本文旨在探討目前退休金會計處理理論及實務上的問題，俾為未來處理相關問題之參 考。 第一章 緒論，敘述研究動機、目的、限制及論文架構。 第二章 退休金概述，說明財務報告之目的；性質與背景；退休金在財務報告中的重 要性；會計準則的演進；及現行處理方式的缺失。 第三章 理論之探討，包括負債承認；費用計算；成本分攤；過去與前期服務成本； 精算損益；以及附註揭露等問題。 第四章 退休金之會計處理，就美國會計準則委員會發佈的準則加以說明，並探討其 獲致結論之原因。 第五章 精算配合問題，介紹我國精算學會的概況，以及技術方面的現況，其提出技 術配合的看法。 第六章 結論的建議。

本文討論退休金長期預期報酬率之設定。根據會計公報規定，公司之退休金長期預期報酬率必須符合其資產配置。但是在本文中我們發現資產配置並無法有效的反應公司的退休金長期預期報酬率，因此我們提出展望理論及盈餘管理兩種理論來解釋退休金長期預期報酬率之設定。我們發現經理人企圖透過改變退休金長期預期報酬率的假設來進行盈餘管理，同時退休金長期預期報酬率也會因為公司的風險態度而有所改變。 / Abstract: We try to find out the considerations for managers to set their assumptions of expected long term rate of return on pension plan assets (ROPA). First, we use the asset allocations of pension funds and historical returns to calculate the expected rate of return based on historical asset returns (EROPA). There is difference between ROPA are EROPA, suggests that asset allocations are not the only consideration when managers setting their ROPA assumption. Two theories are examined in this paper to explain such difference between ROPA and EROPA: earnings management and prospect theory. We use two models to test the earnings management, single accrual model and threshold model. We find that the intentions to smooth the reported income are the main incentives for managers to manipulation their ROPA. The incentive to do earnings management can partly explain the difference between ROPA and EROPA. However, in threshold model, we can not observe any evidence in our research. We also introduce prospect theory to examine the risk attitude. We find that managers’ risk attitude affect the setting of assumptions, too. We conclude prospect theory provides a good explanation of the difference between ROPA and EROPA.