查詢結果分析
來源資料
第1筆 /總和 1 筆
/ 1 筆
頁籤選單縮合
題名 | 以高頻物價數據進行通膨預測=Inflation Nowcasting Using High Frequency Price Data |
---|---|
作者 | 蕭宇翔; 繆維正; | 書刊名 | 經濟論文叢刊 |
卷期 | 49:3 2021.09[民110.09] |
頁次 | 頁371-414 |
分類號 | 561.14 |
關鍵詞 | 即時預報; 通膨預測; 分解預測; 組合預測; Nowcasting; Inflation forecasting; Disaggregate forecasts; Combining forecasts; |
語文 | 中文(Chinese) |
中文摘要 | 食物與能源價格是影響台灣通膨率波動的重要因素,若能即時掌握兩者價格變化應能有助預測當月通膨率。本文從農產品批發市場交易行情站、畜產行情資訊網、漁產品全球資訊網、中油公司油品行銷事業部等相關網站,擷取蔬菜、水果、毛豬、家禽、漁產品與汽油之每日批發或零售價格,並運用此高頻資料對台灣消費者物價指數(consumer price index,CPI)年增率進行即時預報(nowcasting)。實證發現,運用主計總處公布上月CPI統計時可取得的即時價格資訊,即時預報模型的樣本外誤差均方根(root mean square error,RMSE)約較AR模型低34-44%,且隨著月中資料的更新,即時預報模型的預測誤差亦進一步降低。 |
英文摘要 | Food and energy prices are the most important determinants of inflation fluctuations in Taiwan, and their high frequency data are anticipated to be effective predictors for real time CPI. We collect the online, daily, retail and wholesale prices of vegetables, fruit, pork, poultry, fish, and gasoline from official websites. Empirical results demonstrate that with the high frequency price data available on the DGBAS release day for the last month's CPI, the Root Mean Square Error (RMSE) of the nowcasting models are lower than the AR model by 34-44%. Moreover, with the updating of the data through the month, the nowcasting errors decline further. Incorporating food and energy high frequency price data would enable policymakers to predict the real time CPI change amid unusual weather conditions (e.g., typhoons, heaven rain) or amid an unusual international crude oil price fluctuation. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。