient and large investors; they can internalize positive externalities of equity investments, diversify idiosyncratic risks, and facilitate firm access to long-term equity capital, promoting R&D, innovation, and TFP.
  - Financial inclusion channel:
    - FICMI funds expand indirect household access to equities and broader asset classes.
  - Intertemporal optimization channel:
    - Forced/incentivized savings sub-channel: increases savings where households save too little.
    - Portfolio rebalancing sub-channel: pension funds shift household savings away from bank deposits or housing toward equities/bonds, potentially raising current consumption if income effects dominate substitution effects.
    - Model simulations: up to a threshold, higher FICMI equity investment and higher equity returns are associated with higher current consumption.
- Rationale for government intervention and governance caveat:
  - Network externalities can produce multiple equilibria; government-designed/incentivized FICMI can coordinate economies toward higher-equilibrium long-term equity investment.
  - Risk of excessive political influence in pension investing; robust governance frameworks are crucial.
- Net effect on household savings:
  - FICMI can increase household savings via forced contributions (in under-saving contexts) or decrease non-pension precautionary savings (in over-saving contexts like some East Asian countries), moving economies closer to optimal household savings and improving savings structure via portfolio rebalancing.

*Source: IMF Working Paper (content unit: wpiea2025049-print-pdf).*

### II. Institutional background, definitions, and typical reforms
- Pension financing and design:
  - Financing: PAYG (general revenues or earmarked levies) or fully funded (individual contributions and accumulated assets).
  - Benefit types: Defined Benefit (DB) vs Defined Contribution (DC).
  - Participation: mandatory or voluntary.
- FICMI precise definitions used in the paper:
  - "Fully-funded": accumulated pension assets fully match expected pension payouts for each individual at retirement.
  - "Individually-owned": full property rights to each individual’s pension contributions and investment returns (with regulated withdrawal ages and penalties for premature withdrawal).
  - "Collectively-managed": professional managers manage accumulated contributions as one or multiple collective schemes.
  - "Mandatory/incentivized": contributions mandated by law or strongly incentivized (tax or other regulatory incentives); auto-enrollment can partially satisfy this criterion.
- Typical reform patterns:
  - Shift from PAYG DB to funded DC supplementary pensions; funded pensions are not always privately managed (examples: Singapore, Hong Kong).
- Fiscal implications:
  - Public pension spending can be "one of the largest single public expenditure items," often double-digit shares of GDP; FICMI schemes can involve relatively low fiscal costs compared with PAYG schemes and effectively complement public pensions.

### III. Empirical analysis — data, specification, and main results
- Data and sample:
  - Pension contribution data: OECD annual data 2001–2020 for 36 OECD countries (Ireland and Sweden missing).
  - Final sample: 30 countries after merging controls (GDP_G, DepRate_ST, Pop_G and lags) and trimming bottom/top 1 percent of contribution-to-GDP values.
  - Equity limit data: manually compiled from OECD (2022) Annual Survey (current as of 2021; changes 2002–2021).
  - Bank financing data: Haver; merged dataset contains 30 countries.
- Main empirical specification:
  - Dependent variable: Stock_Mktcap_GDP_Ratio_{i,t} (stock market capitalization-to-GDP ratio, percentage points).
  - Key regressors: Contribution_GDP_Ratio_{i,t} (annual funded/private pension contributions as percent of GDP), lagged contribution, controls Z_{it} (GDP_G, DepRate_ST, Pop_G and lags), country fixed effect α_i, year fixed effect γ_t.
  - Extended specs add EquityLimit_{i,t} and lagged dependent variables; Arellano-Bond (AB) dynamic panel used to mitigate endogeneity.
- Main empirical findings and magnitudes:
  - Contemporaneous funded pension contribution-to-GDP ratio is positively and statistically significantly associated with stock market capitalization-to-GDP across most specifications.
  - Arellano-Bond magnitude: a one-percentage-point increase in funded pension contribution-to-GDP corresponds to a 1.9 percentage point increase in stock market capitalization-to-GDP.
  - After adding equity investment limit, effect: a one-percentage-point increase in contribution-to-GDP corresponds to a 2.1 percentage-point increase in stock market capitalization-to-GDP.
  - Bank financing (Arellano-Bond): a one-percentage-point increase in pension contribution-to-GDP corresponds to a 1.8 percentage point decrease in bank loan balance-to-GDP (p-values around 11 percent without equity limit and 12 percent with it).
  - Equity investment limits generally statistically insignificant (not binding for most OECD countries in the sample).
- Robustness and endogeneity checks:
  - Arellano-Bond, inclusion of lagged dependent variable, alternative equity limit measure (average vs maximum), local projections, COVID dummy (2020) — main associations remain consistent.
  - Local projections produce qualitatively similar results: a one-percentage-point increase in pension contribution-to-GDP is "approximately associated with a two-percentage-point increase in stock market capitalization-to-GDP."
- Summary sample moments (Appendix Table 1; exact values preserved):
  - Observations: 499
  - Year: Mean 2011; Standard Deviation 5.6; Min 2001; Max 2020
  - Stock_Mktcap_GDP_Ratio: Mean 73.4; Standard Deviation 52.3; Min 2.8; Max 393.0
  - Contribution_GDP_Ratio: Mean 2.2; Standard Deviation 2.5; Min 0.0; Max 9.8
  - GDP_G: Mean 4.1; Standard Deviation 4.2; Min -9.8; Max 20.3
  - DepRate_ST: Mean 2.5; Standard Deviation 2.7; Min -0.8; Max 19.6
  - Pop_G: Mean 0.6; Standard Deviation 0.6; Min -0.7; Max 3.0
  - EquityLimit: Mean 88.6; Standard Deviation 20.8; Min 0.0; Max 100.0

*Source: IMF Working Paper (content unit: wpiea2025049-print-pdf).*

### IV. Stylized model — setup, frictions, and welfare implications
- Model structure:
  - Three-period model; FICMI pension fund collects fraction τ ≥ 0 of agent endowment y in period t=1 and allocates fraction η ≥ 0 to risky asset and 1−η to risk-free asset.
  - Risk-free gross return R realized Periods 2 and 3; perceived risky return R̂_3, actual R_3 = R̂_3 + δ with δ > 0 capturing network externalities/coordination failure.
  - Agents choose private savings and portfolio shares; CRRA utility with risk aversion ρ.
- Key friction:
  - Perceived long-term equity return R̂_3 is lower than actual return R_3 by δ > 0, representing coordination failure / network externalities where higher aggregate equity financing raises actual returns.
- Welfare and policy implications:
  - For low τ and η, FICMI yields no welfare gains (non-binding).
  - Excessively large τ or η can reduce welfare (over-forcing savings or excessive equity risk).
  - Intermediate ranges of τ and η exist where FICMI increases lifetime utility by correcting under-saving and excessive conservatism.
  - Larger friction (higher δ) expands the τ range for which FICMI is welfare-improving.
- Extended mechanism:
  - Financial inclusion variant: agents without direct equity access (η̂_t = 0) obtain welfare gains via indirect access through pension funds, improving consumption and welfare.

*Source: IMF Working Paper (content unit: wpiea2025049-print-pdf).*

### V. Case studies — selected country experiences and lessons
- Chile (1981 reform; equity limit relax to 30 percent in 1985):
  - After 30 percent equity allowance in 1985, "surge in stock prices with an average annual growth rate of over 30 percent"; rapid growth stopped in 1994 one year after 30 percent limit was reached.
  - AFPs held about 50 percent of portfolios in equities as of September 2020 — more than double mutual funds’ equity share.
  - COVID pension withdrawals reduced pension assets by "approximately 20 percent of GDP" (combined July 2020, December 2020, April 2021) and may lead to an average 7 percent reduction in retirement pensions for current members.
  - PGU introduced in 2022 covering "90 percent of population aged 65 and older."
- Uruguay (1995 reform; partial transition):
  - Government bonds outstanding held by pension funds rose from "6.5 percent (end-1998) to 14.8 percent (end-2005)."
  - Pension portfolios end-2005: "59.5 percent in government bonds and 0.1 percent in equities" driven by regulation.
  - Pension expenditure reached "12.9 percent of GDP in 2021."
  - 2023 reforms rebalanced mandatory contributions (66.7 percent to BPS PAYG; rest to AFAP FICMI); mandatory AFAP participation extended to all workers from August 1, 2023.
- Kazakhstan (1997 reform; 2013 nationalization):
  - 1997 design required "≥50 percent in government securities" with severe limits on corporate securities; employer payroll tax 15 percent to residual PAYG; employee mandatory contribution 10 percent.
  - 2013 nationalization consolidated funds into a single pension savings fund managed by the central bank; state-directed use reduced market liquidity and raised governance concerns.
- Singapore (Central Provident Fund, CPF):
  - CPF contributions phased from "5 percent initially to around 25 percent by late 1970s"; as of 2022 total contribution rate "37 percent for workers up to age 55 (17 percent employers, 20 percent employees)."
  - CPF uses contributions to purchase Special Singapore Government Securities (SSGS); complementary PAYG-style social programs for vulnerable groups exist.
- China (dominant PAYG pillar; limited funded presence):
  - UN (2023) projection: total population started declining in 2022 and projected to decline by "about 8 percent to 1,313 million in 2050 (relative to 2023)."
  - Zheng (2019) projection: support ratio decline from "under 2.65 workers per retiree in 2019 to 1.03 in 2050."
  - Funded pension assets end-2021: "4.43 trillion RMB (2.64 trillion RMB Enterprise Annuity; 1.79 trillion RMB Occupational Annuity) ≈ 3.9 percent of 2021 GDP."
  - OECD comparison: funded/private pension plans accumulated "$58.9 trillion at end-2021 ≈ 100.5 percent of OECD 2021 GDP."
  - Payout composition 2021: "Pillar I accounts for 98.9 percent of total pension payouts."
  - Household assets 2019: "real estate 59.1 percent; bank deposits 8.0 percent."
  - Stock market: Shanghai Composite roughly "around 3,000 over last 15 years"; median annual turnover ratio "207 percent during 2001–2020 vs OECD median 61 percent."
- United States (401(k) experience):
  - Tax incentives and automatic enrollment effective: Vanguard (2023) reports participation rate in 401(k) plans reached "83 percent" and automatic enrollment adoption tripled ("76 percent of plans with at least 1,000 participants have automatic enrollment").
  - Regulatory oversight (ERISA) and plan design innovations (Target-Date Funds, catch-up contributions for workers aged 50+) are important.
  - Distributional concerns: tax incentives and employer matching benefit higher-income individuals disproportionately.

*Source: IMF Working Paper (content unit: wpiea2025049-print-pdf).*

### VI. Enabling conditions, implementation considerations, and operational measures
- Enabling conditions emphasized:
  - Strong and lasting political commitment and institutional monitoring arrangements.
  - Well-designed fiscal strategy: mix of debt issuance, tax adjustments, and expenditure reprioritization to avoid pension funds crowding out private investment via government bond issuance.
  - Sound financial infrastructure: payments, custodial services, transparent accounting, independent auditors.
  - Governance safeguards to avoid politically directed investments, particularly where SOEs dominate.
  - Pension reform can help create capital markets; deep capital markets need not be a prerequisite.
- Further implementation considerations:
  - Voluntary vs mandatory pillars: OECD (2018) finds financial incentives effective; total fiscal cost typically in low single digits of GDP; country-specific analysis needed.
  - Government subsidies/bailouts may stabilize crises but carry fiscal costs.
  - Diverting household savings away from SOEs/real estate can raise SOE financing costs and implicit fiscal costs but may reallocate resources toward private-sector investment.
  - Practical measures to expand FICMI coverage: public awareness, financial literacy, transparent regulation, flexible pension products, digital platforms, addressing informality, and strengthening accounting/auditing.

*Source: IMF Working Paper (content unit: wpiea2025049-print-pdf).*

### VII. Conclusion — aggregated findings and policy recommendations
- Empirical and theoretical summary (exact magnitudes preserved where reported):
  - Funded/private pension (FICMI) development is positively associated with stock market capitalization-to-GDP; robust to dynamic AB estimation and robustness checks.
  - A one-percentage-point increase in funded pension contribution-to-GDP linked to:
    - "~1.9 percentage point increase in stock market capitalization-to-GDP (Arellano-Bond specification)."
    - "~2.1 percentage point increase in some specifications (after adding equity limit)."
    - "~1.8 percentage point decrease in bank loan balance-to-GDP (Arellano-Bond bank financing results)."
  - Stylized model: FICMI can correct coordination failure and under-investment in long-term equity, improving welfare for a range of contribution rates and equity allocations; excessive τ or η can reduce welfare.
  - Case studies demonstrate outcomes depend on design, fiscal strategy, investment limits, and governance.
- Five high-level policy considerations:
  1. FICMI pillar must be sufficiently large to realize benefits.
  2. Implement vigorous structural reforms to meet enabling conditions (fiscal strategy, financial infrastructure).
  3. Act promptly, decisively, and persistently with strong political commitment.
  4. Younger nations and some low-income EMDEs can benefit from early development of FICMI pillars due to lower current burdens of supporting older generations.
  5. For emerging markets (e.g., China), pension reform aiding stock market development is important to provide long-term equity financing for technological innovation.
- Specific recommended measures:
  - Re-optimize pillar composition to increase funded pillar weight relative to PAYG.
  - Prefer FICMI design with clear property rights; combine mandatory participation with tax and non-tax incentives as appropriate.
  - Multi-pronged transition strategy: debt issuance, tax adjustments (income, capital, consumption), expenditure re-prioritization; consider alternatives to simply raising retirement age.
  - Strengthen governance and safeguards; expand coverage via education, financial literacy, transparency, digital enrollment, and measures to reduce informality.
- Future research directions:
  - Policy design for setting contribution rates and replacement ratios.
  - Country-specific mixes for transition to FICMI.
  - Optimal mix between non-contributory minimum pensions and FICMI schemes.

*Source: IMF Working Paper (content unit: wpiea2025049-print-pdf).*

### introduction of FICMI pension schemes can impact household savings in two ways: first, FICMI pension

### I. INTRODUCTION

### A. Motivations and Summary of Main Channels
- Demographic and macroeconomic motivation:
  - According to the 2019 World Population Prospects, one in six people in the world by 2050 will be over age 65, up from one in eleven in 2019.
  - In countries like China, weak social security systems can lead to a higher level of household savings, posing challenges to rebalancing from investment to consumption (IMF, 2022a; IMF, 2023a).
  - In a number of emerging market and developing economies (EMDEs), two critical drivers of persistent housing boom/bust cycles and resource misallocations are: weak social security systems (necessitating increased household savings for retirement) and underdeveloped domestic capital markets (particularly stock markets), which result in a lack of attractive non-housing financial products and lead to savings being disproportionately allocated to housing (see Bayoumi and Zhao, 2021).

- Policy proposal:
  - Promote the development of fully-funded, individually-owned, collectively-managed, and mandatory/incentivized (FICMI) pension schemes (example: the 401(k) scheme in the U.S.).

- Three principal contributions of the paper:
  - Empirical and theoretical establishment of a strong connection between developing FICMI pension schemes and developing stock markets, affecting economic growth, income distribution, and household savings.
  - Highlighting challenges of individual impatience, network externalities, and coordination failure in long-term equity investments and stock market development, and showing how FICMI pension funds can ameliorate these issues.
  - Practical policy guidance on enabling conditions, fiscal costs of transition, and design choices for pension reform, including qualitative guidance on optimal contribution rates.

- Major channels through which FICMI pension schemes affect stock market development and household outcomes:
  - Long-term equity capital channel:
    - FICMI pension funds are patient investors (steady cash inflows from pension contributions and relatively predictable/remote cash outflows for future pension payments) and large investors that can (partially) internalize positive externalities of equity investments and diversify idiosyncratic risks.
    - This makes it easier for firms to raise long-term equity capital through capital markets, promoting R&D, innovation, technological advancement, enhancing TFP, and boosting economic growth.
    - The paper focuses on the impact of FICMI pension funds on stock market development, given extensive literature on stock market development’s impact on the macroeconomy.
  - Financial inclusion channel:
    - FICMI pension funds improve access to stock markets and expand investment options for households with limited direct access to corporate equities and government/corporate bonds.
    - By indirectly investing in a broader set of asset classes through FICMI pension funds, such households can benefit more broadly from future economic growth and higher TFP generated by the corporate channel.
  - Intertemporal optimization channel (two sub-channels):
    - Forced/incentivized savings sub-channel:
      - For households with insufficient savings (such as households in many EMDEs), FICMI pension schemes will force/incentivize higher savings and turn households into indirect investors in stock markets, moving savings closer to the optimal level and increasing supply of long-term patient capital.
    - Portfolio rebalancing sub-channel:
      - FICMI pension funds, as patient institutional and collective managers, can promote healthy stock market development, increase attractiveness and access to corporate equities and government/corporate bonds, and reduce over-reliance on bank deposits or housing as savings vehicles.
      - This can produce higher or more sustainable investment returns and reduce the amount of savings needed to reach future consumption targets (in cases where income effect dominates substitution effect, as detailed in the modeling section), freeing up resources for immediate consumption.
      - For countries with excessive household savings (such as some countries in East Asia), this sub-channel could decrease savings and increase current consumption.
      - Model simulations indicate that, upon introducing a FICMI pension scheme and up to a certain threshold, the more a FICMI pension fund invests in corporate equities, the higher current consumption would be; furthermore, the greater the return on these equities, the more pronounced the consumption-boosting effect becomes.

- Frictions and coordination rationale for government intervention:
  - Individual impatience and information/transaction costs make equity investments unattractive for small retail investors unless funds are pooled for adequate diversification.
  - Mutual funds and other non-pension collective schemes could play a coordination role, but redemption pressure and investor impatience can force them to behave impatiently, limiting their effectiveness.
  - FICMI pension funds, by investing collectively and patiently, can alleviate network externalities, coordination failure, and individual impatience in equity investments.
  - The paper offers another rationale for government intervention: due to network externalities of long-term equity investments, multiple equilibria exist in a country’s long-term equity investment level, and the “good” equilibrium is not robust to noisy signals and misperceptions. Government policies designing, implementing, and/or incentivizing FICMI pension schemes can serve as an important coordination device to move the country to a higher-equilibrium level of long-term equity investments, enhancing growth potential and income distribution.

- Governance caveat:
  - Excessive government intervention in pension design and operation could lead to governance issues (e.g., directing pension savings into politically favored investments).
  - Establishing a robust governance framework (e.g., prohibiting political influence over specific investment products) is crucial; these issues are explored further in case studies and policy implications.

- Practical policy considerations:
  - The paper discusses enabling conditions and strategies to manage fiscal costs associated with transition to funded pension schemes.
  - Link to domestic resource mobilization: It is difficult to imagine raising $2.2 trillion of domestic resources every year in EMDEs without developing domestic capital markets; pension reform can help develop domestic capital markets and mobilize private savings.
  - Model results indicate that an excessively high contribution rate can lead to suboptimal outcomes, providing qualitative guidance for determining an optimal contribution rate.

- Focus on stock market development rationale:
  - For young and innovative firms, debt is a poor substitute for equity finance due to information problems, skewed and highly uncertain returns, and lack of collateral value; R&D and innovation require equity finance (Brown, Fazzari, and Petersen, 2009).
  - Some countries (e.g., China) show bond markets can develop with bank support while stock markets cannot.
  - Pension reform also influences bond market development and could bring financial stability benefits (topic left for future research).

- Net effect on household savings:
  - FICMI pension schemes can increase household savings via the forced/incentivized savings channel (where households save too little without FICMI pension, as in many EMDEs) or decrease household savings and increase household consumption by reducing non-pension savings and decreasing precautionary savings (where households save too much without FICMI pension, such as in China).
  - In both cases, FICMI pension schemes can help move the economy closer to the optimal level of household savings and improve the structure of such savings through the portfolio rebalancing sub-channel.

### B. Institutional Background and Terminology
- Pension financing and design:
  - Pension payments can be financed by general government revenues or earmarked levies on a pay-as-you-go (PAYG) basis, or they can be fully funded through individual contributions and assets accumulated from such contributions.
  - Pensions can be defined benefit (DB) with pre-defined payments according to a formula, or defined contribution (DC) where payments depend on accumulated assets while contributions are pre-defined.
  - Pension participation can be mandatory or voluntary.

- Typical reforms and FICMI characterization:
  - Many countries have historically relied on PAYG DB pension plans administered by public agencies; rising longevity and declining worker-to-retiree ratios threaten PAYG DB sustainability.
  - In response, many countries have introduced supplementary pensions—typically funded, defined contribution pensions with individual ownership and professional fund management (public or private)—which are often FICMI schemes.
  - Note: Funded pensions are not always privately managed (examples: Singapore and Hong Kong).

- Fiscal implications:
  - IMF (2022b) finds public pension spending is often one of the largest single public expenditure items, often representing a double-digit share of GDP; even relatively small and gradual changes can materially impact public finances in the long run.
  - FICMI pension schemes involve relatively low fiscal costs compared with PAYG schemes and can effectively complement public pension systems.
  - The paper discusses strategies to handle fiscal costs of transition to funded pensions.

*Recommended citation: Khan, Shujaat, Bo Li, and Yunhui Zhao. 2025. “Pension Reform and Stock Market Development,” IMF Working Paper No. 25/49, International Monetary Fund, Washington, D.C.*

### Chapter 1).

### Chapter 1)

### Pension system features and definitions
- The IMF’s ongoing model refinements for the external balance assessment (EBA) also examine the impact of various pension system features on the current account gap.
- Description of FICMI pension schemes:
  - Not supplementary or complementary to PAYG DB public pension system; instead, they are the primary (or the only) pension available. Examples include Singapore, Chile, and Hong Kong.
  - In other countries like the U.S., FICMI pension schemes (such as 401(k)) play a very important role in supporting stock market development and also providing a significant source of retirement security.
- Precise definitions as used in the paper:
  - “Fully-funded”: for each individual, upon retirement, accumulated pension assets fully match expected pension payouts; expected pension payments will come entirely from such accumulated pension assets, without the expectation for any intra- or inter-generational transfer or any government intervention.
  - “Individually-owned”: full property rights to each individual’s pension contributions and related investment returns, except that governments can regulate investment options and mandate withdrawals only after the individual reaches a certain retirement age; households would not be able to withdraw from FICMI prematurely without some penalty.
  - “Collectively-managed”: one or a number of professional investment managers, possessing the necessary qualifications and governed by sound practices, who manage the accumulated pension contributions either as one pension fund or as multiple collective investment schemes. This contrasts with private retirement savings, where individuals are allowed to have full control over their investments.
  - “Mandatory/incentivized”: pension contributions are mandated by law, or strongly incentivized by tax benefits or other government regulations. Design of auto-enrollment can be regarded as partially satisfying this criterion in the sense that it mandates participation unless an explicit decision is made to opt out.
- Auto-enrollment:
  - Auto-enrollment is one of the successful initiatives in recent decades with a significant impact on supplementary pension savings, as illustrated by the U.S. example in the subsequent case study section.
  - Footnote on auto-enrollment design: aims to leverage behavioral economic principles, particularly inertia. Many people fail to enroll in pension schemes due to procrastination, lack of knowledge, or the perceived complexity of making such decisions. Auto-enrollment simplifies the decision-making process by defaulting employees into saving for retirement, while still allowing them the freedom to choose not to participate if they decide it is not in their best interest.

### Selected literature review — four literature strands
- Broad relation to four strands of literature:
  1. Capital structure literature:
     - Modigliani–Miller Theorem (Modigliani and Miller, 1958): in the absence of taxes, bankruptcy costs, agency costs, and asymmetric information, and in an efficient market, the enterprise value of a firm is unaffected by how that firm is financed.
     - Trade-off Theory (Kraus and Litzenberger, 1973): optimal share of debt financing depends on trade-off between dead-weight costs of bankruptcy and tax-saving benefits of debt.
     - Pecking Order Theory (Myers and Majluf, 1984): firms prefer internal financing, then debt, and lastly raise equity as a “last resort”; grounded on asymmetric information.
     - Empirical evidence for both theories is mixed; Fama and French (2002) criticize both theories. Brealey, Myers, and Allen (2008) argue Pecking Order Theory does not apply to high-tech industries where equity financing is preferable due to high cost of debt as assets are intangible.
     - Relevance: main mechanism in the model involves the impact of equity financing (and pension schemes) on R&D.
  2. Equity financing and innovation:
     - Brown, Fazzari, and Petersen (2009): young, high-tech, publicly-traded firms finance R&D almost entirely through internal cash flow and external equity markets; information problems, skewed and highly uncertain returns, and lack of collateral make debt a poor substitute for equity finance.
     - Implication: well-calibrated pension fund reforms can boost the supply of external equity and alleviate financing constraints.
     - Supporting evidence:
       - Aghion et al. (2004): UK firms reporting positive R&D are more likely to use equity finance than firms reporting no R&D.
       - Garcia-Macia (2013): innovation intensity predicts lower debt financing; equity financing preferentially used by innovation-intensive firms lacking internal funds.
       - Hsu, Tian, and Xu (2014): using data on 32 developed and emerging countries, industries more dependent on external finance and more high-tech intensive exhibit higher innovation levels in countries with better-developed equity markets.
     - Footnote: Acharya and Xu (2017) using US data from 1976 to 2006 find public firms in external finance-dependent industries spend more on R&D and generate a better patent portfolio than private counterparts. Kerr and Nanda (2015) provide a comprehensive review of financing innovation literature.
  3. Pension reforms literature:
     - Amaglobeli et al. (2019): “countries with an enabling macroeconomic and legal environment and sufficiently developed financial systems could consider complementing the public pension scheme with a funded defined contribution scheme,” and the ability of households to save for retirement and diversify retirement-related risks will depend on availability of a wide array of relevant financial products.
     - The paper emphasizes the mutually reinforcing feedback between development of FICMI pensions and that of the stock market.
     - Fouejieu et al. (2021): examine Europe’s reforms on public pension systems and present a novel measure of the balance between pension benefits and contributions.
     - Related literature on transition from PAYG to a funded system: Huang, Imrohoroglu, and Sargent (1997); Rudolph and Rocha (2009); Kudrna, Tran, and Woodland (2019); Heer, Polito, and Wickens (2020).
     - Footnote: Batog et al. (2019) focusing on Central, Eastern, and Southeastern Europe find fiscal pressures from rapidly-shrinking working-age populations could displace public or private investment through increases in taxation and spending compression, or if accommodated by increased public debt and higher borrowing costs.
  4. Patient and long-term capital literature:
     - Cremers and Petajisto (2009): funds with a high “active share” in the U.S. tend to outperform benchmarks.
     - Cremers and Pareek (2016): outperformance is stronger among high-active-share funds identified as “patient”, meaning with holding durations of over two years.
     - Ivashina and Lerner (2019): despite abundance of capital and substantial need for returns, long-term investments have been problematic; they outline key challenges and suggest ways to address them.
     - Beetsma et al. (2024): using a comprehensive Danish dataset, find that following a pension fund equity investment, firm productivity experienced a substantial increase, averaging between 3 percent and 5 percent; effect particularly pronounced for unlisted and small firms. Additionally, pension funds tend to invest for longer periods than other institutional investors, such as private equity. They interpret findings as consistent with pension funds engaging in long-term financing commitments and enabling firms to make productivity-enhancing investments.
     - Footnote: Cremers, Fulkerson, and Riley (2019) review two decades of literature on actively managed mutual funds and conclude conventional wisdom is too negative on value of active management.

### Key methodological and structural notes
- The rest of the paper is structured as follows:
  - Section II conducts empirical analysis using OECD data;
  - Section III presents a stylized model and its extension to illustrate the main propositions.

*Source: wpiea2025049-print-pdf - Chapter 1).*

### Section IV presents several country case studies, which also cover non-OECD countries to

### wpiea2025049-print-pdf - Section IV presents several country case studies, which also cover non-OECD countries to

### III. EMPIRICAL ANALYSIS — Approaches and Specification
- Main empirical approach: cross-country panel regression; complementary method: local projections (see Appendix 1).
- Primary regression specification (stock market capitalization-to-GDP ratio on pension contributions):
  - Dependent variable: Stock_Mktcap_GDP_Ratio_{i,t} — stock market capitalization-to-GDP ratio in country i and year t, in percentage points.
  - Key regressors:
    - Contribution_GDP_Ratio_{i,t} — annual contribution to funded and private pension schemes (FICMI schemes), as a percent of GDP (excludes PAYG pensions and social security taxes).
    - L.Contribution_GDP_Ratio_{i,t} — one-period lag of contribution ratio.
    - Z_{it} — controls: real GDP growth (GDP_G), short-term deposit rate (DepRate_ST), population growth (Pop_G), and first lags of these variables.
    - α_i — country fixed effect; γ_t — year fixed effect; ε_{i,t} — error term.
  - Motivation: use contributions (flows) rather than pension fund assets (stocks) to mitigate reverse causality because pension funds commonly invest in equities.

- Extended specification including investment regulation:
  - Adds EquityLimit_{i,t} — maximum share of pension portfolios allowed in equities; L.EquityLimit_{i,t} — lagged term.
  - In some specifications include lagged dependent variable to account for persistence and potential reverse causality.
- Standard errors: regressions run with and without clustering at the country level to address serial correlation within countries.
- FE vs RE: both fixed-effect and random-effect models estimated; Hausman test generally favors fixed-effect (low p-values).

### III. EMPIRICAL ANALYSIS — Data
- Pension contribution data source: OECD; annual data from 2001 to 2020 for 36 OECD countries.
  - Ireland and Sweden are the only two OECD countries that do not have this data.
- Sample construction:
  - Use contribution data only for funded pensions.
  - Drop bottom 1 percentile and top 1 percentile values of pension contribution-to-GDP data to reduce influence of extremes (does not reduce number of countries with available data).
  - After merging controls from IMF’s WEO database (GDP_G, DepRate_ST, Pop_G and lags), final sample: 30 countries with available data.
- Equity limit data: manually compiled from OECD (2022)’s Annual Survey of Investment Regulation of Pension Funds and Other Pension Providers (2022), covering current (as of 2021) portfolio limits and major changes during 2002-2021.
- Bank financing data: from Haver; final dataset after merging also contains 30 countries.

### III. EMPIRICAL ANALYSIS — Main Results and Magnitudes
- Stock market capitalization:
  - Contemporaneous funded pension contribution-to-GDP ratio exhibits a positive and statistically significant association with stock market capitalization-to-GDP across most specifications (including year dummies and clustered standard errors).
  - Lagged pension contribution-to-GDP ratio is not statistically significant (one-year lag may be long enough for funds to adjust allocations).
  - Arellano-Bond (AB) dynamic panel (addressing reverse causality):
    - Results qualitatively similar to baseline.
    - Magnitude: a one-percentage-point increase in funded pension contribution-to-GDP ratio corresponds to a 1.9 percentage point increase in the stock market capitalization-to-GDP ratio (Arellano-Bond specification).
    - Interpretation: effect more than one-for-one; potential channels include equity investments by pension funds improving market functionality, prompting equity issuance and household equity investment (positive spillovers).
- Equity investment limits:
  - Generally statistically insignificant across specifications, except Column (4) where lagged equity investment limit is positively correlated with stock market capitalization-to-GDP.
  - Inclusion of equity investment limit leaves effect of contribution-to-GDP ratio almost unchanged: a one-percentage-point increase in contribution-to-GDP corresponds to a 2.1 percentage-point increase in stock market capitalization-to-GDP (reported result after adding equity limit).
  - Suggests equity investment limits were not binding constraints for most OECD countries over much of the sample.
- Bank financing (Arellano-Bond results):
  - Focus on bank loan balance-to-GDP ratio as left-hand-side variable.
  - Columns (4) and (8) (which control for year dummies) show a negative correlation between funded pension contributions and bank-loan-to-GDP ratio.
    - Coefficient magnitudes: a one-percentage-point increase in pension contribution-to-GDP corresponds to a 1.8 percentage point decrease in bank loan balance-to-GDP (Columns cited).
    - Statistical significance: p-values around 11 percent without equity investment limit and 12 percent with it (i.e., almost at the 10 percent level).
  - Equity investment limit remains statistically insignificant in the bank financing regressions.
  - Interpretation: funded pension development associated with reduced reliance on bank loans and increased importance of equity financing; household savings shift from bank deposits toward equities (improved saving composition given typically higher stock returns).

### III. EMPIRICAL ANALYSIS — Endogeneity and Robustness
- Endogeneity / reverse causality:
  - Arellano-Bond estimation used to mitigate endogeneity by differencing and instrumenting with lagged levels.
  - Including lagged dependent variable reduces magnitude of main variable but positive association remains.
- Robustness checks:
  - Alternative measure of equity investment limit: simple average across multiple limits (instead of maximum) — results qualitatively and quantitatively similar.
    - Reported comparisons: a one-percentage-point increase in pension contribution-to-GDP corresponds to a 2.1 percentage-point increase in stock market capitalization-to-GDP in both Table 3 (Column (4)) and Appendix Table 1 (Column (4)).
    - For bank financing, the 1.8 percentage point decrease in bank loan-to-GDP is consistent across Table 4 (Column (8)) and Appendix Table 5 (Column (4)).
  - Local projections (Approach #2) produce qualitatively similar results (Appendix Figures 1–5 referenced).
  - COVID dummy (2020): including a COVID dummy does not change coefficients or significance of main variables (coefficients remain the same as without COVID dummy); likely year dummies already capture COVID effects.
  - Combined checks (average equity limit + COVID dummy): main variables remain consistent.
- Standard errors: all regressions performed with and without clustering at country level.

### IV. STYLIZED MODEL — Benchmark and Mechanisms
- Model setup:
  - Three-period model with FICMI pension fund that collects fraction τ ≥ 0 of agent’s endowment y in period t=1 and allocates fraction η ≥ 0 to risky asset (equity) and 1−η to risk-free asset (debt).
  - Risk-free gross return denoted R realized at beginning of Period 2 and Period 3.
  - Risky asset return perceived by agent is R̂_3 while actual return is R_3 = R̂_3 + δ with δ > 0 (captures network externalities/coordination failure).
  - Agents choose private savings shares and portfolio shares for periods 1 and 2; consumption in period 3 depends on compounded returns of risky and risk-free investments.
  - CRRA utility with risk aversion ρ assumed.
  - Model abstracts from bank financing on corporate side; firm finances via bonds and equities; assume no default for simplicity.
- Key friction: perceived return on long-term equity (R̂_3) is lower than actual return (R_3) by δ > 0, reflecting coordination failure and network externalities where higher aggregate equity financing can unlock higher actual returns.
- Welfare implications:
  - For low τ and η (low contributions and conservative pension equity share), FICMI yields no welfare gains relative to no-pension case (non-binding).
  - Excessively large τ or η can reduce welfare (over-forcing savings or excessive equity risk).
  - There exists an intermediate range of τ and η where FICMI increases lifetime utility by correcting under-saving and excessive conservatism due to perceived low equity returns.
  - Larger friction (higher δ) expands the range of τ for which the pension scheme improves welfare.
- Channels emphasized:
  - Forced/incentivized savings: pension contributions raise aggregate long-term capital.
  - Portfolio rebalancing: pension funds allocate to equities, correcting under-investment in long-term assets.
  - Financial inclusion (extended model): agents without direct access to equities (η̂_t = 0) benefit from indirect access via pension funds, improving welfare and potentially boosting consumption.

### IV. CASE STUDIES — Key country experiences and lessons
- Chile (1981 reform; equity limit relaxed in 1985 to 30 percent):
  - After allowing up to 30 percent equities in 1985, Chile experienced a surge in stock prices with an average annual growth rate of over 30 percent; rapid growth stopped in 1994 one year after 30 percent limit was reached.
  - Pension Fund Associations (AFPs) held about 50 percent of portfolios in equities as of September 2020 — more than double mutual funds’ equity share.
  - Lessons:
    - Complement pension reform with fiscal adjustments to avoid excessive government bond issuance crowding out private sector investments.
    - Transition toward multi-pillar system in 2008 (introduction of tax-funded solidary pension system, SPS) to protect vulnerable groups.
    - Challenges: low contribution rates, high informality, low job tenure, and large pension withdrawals during COVID that reduced pension assets by approximately 20 percent of GDP (combined withdrawals in July 2020, December 2020, and April 2021) and may lead to an average 7 percent reduction in retirement pensions for current members.
    - Introduction of minimum guaranteed pension (PGU) in 2022 covering 90 percent of population aged 65 and older.

- Uruguay (1995 reform; partial transition, multi-pillar from the outset):
  - Participants below 40 were required into FICMI mandatory private individual accounts; older cohorts offered choice.
  - High concentration in government bonds after reform: share of government bonds outstanding held by pension funds rose from 6.5 percent (end-1998) to 14.8 percent (end-2005); pension portfolios: 59.5 percent in government bonds and 0.1 percent in equities (end-2005) driven by regulation (40–60 percent to government securities).
  - Lessons:
    - Transition fiscal costs higher than expected due to shifts in member choices and economic crises (Uruguay 2002).
    - Pension expenditure reached 12.9 percent of GDP in 2021; 2023 reforms raised retirement age and rebalanced mandatory contributions between PAYG and FICMI (66.7 percent to BPS PAYG; rest to AFAP FICMI); mandatory AFAP participation extended to all workers from August 1, 2023.

- Kazakhstan (1997 reform; 2013 nationalization):
  - 1997 reform created a FICMI pillar with mandatory employee contribution (10 percent) and employer payroll tax (15 percent to residual PAYG); strict investment regulations required ≥50 percent in government securities and severe limits on corporate securities (state funds not allowed in corporate securities).
  - 2013 nationalization consolidated private pension funds into a single pension savings fund managed by the central bank; additional pension taxes imposed; retirement age increases implemented.
  - Lessons:
    - Tight portfolio restrictions produced over-concentration in government securities, limiting capital market development.
    - Favorable external conditions (commodity boom) reduced government debt issuance, limiting domestic investment opportunities and pushing funds to foreign assets with potential negative local-currency returns.
    - Consolidation and state-directed use of pension assets reduced domestic market liquidity and raised governance/conflict-of-interest concerns; authorities sought to remove UAPF from central bank control since December 2015.

- Singapore (Central Provident Fund, CPF; established 1955, expanded 1968 onward):
  - CPF operates as a funded pension system with individual accounts and multiple sub-accounts (Ordinary, Special, Medisave) and supplements (Silver Support Scheme in 2016).
  - CPF uses contributions to purchase Special Singapore Government Securities (SSGS); proceeds go into Government’s past reserves.
  - Lessons:
    - Phased contribution increases: from 5 percent initially to around 25 percent by late 1970s; as of 2022 total contribution rate was 37 percent for workers up to age 55 (17 percent employers, 20 percent employees).
    - Compulsory savings, individual property rights, diversified investment options (CPF Investment Scheme), and complementary PAYG-style social programs for vulnerable groups.

- China (limited funded pension presence; dominant PAYG pillar):
  - Demographics: UN (2023) projection — total population started declining in 2022 and projected to decline by about 8 percent to 1,313 million in 2050 (relative to 2023). Zheng (2019): support ratio likely to decrease from under 2.65 workers per retiree in 2019 to 1.03 in 2050.
  - BOAI (Basic Old Age Insurance) PAYG pillar projected to have flow deficit in 2028 and stock deficit in 2035 under Zheng (2019) baseline.
  - Funded pension assets at end-2021: 4.43 trillion RMB (2.64 trillion RMB Enterprise Annuity; 1.79 trillion RMB Occupational Annuity) ≈ 3.9 percent of 2021 GDP.
  - Comparison: OECD funded/private pension plans accumulated $58.9 trillion at end-2021 ≈ 100.5 percent of OECD 2021 GDP.
  - Payout composition in 2021: Pillar I accounts for 98.9 percent of total pension payouts (excludes Occupational Annuity payouts due to data gaps).
  - Compliance issues: widespread underpayment of social insurance premiums; in many regions real contribution rate much lower than statutory rate; 70 percent firms paid less than required rate in cited evidence.
  - Household asset composition (2019): real estate 59.1 percent of household assets; bank deposits 8.0 percent.
  - Stock market features: Shanghai Composite index roughly around 3,000 over last 15 years despite large GDP growth; median annual turnover ratio 207 percent during 2001–2020 vs OECD median 61 percent — indicating prevalence of short-term trading and lack of long-term institutional investors.

- United States (401(k) system since 1981):
  - 401(k) as FICMI example: fully funded employee accounts often with employer matching and tax incentives.
  - Lessons:
    - Tax incentives effective at encouraging participation; participation rates increased with automatic enrollment and plan design innovations.
    - Automatic enrollment adoption increased after Pension Protection Act (2006); Vanguard (2023) reports participation rate in 401(k) plans reached 83 percent and automatic enrollment adoption tripled (76 percent of plans with at least 1,000 participants have automatic enrollment).
    - Regulatory oversight important (e.g., ERISA) to maintain fiduciary standards.
    - Plan design innovations: Target-Date Funds; catch-up contributions for workers aged 50+.
    - Distributional concerns: tax incentives and employer matching benefit higher-income individuals disproportionately.

### V. ENABLING CONDITIONS AND IMPLEMENTATION CONSIDERATIONS
- Enabling conditions emphasized:
  - Strong and lasting political commitment to reform; institutional arrangements for monitoring (e.g., Chile’s Capital Markets Committee).
  - Well-designed fiscal strategy for transition: mix of debt issuance, tax adjustments, and expenditure reprioritization to avoid vicious cycle of pension funds over-investing in government bonds.
    - Note: McGrattan and Prescott (2017) highlight a “Laffer-like effect” where reducing payroll tax rate can boost labor supply and raise tax revenue, aiding transition.
  - Sound financial infrastructure: payments system, custodial services, transparent accounting and valuation, independent skilled auditors.
  - Governance safeguards to prevent government-directed investment choices and to mitigate conflicts of interest (especially where SOEs are dominant).
  - Do not treat deep capital markets as a prerequisite; pension reform can help create a virtuous cycle between pensions and capital market development.
- Further considerations:
  - Voluntary vs mandatory pillars and financial incentives:
    - OECD (2018) finds financial incentives effective; total fiscal cost typically in low single digits of GDP.
    - Country-specific analysis needed to assess desirability of mandatory versus tax-incentivized arrangements.
  - Government subsidies/bailouts: potential role in crisis periods to stabilize contributions and corporate equity financing, but fiscal costs must be weighed.
  - More complete financial markets (additional saving channels) may reduce the quantitative role of pension funds in consumption smoothing but not qualitatively change main results.
  - Diverting household savings away from SOEs/real estate could raise SOE financing costs and implicit fiscal costs but can also reallocate resources toward more productive private-sector investment.
  - Practical measures to expand FICMI coverage: public awareness, financial literacy, transparent regulation, flexible pension products, digital platforms, addressing informality, and strengthening accounting/auditing systems.

### VI. CONCLUSION — Findings and Policy Implications
- Summary of empirical and theoretical findings:
  - Funded and private pension (FICMI) development is positively associated with stock market capitalization-to-GDP; robust to dynamic AB estimation and multiple robustness checks.
  - A one-percentage-point increase in funded pension contribution-to-GDP linked to:
    - ~1.9 percentage point increase in stock market capitalization-to-GDP (Arellano-Bond specification).
    - ~2.1 percentage point increase in some specifications (after adding equity limit).
    - ~1.8 percentage point decrease in bank loan balance-to-GDP (Arellano-Bond bank financing results).
  - Stylized model shows FICMI can correct coordination failure and under-investment in long-term equity, improving welfare for a range of contribution rates and equity allocations; excessive τ or η can reduce welfare.
  - Case studies illustrate varied outcomes depending on design, fiscal strategy, investment limits, and governance.
- Five high-level considerations for FICMI policy design:
  1. FICMI pillar must be sufficiently large to realize benefits.
  2. Implement vigorous structural reforms to meet enabling conditions (fiscal strategy, financial infrastructure).
  3. Act promptly, decisively, and persistently with strong political commitment.
  4. Younger nations and some low-income EMDEs can benefit from early development of FICMI pillars due to lower current burdens of supporting older generations.
  5. For emerging markets (e.g., China), pension reform aiding stock market development is important to provide long-term equity financing crucial for technological innovation.
- Specific recommended policy measures:
  - Re-optimize pillar composition to increase weight of funded pillar relative to PAYG.
  - Prefer FICMI design for funded pillar with clear property rights; combine mandatory participation with tax and non-tax incentives as appropriate.
  - Pursue a multi-pronged transition strategy: debt issuance, tax adjustments (income, capital, consumption), expenditure re-prioritization, and consider alternatives to simply raising retirement age.
    - Example country-specific measures: channeling a larger share of SOE IPO proceeds to pension funds; increasing SOE dividend payout ratios; transferring some government-owned equities to pension funds.
  - Strengthen governance and safeguards against politically motivated allocations.
  - Expand coverage via education, financial literacy, transparency, digital enrollment, and measures to reduce informality.
- Future research directions:
  - Policy design for setting contribution rates and replacement ratios.
  - Country-specific policy mixes for transition to FICMI.
  - Optimal mix between non-contributory minimum pensions and FICMI pension schemes.

*Source: IMF Working Paper (content unit: wpiea2025049-print-pdf).*

### Appendix 1. Local Projections

### Appendix 1. Local Projections

### Methodology
- Empirical approach: local projections a la Jorda (2005).
- Main specification:
  - 푆푡표푐푘_푀푘푡푐푎푝_퐺퐷푃_푅푎푡푖표
    푖,푡+ℎ
    = 훽0 + 훽1 퐿.푆푡표푐푘_푀푘푡푐푎푝_퐺퐷푃_푅푎푡푖표
    푖,푡
    + 훽2 퐶표푛푡푟푖푏푢푡푖표푛_퐺퐷푃_푅푎푡푖표
    푖,푡
    + 훽3 퐸푞푢푖푡𝑦퐿푖𝑚𝑖𝑡
    푖,푡
    + 휂 Z
    푖𝑡
    + 훼
    푖
    + 훾
    푡
    + 휖
    푖,𝑡
  - 푆푡표푐푘_푀푘푡푐푎푝_퐺퐷푃_푅푎푡푖표
    푖,푡+ℎ
    is the h-year-ahead stock market capitalization-to-GDP ratio (h = 0, 1, 2, 3).
  - 퐶표푛푡푟푖푏푢푡푖표푛_퐺퐷푃_푅푎푡푖표
    푖,푡
    is contribution to funded and private pension schemes in country i and year t, as a percent of GDP.
  - Lagged dependent variable included to account for potential reverse causality; lagged 퐶표푛푡푟푖푏푢푡푖표푛_퐺퐷푃_푅푎푡푖표
    푖,푡
    not included because left-hand-side variable already accounts for dynamic responses of 퐶표푛푡푟푖푏푢푡푖표푛_퐺퐷푃_푅푎푡푖표
    푖,푡.

### Results — Stock Market Capitalization
- Benchmark local projection (without equity investment limit):
  - As funded pension contribution increases, stock market capitalization in the same year also increases, although this effect is insignificant for outer years.
  - A one-percentage-point increase in pension contribution-to-GDP is approximately associated with a two-percentage-point increase in stock market capitalization-to-GDP.
  - Magnitude is very similar to panel regression result obtained from the Arellano-Bond method.
- Including equity investment limit:
  - Statistical significance and economic magnitude of the pension fund contribution-to-GDP do not change much.
  - The equity limit itself is statistically insignificant.
  - Results align closely with Arellano-Bond panel regressions.
- Presentation details:
  - Appendix Figure 1: local projections of stock market cap onto funded pension contribution; note indicates empirical specification underlying the chart does not control for the equity investment limit.
  - Confidence intervals shown: Shaded area = 90 percent confidence intervals.

### Results — Bank Loans
- Local projections using bank loan-to-GDP ratio as dependent variable:
  - Bank loan-to-GDP drops somewhat in the same year when the pension contribution increases and drops more in the subsequent year.
  - These results are not statistically significant at the 10 percent level.
  - Appendix Figure 3 presents these responses; shaded area = 90 percent confidence intervals.

### Summary Statistics (Appendix Table 1)
- Sample size and variable moments (exact values preserved):
  - Observations: 499
  - Year: Mean 2011; Standard Deviation 5.6; Min 2001; Max 2020
  - Stock_Mktcap_GDP_Ratio: Mean 73.4; Standard Deviation 52.3; Min 2.8; Max 393.0
  - Contribution_GDP_Ratio: Mean 2.2; Standard Deviation 2.5; Min 0.0; Max 9.8
  - GDP_G: Mean 4.1; Standard Deviation 4.2; Min -9.8; Max 20.3
  - DepRate_ST: Mean 2.5; Standard Deviation 2.7; Min -0.8; Max 19.6
  - Pop_G: Mean 0.6; Standard Deviation 0.6; Min -0.7; Max 3.0
  - EquityLimit: Mean 88.6; Standard Deviation 20.8; Min 0.0; Max 100.0
- Sources: OECD; IMF WEO; Authors’ calculations.

### Sample Composition (Appendix Tables 2 and 3)
- Year distribution (selected exact frequencies and percents preserved; total 499 observations, 100.0 percent):
  - 2001: Frequency 17; Percent 3.4; Cumulative 3.4
  - 2002: Frequency 21; Percent 4.2; Cumulative 7.6
  - ...
  - 2020: Frequency 26; Percent 5.2; Cumulative 100.0
- Country distribution (selected exact frequencies and percents preserved; total 499 observations, 100.0 percent):
  - Australia: Frequency 19; Percent 3.8; Cumulative 3.8
  - Austria: Frequency 17; Percent 3.4; Cumulative 7.2
  - Belgium: Frequency 20; Percent 4.0; Cumulative 11.2
  - ...
  - United States: Frequency 20; Percent 4.0; Cumulative 100.0
- Sources: Authors’ calculations.

### Game-Theoretical Models Illustrating FICMI as a Coordination Device
- Model #1: Simple game with two Nash equilibria
  - Players: individual investors; strategies: Stay or Run.
  - Payoff matrix (Investor A, Investor B):
    - Run / Run: (1, 1)
    - Run / Stay: (2, 0)
    - Stay / Run: (0, 2)
    - Stay / Stay: (2, 2)
  - Two Nash equilibria: (Run, Run) and (Stay, Stay); (Stay, Stay) is the “good” equilibrium with higher payoffs (2 vs. 1).
- Model #2: Small probability of deviation (ε)
  - Introduce probability 휖 that a player deviates from intended strategy (noise/misperception).
  - Expected payoffs (examples preserved exactly):
    - From “Stay” (when other may deviate): 휖⋅0 + (1- 휖)⋅2
    - From deviating to “Run”: 휖⋅1 + (1- 휖)⋅2
  - Conclusion: For all positive values of the other player’s deviation probability 휖, (Stay, Stay) is no longer an equilibrium in the sense of the trembling hand perfect equilibrium (THPE); (Run, Run) remains THPE-robust.
  - Footnote example: If payoffs of (Run, Stay) are (1.8, 0) instead of (2, 0), and (Stay, Run) are (0,1.8), (Stay, Stay) will not be an equilibrium as long as the deviation probability 휖> 1/6.
- Model #3: Small probability of deviation with FICMI pension
  - Add term γ to payoff for staying, representing FICMI-induced higher stock returns.
  - Revised payoff matrix:
    - Run / Run: (1, 1)
    - Run / Stay: (2, 0+γ)
    - Stay / Run: (0+γ, 2)
    - Stay / Stay: (2+γ, 2+γ)
  - Expected payoffs:
    - From “Stay”: 휖⋅γ + (1- 휖)⋅(2+γ)
    - From deviating to “Run”: 휖⋅1 + (1- 휖)⋅2
  - Condition for “Stay” to dominate deviating: γ > 휖.
    - This condition reduces to γ > 0 when 휖 approaches 0.
  - Conclusion: Introduction of FICMI (γ > 0, per main text assumption that FICMI increases stock return) can restore the (Stay, Stay) equilibrium and make it robust to small probabilities of deviation (THPE robustness). FICMI thus acts as a coordination device enabling the “good” equilibrium.

### Additional Empirical Materials (figures and tables noted)
- Appendix Figures:
  - Appendix Figure 1: Local projections of stock market cap onto funded pension contribution (no equity limit), shaded area = 90 percent confidence intervals.
  - Appendix Figure 2: Responses when equity investment limit included; left chart response to pension contribution-to-GDP; right chart response to equity limit; shaded area = 90 percent confidence intervals.
  - Appendix Figure 3: Local projections of bank loan onto funded pension contribution and equity investment limit; shaded area = 90 percent confidence intervals.
  - Appendix Figures 4–5: Analogous charts using average equity investment limit; shaded area = 90 percent confidence intervals.
- Appendix Tables (titles preserved; detailed regression tables and p-values appear in the appendix):
  - Appendix Table 4: Average Equity Investment Limit, Funded Pension Contribution-to-GDP, and Stock Market Capitalization with Arellano-Bond (AB) Estimation. Note: P-values in parentheses. *** p<0.01, ** p<0.05, * p<0.1. AB=Arellano-Bond.
  - Appendix Table 5: Average Equity Investment Limit, Funded Pension Contribution-to-GDP, and Bank Loan-to-GDP with Arellano-Bond (AB) Estimation. Note: P-values in parentheses. *** p<0.01, ** p<0.05, * p<0.1. AB=Arellano-Bond.
  - Appendix Table 6: Results for Stock Market Capitalization with COVID Dummy. Note: P-values in parentheses. *** p<0.01, ** p<0.05, * p<0.1. AB=Arellano-Bond.
  - Appendix Table 7: Results for Bank Loan-to-GDP with COVID Dummy. Note: P-values in parentheses. *** p<0.01, ** p<0.05, * p<0.1. AB=Arellano-Bond.
- Data sources cited in appendix notes: OECD; Haver; IMF WEO; Authors’ calculations.

*Source: Authors’ calculations.*

### 20. Securities Times.

### 20. Securities Times

### Major themes covered by the cited literature
- Pension system design, reform, and sustainability
- Funded pensions and long-term capital formation
- Effects of pension policy changes on savings and financial markets
- Development of local securities and derivatives markets
- Retirement financing with an aging population
- Informality and social security reform
- Investor behavior and pension participant attention
- Regulation and investment practices of pension funds
- Country case studies: China, Chile, Kazakhstan, Uruguay, India, Latin America

### Key cited works and focal points (by topic)
- Funded pensions and long-term capital formation
  - Li, Bo, Andi Chen, and Bingwen Zheng, 2023 (in Chinese). Funded Pension and Long-term Capital Formation, China Finance 40 Institute Working Paper.
  - Zheng, Bingwen, 2019 (in Chinese). China Pension Actuarial Report: 2019-2050. China Labor and Social Security Publishing House, Beijing.
  - People’s Bank of China, 2020. Chinese Urban Household Survey of Assets and Liabilities in 2019. Journal of Chinese Finance (in Chinese) (available in Sina Finance).
- Pension withdrawals and savings behavior
  - Madeira, Carlos, 2022. The impact of the Chilean Pension Withdrawals during the Covid Pandemic on the Future Savings Rate, Journal of International Money and Finance, 126: 102650–102670.
- Pension reform, macroeconomic stability, and optimal systems
  - Roldos, Jorge, 2007. Pension Reform and Macroeconomic Stability in Latin America. IMF Working Paper 07/108.
  - Westerhout, Ed, Lex Meijdam, Eduard Ponds, Jan Bonenkamp, 2022. Should We Revive PAYG? On the Optimal Pension System in View of Current Economic Trends. European Economic Review, 148, 104227.
  - McKiernan, Kathleen, 2021. Social Security Reform in the Presence of Informality, Review of Economic Dynamics, 40, Pages 228–251.
- Pension markets, regulation, and investment practices
  - OECD, 2016. OECD Pensions Outlook 2016. OECD Publishing, Paris.
  - OECD, 2020. OECD Pensions Outlook 2020. OECD Publishing, Paris.
  - OECD, 2021. Pension Markets in Focus. OECD Publishing, Paris.
  - OECD, 2022. Annual Survey of Investment Regulation of Pension Funds and Other Pension Providers – 2022 Edition. OECD Publishing, Paris.
  - Superintendencia de Pensiones, 2022. “Inversiones y Rentabilidad de los Fondos de Pensiones,” September.
  - Rudolph, Heinz, and Roberto Rocha, 2009. Enabling Conditions for Second Pillars of Pension Systems. World Bank Policy Research Working Paper 4890.
- Local securities markets and financial development
  - Mathieson, Donald J., Jorge Roldos, Ramana Ramaswamy, and Anna Ilyina, 2004, “Emerging Local Securities and Derivatives Markets,” World Economic and Financial Surveys, IMF, Washington, D.C.
  - OECD, 2017. Mobilising Financing to Transform Kazakhstan’s Economy. Chapter 3, OECD Development Pathways, Multi-dimensional Review of Kazakhstan. OECD Publishing, Paris.
- Behavioral aspects of pension investing and market performance
  - Mitchell, Olivia S., Gary R. Mottola, Stephen P. Utkus, and Takeshi Yamaguchi, 2011. The Inattentive Participant: Portfolio Trading Behavior in 401(K) Plans, SSRN, No. 2006–115, Michigan Retirement Research Center.
  - Villatoro, Félix, Olga Fuentes, Julio Riutort, and Pamela Searle, 2019. Mislearning and (Poor) Performance of Individual Investors. Journal of Pension Economics and Finance, First View, Pages 1–23.
- Theoretical foundations of corporate finance and investment decisions
  - Modigliani, Franco, and Merton H. Miller, 1958. The Cost of Capital, Corporation Finance and the Theory of Investment, The American Economic Review, Vol. 48, No. 3 (Jun., 1958), Pages 261–297.
  - Myers, Stewart C., and Nicholas S. Majluf, 1984. Corporate Financing and Investment Decisions When Firms Have Informationthat Investors Do Not Have, Journal of Financial Economics, Volume 13, Issue 2, Pages 187–221.
  - McGrattan, Ellen R., and Edward C. Prescott, 2017. On Financing Retirement with An Aging Population, Quantitative Economics, 8, Pages 75–115.
- Country-specific analyses and historical studies
  - Saldain, Rodolfo, 2007. Chapter 14, Uruguay: A Mixed Reform. In Stephen J. Kay (ed.), and Tapen Sinha (ed.): Lessons from Pension Reform in the Americas, Pages 359–378.
  - World Bank, 2004. Uruguay: Poverty Update 2003. Washington D.C.
  - World Bank, 2012. China 2030: Building a Modern, Harmonious, and Creative High-Income Society. Washington D.C.
  - United Nations, 2023. India Overtakes China as the World’s Most Populous Country. Policy Brief No. 153.
  - Zhou, Qian. 2021. China’s Social Security System: An Explainer. China Briefing.
- Empirical and methodological contributions relevant to pensions and markets
  - McGrattan, Ellen R., and Edward C. Prescott, 2017. On Financing Retirement with An Aging Population, Quantitative Economics, 8, Pages 75–115.
  - Roldos, Jorge, 2007. Pension Reform and Macroeconomic Stability in Latin America. IMF Working Paper 07/108.

*20. Securities Times*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025049-print-pdf.pdf_