本文探討少數可觀測總體資訊集合，是否足以代表大型總體追蹤資料 (panel
data) 中的共同波動，以 2010 年 1 月至 2025 年 1 月之月資料為樣本，先由 89 個總體變數透過主成分分析（Principal Component Analysis, PCA）萃取潛在總體因子，再將其與三組皆為 6 個變數的小型可觀測資訊集合進行比較。三組小型資訊集合分別為核心總體組、景氣循環組與總體金融條件組。本文採用典型相關分析（Canonical Correlation Analysis, CCA），並結合潛在因子與可觀測因子之涵蓋能力檢定（spanning test），評估各小型資訊集合與大型追蹤資料因子空間之共享資訊。實證結果顯示，三組小型資訊集合都能有效捕捉大型追蹤資料的主要共同方向，尤其對前三個潛在共同因子具有高度對應能力；然而若以整體涵蓋強度來衡量，總體金融條件組表現最佳，核心總體組次之，景氣循環組相對較弱。另一方面在通膨、就業與產出之外，加入期限利差、信用利差與貨幣數量有助於提高小型追蹤資料對大型資訊集合的代表能力。不過高階共同資訊依舊難以由 6 個可觀測變數來完全取代，所以表示小型追蹤資料可作為核心共同資訊的簡單摘要，但不適合作為大型追蹤資料的完整替代。

This paper evaluates whether small observable macroeconomic information sets can represent the common fluctuations embedded in a large macroeconomic and financial panel. Using 89 monthly variables from January 2010 to January 2025, the paper extracts latent macroeconomic factors from the large panel by principal component analysis and compares them with three six-variable observable sets: a core macroeconomic set, a business-cycle set, and a macro-financial conditions set. The comparison is based on canonical correlations and a spanning test for latent and observable factors. The empirical results show that all three small panels capture the main common movements of the large panel, especially the first few latent factors. However, the macro-financial conditions set exhibits the strongest overall spanning performance, followed by the core macroeconomic set and the business-cycle set. The results suggest that financial conditions and monetary aggregates add meaningful information beyond inflation, employment, and output. At the same time, none of the six-variable sets fully replaces the large information set, since higher-order common components remain difficult to capture with a small panel alone.

Andreou, E., P. Gagliardini, E. Ghysels, and M. Rubin (2024), “Spanning latent and observable factors,” Journal of Econometrics, 248, Article 105743.

Bai, J., and S. Ng (2002), “Determining the number of factors in approximate factor models,” Econometrica, 70, 191–221.

Bai, J., and S. Ng (2008), “Forecasting economic time series using targeted predictors,” Journal of Econometrics, 146, 304–317.

Bernanke, B. S., J. Boivin, and P. Eliasz (2004), “Measuring the effects of monetary policy: A factor-augmented vector autoregressive (FAVAR) approach,” NBER Working Paper, No. 10220.

Boivin, J., and S. Ng (2006), “Are more data always better for factor analysis?”Journal of Econometrics, 132, 169–194.

Christiano, L. J., M. Eichenbaum, and C. L. Evans (1999), “Monetary policy shocks: What have we learned and to what end?” In J. B. Taylor and M. Woodford (eds.), Handbook of Macroeconomics, Vol. 1A, 65–148, Amsterdam: Elsevier.

Hatzius, J., P. Hooper, F. S. Mishkin, K. L. Schoenholtz, and M. W. Watson (2010),“Financial conditions indexes: A fresh look after the financial crisis,” NBER Working Paper, No. 16150.

Mammen, E. (1993), “Bootstrap and wild bootstrap for high dimensional linear models,” The Annals of Statistics, 21, 255–285.

Sims, C. A. (1992), “Interpreting the macroeconomic time series facts: The effects of monetary policy,” European Economic Review, 36, 975–1000.

Stock, J. H., and M. W. Watson (2002), “Forecasting using principal components from a large number of predictors,” Journal of the American Statistical Association, 97, 1167–1179.