Working Capital Efficiency Across 15 Markets: Where It Works and Where It Fails

We tested working capital efficiency as a stock selection signal on 15 exchanges worldwide. 11 outperformed their local benchmark. 4 underperformed. 2 had too many cash periods to draw conclusions.

Contents

1. The Signal
2. All 15 Exchanges
3. The Outperformers
4. UK: The Surprise Winner (+7.75%/yr vs FTSE 100)
5. Canada: Most Defensive (+5.44%/yr vs TSX Composite)
6. Hong Kong: Local Alpha (+4.07%/yr vs Hang Seng)
7. Germany: Strong vs DAX (+3.47%/yr)
8. India: Highest Absolute Return (15.2% CAGR, +2.88%/yr vs Sensex)
9. US: Best Risk-Adjusted (Sharpe 0.515, +2.47%/yr)
10. China, Australia, Sweden, Japan: Moderate Outperformers
11. The Underperformers
12. Switzerland (-1.08%/yr vs SPY)
13. Taiwan (-1.52%/yr vs TAIEX)
14. Thailand (-3.29%/yr vs SPY)
15. Saudi Arabia (-6.70%/yr vs SPY)
16. Why It Works in Most Markets
17. The High-Cash Exchanges
18. Run It Yourself
19. Takeaway

The key change from our earlier analysis: we now benchmark each exchange against its own local index (Sensex for India, DAX for Germany, FTSE for the UK) instead of comparing everything to the S&P 500. This is more honest because it compares same-currency returns. The picture looks different.

The Signal

Working capital divided by revenue. Lower means the company converts revenue to cash with less capital tied up in receivables and inventory. We screen for WC/Revenue < 50%, revenue growth > 0, ROE > 8%, operating margin > 10%, and market cap > $500M. Top 30 by WC/Revenue ascending, equal weight, annual rebalance in June, with transaction costs.

The academic basis: Sloan (1996) on accruals, Hirshleifer et al. (2004) on bloated balance sheets. Companies with lean working capital tend to have higher-quality earnings and better subsequent stock returns.

All 15 Exchanges

Each exchange is benchmarked against its local index (same currency). Exchanges without a local index in FMP use the S&P 500.

*Korea and Saudi Arabia had >30% cash periods. The quality filters are too strict for these markets. Their results are included for completeness but shouldn't be compared directly.

Note: Sweden, Thailand, Switzerland, and Saudi Arabia use S&P 500 as benchmark because their local indices lack price data in FMP's stock_eod table.

The Outperformers

Eleven exchanges beat their local benchmark. Each tells a different story.

UK: The Surprise Winner (+7.75%/yr vs FTSE 100)

The UK was previously excluded from our analysis due to data gaps. FMP's coverage has since improved. The result: 8.87% CAGR vs just 1.13% for the FTSE 100. That's +7.75% annual excess and an 84% win rate, the highest of any exchange. Max drawdown of -22.68% is the second-lowest globally.

The FTSE 100 has been one of the weakest major indices over this period (1.13% CAGR), so the working capital screen is partly benefiting from stock-picking in a market where the index drags. But 21 of 25 years beating the index is hard to dismiss.

Canada: Most Defensive (+5.44%/yr vs TSX Composite)

Canada had the shallowest max drawdown: -22.29%. The TSX Composite's was -28.80%. The CAGR of 9.51% beats the TSX by 5.44% per year. Zero cash periods, 24.8 average stocks, and a down capture of just 10.5%.

The trade-off: Canada's TSX Composite returned only 4.06% over this period. The large excess partly reflects a weak benchmark.

Hong Kong: Local Alpha (+4.07%/yr vs Hang Seng)

Hong Kong previously appeared to underperform when compared to SPY. Against the Hang Seng (which returned just 1.68% CAGR over 25 years), the portfolio's 5.74% CAGR looks much better: +4.07% annual excess with 56% win rate.

Germany: Strong vs DAX (+3.47%/yr)

Germany's +3.47% excess vs the DAX is much stronger than the old +0.77% vs SPY. The DAX returned 4.78% over this period, and German quality companies with lean working capital added meaningful alpha. Sharpe of 0.422, zero cash periods, 20.8 average stocks.

India: Highest Absolute Return (15.2% CAGR, +2.88%/yr vs Sensex)

India still has the highest absolute CAGR at 15.20%. But against the Sensex (which returned 12.32%), the excess is a more modest +2.88%. Five cash periods (2000-2004) when too few Indian stocks qualified. The Sensex itself was strong, so the bar was higher.

US: Best Risk-Adjusted (Sharpe 0.515, +2.47%/yr)

The US returned 10.10% CAGR with +2.47% excess vs SPY. The Sharpe ratio of 0.515 is the best of any exchange. Down capture of 22%: for every dollar the market lost, this portfolio lost 22 cents. Zero cash periods, fully invested for 25 years.

China, Australia, Sweden, Japan: Moderate Outperformers

China (+1.46% vs SSE Composite), Australia (+2.99% vs ASX 200), Sweden (+1.17% vs SPY), and Japan (+1.06% vs Nikkei) all showed positive excess. These were previously classified as underperformers when everything was compared to SPY. Using local benchmarks reveals that the signal works in these markets too, just not as strongly.

The Underperformers

Four exchanges had negative excess returns vs their local benchmark. This is expected. No strategy works everywhere.

Switzerland (-1.08%/yr vs SPY)

Only 14.2 stocks on average (the smallest of any exchange). Switzerland has a concentrated market dominated by Nestle, Novartis, and Roche. The quality filters are too narrow for such a concentrated index. No local benchmark available in FMP.

Taiwan (-1.52%/yr vs TAIEX)

Taiwan had 28% cash periods and -1.52% excess vs TAIEX. The TSE is semiconductor-heavy. TSMC and its supply chain dominate the index. Working capital efficiency in a hardware-heavy market works differently than in services or consumer sectors.

Thailand (-3.29%/yr vs SPY)

The SET had thin early data and higher transaction costs for mid-cap Thai stocks. No local benchmark available in FMP, so this compares to SPY (unfavorable due to currency and market differences).

Saudi Arabia (-6.70%/yr vs SPY)

Saudi had 36% cash periods and -6.70% excess. The quality filters are too strict for this market. Thin early data and no local benchmark.

Why It Works in Most Markets

The switch to local benchmarks changed the story. The signal works in 11 of 15 markets. Three patterns explain the variation:

Benchmark quality matters most. The biggest driver of excess returns is how the local index performed. The UK (+7.75%) and Canada (+5.44%) have high excess partly because the FTSE 100 and TSX Composite were weak indices over this period. The US (+2.47%) has lower excess because SPY is a hard benchmark to beat.

Market composition matters. The strategy works best in markets with a broad mix of sectors (US, UK, Canada, Germany). It still struggles in markets dominated by a single sector (Taiwan = semiconductors), but this effect is smaller than the benchmark quality effect.

Coverage depth matters. Less analyst coverage creates more mispricing. India and the UK, with thinner mid-cap coverage, produced strong excess returns. The US, the most covered market, still generated +2.47% excess, suggesting the signal has genuine economic content beyond mispricing.

The High-Cash Exchanges

Saudi Arabia (36% cash) and Korea (36% cash) had too many periods without enough qualifying stocks. The market cap thresholds combined with strict quality filters eliminated too many companies. Korea's 3.57% CAGR vs KOSPI's 5.20% includes 9 cash periods at the start.

These results aren't comparable to the other 13 exchanges. They're included to document what happens when the signal's universe is too narrow.

Run It Yourself

# Run all exchanges
python3 working-capital/backtest.py --global --output results/exchange_comparison.json --verbose

# Run a specific exchange
python3 working-capital/backtest.py --preset us --output results/us.json --verbose
python3 working-capital/backtest.py --preset india --output results/india.json --verbose

Takeaway

Working capital efficiency works as a stock selection signal in most markets when measured against the right benchmark. 11 of 15 exchanges beat their local index. The strategy's alpha is broader than we originally thought.

The biggest lesson: benchmark choice matters. When we compared everything to SPY, 8 exchanges appeared to underperform. When we switched to local benchmarks (Sensex, DAX, FTSE, etc.), 6 of those 8 showed positive excess returns. The signal was working. It just wasn't beating the US market from abroad.

The 4 underperformers (Switzerland, Taiwan, Thailand, Saudi Arabia) have specific structural reasons: concentrated markets, semiconductor dominance, missing local benchmarks, or too-strict filters.

Data: Ceta Research (FMP financial data warehouse), 15 exchanges tested, 2000-2025. Each exchange benchmarked against its local index. Full methodology: METHODOLOGY.md