本文應用「葛蘭吉直接測試法」，檢證1961至1992會計年度臺灣地區政府歲入與 歲出之間的因果關係。分析結果顯示，歲入與總歲出二者之間，具有「量入為出」的關係， 即總支出規模的變動，受歲入額度變動影響甚大，尤其是非國防支出中的「一般政務」支出 與「教育科學文化」支出部分，遵循「量入為出」原則的特徵最為明顯。國防支出則不受此 一原則的拘束。此外，債務支出與其他政事別支出不同，顯現「量出為入」的傾向。這種「 歲入多寡，決定總歲出規模」的困果關係若持續不變，則欲以增稅來降低我國政府預算赤字 的作法，可能只會引起政府預算規模的擴張，對預算赤字的控制，助益不大。

　　　　　This paper uses the direct Granger test to examine the intertemporal relationships between public spending and revenues in Taiwan for fiscal years 1961-1992. Revenues are found to Granger-cause spending outcomes, with the strongest link occurring in the expenditures for general administration, and for education, science, and culture. The reverse causal relationship, spend-then-tax, finds little empirical support, with the exception of obligations Granger-causing revenue outcomes. The lesson learned from past data thus appears to be that raising taxes is more likely to result in more government spending than to reduce the size of budget deficit.