The Mandatory Provident Fund has served Hong Kong citizens for nearly 25 years. Soon, Hong Kong will launch the e-MPF, integrating disparate savings schemes into a single digital system. This presents a well-timed opportunity to drastically improve Hong Kong’s primary retirement savings system. This paper provides empirical examination of the MPF’s performance and fee structure, addressing longstanding criticisms. Our evidence shows that suboptimal asset allocation is the primary problem, followed by excessive fees. MPF assets have grown faster than fees have been reduced, making the system substantially more expensive over time, with total costs reaching about $15 billion HKD annually. Finally, drawing on empirical evidence and considering the transformative potential of the upcoming digital platform, we make several policy recommendations to enhance the MPF’s efficiency and investment outcomes.

1. Introduction

Ensuring that households accumulate sufficient financial assets to support their retirement is a major challenge for policymakers, particularly as populations age globally. While some countries initially adopted defined benefit schemes, which placed citizens’ retirement savings under direct government management, concerns over their long-term fiscal viability prompted many governments to shift toward defined contribution schemes. The Hong Kong Mandatory Provident Fund was one of the first such government-established schemes, following Chile’s program in 1981 and Australia’s Superannuation Fund in 1992. By exposing participants to market-based returns, the MPF sought to compel individuals to save for retirement while minimizing balance sheet risks for the Hong Kong government.

The MPF has succeeded wonderfully in encouraging household participation in securities markets. As of September 2024, it has grown to about $165 billion USD, representing a substantial portion of Hong Kong’s total net wealth of about $3.5 trillion, according to UBS estimates. This roughly 3-4% share of aggregate wealth likely understates the MPF’s importance to less wealthy households, given the large disparity between average and median net wealth in Hong Kong.[2] The MPF represents a clear success in advancing financial inclusion – an accomplishment policymakers can take pride in.

However, despite its sizeable role in improving financial inclusion, the MPF has come under fire for its high fees and performance. Meanwhile, the launch of the e-MPF and Beijing’s plan to promote Hong Kong as a wealth management hub present an opportune moment to reassess the MPF as it reaches a critical juncture in its history. In this green paper, we analyze the fees, performance, and design of the MPF. Our findings uncover three primary drivers of its underperformance: (1) overly conservative asset allocation, (2) inferior products, even after accounting for asset allocation, and (3) elevated fees. To our surprise, among the three, fees are not the greatest factor in its underperformance. Finally, we provide several policy recommendations to help guide the MPF toward a more effective and sustainable future.

2. Data sources

For this study, we use several data sources. We hand-collect data on the MPF and aggregate fund expense ratios and asset allocations over time from annual reports by the Mandatory Provident Fund Schemes Authority (MPFA). Second, we collect returns data from Morningstar, which covers the majority of MPF funds. We compare these returns to those of exchange-traded funds and mutual funds in the United States, which come from CRSP, a commonly used academic database. We have snapshots of fund assets over time, but they are incomplete, so we refrain from conducting any fund-flow analysis.[3]

3. Assessing the MPF’s Performance

The MPF has faced criticisms over the years for its low annualized rate of return, averaging about 2.9% since its inception as of its most recent annual report. This performance could be attributed to several factors, not all of which warrant criticism. We perform a systematic analysis of three key drivers behind the MPF’s low returns.

• Asset Allocation: Asset allocation may explain the MPF’s performance in two ways. First, allocation toward lower-risk investments could drive returns lower, reflecting the fundamental risk-return tradeoff. Second, a heavier allocation to underperforming markets, such as Hong Kong relative to the United States, could also play a role. This does not necessarily indicate an error in investment choices. In theory, in an efficient market, assets are bought and sold until they offer a similar expected risk-return tradeoff. Under this assumption, investors who allocated to Hong Kong disproportionately may have simply had bad luck over the past 15 years as Hong Kong’s stock market lagged behind global peers. Of course, one can debate whether markets are truly efficient and whether the underperformance of Hong Kong’s equity market was predictable.
• High Fees: MPF fees are notoriously high, with fund expense ratios among the highest in the world and often criticized in the media. As of 2008, fees stood at around 2%, and remain above 1% in 2024. In comparison, low-cost ETFs and index funds in the United States charge between 0.03-0.15%. These high fees significantly erode returns over time, particularly given the compounding effect.
• After accounting for asset allocation and fees, MPF funds underperform: MPF managers often emphasize their “active” management approach. However, some active managers may further underperform despite their best efforts, due to factors such as poor stock selection or poor market timing.

3.1 Asset Allocation

We first investigate asset allocation. As shown in Figure 1, the MPF’s Hong Kong asset allocation has consistently skewed toward lower-risk assets over the years. Figure 2 shows assets being largely overweight on Hong Kong, a phenomenon economists often call “home bias”. The graph shows that the MPF ’s allocation to Hong Kong stocks is about 10 times greater than expected based on global equity market capitalization.

The optimal asset allocation is arguably subjective. It may vary based on individual investor preferences and the range of products available in the market. However, mainstream financial economists offer two key critiques of this asset allocation. First, Hong Kong’s realized asset allocation is conservative, with 20% in bonds, 20% in cash, and the remainder in equities. A popular model for asset allocation is the lifecycle model, which recommends that households invest in riskier assets (equities) during earlier stages and shift to less risky assets (bonds) as retirement approaches. TIAA-CREF, a prominent financial organization in the United States that specializes in retirement planning solutions, advocates a “glide path” from risky to less risky assets. Under this approach, individuals start with a portfolio of 90% equities and 10% bonds at age 20 (assuming retirement at 65), gradually transitioning to 45% equities at retirement. For Hong Kong, adjusting for its target allocation over this period, an optimal equity allocation would be 80% in equities and 12.7% in fixed income (with the rest in cash and real estate).[4] By comparison, the MPF’s current asset allocation holds far too little in equities and far too much in cash and bonds.[5]

Second, given that Hong Kongers are already disproportionately exposed to the local market through their labor income, it does not make sense to overinvest in Hong Kong via their investment income. Deviating from the global market capitalization-weighted portfolio constitutes an active bet on the prospects of Hong Kong. Economists often describe this as “home bias”, criticizing investors for improper diversification of risk and calling this among the most basic cardinal sins in investing. If anything, it makes sense to underinvest in Hong Kong to diversify away from local economic shocks, given the MPF provides such access.  The benefits of international diversification are well-documented in financial literature, as a broader pool of assets generally offers superior risk-adjusted returns.[6]

Figure 3: Hypothetical AUM of MPF Under Different Asset Allocations

What is the impact of imbalanced asset allocation? We simulate the growth of the MPF over time to calculate how much returns would have increased. We hand-collect contributions made to the MPF as well as the returns reported by the MPF, and simply ask: assuming the MPF behaved the exact same way except for asset allocation, how much would its performance have improved? To do so, we simulate the growth of the MPF in dollar terms. We pull annual contributions and annual returns by risk class (i.e. the five risk classes into which funds are categorized) from MPFA reports and fund-level returns from Morningstar. We then ask, if participants invested in the equity funds of the MPF instead of a mix of stocks and bonds, how would the MPF’s growth be different? Or what if the MPF had only invested in U.S., Chinese or global equities?[7]

This analysis has the advantage of being simple and transparent, requiring limited technical knowledge. However, there are several important caveats. First, we compare only total returns without accounting for differences in risk. For instance, the MPF targeting North America might operate at a lower risk level than an all-North American equity fund such as SPY. We address this issue later with more formal regression analysis. Second, we are aiming to decompose historical underperformance only. Past underperformance of a particular asset does not mean it will persist in the future, as even the 25 years we study may represent a relatively short observation period.

Figure 3 presents our results since 2008, the first year we have complete data on fund expense ratios, returns, and investment allocations by risk class, as reported by the MPFA annual reports. This allows us to simulate what would happen if the specifically chosen MPF funds remained the same except for changes in asset allocation. Appendix Figure A shows how MPF funds performed over a longer period of time, but assumes equal weighting of returns within Morningstar categories since we do not know how participants allocated across different types of funds.

The overall takeaway of these graphs is that asset allocation matters. Any sort of allocation away from Hong Kong equities toward a more balanced global equity allocation would have led to an MPF 1.5 times larger. Allocating to U.S. equities, which have delivered exceptional performance, would have resulted in an MPF 2.5 times larger, with an annualized return of 5.1%. Since 2008, a significant driver of underperformance in the MPF is clearly the overweight allocation to Hong Kong stocks. While fixed income allocation made little difference over the same period, Figure A1 show that over longer periods, fixed income significantly lags behind equity. This is sensible, because equities – being riskier than bonds since bondholders are paid first – demand a higher premium.

In short, the recent underperformance of Chinese equities, coupled with an over-allocation to bonds and cash-like instruments, likely accounts for much of the MPF’s lagging performance. At the same time, the Hong Kong and Chinese economies have experienced slower growth in the last few years. Rather than diversifying away from local economic risks, the MPF’s allocation compounds Hong Kongers’ exposure to domestic conditions through the stock market.[8]

3.2 High Fees

Next, we examine the impact of fees by removing them, i.e. assuming they are zero. Our findings suggest that while fees are a significant factor, asset allocation likely plays a more dominant role in explaining the MPF’s underperformance. Adjusting asset allocation would have been roughly 3-4 times more effective in generating profit than eliminating fees alone. Currently, total MPF assets stand at 1.14 trillion HKD. The terminal value of the MPF with fees added back in would be 1.31 trillion HKD. This is an extra 15% risk-free. While significant, a more balanced global equity allocation would have delivered a terminal MPF value of $1.75 trillion HKD by 2023, which suggests asset allocation has had a greater impact. Assuming the fees were similar to those in 2008, the cumulative impact of fees on the MPF has been about 21%.[9]

One caveat about this analysis is that the fund expense ratio does not account for trading costs borne by investors. If transaction costs paid to brokers are high or if MPF managers excessively churn, additional implicit fees might have been charged, which we will count as excess underperformance in Section 3.3. However, without knowing more about the trading behaviors of MPF managers and the prices at which transactions occur, it is difficult to draw definitive conclusions.

Figure 4: No fees

3.3 Excess Underperformance / Performance Drag

The final possibility is that MPF products underperform their respective benchmarks, even after accounting for fees and asset allocation. To evaluate this, we must compare performance against benchmarks that investors could have plausibly accessed as alternatives. This nets out the effect of asset allocation by enabling apples-to-apples comparisons in terms of risk and asset class. To account for fees, we ask whether the underperformance relative to the benchmark was in the ballpark of fees. We do not find overwhelming support for the notion that MPF funds dramatically underperform their benchmark for any reason other than fees, although our tests are inconclusive because we benchmark the MPF conservatively.  

We use linear regression, a statistical tool designed to correlate the returns of funds like the MPF against benchmarks, to see (1) how similar the fund is to the benchmark, and (2) how much extra return (or underperformance) there is left over, after accounting for the benchmark. There are two complicating factors to note. The first complication is benchmark selection, so we will present results from multiple benchmarking exercises. The other complication is accounting for taxes. MPF returns are post-tax, but returns quoted in databases often assume tax-free status. For example, in the United States, a foreign investor pay no capital gains tax but face a 30% dividend tax. Given the U.S. dividend yield of about 2%, this translates to 0.66% per year that could not be reinvested. This number varies across countries, complicating cross-border comparisons because a proper comparison must account for the total returns after taxes in various jurisdictions. Tax rates in the U.S. are high. China’s are roughly 10% or less. Rates are often zero or lower in jurisdictions with which Hong Kong has a tax treaty. To simplify this analysis and make our conclusions very conservative, we restrict ourselves to the U.S. equity index fund and ETF universe, and subtract 30% from dividends.

Linear regression models the relationship between a dependent variable (MPF returns) and explanatory variables (in this case, benchmark fund returns). Specifically, we estimate the following relationships.

             is the global money market rate, and is the return of a benchmark asset.  is what’s sometimes called the “beta” of an asset, or its covariance with the asset. The term   refers to the average outperformance above the benchmark (negative alpha indicates underperformance). As stated before, MPF funds may target lower-risk assets than the market, thus earning lower returns. For example, MPF managers may shift to cash or bonds to preserve investor capital over time, resulting in smaller fund movements relative to market swings. Suppose we want to estimate the relationship between an MPF fund and the S&P 500. For example, =0.5 implies that for every 1% movement in the S&P 500, the asset moves by 0.5%. If the asset performs exactly like an asset half invested in the market, its α should be zero. If its α is less than zero, it means it underperforms the S&P 500 after accounting for its similarity to the S&P 500. Finally,  are returns explained neither by the average nor the benchmark. A large absolute value of may indicate tracking error relative to the benchmark.

Table 1 presents our results across four benchmarks, each using a slightly different set of assumptions to help sharpen our interpretation of whether the MPF is underperforming or not. First, we obtain all mutual funds in CRSP by objective code, using average returns after fees. Since 2008, the average mutual fund has charged around 0.9%, which is similar to MPF fund management fees. Index funds and ETFs have driven this cost lower. In Panel B, we benchmark post-fee returns against ETFs. Both exercises assume a Hong Kong investor went to the United States and was taxed 30% for distributions (a conservative assumption). In Panel C, we hand-select ETFs for corresponding markets (VOO for the U.S., VGK for Europe, EWJ for Japan, VT for global, LQD for corporate bonds, and the Hong Kong tracker for China funds).

Table 1: The MPF Against Benchmarks

In Panels A-C, we run univariate regressions and report the alpha of the chosen benchmark, where each alpha is the alpha of a single time series regression of the following form:

In Panel D, we report the average Morningstar alpha, if available, using their MPT Index, which finds the index that fits the asset well.

Benchmark name	MPF Fund Category	Average
Panel A
Lipper Objective Code, Category Average	Equity	-0.73%
	Fixed Income	-0.88%
	Money Market	-0.40%
	China Equity Subset	-0.93%
	US Equity Subset	-1.84%
Panel B
Lipper Objective Code, Three Cheapest	Equity	-0.89%
	Fixed Income	-0.79%
	Money Market	-0.35%
	China Equity Subset	-1.90%
	US Equity Subset	-1.74%
Panel C
Single Well-Known ETF Benchmark	Equity	-0.91%
	Fixed Income	-1.11%
	Money Market	-0.48%
	China Equity Subset	-0.81%
	US Equity Subset	-1.99%
Panel D
Morningstar Reported Benchmarks	Equity	-1.84%
	Fixed Income	-1.3%
	Money Market	-0.54%
	China Equity Subset	-2.05%
	US Equity Subset	-1.89%

In Panel A, where we use the Lipper Objective Code to categorize average returns, we see that MPF funds underperform in every category, with average alphas of -0.73% for Equity, -0.88% for Fixed Income, -0.40% for Money Market, and -0.93% for the China Equity subset. This initial comparison demonstrates a general underperformance across these categories. Panel B sharpens this comparison by benchmarking against the three cheapest available ETFs, further illuminating the performance gap. Here, MPF funds show increased underperformance, which is particularly stark in the China Equity subset with an average alpha of -1.90%. Equity funds underperform by -0.89%, Fixed Income by -0.79%, and Money Market by -0.35%. Panel C continues this comparison, using the single most competitive ETF in each category as the benchmark. MPF returns here also lag behind, with alphas of -0.91% in Equity, -1.11% in Fixed Income, -0.48% in Money Market, and -0.81% in the China Equity subset. This benchmark shows that a single ETF outperforms MPF funds, reinforcing the competitive advantage of low-cost ETFs in various markets.

These findings lead to a clear conclusion: net of fees, the MPF underperforms compared to what investors could achieve by opening a discount brokerage account and managing their own asset allocation. When adding back in the scheme administration costs of roughly 0.73%, MPF funds are generally in line with, though perhaps a little worse than, the post-fee performance of the average U.S. mutual fund. However, there are pockets of underperformance in both Chinese markets (where Hong Kong investors overbet) and U.S. equities (which represent the largest portion of the global portfolio).

While one might wonder if the authors’ calculations are being relatively unfavorable to the MPF, those using industry databases paint a similar picture. In Panel D, we report our Morningstar results, which use the best-fitting index. Morningstar does not benchmark all funds, such as target date funds or allocation funds, instead focusing solely on Equity, Fixed Income and Money Market funds. Also, they do not consider dividend taxes, which is important for U.S. funds. In this case, MPF funds display the largest underperformance across all categories, with an alpha of -1.84% in Equity, -1.3% in Fixed Income, -0.54% in Money Market, and a significant -2.05% in the China Equity subset. These results indicate that funds underperform the Morningstar benchmark by approximately the amount of fees, or slightly more. This suggests that MPF funds are likely no better than – but not blatantly uncompetitive with – retail funds gross of fees, though they clearly underperform low-cost passive index funds after fees. That said, our benchmarking exercise is conservative, and if we made less conservative assumptions perhaps such evidence would emerge.[10]

3.4  An Aside on the High Fees

While not the sole driver of the MPF’s underperformance, fees remain a significant factor. They are also an emotive issue for many MPF participants, based on the tone of numerous media reports we have read on the subject as well as the focus of many policy efforts by the MPFA. Given their importance, we found it particularly interesting to systemically analyze the dynamics of fees over time and their determinants.

The MPFA often emphasizes that fund expense ratios (FERs) have fallen over time. By mandating the disclosure of FERs, which were first published in 2008, the MPFA likely put significant pressure on trustees and fund managers to lower fees and implemented a number of policies that have effectively reduced fees. However, despite these great efforts, the aggregate and per-capita nominal and inflation-adjusted fees of the MPF system have actually risen over time. Figure 6 clearly illustrates the increase in the MPF’s total fees. This is because fund expense ratios have slowed faster than the asset base has risen. To put this into perspective, the MPF system charges roughly $2 billion USD per year, or $500 USD per person, representing around 2-3% of the financial industry’s revenue relative to GDP.

A common defense by banks for the MPF’s high fees is high operational costs. These include substantial setup expenses and the labor-intensive manual processing of documents. In this context, the e-MPF promises to be a step forward. However, our findings suggest that these two factors alone do not fully explain the persistently high fees of the system.

By definition, fixed costs do not explain the rising total fees charged by the MPF system, as they occur at the beginning. However, one could argue that variable costs are behind the rising expenses. Indeed, people change jobs, retire, or leave Hong Kong. To examine this, we hand-collect the number of participants in the system and deflate this number by multiple population denominators. The results are directionally consistent – the per-person cost of the system (or a proxy for the variable cost) has risen over time. From 2008 to 2023, total fees increased by 2.6 times (1.8 inflation adjusted), per-account cost doubled (or increased 1.4 times inflation adjusted), and the HKD cost per employee rose by 2.84 times (1.95 times inflation adjusted). If Hong Kong banks have been implementing the financial technology they trumpet, both fixed and variable costs should have reduced over time, rather than risen. Instead, if fees are any indication, the opposite has happened.

Figure 5: Expenses of MPF Over Time (Nominal)

We also conduct fund-level analysis. One hypothesis is that the emergence of new, smaller funds has contributed to rising costs. However, our analysis suggests otherwise: older, bigger funds charge higher fees, likely reflecting their established market positioning. In Figure 7, simple univariate bin plots show that older, larger funds charge more rather than less.

In the Appendix, we perform regression analysis – a statistical model to estimate the relationship between fees and certain variables, while controlling for other confounding variables. This approach allows us to test the same relationships while accounting for risk ratings, the types of assets traded, and other relevant factors. The regression results suggest that age, fund size and trustee size are positively correlated with fees. Unless the variable costs of such institutions – which remain unobservable as we do not know the number of scheme participants – increase with the age and size of the funds, schemes or trustees, these results suggest the opposite of the cost-based hypothesis. This is after controlling for returns and the type of assets traded.

Figure 6: Size/Age and Fees

One remaining explanation for the MPF’s high fees is the lack of competition and the presence of market power. Quantifying the cost of this market power is challenging. Former HKUST student Claire Hong’s dissertation shows that once participants are allowed to switch MPFs, there appears to be a decline in MPF fees relative to Hong Kong retail funds of about 0.2% per year. However, this is obviously a lower bound. Just because people can switch MPFs, it is not a frictionless process as one has to start a new account at another institution, and scheme operators hold significant market power in defining the investment menu. In addition, the comparison group, Hong Kong retail funds, are among the least fee-competitive in the world. Therefore, Hong’s (2019) analysis likely significantly underestimates the true cost of market power.

4. Additional discussion, commentary, and policy recommendations

We now turn to policy recommendations. Our previous findings highlight a few problems: (1) the fees of MPF schemes are high, owing partly to low competition, and (2) asset allocations are imbalanced. Some of our recommendations tackle these two problems, while others focus on broader design improvements to the MPF system, including the e-MPF. Scheduled to roll out in the next year and a half, the new system aims to digitize the labor-intensive administrative processes often blamed for driving fund administration costs. The platform will also allow employees to freely switch between schemes, thereby increasing competition by reducing customer stickiness. At launch, the e-MPF will charge a fee of 0.37%, roughly half of the fees currently charged by trustees and scheme operators, with plans to further reduce costs over time as assets grow. Given the system is rolling out soon, we think this is the right time to provide input before its details are set in stone.

The government’s approach to the MPF is to use a market-based framework whereby the government merely acts as a facilitator and regulator. However, creating a well-functioning free market is not easy and requires three key elements: (1) diverse and competitive market offerings, (2) sufficiently sophisticated investors to discern these offerings, and (3) regulatory oversight. Under this market-based approach, whereby the government operates within certain self-imposed constraints, we offer four recommendations. We then discuss the additional measures the government may consider, should it choose to take a more proactive and tougher approach.

Recommendation #1: Drive Down Fees

If the government chooses to maintain its current approach, it has several tools at its disposal. To directly reduce fees, the government’s only option is to revise the default investment strategy. The DIS allocates assets based on participants’ age, starting mostly with equities and following a “glide path” into bonds as retirement nears. This diversification strategy follows the MSCI All Country World Index allocation, an international standard and a clear improvement over the allocations currently adopted by MPF participants. The DIS should re-adjust its fee cap of 0.75%. Subject to Legislative Council approval, the MPFA can mandate a lower cap. Introduced in 2017, the 0.75% fee cap was high even by 1990s standards. Fund managers and scheme administrators have now had seven years to adapt to this competitive threshold. In aggregate, total fees have kept rising, so it is unclear if the profitability of funds has suffered materially. This suggests it would be appropriate to further drive down fees to even 10 basis points. Further, to elevate the prominence of the DIS in the investment menu and to attract more participants with different preferences, the government could mandate the inclusion of a few alternative glide paths, such as options more aggressively invested in equities even as one gets older.[11]

The MPFA could also invite new service providers that charge lower fees into the market. Once the e-MPF platform is operational, it will handle scheme administration, leaving only marketing, sales or client service and product development to scheme operators. This will make it more viable for new entrants to establish themselves or take over less technologically savvy incumbents. We think Vanguard, State Street, or Ping An would be great candidates. Alternatively, this could be a great opportunity for Hong Kong to attract low-fee fintech service providers to set up operations and offer MPFA services in the city. Such providers could be approved on the basis that they push certain product recommendations such as low-cost index funds.

Recommendation #2: Fix Asset Allocations by Being Prescriptive, Promoting Investor Education and Curating Information

We argue that MPF participants’ imbalanced asset allocation is at least partly responsible for the system’s historically low returns. While one might contend that asset allocation is the responsibility of individual participants, the market is a highly regulated system that the Hong Kong government had paternalistically imposed in the first place. The government should therefore actively monitor asset allocation.

First, the government should be unafraid to be prescriptive. It should openly endorse a global investment allocation and educate people on its merits. This allocation strategy aligns with asset allocation standards endorsed by international labor and financial organizations and should therefore be politically uncontroversial.[12] Academics have long argued that international diversification offers many benefits to local investors. Further, we argue  investors’ overconcentration in local markets could lead to unintended consequences.[13] This would simply require the MPFA to double down on its efforts with the DIS and be more prescriptive through public education and marketing.  

In particular, the MPFA, in partnership with the Investor and Financial Education Council, should consider taking a more proactive role in educating investors and enhancing citizens’ financial literacy. Currently, the IFEC is focused on generic financial planning, budgeting, avoiding scams, and producing entertaining gamified content – not on teaching people how to invest. From our personal interactions with the IFEC, we surmise that the IFEC is hesitant to prescribe risk-taking or perhaps even discourages it. However, taking risks on fairly priced or undervalued assets – particularly for assets set aside early in one’s career – is a strategy for earning return-on-risk and should be encouraged.

Second, the government can curate information to help market participants navigate the complex product space. The e-MPF will allow participants to freely switch between products, which has the benefit of increasing competition. However, the MPF ecosystem currently includes over 450 funds, and participants may struggle to make optimal choices. Many schemes offer virtually identical funds but with varying fee structures. Egan (2019) shows that in the U.S., brokers are incentivized to steer consumers toward ‘dominated,’ higher-fee products with completely identical pre-fee returns, underscoring how search costs and the complexity inherent in financial products lead consumers to suboptimal decisions.

The basis of this phenomenon is a well-studied problem called “choice overload.[14] Specifically, when consumers have too many choices and a limited attention budget, they may choose products with salient attributes. This may explain why the oldest and largest products in the MPF currently have higher fees. As the e-MPF expands the available choice set (now hundreds of funds), participants may face even greater difficulty navigating their options absent an information architecture that reduces search costs and complexity.

How can the government curate information? To begin with, the MPFA can implement a more modern risk-and-performance rating system grounded in academic practice. The existing ratings, which use six risk classes to describe MPFA funds, have not been recalibrated since the system’s launch. Moreover, the MPFA platform’s website is outdated. Academic research has found that that the way information is framed significantly impacts investor behavior.[15] Currently, the ratings do not seem particularly useful.[16] Furthermore, many of these products are redundant, further complicating consumers’ decision-making. Basic clustering analysis can identify similar assets to form proper peer group comparisons, simplifying choices from hundreds of funds to a few broad buckets. Metrics like those created for this paper (alphas, betas, etc.) can be easily interpreted and convey to participants more meaningful benchmarks than simply returns displayed on the MPFA platform, while applying pressure on MPF providers to improve their offerings.

The MPFA’s investor education initiatives can benefit from collaboration with Hong Kong’s many business academics. Local finance academics can design dashboards displaying more thorough performance metrics and optional MPF-specific financial literacy tests. Marketing academics at local business schools may contribute insights on how to frame choice sets to encourage more optimal investment behavior, or develop scientifically grounded assessments for participants to judge their own risk tolerance. Many Hong Kong academics would be willing to work pro bono or for modest grants provided by the MPFA, as contributing to public knowledge and winning competitive grants both align with the key performance indicators (KPIs) for their career advancement. 

Recommendation #3: Boost the Product Space of the MPF

We hope the government can enhance the product space of the MPF. The first thing missing in the MPF system is low-cost index funds. Many MPF managers are active, and active managers tend to charge higher fees. However, their net returns clearly underperform low-cost passive index funds, which remain surprisingly few in the MPF schemes. We hope the MPF’s current funds can be supplanted with cheaper alternatives, enabling the majority of investors to default into these cost-effective passive funds. We also hope the MPF could consolidate many of its similar or redundant offerings, or at least curate information about them, to reduce the complexity of navigating the space.

Once participants are equipped with sufficient investor education and access to curated information, the MPF could also consider promoting greater product innovation and loosening restrictions to allow for more diversified asset allocations and aggressive investment options that would typically be out of reach for retail investors. For example, the default investment option follows a glide path toward more bonds as one gets older; as mentioned earlier, the government could mandate DIS options with alternative glide paths that invest more heavily in equities even as retirement nears. Alternatively, the MPF has been limited to vanilla products such as developed market equities, fixed income, and money market instruments. While the MPF now allows REITs, which is a step in the right direction, it still excludes gold and commodities, which serve as inflation hedges, and emerging market equities or sector-specific funds. For investors seeking potentially outsized returns, the MPF could explore alternatives such as private equity or performance-sensitive hedge funds, which are difficult to access through discount retail brokers. The MPF could adopt a fee structure similar to the Hong Kong Monetary Authority’s (HKMA) pay-for-performance model with external managers currently. Australia’s Superannuation Fund allows a broader range of options, including those mentioned above. However, to prevent the product space from becoming too large for consumers to navigate and difficult to govern, the government can decide whether to allow participants to invest in a select number of these alternatives directly, or simply as part of a vehicle with a balanced asset allocation to improve diversification.

Recommendation #4: Harvest Data, the New Oil

The MPFA should leverage its data better. As an immediate step, the MPFA could publish return series, assets under management (AUM) series, and other key metrics for every fund on the platform, accompanied by standardized identifiers. As noted before, requiring funds to disclose their holdings in a systematic manner would enable better scrutiny and analysis. In the U.S., mutual funds and other institutional investors have been reporting their holdings regularly since 1979, demonstrating that such disclosures are neither onerous nor impractical. Greater transparency would make it easier to replicate and automate analyses like those presented in this paper, benefitting the public. Investor letters and trustee annual reports, as they currently stand, offer limited utility.

 The upcoming e-MPF platform represents a wonderful opportunity to digitize the system and harness valuable data on Hong Kong investors’ behaviors. For example, this “digital exhaust” could provide insights into the characteristics of individuals who make suboptimal investment allocations, choose poor-performing products, or don’t monitor their investments effectively. Moreover, the e-MPF could evolve into a more powerful tool by offering additional features over time, such as investor surveys and account aggregation, allowing the Hong Kong government to develop a more comprehensive understanding of its constituents’ financial behaviors and better serve their needs. These analytics may serve as a public good, particularly by leveraging social proof and peer effects. For example, D’Acunto, Rossi and Weber (2022) shows how crowdsourced spending analytics could induce those who spend more than the peer average to save more. Such analytics could create broad-based awareness of how others are doing, encouraging one to reflect on their own investment behaviors. A/B testing could also provide insights on ways to optimize the platform’s information architecture. Given that the e-MPF’s cost is not low – 0.37% of 1.2 trillion HKD in assets represents about one-fourth of State Street’s IT expenditure – the MPFA should set ambitious targets for the e-MPF platform.[17]

Finally, we offer a thought experiment on how the government could play a more active role

The recommendations outlined above assume a market-based framework, where the government simply acts as a facilitator and regulator. In this approach, the government faces the trilemma of  (1) establishing a robust market, (2) educating participants, and (3) developing proper oversight. While necessary, these efforts require significant government resources and attention, and there is a long way to go from here. Our recommendations, even if implemented, serve as somewhat blunt instruments for fee reductions as they operate indirectly through market mechanisms. If the government is willing to exercise less self-restraint, we believe it could redirect resources to achieve better outcomes for market participants.

What if, for example, the government were to create its own scheme? Such a scheme could offer a selection of vanilla products like low-cost index tracking funds – products notably absent from the current MPF market. Alternatively, the government could provide its own retirement products and asset allocation services, following basic glide paths and made up of the lowest-cost investment products. The government’s credibility would likely attract customers, while its lack of profit motive would keep costs minimal. One apprehension government officials may have is whether they can serve MPF participants well, but it is unlikely that they would make worse decisions than the least sophisticated participants the MPF is precisely designed to protect – especially when operating with lower fees. For those seeking riskier and more exotic products, private MPF schemes and privately managed funds can still exist if they demonstrate clear value propositions. This hybrid model would prevent investors from defaulting into today’s high-cost, mediocre-performing products.

Another possibility is for the government to revisit the sometimes discussed idea of offering a real return instrument, providing returns superior to money market funds. This recommendation draws inspiration from Singapore’s Central Provident Fund, which channels funds into the Government Investment Corporation (GIC), significantly strengthening the financial position of the Singapore government through robust investment returns. The hurdle rate for such an instrument could vary based on prevailing economic conditions, passing through additional upside in good times. Given that Hong Kong faces a looming fiscal deficit, the government will have to issue debt at some point. A government-managed MPF scheme could simultaneously address the market’s mediocre product offerings and help mitigate fiscal shortfalls.

We think a government-managed scheme offering superior returns and lower fees would likely garner widespread support. Money market funds within the MPF system currently offer embarrassingly low returns paired with extremely high fees. The guaranteed funds, per the most recent MPFA annual report, have provided on average 0.9% per year.[18] This would not be difficult to beat. Such a scheme would also be relatively easy to administer. Hong Kong has a deep pool of investment professionals, many of whom would welcome the opportunity to serve. Alternatively, much of the infrastructure needed for a government scheme already exists within the HKMA. The HKMA is staffed with investment professionals and already manages the Exchange Fund, money for various government departments, and the Hong Kong Mortgage Corporation (HKMC), known for offering one of the best annuities on the market.

5. Conclusion

All in all, we believe the MPF can implement a series of straightforward adjustments to position it better for future success. However, we argue that there is more at stake than just the MPF. We all go to the Wan Chai or Sham Shui Po computer centers, knowing we will find fair prices and reasonable quality. Unfortunately, the same cannot always be said about Hong Kong’s retail asset management industry. Hong Kong’s financial industry is notorious for its high fees, with various Morningstar reports ranking the city among the most expensive major financial centers. Improving the product space and reducing MPF fees through targeted government action would not only benefit participants but also enhance the overall competitiveness, standards, and reputation of Hong Kong’s broader financial industry. By more aggressively addressing inefficiencies in the MPF system, the government has an opportunity to strengthen Hong Kong’s financial services ecosystem, draw in more capital, and advance the Central Government’s strategic vision of Hong Kong as the wealth management hub for the Greater Bay Area and a preeminent one globally.

References

Ayres, I., & Nalebuff, B. (2008). Buying stock on margin can reduce retirement risk. Working paper.

Briere, M., Poterba, J. M., & Szafarz, A. (2021). Choice overload? Participation and asset allocation in French employer-sponsored saving plans (No. w29601). National Bureau of Economic Research.

Bordalo, P., Gennaioli, N., & Shleifer, A. (2013). Competition for attention (No. w19076). National Bureau of Economic Research.

Campbell, J. Y. (2002). Strategic Asset Allocation: Portfolio Choice for Long-Term Investors. Oxford University Press.

Cocco, J. F., Gomes, F. J., & Maenhout, P. J. (2005). Consumption and portfolio choice over the life cycle. The Review of Financial Studies, 18(2), 491-533.

D’Acunto, F., Rossi, A. G., & Weber, M. (2023). Crowdsourcing peer information to change spending behavior. Chicago Booth Research Paper, (19-09).

Egan, M. (2019). Brokers versus retail investors: Conflicting interests and dominated products. The Journal of Finance, 74(3), 1217-1260.

Hong, Y. (2021). Freedom of choice in pension plans: Evidence from a quasi-natural experiment. Hong Kong University of Science and Technology (Hong Kong).

Hwang, B. H., Liu, B., & Xu, W. (2019). Arbitrage involvement and security prices. Management Science, 65(6), 2858-2875.

Levi, Y. (2021). Personal financial information design and consumer behavior. Available at SSRN 3886082.

Iyengar, S. S., & Kamenica, E. (2006). Choice overload and simplicity seeking. University of Chicago Graduate School of Business Working Paper, 87, 1-27.

Rösch, D. (2021). The impact of arbitrage on market liquidity. Journal of Financial Economics, 142(1), 195-213.

Appendix

Appendix A – MPF Under Full Sample

Appendix B – Determinants of the FER in 2024 October

We thank Elvin Yu of Goji Consulting for many discussions. However, all opinions and mistakes are our own.

[2] According to the 2024 UBS Wealth Report, the average wealth per adult in Hong Kong was $582,000, while the median stood at $206,859. Based on Census data, the number of adults in Hong Kong, excluding those aged 15-19 and younger, was approximately 6.47 million.

[3] Although the data had been retained before and retrieved for prior studies (e.g. Hong (2017)), the MPFA, when asked, indicated the data on fund assets historically may have unfortunately been deleted.

[4] We collect data from here: https://www.tiaa.org/public/pdf/lifecycle_funds_at_a_glance.pdf. We also collect data from the Hong Kong Census.

[5] Some might argue that even TIAA-CREF’s glide path recommendation is too conservative. The MPF is not withdrawable until age 65 unless one faces terminal illness or leaves the system, resulting in less need for cash on hand. Moreover, labor income is more stable than investment income. Hong Kong boasts one of the lowest unemployment rates in the world. With stable, fixed labor income, one should take up a stronger equity allocation, perhaps up to 100% (see Cocco, Gomes and Maenhout (2005)), or even larger than 100% when young (e.g. Ayres and Nalebuff (2000), Campbell and Viceria (2002)).

[6] According to data from AQR, during the period from 1986 to 2004—when Hong Kong markets performed relatively well—the Sharpe ratio for Hong Kong markets was 0.41, compared to the global Sharpe ratio of 0.405. However, since 2005, the global portfolio has reached a Sharpe ratio of 0.465 while Hong Kong markets delivered a Sharpe ratio of just 0.292.

[7] We make a number of assumptions. First, in effect, this assumes no market impact nor fee reduction. Given the size of the MPF relative to other pensions, and the large liquidity of the markets in which MPF funds tend to participate, we think this assumption is plausible. Second, we assume that participants do not change their contribution amounts in response to higher or lower returns.

[8] One can of course argue that given the recent performance of the Chinese market, now is the time to buy Chinese stocks. However, this constitutes a market timing bet on the assumption the market is not efficient. All else equal, if markets are efficient, assets are bought and sold until their prices reflect market expectations, making the capitalization-weighted portfolio is the optimal choice. Under an efficient markets view, a broader set of options (i.e. global equities) provide more diversification benefits. Market timing is a difficult thing to do – while fund managers with a timing mandate, market-timing is difficult and is not an advisable strategy for fund participants.

[9] This impact, while significant, may appear smaller than some would expect at first sight. This is because the MPF receives contributions over time, which grow nominally due to inflation and population growth, so later cash flows have lower cumulative loss from fees. However, the cumulative effect of fees over a 25-year period is approximately equivalent to fees multiplied by the number of years. For contributions made in 2000, this would have translated to a cumulative impact of 45% (i.e. ).

[10] For example, it is commonly assumed in the academic financial literature that a foreign investment manager could sell before the ex-dividend date, incurring a transaction cost but avoiding dividend tax, so we suspect a 30% withholding is a very conservative assumption. See Roesch (2021) for a discussion.

[11] As discussed before, the glide path prescribed by TIAA-CREF is likely too conservative. The glide path used by the Default Investment Strategy starts at 60/40 equities to bonds, which is even more conservative than TIAA-CREF.

[12] Even if one were to view the global investment allocation as too U.S.-centric, diversifying away from Hong Kong toward pan-Asia, Europe, and emerging markets would provide a counterbalance to local domestic conditions.

[13] For example, one potential catch-22 of investing too much in the local market is that valuations may be too high, meaning future returns will be too low.

[14] See Bordalo, Gennaioli, Shleifer (2013), Briere, Poterba, Szafarz (2021), Iyengar, Kalmenica (2006).

[15] Running a field experiment on a fintech platform, Levi (2021) shows that simply framing past consumption and savings under a “risk frame” in certain ways affected user savings by 15% for over 6 months after the experiment.

[16] For example, the so called risk class guaranteed fund is not guaranteed in the sense of a money market fund. It is principal protected. While we understand there is a specific legal meaning, it is hard already for time-constrained retail investors to understand these subtle differences.

[17] State Street’s 2023 Annual Report suggests it spends more than $2 billion per year on technology and operational efficiency, the vast majority of which is purely technology.

[18] While money market funds also seem in the data to be quite popular, even though they offer a substantial haircut of around 1-1.2% over the money market rate. The existence, size and high price tags of these options loosely suggest that a fixed return instrument would be extremely popular, a notion affirmed by anecdotes we received in writing this article.