為探討勞工保險以被保險人退休當月起前3年的平均月投保薪資為老年一次給付計算基礎的做法，是否使被保險人有誘因在退休前3年提高投保薪資，本研究採用difference-in-differences模式，以民國96年符合退休條件且已實際退休的被保險人為實驗組；符合退休條件但當年未退休、且在未來三年內也未退休的被保險人為對照組。分析顯示，控制了其他重要影響因素後，誘因機制呈現顯著的正向效果；退休者在退休前3年其月投保薪資顯著上升，平均而言上升10.7%；而且以最高60個月的平均月投保薪資為給付計算基礎的改變，效果有限。需以更長期間為基礎，才能消除制度的不當誘因。

Using Taiwan's Labor Insurance dataset and the difference-in-differences model, this study investigates the possible incentives of utilizing the average insured monthly earnings for the period beginning 3 years prior to retirement as the basis for old-age benefit calculations. In the study, those who had retired in 2007were categorized into the experimental group, while those who were qualified to retire but had not retired for the ensuing 3 years were categorized into the control group. The results indicate that after controlling for all influential factors such as gender and years of service, the incentive displayed a positive impact. On average, it had increased the insured earnings by 10.7%. Changing the benefit calculation formula to using the average monthly amounts of the highest 60 months of insured earning would not generate much impact. Using a much longer period as the basis for benefit calculation can help to alleviate or even eliminate the incentive impact, thus removing the extra financial burden caused by the incentive mechanism embedded in the benefit calculation formula of the labor insurance system.