2026年2月社融预测：23417亿元

Quantitative Models and Construction Methods Model Name: Bottom-up Social Financing (社融) Forecasting Framework - Model Construction Idea: The model is built to predict the total and structural components of social financing (社融) by breaking it down into sub-items. This bottom-up approach leverages the unique characteristics of each sub-item to enhance prediction accuracy[10][11]. - Model Construction Process: 1. Sub-item Decomposition: Social financing is divided into sub-items such as RMB loans, government bonds, corporate bonds, etc. Each sub-item is analyzed based on its economic logic, high-frequency data, and seasonal characteristics[10][11]. 2. Prediction Methods for Sub-items: - RMB Loans: - Decomposed into corporate loans and short-term residential loans, which are predicted using PMI and Tangshan steel plant capacity utilization as independent variables in rolling regression models. - Long-term residential loans are predicted using a three-stage model based on commodity housing sales data[11]. - Corporate Bill Financing: Predicted using rediscount rates as exogenous variables in a rolling regression model with a 5-year window[11]. - Government Bonds: High-frequency issuance and maturity data are tracked, with adjustments made for discrepancies in reporting periods[11]. - Corporate Bonds: A 5-year rolling regression model is used to reweight sub-items, reducing discrepancies in reporting standards[11]. - Other Sub-items: - Foreign currency loans: Predicted using a 3-month moving average[11]. - Trust loans: Estimated by tracking the issuance and maturity of collective and single trust products[11]. - Entrusted loans: Predicted using a 12-month moving average, with additional judgment for infrastructure-related increments[11]. - Undiscounted bank acceptance bills: Estimated using the average of the same period over the past three years due to the cessation of high-frequency data publication[11]. - Non-financial corporate domestic equity financing: Derived by subtracting financial sector data from high-frequency equity financing data (e.g., IPOs, additional issuances)[11]. - Loan write-offs: Predicted using the same period value from the previous year due to significant seasonal effects[11]. - Asset-backed securities (ABS): Tracked using high-frequency data on credit ABS net financing[11]. Model Evaluation - The bottom-up framework provides detailed structural insights into social financing while maintaining high accuracy in total volume predictions. It effectively captures the unique characteristics of each sub-item, enhancing the model's robustness and reliability[10][11]. --- Model Backtesting Results Social Financing Forecasting Framework - February 2026 Total Social Financing Prediction: 23,417 billion RMB, a year-on-year increase of 110 billion RMB[10][20]. - TTM Growth Rate: 0.29% month-on-month[10][20]. - Structural Predictions: - RMB Loans: 9,286 billion RMB (Corporate loans and short-term residential loans: 8,503 billion RMB; Long-term residential loans: -747 billion RMB)[20]. - Corporate Bill Financing: 1,530 billion RMB[20]. - Government Bonds: 14,199 billion RMB[20]. - Corporate Bonds: 1,829 billion RMB[20]. - Other Sub-items: - Foreign Currency Loans: -143 billion RMB[20]. - Trust Loans: -212 billion RMB[20]. - Entrusted Loans: 47 billion RMB[20]. - Undiscounted Bank Acceptance Bills: -2,247 billion RMB[20]. - Non-financial Corporate Domestic Equity Financing: 357 billion RMB[20]. - Loan Write-offs: 542 billion RMB[20]. - Asset-backed Securities: -240 billion RMB[20].