We investigate the cross-sectional pricing power of factors in the Chinese A-share market under a rigorous multiple-testing framework. We construct 139 characteristicmanaged portfolios and 30 principal components (PCs) derived from instrumented PCA (IPCA), risk-premium PCA (RPPCA), and traditional PCA. Using individual stocks as test assets and a bootstrap-based panel regression approach, we evaluate these candidates' incremental explanatory power while addressing data mining and p-hacking concerns. We find that latent factors constructed via IPCA and RPPCA-notably IPC9 and RPPC1-exhibit significant pricing power, whereas traditional PCs fail to provide incremental information. Among characteristic-based portfolios, three-year capital expenditure (CAPEX) growth, debt growth, and eight-quarter earnings consistency stand out by outperforming PCA-derived factors. Beyond statistical robustness, we provide new economic insights by linking factor efficacy to China-specific market frictions. Specifically, we show that IPC9 proxies for retail-driven noise trader risk and RPPC1 captures a flight-to-quality premium in large-cap stocks, yet the pricing power of both factors is strictly confined to market segments accessible to arbitrage capital. Furthermore, we document a structural migration in the CAPEX factor's efficacy from state-owned to private enterprises, mirroring the secular evolution of China's industrial policy. Further analysis under value-weighted schemes, small-cap exclusions, and sectoral filters underscores market heterogeneity: large-cap investors prioritize profitability and liquidity, mid-caps exhibit sensitivity to institutional trading dynamics, and non-financial stocks reward long-term investment efficiency. These findings contribute to a better understanding of conventional "factor zoo" paradigms and advocate for hybrid models that integrate standalone characteristics with synthetic risk proxies.
展开