近幾年提升稅收預測效能成為財政政策制定者關切的議題，其中營利事業所得稅(簡稱營所稅)尤為我國主要財政稅收之一，由於台灣為高度出口導向型經濟體，企業獲利易受國際市場外部衝擊以及景氣循環影響，導致稅收呈現較劇烈之波動性。因此，建構高預測精準度之稅收預測計量模型以優化營所稅稅收推估品質具實務上之迫切性。
本研究應用自我迴歸分佈滯後混合頻率資料抽樣模型(ADL-MIDAS模型)預測營所稅稅收，在權重函數設定上採用經標準化之指數艾蒙滯後（Exponential Almon Lag）權重函數。欲透過個別納入出口、台灣平均加權股價指數TAIEX、實質GDP成長率、工業生產指數、失業率等高、中頻經濟指標滯後資訊，搭配擴張視窗(Expanding Window)樣本外預測方式以預測營所稅稅收。藉此檢視無須將自變數、應變數進行同頻聚合之ADL-MIDAS模型是否能夠更敏銳地捕捉營所稅稅收變動之趨勢。
實證結果顯示，於ADL-MIDAS模型架構下，出口指標具優良的預測表現，其預測效能不僅優於AR模型，且在無疫情衝擊之常態平穩景氣情勢下優於政府官方稅收預測，從中也可看出本文主要實證模型在面臨極端衝擊時之預測侷限，有鑑於此，本研究建議於營所稅稅收預測可採行雙階段預測架構：第一階段先以計量模型建立客觀之預測基準；第二階段則導入專家評估機制，結合即時經濟情勢、最新稅制變革與政策走向進行權衡性調整，以兼顧模型估計之嚴謹性與實務決策之彈性，以提升營所稅稅收預測之精度。

In recent years, enhancing the performance of tax revenue forecasting has emerged as a critical concern for fiscal policymakers. Among the various tax categories, the Corporate Income Tax (CIT) stands out as one of the most critical sources of fiscal revenue. Since Taiwan operates as a highly export-oriented economy, corporate profits are highly susceptible to the shocks of international market fluctuations and business cycles, which consequently leads to more sharp volatility in tax revenues. Therefore, constructing a high-precision econometric model to optimize the estimation quality of the CIT revenues has a practical urgency.
This study applies the Autoregressive Distributed Lag Mixed-Data Sampling (ADL-MIDAS) model to forecast CIT revenues. The model utilizes a normalized Exponential Almon Lag weight function. By individually incorporating the lagged information of various high- and medium-frequency economic indicators—including exports, the Taiwan Capitalization Weighted Stock Index (TAIEX), real GDP growth rate, the industrial production index, and the unemployment rate—we conduct tax revenue forecasting via an expanding window out-of-sample approach. Ultimately, this study examines whether the ADL-MIDAS model, characterized by its exemption from temporal aggregation, can more sensitively capture the nuanced and dynamic trends of the CIT revenue’s fluctuations.
The empirical results indicate that, under the ADL-MIDAS model framework, the export indicator exhibits excellent forecasting performance. Its forecasting efficacy not only outperforms the baseline model, AR model , but also excels the government forecasts under normal economic conditions. Meanwhile, these results reveal the forecasting limitations of econometric models when confronting extreme shocks. Therefore, we recommend adopting a two-stage forecasting framework for predicting CIT revenues. In the first stage, econometric model is used to establish an objective benchmark. In the second stage, an expert assessment mechanism is introduced to make flexible adjustments based on immediate economic trends, the latest tax reforms, and policy shifts. This approach balances the rigor of model estimation with the flexibility of practical decision-making, ultimately improving forecasting accuracy.

一、中文文獻
王健全、陳厚銘(1995)。 「促進產業升級條例」有關研究發展投資抵減之獎勵效果評析。 https://www.cier.edu.tw/publish/%E3%80%8C%E4%BF%83%E9%80%B2%E7%94%A2%E6%A5%AD%E5%8D%87%E7%B4%9A%E6%A2%9D%E4%BE%8B%E3%80%8D%E6%9C%89%E9%97%9C%E7%A0%94%E7%A9%B6%E7%99%BC%E5%B1%95%E6%8A%95%E8%B3%87%E6%8A%B5%E6%B8%9B%E4%B9%8B%E7%8D%8E/
立法院(2025)。 2024年稅收3.7619兆、超徵5283億元 雙創史上新高，藍推還稅於民立法，普發現金1萬。 https://www.ly.gov.tw/Pages/Detail.aspx?nodeid=54301&pid=248666
立法院預算中心(2023)。 自財政紀律觀點檢視稅收估列偏差與改進方案初探。 https://www.ly.gov.tw/Pages/Detail.aspx?nodeid=46347&pid=230658&utm_
立法院預算中心(2024)。 114 年度中央政府總預算案整體評估報告。 https://www.ly.gov.tw/Pages/List.aspx?nodeid=55648
姚名鴻、許雅柔(2024)。 小規模營業人導入行動支付適用租稅優惠對營業稅稅收影響之實證分析 [An Empirical Analysis of the Impact of Tax Incentives for Small Business Adopting Mobile Payments on Business Tax Revenues]。 管理資訊計算, 13, 10-19.
財政部統計處(2019)。 稅收執行差距及稅收預測議題之探討。https://www.mof.gov.tw/multiplehtml/1419
財政部統計處(2024a)。 113年8月海關進出口貿易初步統計。 https://www.mof.gov.tw/singlehtml/384fb3077bb349ea973e7fc6f13b6974?cntId=ce4b73798ef1499e960ad2227b3dc168
財政部統計處(2024b)。 113年9月海關進出口貿易初步統計。 https://www.mof.gov.tw/singlehtml/384fb3077bb349ea973e7fc6f13b6974?cntId=89a6f2f0c8bd45999d79a33fe21e7e3d
財政部統計處(2025)。 114年稅收徵起情形分析。 https://www.mof.gov.tw/multiplehtml/1474?categoryCode=FIN
國家發展委員會(2017)。 台灣經濟發展歷程與策略。 https://www.ndc.gov.tw/News_Content.aspx?n=9D32B61B1E56E558&sms=9D3CAFD318C60877&s=B47DCAA62A4536A7
國家發展委員會(2020)。 六大核心戰略產業推動方案（110年～113年）。 https://www.ndc.gov.tw/Content_List.aspx?n=9614A7C859796FFA
陳國樑、楊佩烜、黃勢璋(2022)。 臺灣稅收預測表現之探討. 人文及社會科學集刊, 32(2), 271-319。
經濟部(2016)。 產業創新研發-推動五加二產業創新計畫。 https://www.moea.gov.tw/Mns/populace/Policy/Policy.aspx?menu_id=32800&policy_id=4
經濟部統計處(2021)。 疫情對我國產業之影響。https://www.moea.gov.tw/Mns/dos/bulletin/Bulletin.aspx?kind=9&html=1&menu_id=18808&bull_id=9117
監察院. (2011). 我國賦稅結構與政府稅制改革措施-專案調查研究報告。https://www.cy.gov.tw/News_Content.aspx?n=773&s=28251
二、英文文獻
Asimakopoulos, S., Paredes, J., & Warmedinger, T. (2013). Forecasting fiscal time series using mixed frequency data (Working Paper No. 1550). European Central Bank.
Ball, L. M. (2009). Hysteresis in unemployment: old and new evidence(NBER Working Paper No. 14818). National Bureau of Economic Research.
Beer, S., Erard, B., & Hanappi, T. (2025). How to Forecast Corporate Income Tax Revenues(How-To Note No. 2025/010).International Monetary Fund.
Box, G. E., & Pierce, D. A. (1970). Distribution of residual autocorrelations in autoregressive-integrated moving average time series models. Journal of the American statistical association, 65(332), 1509-1526.
Burguete, J. F., Ronald Gallant, A., & Souza, G. (1982). On unification of the asymptotic theory of nonlinear econometric models. Econometric reviews, 1(2), 151-190.
Casey, E., & Hannon, A. (2016). Challenges forecasting Irish corporation tax. Irish Fiscal Advisory Council Analytical Note(10).
CFA Institute. (2026). Model misspecification[CFA Program Level II Curriculum]. https://www.cfainstitute.org/insights/professional-learning/refresher-readings/2026/model-misspecification
Chai, T., & Draxler, R. R. (2014). Root mean square error (RMSE) or mean absolute error (MAE)?–Arguments against avoiding RMSE in the literature. Geoscientific model development, 7(3), 1247-1250.
Clark, T. E., & West, K. D. (2007). Approximately normal tests for equal predictive accuracy in nested models. Journal of econometrics, 138(1), 291-311.
Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1995). Detecting earnings management. Accounting review, 70(2), 193-225.
Dickey, D. A., & Fuller, W. A. (1979). Distribution of the estimators for autoregressive time series with a unit root. Journal of the American statistical association, 74(366a), 427-431.
Dickey, D. A., & Fuller, W. A. (1981). Likelihood ratio statistics for autoregressive time series with a unit root. Econometrica: Journal of the econometric society, 1057-1072.
Diebold, F. X., & Mariano, R. S. (1995). Comparing predictive accuracy I: An asymptotic test, Journal of Business & Economic Statistics, 13(3), 253–263.
Diebold, F. X. (1998). Elements of forecasting.South-Western. https://www.scirp.org/reference/referencespapers?referenceid=3258882
Đurović Todorović, J., Đorđević, M., Mirović, V., Kalaš, B., & Pavlović, N. (2024). Modeling tax revenue determinants: the case of visegrad group countries. Economies, 12(6), 131.
Economics, T. (2026). Taiwan Exports. https://tradingeconomics.com/taiwan/exports
Fama, E. F. (1990). Stock returns, expected returns, and real activity. The journal of finance, 45(4), 1089-1108.
Ghysels, E., & Ozkan, N. (2012). Real-Time Predictions of the US Federal Government Budget: Expenditures, Revenues and Deficits. UNC Kenan-Flagler Research Paper.
Ghysels, E., Santa-Clara, P., & Valkanov, R. (2004). The MIDAS touch: Mixed data sampling regression models (CIRANO Working Paper No. 2004s-20). CIRANO.
Ghysels, E., Sinko, A., & Valkanov, R. (2007). MIDAS regressions: Further results and new directions. Econometric reviews, 26(1), 53-90.
Ghysels, E., Kvedaras, V., & Zemlys, V. (2016). Mixed frequency data sampling regression models: The R package midasr. Journal of statistical software, 72, 1-35.
Ghysels, E., Kvedaras, V., & Zemlys-Balevičius, V. (2020). Mixed data sampling (MIDAS) regression models. In Handbook of statistics (Vol. 42, pp. 117-153). Elsevier.
Gomez-Zamudio, L. M., & Ibarra, R. (2017). Are daily financial data useful for forecasting GDP? Evidence from Mexico. Journal of the Latin American and Caribbean Economic Association, 17(2), 173-203.
Gunay, S., Can, G., & Ocak, M. (2021). Forecast of China’s economic growth during the COVID-19 pandemic: a MIDAS regression analysis. Journal of Chinese Economic and Foreign Trade Studies, 14(1), 3-17.
Harvey, D., Leybourne, S., & Newbold, P. (1997). Testing the equality of prediction mean squared errors. International Journal of forecasting, 13(2), 281-291.
Kvedaras, V., & Zemlys, V. (2012). Testing the functional constraints on parameters in regressions with variables of different frequency. Economics Letters, 116(2), 250-254.
Kvedaras, V., Zemlys, V. (2025). midasr: Mixed data sampling regression models. (Version 0.9) [Computer software].CRAN. https://cran.r-project.org/web/packages/midasr/index.html
Hyndman, R. J., & Athanasopoulos, G. (2013). Forecasting: Principles and Practice.(2nd ed). OTexts. https://otexts.com/fpp2/stl.html
Hyndman, R. J., & Athanasopoulos, G. (2013). Forecasting: Principles and Practice.(2nd ed). OTexts. https://otexts.com/fpp2/accuracy.html
Ljung, G. M., & Box, G. E. (1978). On a measure of lack of fit in time series models. Biometrika, 65(2), 297-303.
Marcellino, M., Stock, J. H., & Watson, M. W. (2006). A comparison of direct and iterated multistep AR methods for forecasting macroeconomic time series. Journal of econometrics, 135(1-2), 499-526.
National Institute of Standards and Technology. (2002). NIST engineering statistics handbook.(Section 1.3.5.11). https://www.itl.nist.gov/div898/handbook/eda/section3/eda35b.htm
Shahnazarian, H., Solberger, M., & Spånberg, E. (2017). Forecasting and analysing corporate tax revenues in Sweden using Bayesian VAR models. Finnish economic papers, 28(1) , 50-74.
Tashman, L. J. (2000). Out-of-sample tests of forecasting accuracy: an analysis and review. International Journal of forecasting, 16(4), 437-450.