大型總體經濟計量模型能描繪出各變數之間相互影響的關係，設定多個方程式並聯立求解後得到多變量的預測值，此模型一大優點是可以進行敏感性分析，模擬外在衝擊或是政策施行的影響效果，以提供政策制定的參考依據。而文獻提及到總體經濟計量模型設定的困難之處，在於各迴歸的解釋變數是以經濟理論進行挑選，再剔除不顯著的變數，可能有模型誤設的問題。另一類經濟預測模型運用特徵萃取的降維方式，將經濟變數以市場作區分，再組合成各市場的擴散指標，以進行經濟成長率的單變數預測模型；雖降低了模型設定上的困難，但只能解釋市場的影響效果，無法如總體經濟計量模型可以進行敏感性分析。
因此本文為了降低總體經濟計量模型設定的難度以及提升預測的準確度，使用Python建立模型，並嘗試採用另一種降維方式，使用兩階段變數選擇方法，應用在模型的迴歸解釋變數挑選問題上，第一階段經 Lasso 迴歸交叉驗證法選出變數組合後；第二階段再透過逐步迴歸法找出調整後 R^2 最高的組合，能藉此淘汰掉不太顯著的變數，得到國內均衡 (m1) 模型的各迴歸變數組合。另外，利用國外經濟資料組成國外市場的擴散指標，在 m1 模型設定為基礎之下，再針對擴散指標進行一輪兩階段變數選擇後，便得到經國外調整 (m2) 模型的各迴歸變數組合。
研究發現，比較 m1 模型與透過經濟理論挑選變數的基準模型，模型整體預測有獲得改善。另一結果為 m2 模型相較於 m1 模型的預測表現，亦有部分變數提升其準確度。兩個結果顯示變數選擇方法以及模型加入擴散指標進行變數選擇，皆有改善模型預測的可能性。未來針對總體經濟計量模型探討高維度資料分析的方式，可將本文作為參考。

關鍵字: 總體經濟計量模型、擴散指標、變數選擇、Lasso迴歸、逐步迴歸法

For the propose of improving the performance of macroeconometric model in predicting domestic economic indicators and reducing the difficulty of model identification, the article not only considers the domestic variables of relevant important macroeconomic and financial data, but also trains the foreign domestic variables as the diffusion index by using principal component analysis. Furthermore, we use the two-stage variable selection method to select the explanatory variables. In the first stage, we shrink the coefficients of unimportant variables to zero with Lasso (least absolute shrinkage and selection operator) method. In the second stage, the stepwise regression is used to find the variable combination with the highest coefficient of determination after adjustment. The empirical result indicates that the forecasting performance of multivariate is better than the previous models that select variables by using economic theory. In the future, this article can be used as a reference for discussing high dimensional data analysis with macroeconometric models.

Key words: macroeconometric model, diffusion index, variable selection, Lasso regression, stepwise regression

吳中書、單易、鄭淑如、梅家瑗、蘇文瑩、高志祥、羅雅惠、黃純宜與王淑娟 (2001)，「臺灣總體經濟計量動態季模型」，《臺灣經濟預測與政策》，31(1)，111-159。
林建甫 (2006)，「台灣總體經濟金融模型之建立」，《中央銀行季刊》，28(1)，5-41。
林建甫 (2010)，「總體經濟計量模型的建立與應用」，《經濟論文叢刊》，38(1)，1-64。
徐士勛、管中閔與羅雅惠 (2005)，「以擴散指標為基礎之總體經濟預測」，《臺灣經濟預測與政策》，36(1)，1-28。
陳宜廷、徐士勛、劉瑞文與莊額嘉 (2011)，「經濟成長率預測之評估與更新」，《經濟 論文叢刊》，39(1)，1-44。
Ahlburg, D. A. (1984), “Forecast Evaluation and Improvement Using Theil's Decomposition,” Journal of Forecasting, 3(3), 345-351.
Arlot, S. and A. Celisse (2010), “A Survey of Cross-Validation Procedures for Model Selection,” Statistics Surveys, 4, 40-79.
Fair, R. C. (1984), Specification, Estimation, and Analysis of Macroeconometric Models, Harvard University Press.
Fair, R. C. (1996), “Computational Methods for Macroeconometric Models,” Handbook of Computational Economics, 1, 143-169.
Garratt, A., K. Lee, M. H. Pesaran and Y. Shin (2003), “A Long Run Structural Macroeconometric Model of the UK,” Economic Journal, 113(487), 412-455.
Hocking, R. R. and R. Leslie (1967), “Selection of The Best Subset in Regression Analysis,” Technometrics, 9(4), 531-540.
Hwang, J. S. and T. H. Hu (2015), “A Stepwise Regression Algorithm for High-Dimensional Variable Selection,” Journal of Statistical Computation and Simulation, 85(9), 1793-1806.
Li, J. and W. Chen (2014), “Forecasting Macroeconomic Time Series: LASSO-Based Approaches and Their Forecast Combinations with Dynamic Factor Models,” International Journal of Forecasting, 30(4), 996-1015.
McCracken, M. W. and S. Ng (2016), “FRED-MD: A Monthly Database for Macroeconomic Research,” Journal of Business & Economic Statistics, 34(4), 574-589.
Meinshausen, N. (2007), ”Relaxed Lasso,” Computational Statistics & Data Analysis, 52(1), 374-393.
Pindyck, S. and L. Rubinfeld (1998), Economentric Models and Economic Forecasts, McGraw-Hill Higher Education.
Sargent, T. (1979), “Estimating Vector Autoregressions Using Methods Not Based on Explicit Economic Theories,“ Federal Reserve Bank of Minneapolis Quarterly Review, 3(Summar), 8-15.
Sims, C. A. (1980), “Macroeconomics and reality,“ Econometrica, 48, 1-48.
Sims, C. A. (1986), “Are Forecasting Models Usable for Policy Analysis?, “ Federal Reserve Bank of Minneapolis Quarterly Review, 10(1), 2-16.
Stock, J. H. and M. W. Watson (1998), “Diffusion Indexes,” NBER working paper No. 6702.
Stock, J. H. and M. W. Watson (2002), “Macroeconomic Forecasting Using Diffusion Indexes,” Journal of Business & Economic Statistics, 20(2), 147-162.
Tibshirani, R. (1996), “Regression Shrinkage and Selection via the Lasso,” Journal of the Royal Statistical Society: Series B (Methodological), 58(1), 267-288.
Zou, H. (2006), “The Adaptive Lasso and Its Oracle Properties,” Journal of the American statistical association, 101(476), 1418-1429.
資料來源
中華民國統計資訊網
https://statdb.dgbas.gov.tw/pxweb/Dialog/statfile9L.asp
AREMOS 台灣經濟統計資料庫
http://net.aremos.org.tw/
ECONOMIC RESEARCH
https://research.stlouisfed.org/econ/mccracken/fred-databases/