Investing a share of the Shenzhen salary in US index funds: a 20-year study
10% of salary a month over the past 20 years: QQQ ¥1.92m, SPY ¥1.05m

1. Historical backtest: investing 10% of salary a month (¥1,129 in today's money) from 2006-10 to 2026-09, ¥0.27m in total, would have produced ¥1.92m in QQQ (7.1x), ¥1.44m in the 50/50 mix (5.3x) and ¥1.05m in SPY (3.9x). Real money-weighted returns in yuan were 17.0%, 14.7% and 12.1% a year.
2. Savings ratio: results scale with the amount invested. At 20% of salary the end values double, and at 30% they triple: ¥5.76m in QQQ and ¥3.15m in SPY.
3. Starting date: across 224 rolling 20-year windows starting monthly from 1988-02 (real US dollars), the median ratio of end value to amount invested was 1.90x for SPY and 2.59x for QQQ, with ranges of 0.98x–3.34x and 1.29x–6.27x. The latest window beat 99% of all windows; for the window ending at the financial-crisis low of 2009-02, SPY was worth less than the amount invested in real terms.
4. Forward scenarios: at base-case real returns of 6% for SPY and 7% for QQQ, 10% of salary a month grows to ¥0.51m in SPY, ¥0.58m in QQQ and ¥0.54m in the mix after 20 years, equal to 46–51 months of salary and well below the backtest.

• Salary: Shenzhen's 2025 pension contribution base of ¥11,293 a month, which equals the city's average wage (contribution index 1.0).
• Savings ratio: 10%, 20%, 30% of salary, i.e. ¥1,129, ¥2,259, ¥3,388 a month, invested on the first trading day of each month for 240 months.
• Purchasing power: the monthly amount is constant in today's money, with no real wage growth. In the backtest, past contributions are deflated to the prices of the time by China's CPI, so end values are in today's yuan.
• Funds: SPY tracks the S&P 500 and QQQ the Nasdaq-100, using total-return (dividends reinvested) prices. The mix invests half of each contribution in each fund and rebalances to 50/50 every December.
• Exchange rate: purchases convert yuan at that day's USD/CNY rate; end values convert back at 6.7030 on 2026-09-30.
• Wage growth: the main estimates assume no real wage growth; section 06 adds a sensitivity case with real wages rising 3% a year.
• Not included: dividend withholding tax, conversion and trading costs, premiums on onshore QDII funds and income tax.

10% of salary a month (¥1,129 in today's money). All amounts in today's yuan. The largest paper loss is the deepest shortfall of account value below the amount invested so far, with the month it occurred.

Early contributions coincided with the 2008 financial crisis, and all three options showed paper losses of more than 40% around 2009-02. The long advance that followed determined the end result.

Exchange rate: the yuan rose from 7.90 to 6.70 per dollar over the period, but purchase rates ranged from 6.04 to 7.90, averaging 6.75, and 75% of contributions were made below 7.0. Had the rate stayed at 7.90 throughout, end values would differ by no more than 0.2%. Regular purchases spread the conversion dates, so the exchange rate had little effect in this period.

Results are proportional to the monthly amount; multiples and returns do not change with the ratio. The table shows end values from the 20-year backtest (today's money).

A new 20-year plan starts every month from 1988-02 to 2006-09, 224 windows in all, measured in real US dollars (adjusted by US CPI) and excluding exchange-rate effects. Before SPY's launch (1993-01) the S&P 500 total-return index is used, and before QQQ's launch (1999-03) the Nasdaq-100 price index. The multiple is end value over amount invested.

Outcomes depend mainly on the end date. The lowest values come from the window that began in 1989-03 and ended in 2009-02, at the low of the financial crisis, where SPY returned 0.98x in real terms, below the amount invested. The highest come from windows that began in 2005–2006. The latest window (starting 2006-09, nearly the same period as the backtest in section 03) outperformed 99% of all windows. The windows overlap, so there are far fewer independent observations than 224.

Scenarios assume real annual returns after fees, consistent with this site's pension comparison: SPY 4.0%/6.0%/8.0%; QQQ 4.0%/7.0%/10.0%; 50/50 mix 4.0%/6.5%/9.0% (conservative/base/optimistic; the mix uses the average of the two). The conservative case uses the same 4% for all three, so the results are identical. For reference, the median real return across the 224 historical windows in section 05 was QQQ 8.7%, 50/50 mix 7.5%, SPY 6.0%, close to the base case. These are assumptions, not forecasts.

Real returns over the past 20 years (12.1%–17.0%) exceeded every scenario assumption, including the optimistic one. If returns revert to long-run averages, the same contributions would be worth about 49% of the backtest result for SPY and 30% for QQQ after 20 years.

If real wages rise 3% a year and the savings ratio stays at 10%, the monthly amount grows from ¥1,129 to ¥2,040 in year 20, and the total invested rises from ¥0.27m to ¥0.37m. The table shows base-case end values; against the year-20 monthly salary of ¥20,396, they equal 32–36 months of salary.