经济指数系列报告（二）：基于高频数据的PMI预测方法

Group 1: Report Industry Investment Rating - No relevant information provided Group 2: Core Viewpoints of the Report - The report constructs a fitting prediction system based on high - frequency data for the manufacturing PMI. The historical back - test shows that the directional prediction win - rates of this method for the production, new orders, and comprehensive PMI are over 82%, 62%, and 70% respectively, providing an effective quantitative analysis framework for predicting manufacturing trends [4]. Group 3: Summaries According to the Directory 1. Basic Definition and Components of the Purchasing Managers' Index - Definition of PMI: The Purchasing Managers' Index (PMI) is an internationally recognized leading indicator for macro - economic monitoring. The manufacturing PMI, with 50% as the "boom - bust watershed", reflects the manufacturing economic situation and is an important barometer for assessing manufacturing prosperity and predicting macro - economic trends [8]. - Components of Manufacturing PMI: It is calculated by weighting five key sub - items: new order index (30%), production index (25%), employment index (20%), (100 - supplier delivery time index) (15%), and main raw material inventory index (10%). Demand and supply sub - items have a combined weight of 55% and are the core factors for judging the PMI trend [9][11][12]. 2. Prediction Ideas and Indicator Screening for Manufacturing PMI - Prediction Ideas: A fitting prediction system centered on high - frequency data is constructed. The process includes screening high - frequency data affecting supply and demand, fitting the PMI production index and new order index with these data, and then fitting the manufacturing PMI [13][14]. - High - frequency Indicator Screening: For the PMI production index, 14 supply - side high - frequency indicators are selected, focusing on the steel and petrochemical industries. For the PMI new order index, high - frequency indicators cover bulk consumption and investment demand, personnel flow, external demand, and intermediate product prices [13][15]. 3. Fitting of the Manufacturing PMI Index - Fitting of PMI Production Index: After monthly averaging the selected high - frequency indicators, a regression analysis model is built. The historical prediction win - rate for the direction of the production index is over 82%, and the prediction for January 2026 is 50.36, consistent with the official data's directional judgment [16]. - Fitting of PMI New Order Index: Using the same framework, a high - frequency simulation combination with 16 core indicators is built. The historical prediction win - rate for the direction of the new order index is over 62%, and the prediction for January 2026 is 44.27, consistent with the official data's directional judgment [18]. - Fitting of Manufacturing PMI Index: After fitting the production and new order indices, the manufacturing PMI is predicted. The historical win - rate for the direction judgment of the manufacturing PMI is 70%, and the prediction for January 2026 is 46.2, consistent with the official data's directional judgment [20]. 4. Summary - The research selects representative daily or weekly indicators from supply and demand sides, simulates the changes of PMI production and new order indices, and finally fits the comprehensive PMI. The system shows high fitting accuracy and reliable prediction ability for the "boom - bust line" direction, enhancing the timeliness and forward - looking of macro - economic monitoring [22].