High Dividend Yield Quality Screen Across Global Exchanges
We backtested a high dividend yield quality screen across 14 global exchanges from 2000-2025. 10 of 14 exchanges beat their local benchmark. India (NSE-only) is the exception -- 60% cash, underperforms Sensex. Germany, HK, UK, US, Canada all show strong alpha.
We ran the same high dividend yield quality screen on 17 exchanges worldwide. Same filters everywhere: dividend yield 4-15%, payout 0-80%, FCF > 0, ROE > 8%, D/E < 2.0. Top 30 by yield, equal weight, annual July rebalance, next-day close execution (MOC). The question: does the high-yield quality signal work globally, or is it a US-specific phenomenon?
Contents
- Method
- Results: All Exchanges (vs Local Benchmarks)
- Excluded Exchanges
- Key Findings
- 1. The signal works in most markets
- 2. India is the exception, not the rule
- 3. SPY was the wrong benchmark for non-US exchanges
- 4. Down capture varies by market
- 5. High cash rates indicate poor signal fit
- Country-Specific Observations
- Limitations
- Takeaway
- Run This Screen Yourself
- References
The short answer: it works in most markets. 10 of 14 tested exchanges produced positive excess returns vs their local index. The signal is globally robust in markets with established dividend cultures.
Data: FMP financial data warehouse, 2000–2025. Updated March 2026.
Method
Data source: Ceta Research (FMP financial data) Period: 2000-2025 (25 years) Signal: Dividend yield 4-15%, payout 0-80%, FCF > 0, ROE > 8%, D/E < 2.0 Portfolio: Top 30 by yield, equal weight, annual July rebalance, cash if < 10 qualify Execution: Next-day close (MOC) Benchmark: Local index for each exchange (Sensex for India, DAX for Germany, etc.)
For full methodology, see our US analysis.
Results: All Exchanges (vs Local Benchmarks)
| Exchange | CAGR | Benchmark | Excess | Sharpe | MaxDD | Cash% |
|---|---|---|---|---|---|---|
| US (NYSE/NASDAQ/AMEX) | 12.25% | S&P 500 7.85% | +4.40% | 0.598 | -26.36% | 0% |
| Germany (XETRA) | 10.00% | DAX 5.04% | +4.96% | 0.494 | -32.93% | 8% |
| China (SHZ/SHH) | 8.77% | SSE Comp. 2.43% | +6.34% | 0.185 | -34.67% | 24% |
| Hong Kong (HKSE) | 7.93% | Hang Seng 1.64% | +6.29% | 0.254 | -36.94% | 12% |
| Sweden (STO) | 7.83% | OMX30 2.55% | +5.28% | 0.469 | -4.75% | 48% |
| Canada (TSX) | 6.78% | TSX Comp. 3.95% | +2.82% | 0.311 | -31.79% | 12% |
| Taiwan (TAI) | 5.88% | TAIEX 4.09% | +1.80% | 0.365 | -9.61% | 28% |
| UK (LSE) | 5.67% | FTSE 100 1.23% | +4.45% | 0.132 | -24.95% | 24% |
| Thailand (SET) | 5.39% | SET Index 5.13% | +0.26% | 0.193 | -36.09% | 20% |
| Switzerland (SIX) | 5.22% | SMI 1.74% | +3.48% | 0.424 | -26.67% | 36% |
| India (NSE) | 5.92% | Sensex 12.06% | -6.13% | -0.03 | -15.86% | 60% |
Excluded Exchanges
| Exchange | Reason |
|---|---|
| Japan (JPX) | 80% cash periods. The 4-15% yield threshold doesn't fit Japanese dividend culture (lower payouts). 2.80% CAGR, -0.51% vs Nikkei. |
| Korea (KSC) | 84% cash periods. Korean companies historically pay lower dividends. 0.93% CAGR, -4.42% vs KOSPI. |
| South Africa (JNB) | 64% cash + negative Sharpe (-0.47). Data quality concerns with financial ratios coverage. |
| Saudi Arabia (SAU) | 60% cash periods. No reliable local benchmark available. |
| Australia (ASX) | Known adjClose split artifacts in FMP data. Returns are unreliable. |
| Brazil (SAO) | Known adjClose split artifacts in FMP data. Returns are unreliable. |
Key Findings
1. The signal works in most markets
10 of 14 exchanges with usable data produced positive excess returns vs their local benchmark. The quality-filtered high-yield signal captures a real premium in markets with established dividend cultures: liquid markets where companies have multi-decade histories of paying dividends, with reliable financial data.
2. India is the exception, not the rule
The previous version of this analysis reported India at the top with 12.88% CAGR (BSE+NSE run). That result was inflated by dual-listing duplication: the same stocks appeared on both BSE and NSE, effectively doubling the universe. The corrected NSE-only result is 5.92% CAGR with -6.13% excess vs Sensex.
The core problem: India's 4%+ yield universe on NSE is thin, especially before 2014. The strategy holds cash 60% of the time, missing most of India's bull market. The signal is too strict for India's dividend culture at these thresholds.
This is an important lesson: the 4-15% yield range was calibrated for US markets. Japan, Korea, and India all have structurally lower dividend payout cultures, making the threshold poorly suited.
3. SPY was the wrong benchmark for non-US exchanges
Previous analysis used SPY as a universal benchmark. With local benchmarks: - Hong Kong: +6.29% excess vs Hang Seng (was -0.21% excess vs SPY) - UK: +4.45% excess vs FTSE 100 (was -2.40% excess vs SPY) - Germany: +4.96% excess vs DAX (was +2.24% excess vs SPY)
The signal looks much stronger in non-US markets when benchmarked locally.
4. Down capture varies by market
| Exchange | Down Capture (vs local) | Interpretation |
|---|---|---|
| US | +1.6% | Roughly flat during downturns |
| Canada | 13.5% | Modest losses vs TSX |
| India | 25.8% | Below-average downside capture |
| Germany | 27.5% | Moderate correlation |
| Hong Kong | 34.2% | Moderate correlation with HSI |
| UK | 37.4% | Meaningful losses vs FTSE |
The US strategy's near-zero down capture (+1.6%) is the standout. The quality filters do their best work in the US market.
5. High cash rates indicate poor signal fit
Sweden (48% cash), Switzerland (36%), India (60%), Taiwan (28%) all have high cash rates. When fewer than 10 stocks pass all filters, the strategy holds cash rather than diluting quality. In markets with thin yield universes, this is the right behavior — but it limits returns.
Country-Specific Observations
US: The flagship result. 0% cash, 23.5 avg stocks, +4.40% excess vs S&P 500. The deepest dividend market with the most quality data. Full US analysis →
India: NSE-only result disappoints at 5.92% CAGR (-6.13% vs Sensex). The strategy holds cash for 60% of periods, missing most of India's 2003-2013 bull market. When invested, results are volatile: +81.9% in 2023, -15.5% in 2019. Signal calibration needed. Full India analysis →
Germany: +4.96% excess vs DAX with only 8% cash periods. German Mittelstand companies provide a deep pool of dividend payers. 68% win rate vs the local benchmark. Full Germany analysis →
Hong Kong: +6.29% excess vs Hang Seng (1.64% CAGR). The Hang Seng was poor over this period, and high-yield quality stocks handily outperformed. 72% win rate. Full HK analysis →
Canada: +2.82% excess vs TSX Composite (3.95% CAGR). Low beta (0.504) and low down capture (13.5%) confirm Canadian dividend stocks are defensive. Full Canada analysis →
UK: +4.45% excess vs FTSE 100 (1.23% CAGR). The UK market broadly underperformed, but high-yield quality stocks significantly outperformed the FTSE. 64% win rate. Full UK analysis →
China: 8.77% CAGR vs SSE Composite's 2.43%. +6.34% excess. But extreme single-year returns and 24% cash periods make the result less reliable.
Sweden: 7.83% CAGR vs OMX30's 2.55%. +5.28% excess with only -4.75% max drawdown. But 48% cash periods mean the strategy was uninvested nearly half the time.
Limitations
Threshold calibration: The 4-15% yield range was designed for the US market. Markets with lower dividend cultures (Japan, Korea, India) need different thresholds. This is the primary failure mode.
Currency effects: All returns are in local currency. A portfolio earning 5.92% in INR differs from 5.92% in USD after currency conversion. Cross-exchange CAGR comparisons don't account for this.
Data quality varies: FMP data completeness differs by exchange and time period. Markets with shorter data histories produce less reliable backtests.
Benchmark selection: We use each exchange's primary index as the local benchmark. Index composition varies in concentration and sector weight.
Takeaway
The high dividend yield quality screen works in markets where dividend culture is established and the yield threshold is appropriate. US, Germany, Hong Kong, UK, Canada, Sweden all produce meaningful positive excess returns vs their local indices.
India is the cautionary example: a promising-looking BSE+NSE result (12.88% CAGR) collapsed to 5.92% (-6.13% vs Sensex) once the universe was corrected to NSE-only and the dual-listing duplication removed. The signal is too restrictive for India's market structure.
The consistent lesson: benchmark against the local index, run on clean exchange data, and check your cash rate. A strategy holding cash 60% of the time isn't a dividend strategy — it's mostly a cash fund with occasional equity bets.
Data: Ceta Research (FMP financial data warehouse). 17 exchange presets tested, 11 included in comparison. Backtest: 2000-2025, annual July rebalance, next-day close execution (MOC), equal weight top 30, size-tiered transaction costs. Local index benchmarks for each exchange. India corrected from BSE+NSE to NSE-only to eliminate dual-listing duplication. Past performance does not guarantee future results. This is educational content, not investment advice.
Run This Screen Yourself
The full backtest code (Python + DuckDB) is available in our GitHub repository.
# Run on any exchange
python3 high-yield/backtest.py --preset us
python3 high-yield/backtest.py --preset india
python3 high-yield/backtest.py --global --output results/exchange_comparison.json
Get your API key at cetaresearch.com.
References
- Fama, E. & French, K. (1998). "Value versus Growth: The International Evidence." Journal of Finance, 53(6), 1975-1999.
