“学海拾珠”系列之二百五十四：海外主动基金业绩基准的设置与纠偏

Quantitative Models and Construction Methods - Model Name: Benchmark Mismatch Definition Model Construction Idea: Define benchmark mismatch as the inconsistency between a fund's self-declared benchmark and its Morningstar investment category [3][28] Model Construction Process: 1. Use Morningstar's 3×3 style box classification to categorize funds and benchmarks based on size (large-cap, mid-cap, small-cap) and investment style (growth, blend, value) [28] 2. If a fund's declared benchmark does not align with its style box classification, it is considered mismatched [28] Model Evaluation: Provides a clear framework to identify mismatched benchmarks, but sensitive to style box boundaries [28] - Model Name: IS Breadth (Investment Strategy Breadth) Model Construction Idea: Measure the extent to which a fund's holdings deviate from its core investment category [3][32] Model Construction Process: 1. Categorize funds into nine Morningstar style boxes based on market capitalization and valuation metrics [32] 2. Calculate the proportion of holdings outside the fund's core category [32] 3. Standardize the IS Breadth metric to have a mean of zero and a standard deviation of one [32] Model Evaluation: Successfully captures the flexibility and breadth of investment strategies, validated against alternative metrics [33][37] Model Backtesting Results - Benchmark Mismatch Model: - Benchmark mismatch probability decreases by 0.762% annually on average [53] - For specialized funds (low IS Breadth), the decline rate is 1.06% annually, while for broad strategy funds (high IS Breadth), the rate is 0.482% annually [53] - IS Breadth Model: - IS Breadth positively correlates with fund name broadness (+8.4% probability per standard deviation increase) [35] - Higher IS Breadth increases the likelihood of style drift (+5-6% probability per standard deviation increase) [35] Quantitative Factors and Construction Methods - Factor Name: Bias and Variance in Benchmark Mismatch Factor Construction Idea: Assess the performance manipulation or risk hedging motives behind mismatched benchmarks [60] Factor Construction Process: 1. Calculate bias as the average monthly return difference between the most matched benchmark and the self-declared benchmark over 36 months [60] 2. Calculate variance as the average squared return difference between the two benchmarks over 36 months [60] Factor Evaluation: Specialized funds show higher bias and unchanged variance, indicating performance manipulation, while broad strategy funds exhibit lower bias and variance, suggesting risk hedging motives [61][63] Factor Backtesting Results - Bias and Variance Factor: - IS Breadth reduces bias by 0.0132% per standard deviation increase [63] - Variance decreases slightly for broad strategy funds, supporting risk hedging motives [63] - Factor Name: Systematic Risk Loadings and Return Differences Factor Construction Idea: Compare initial and final benchmarks to analyze systematic risk exposure and return differences [65] Factor Construction Process: 1. Use Fama-French three-factor regression to calculate beta differences for market, SMB, and HML factors between initial and final benchmarks [65] 2. Analyze 36-month return differences between initial and final benchmarks [65] Factor Evaluation: Specialized funds tend to choose initial benchmarks with lower market and SMB exposure but higher HML exposure, aligning with performance manipulation motives [66] Factor Backtesting Results - Systematic Risk Loadings Factor: - Initial benchmarks show lower market beta (-0.0387) and SMB beta (-0.131) compared to final benchmarks [66] - HML beta is higher for initial benchmarks (+0.0523), reflecting value tilt during periods of negative value premium [66] - Return Differences Factor: - Initial benchmarks underperform final benchmarks by 4.43% over 36 months, driven by systematic risk differences [66] Economic Channels and Observations - Investor Learning: - Investors react more strongly to new benchmark-adjusted returns after benchmark switches, especially for specialized funds [70][72] - Institutional Supervision: - Funds with institutional "twin pairs" are 4% less likely to have mismatched benchmarks [74] - Market Competition: - Higher competition from index funds increases the likelihood of benchmark switches by 1.4% per standard deviation increase in competition intensity [77][79] - Relative Performance and Risk: - Funds switch benchmarks to improve relative performance and reduce tracking error, with stronger effects for specialized funds [82][83]