In March 2017, China’s Ministry of Finance revised CAS 22 to replace the incurred loss model with the expected credit loss (ECL) model, requiring firms to use forward-looking information in credit risk assessment. This reform aligns with IFRS convergence. Prior literature focuses on historical information-based policies or examines the ECL model’s effects on banks’ loan loss provisions and market reactions, but little is known about its real impact on non-financial firms’ innovation from a forward-looking perspective. This paper investigates whether and how ECL implementation affects corporate innovation in China.

Using a sample of A-share non-financial listed firms from 2019 to 2023, this paper manually analyzes annual report texts to construct a novel measure of ECL implementation quality (ECLDegree), which counts the number of forward-looking information dimensions actually used by firms. The main dependent variables are patent applications and patent grants. The empirical strategy employs industry and year fixed effects, along with robustness tests including DID based on staggered adoption, IV using distance to optical cable backbone cities, Heckman two-stage, lagged dependent variables, alternative patent measures, and various fixed-effect specifications. The findings reveal that ECL implementation significantly increases both the quantity and quality of innovation. Mechanism testing identifies three channels: promoting trade credit financing, reducing supply chain turnover, and increasing R&D investment. Heterogeneity analysis shows a more prominent effect for firms with more diversified suppliers, stronger accounts receivable management, no financial restatements, and lower levels of financialization.

This paper makes the following contributions: First, it extends the literature on economic consequences of the ECL model from financial institutions to non-financial firms, particularly focusing on supply chain activities. Second, it complements the innovation literature by shifting attention from historical accounting information to forward-looking information embedded in accounting standards, revealing a new mechanism through which forward-looking information shapes corporate innovation decisions. Third, it develops a novel text-based measure of ECL implementation quality using manual annual report analysis, offering a new empirical tool for future research on credit risk management and accounting standard implementation.