A Quantitative Evaluation of the Housing Provident Fund Program in China

Introduction & Motivation

The study examines the Housing Provident Fund Program in China, a large-scale public housing policy designed to promote homeownership via mandatory savings and subsidized mortgages. IDEAS/RePEc+3Federal Reserve Bank of Dallas+3SSRN+3

The author notes that governments often intervene in housing markets to promote homeownership, stabilize housing demand, or assist households with financing. In China, the policy instrument chosen was the Housing Provident Fund Program, first implemented in 1999 and gradually extended across regions. ResearchGate+4Federal Reserve Bank of Dallas+4IDEAS/RePEc+4

The two core components of the Housing Provident Fund Program are:

1. A mandatory savings scheme: urban workers (employed by state, collective, or private firms) are required to deposit a portion of their labor income into the fund until they buy their first home or retire. Past deposits are refunded upon home purchase or retirement. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

2. Subsidized mortgages: participants are eligible for mortgage loans at interest rates below market rates, making housing finance cheaper for fund participants. SSRN+3Federal Reserve Bank of Dallas+3ScienceDirect+3

The evaluation of the housing provident fund program in China.  The Housing Provident Fund program is the largest public housing program in China. It was created in 1999 to enhance home ownership and to make housing more affordable. This program involves a mandatory savings scheme that requires participating workers to deposit a fraction of their income into the program. Past deposits are refunded when the worker purchases a house or retires. The program provides mortgages at subsidized rates to facilitate these home purchases. Given the empirical challenges in evaluating the success of this program, I use a calibrated life-cycle model to quantify the effectiveness of these polices. My analysis shows that a housing program with these features is expected to increase the rate of home-ownership by 4 percentage points in steady state. In addition, the average home size increases by 21% relative to the baseline model. These results are largely unaffected by the existence of employer contributions. I discuss the economic mechanisms by which these outcomes are achieved. The governments around the world take measures to support homeownership. These actions are driven by the belief that housing, for most households, is both an important investment asset and a necessary consumption good, and that homeownership promotes social and economic stability. The U.S. government, for example, has fostered homeownership by encouraging subprime lending and expanding secondary mortgage markets (see, e.g., Main and Sufi (2009) and Gabriel and Rosenthal (2010)). Arguably as a result, the U.S. homeownership rate reached 70% in 2004, compared to 60% in 1960 and 40% in 1940. Since the mortgage crisis of 2008, however, this rate has dropped back to 65%. Many Asian governments, in contrast, have adopted more centralized, mandatory savings plans that aim to fund households’ housing needs. The Housing Provident Fund (HPF) in China is one such example. Table 1 provides examples of similar programs in other countries.

Zhou argues that estimating the causal effect of the Housing Provident Fund Program on homeownership and housing consumption is difficult using standard empirical methods because of selection bias, regional rollout timing, and short duration of implementation. Federal Reserve Bank of Dallas+2SSRN+2

To overcome these challenges, Zhou constructs a heterogeneous-agent life‐cycle model calibrated to Chinese household survey data and conducts counterfactual policy experiments to quantify the effect of the Housing Provident Fund Program on two main outcome variables: the rate of homeownership and the average home size. Federal Reserve Bank of Dallas+2SSRN+2

Key results show that the Housing Provident Fund Program is predicted to raise homeownership by 8.7 percentage points (around a 14 % increase relative to baseline) and increase average home size by about 20 %. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

The remainder of the paper is structured as follows: model setup, calibration, baseline results, policy experiments, extensions (employer contributions, withdrawal by renters, housing supply), and policy implications. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

---

Model Framework

Baseline Model (No Housing Program)

Zhou begins with a life-cycle model in which households choose consumption, savings, and housing decisions over time. Households differ (heterogeneous agents) and face income paths, transaction costs, credit constraints, and liquidity constraints. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

Key features include:

• At each age $j$, a household has assets $a$ and housing status $h$.

• If a household does not yet own a home, it decides whether to remain a renter, or purchase a home of size h0h_0h0​, incurring a transaction cost $f$.

• The budget constraint: consumption plus next-period assets equals income plus returns on assets minus any housing purchases.

• A collateral constraint: borrowing (mortgage) is limited to a fraction $\gamma$ of home value. That is, down payments must be made, and mortgages are constrained by the house collateral.

• A liquidity constraint: liquid savings must be nonnegative.

• The decision to purchase arises when the value of owning surpasses that of not buying.

• The model generates age profiles for homeownership (rising with age) and roughly flat average home sizes. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

This baseline model is used as a counterfactual scenario (in which the Housing Provident Fund Program does not exist).

Incorporating the Housing Provident Fund Program

To analyze the Housing Provident Fund Program, the author modifies the baseline model by embedding two policy levers:

1. Mandatory saving parameter $\theta$: a fraction of labor income is compulsorily deposited into the fund. These savings accumulate until home purchase or retirement.

2. Mortgage subsidy spread: the interest rate offered through the fund is lower than the market rate. The difference ( rmarket−rHPFr_{\text{market}} - r_{\text{HPF}}rmarket​−rHPF​ ) captures the subsidy.

Under the Housing Provident Fund Program, households’ decisions are altered because:

• They accumulate housing-specific savings (the HPF deposits),

• Their borrowing costs are reduced for mortgages (making house financing cheaper),

• The timing and size of home purchases adjust.

Zhou also explores variants such as employer contributions (matching deposits) and allowing renters to withdraw part of deposits to cover rent or housing expenses. SSRN+3IDEAS/RePEc+3IDEAS/RePEc+3

---

Calibration and Data

Zhou calibrates the model using household-level micro data from surveys such as the Chinese Household Income Project (CHIPS) and other sources. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

He picks parameters so that the model closely matches empirical moments of Chinese households: income profiles by age, homeownership rates across ages, distribution of housing sizes, transaction cost parameters, discount factor, risk aversion, etc. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

For the policy parameters:

• The mandatory saving rate $\theta$ is set according to actual deposit rules in existing HPF regulations.

• The mortgage subsidy is set by the typical spread between market and HPF rates.

• In extensions, employer matching rates are varied.

• The housing supply elasticity is drawn from empirical estimates (e.g., Wang et al. (2012) estimate price elasticity of supply for Chinese housing). SSRN+3Federal Reserve Bank of Dallas+3IDEAS/RePEc+3

With this calibration, the author simulates the life-cycle paths of homeownership and home size under baseline (no HPF) and under the Housing Provident Fund Program.

---

Baseline vs. With the Housing Provident Fund Program: Key Results

Increase in Homeownership

Zhou finds that introducing the Housing Provident Fund Program increases the steady-state homeownership rate by 8.7 percentage points, which corresponds to a 14 % rise relative to the baseline no-program scenario. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

Mechanisms behind this uplift:

• Some households who would otherwise purchase later bring forward their purchase to an earlier age.

• Some households who would have remained renters under the baseline now decide to become homeowners (i.e., new entrants to homeownership).

• The mortgage subsidy lowers borrowing costs, relaxing constraints and allowing more households to qualify for mortgages.

• The accumulation of HPF mandatory savings helps households with down payments or reduces the financing burden at purchase time.

Thus, the Housing Provident Fund Program not only accelerates purchases but also expands the set of people able to own homes.

Increase in Home Size

In addition to more households owning homes, the model predicts that the Housing Provident Fund Program leads to a 20 % increase in average home size relative to baseline. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

This increase arises because:

• The subsidy and forced savings effectively raise households’ “housing budget,” enabling them to afford larger homes.

• Purchasers tend to select larger homes across all age groups, not just older ones.

• As new home sizes increase and existing homeowners upgrade over time, the cross-sectional average home size rises.

Savings Behavior & Distortions

Zhou studies how the Housing Provident Fund Program influences aggregate savings patterns.

• Under the Housing Provident Fund Program, households must deposit $\theta$ fraction of income into the HPF. This reduces their free (private) savings capacity.

• Some savings distortion arises: non-homeowners might reduce non-HPF savings because a portion is locked in the fund.

• For homeowners, since they no longer contribute to HPF after purchase (in some model versions), the timing and distribution of savings are altered.

• Interestingly, the program does not uniformly reduce welfare: the benefits from increased housing consumption and subsidized credit outweigh some distortions.

Zhou quantifies how much of the total savings change is due to mandatory deposits versus changes in private savings. He also shows age profiles of total savings before and after implementation. SSRN+3Federal Reserve Bank of Dallas+3IDEAS/RePEc+3

---

Policy Experiments & Extensions

Zhou conducts multiple counterfactual experiments to understand which features of the Housing Provident Fund Program are most influential, and how modifications (employer contributions, renter withdrawal, housing supply) affect results.

Employer Matching / Contributions

One extension considers having employers also contribute or match deposits (rather than only employees). Key findings:

• Employer contributions effectively raise household total wealth, further easing the constraints on housing purchases and consumption.

• When employers match for both renters and homeowners, the average home size and welfare further increase (beyond the base HPF design).

• In extreme cases, if matching is very generous (e.g., $x=1$ matching ratio), homeownership can decline slightly (because the deposit burden may lead to distortions), but home size and welfare rise significantly.

• Matching homeowners’ contributions leads to additional disposable income and utility enhancement.

• The strongest impacts are seen when both renters' and homeowners' deposits are matched, rather than only renters. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

Allowing Renters to Withdraw HPF Deposits

Some Chinese cities allow non-homeowners (renters) to withdraw part of their accumulated HPF deposits to pay for rent or cover housing-related costs. Zhou models this option to see how it affects behavior under the Housing Provident Fund Program:

• If renters can withdraw deposits, they have more flexibility, which may encourage them to maintain or increase contributions or improve welfare.

• Without employer matching, allowing withdrawals does not meaningfully change the baseline HPF effects (on homeownership and home size).

• But when combined with employer matching, interactive effects strengthen: allowing withdrawal under matching yields further increases in homeownership and average home size, and also benefits renters’ utility.

• This design helps mitigate liquidity constraints for renters under HPF.

• However, the more generous withdrawal rights may reduce the incentive to use HPF savings strictly for home purchases, potentially weakening some of the original objectives. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

Housing Supply & General Equilibrium Effects

Thus far, analysis assumes housing supply is perfectly elastic (i.e., additional demand does not raise price). Zhou introduces housing supply constraints to assess general equilibrium effects of the Housing Provident Fund Program:

• He estimates demand curves (before and after HPF) using simulated data and assumes a linear-log specification.

• He uses external empirical estimates of housing supply elasticity (from Wang et al. (2012)) to link demand shifts to price changes.

• With a lower bound supply elasticity, the model predicts that the Housing Provident Fund Program would increase house prices by 4.7 % and quantity (units sold) by 13.7 %.

• With a higher bound supply elasticity, house prices rise by 2.7 %, and quantity rises by 15.8 %.

• Thus, much of the effect of HPF may be absorbed into higher prices, reducing the net gains in housing quantity and welfare.

• The relative elasticity of supply matters: more inelastic supply magnifies price increases and dampens quantity responses.

• Nonetheless, the Housing Provident Fund Program retains substantial positive effects on quantity and homeownership even under realistic supply conditions. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

---

Mechanisms and Comparative Importance of Features

Zhou dissects which part of the Housing Provident Fund Program drives the results and how the mechanisms operate.

• The mortgage subsidy (lower interest rate) is found to be a dominant driver: when only the subsidy is applied (without mandatory savings), much of the increase in homeownership is preserved.

• The mandatory savings component also contributes, especially in aiding down payments and smoothing liquidity constraints.

• Employer matching enhances the impact further.

• The ability to withdraw (for renters) works best in synergy with matching.

• In sum, the Housing Provident Fund Program works best when combining mandatory savings, subsidized mortgages, employer participation, and flexible withdrawal options (carefully designed).

• However, when housing supply is constrained, a large share of gains may flow into housing prices rather than additional home supply or consumer welfare.

• The model also highlights the trade-off between savings distortions (lock-in of HPF deposits) and housing gains.

Zhou compares his results with previous literature (e.g. Tang & Coulson (2017), Buttimer et al. (2004)) that used representative-agent or partial-equilibrium models, noting those tend to understate heterogeneity and misestimate distributional effects. Federal Reserve Bank of Dallas+2IDEAS/RePEc+2

---

Policy Implications and Caveats

Zhou draws several policy lessons and caveats relevant to China and other countries considering similar programs.

Policy Lessons

1. Effectiveness of HPF-style programs
   The simulation suggests the Housing Provident Fund Program can significantly promote homeownership and housing consumption. The combined effects of mandatory savings and subsidized credit are powerful.

2. Design matters
   The details of the program—subsidy levels, matching contributions, withdrawal flexibility—strongly influence outcomes. Policymakers must balance between incentivizing housing investment and avoiding distortions or misuse.

3. Supply-side considerations
   Policies like HPF must be coordinated with housing supply responses. If supply is inelastic, much of the benefit may show up in price inflation rather than quantity or welfare improvements.

4. Redistribution and inequality
   The benefits of the Housing Provident Fund Program may accrue differently across income groups. In related literature, participation tends to help poorer households more (in terms of housing wealth) because they otherwise face binding constraints. (See, e.g., Xu (2023) on HPF’s distributional effects) SSRN

5. Institutional capacity & enforcement
   Effective administration, enforcement of mandatory savings, monitoring, and transparent allocation of funds are essential for HPF success.

6. Side effects and distortions
   Policymakers should guard against distortions in private savings, rigidities caused by locked-in deposits, and reduced flexibility for households.

Caveats & Limitations

• The model is structural and relies on calibration; actual behavior may diverge due to unmodeled frictions or preferences.

• The assumption that households act optimally and have full foresight may overstate responses.

• Supply-side modeling is simplified; in reality, local land use, zoning, regulatory constraints, and local heterogeneity matter.

• Some behavioral, institutional, or financial-market features (e.g. credit market imperfections, default risk) are abstracted away.

• The program effects take time to materialize; transitional dynamics are complex but not fully explored.

• Local heterogeneity in China (urban vs rural, provincial differences) is not deeply modeled.

Zhou acknowledges these limits and suggests further work to include more realism, local variation, and transitional dynamics.

---

Concluding Remarks

In summary, this paper provides an in-depth structural evaluation of the Housing Provident Fund Program in China. By embedding the mandatory savings and subsidized mortgage features into a life-cycle heterogeneous-agent model, Zhou quantifies how the program would affect homeownership, housing size, savings behavior, and welfare.

The key findings are:

• The Housing Provident Fund Program increases homeownership by about 8.7 percentage points (≈14 %) compared to a no-program baseline.

• The program also boosts the average home size by around 20 %.

• Mortgage rate subsidies are the primary driver of impact, with mandatory savings and employer contributions augmenting effects.

• Allowing renters to withdraw deposits has moderate additional benefits, especially when matched by employers.

• In a general equilibrium with limited housing supply, the program leads to price inflation (2.7–4.7 %) and a smaller net gain in quantity, but still yields positive welfare and homeownership effects.

• Design features (subsidy size, matching, flexibility) and supply-side constraints are crucial to effectiveness.

Overall, the study offers strong support for the Housing Provident Fund Program (or analogues) as a policy tool to promote homeownership, provided the institutional and market conditions are favorable. It also provides caution about pricing effects, distortions, and the importance of supply coordination.

Also Read: Mitigating the Portland Housing Crisis: Analyzing How the City of Portland's Policies Can Better Facilitate Continuous Affordable Housing Development