本文採用和過去文獻不同的二階段分量迴歸法（Two-stage quantile regression, 2SQR）來分析美國產物保險公司再保險需求的決定因素，實證結果顯示公司流動性及損失準備的提撥分別在高及低分位數再保險時有顯著相反的影響，而此結果明顯與之前文獻所發現的結果不同。再者，有些變數的結果即使在不同再保險分位數下的實證結果與之前文獻的結果是一致的，但其影響程度及大小卻顯著的不相同。所以，研究證據顯示傳統的二階段最小平方法（Traditional two-stageleast squares regression, 2SLS）在解釋保險公司的再保險需求時可能存在偏誤及有不足之處，而本文所使用的2SQR法可以補足其不足及偏誤之處。因之，本文的實證結果對保單持有人、政策制定者或監理人員而言存在重要的政策意涵，亦即在評估保險公司的再保險需求時，整合2SQR法及2SLS法所隱含的實證意涵是相當重要的。

Unlike previous studies, this study applies two-stage quantile regression (2SQR) to analyze how insurers' demands for reinsurance are determined. The evidence shows that liquidity and loss development present contradictory influences at the lower and higher reinsurance quantiles. These interesting findings exactly differ from the results reported in previous literature. In addition, even though prior studies predict consistency among quantiles, this study reveals the magnitude of demand for reinsurance impacts, and the significant differences across various reinsurance quantiles. Thus, the evidence proposes that the traditional two-stage least squares regression (2SLS) may be insufficient, or produce a biased explanation on the determinants of insurers' demand for reinsurance. Additionally, the complementary function of the 2SQR approach on demand for reinsurance analysis is enriched as a whole. For policyholders, policy makers, and/or regulators, it is critical to integrate the empirical implications of the 2SLS and 2SQR approaches while evaluating an insurer's demand for reinsurance.