使用Google 搜尋工具改善臺灣消費模型推估能力

Use Google Searches for Nowcasting Taiwan's Consumption

國外研究指出網路行為與現實世界的經濟指標之間有很強的關聯，且每人平均所得較高國家之網路使用者愈可能用網路搜尋未來的資料，亦有許多國外文獻發現增加Google搜尋指標，有助於民間消費的推估。本研究運用Google搜尋趨勢的資料建構指標，來測試是否能對臺灣消費的推估有助益。為更即掌握消費動向，本文選擇商業營業額作為消費的高頻率指標，進行實證研究，結果如下：1.Google搜尋指標對商業營業額的推估有助益，但程度有限，可能與受限於資料長度、及選擇搜尋關鍵字有關。2.商業營業額與前期顯著正相關，表示商業營業額存在平滑性。3.股價指數、長短期利差與商業營業額顯著相關，且這2項指標有助於商業營業額推估。4.消費者信心指數、工業及服務業經常性平均薪資、失業率等指標未領先商業營業額。5.工業及服務業經常性平均薪資與商業營業額的關係較不顯著。因應網路資訊時代，指標的建構方式與資料來源可更多元，建議未來可運用Google搜尋工具，建構反映不同總體資訊的Google搜尋指標，作為重要總體指標推估之參考。

Many studies indicate that there is a strong link between network behavior and economic indicators in the real world. The country’s higher per capita GDP users may search more future data on the Internet. This study uses Google Trend’s data to construct Google indicators and tests to see if it will be effective for Taiwan’s consumption estimation. This article chooses sales of trade and food services as consumption’s high frequency indicator. The empirical results are as follows: 1. Google indicators are useful for sales of trade and food services estimation, but its forecasting performance may be limited by the data length and the search keywords. 2. Trade and food services have a high positive autocorrelation. 3. TAIEX’s average closing price and interest rate spread have a significant relationship with trade and food services. 4. Consumer confidence, average regular earnings in industry and services, and the unemployment rate don’t have lead than trade and food services. 5. The connection between average regular earnings in industry and services and trade and food services is not significant. In response to the Internet time, constructing indicators methods and data sources can lead to more diversity. This study recommends using the Google search tool to construct Google indicators reflecting the macroeconomic situation in the future.