本文探討 COVID-19 疫情是否惡化台灣家庭所得不平等，並進一步檢驗此一影響 是否在不同族群間存在差異。本文使用中央研究院調查研究專題中心學術調查研究資料庫（Survey Research Data Archive， SRDA）所提供之家庭收支調查（Survey of Family Income and Expenditure）2017 年至 2024 年資料，將 2017–2019 年界定為疫情前、2020–2021 年界定為疫情中、2022–2024 年界定為疫情後，並以經濟戶長作為家庭所得之觀察單位，以戶長實質可支配所得與戶長實質淨市場所得代表家庭所得結果，作為主要結果變數。研究方法上，本文採用差異中的差異法（DID）與事件研究架構，並進一步以三重差分法（DDD）分析區域、教育與性別之異質性效果，同時呈現對數形式與水準形式之估計結果，以分別比較相對差距與絕對差距之變化。實證結果顯示，疫情期間低所得分位家庭相對高所得分位家庭的所得變化較不利，所得差距有擴大跡象；且相較於可支配所得，淨市場所得下的差距擴大幅度更為明顯，顯示市場面所得衝擊較為直接，而政府移轉與制度調整可能對可支配所得具有一定緩衝作用。分位切點與控制組設定之敏感度分析顯示，主要結論在不同規格下方向大致一致，但穩健程度並不完全相同；整體而言，以最高 20% 家庭作為控制組時，結果較為穩定，而改以最高 10% 家庭作為控制組時，部分規格下之顯著性有所減弱。異質性分析方面，區域與性別差異未獲一致且穩健支持，教育異質性則主要出現在可支配所得之特定期間與尺度下。整體而言，COVID-19 疫情對台灣家庭所得分配的衝擊較集中於所得分布底端，並在較接近市場面的所得概念下表現得更為明顯。;This study examines whether the COVID-19 pandemic worsened household income inequality in Taiwan and whether such effects differed across groups. It uses 2017–2024 data from the Survey of Family Income and Expenditure provided by the Survey Research Data Archive (SRDA)， defining 2017–2019 as the pre-pandemic period， 2020–2021 as the pandemic period， and 2022–2024 as the post-pandemic period. The main outcome variables are the household head’s real disposable income and real net market income， which are used as proxy measures for household income outcomes in this study. This study adopts a difference-in-differences (DID) design and an event-study framework to identify the impact of the pandemic on income inequality. In addition， a triple-difference (DDD) specification is used to examine heterogeneity by region， educational attainment， and gender. Both logarithmic and level specifications are reported to capture changes in relative and absolute income gaps. The results show that， during the pandemic， lower-income households experienced more unfavorable income changes than higher-income households， indicating a widening income gap. The increase in inequality is more pronounced for net market income than for disposable income， suggesting that market-income shocks were stronger， while taxes， transfers， and related institutional adjustments may have partially mitigated inequality in disposable income. Sensitivity analyses based on alternative income cutoffs and control-group definitions show that the main findings remain broadly consistent in direction across specifications， although the degree of robustness is not identical across models. Overall， the evidence is more stable when the top 20% group is used as the control group， whereas some specifications become less statistically robust when the top 10% group is used instead. Regarding heterogeneous effects， no consistent and robust evidence is found for regional or gender differences， whereas educational heterogeneity appears only in specific periods and specifications for disposable income. Overall， the findings suggest that the COVID-19 pandemic had a more adverse impact on households at the lower end of the income distribution in Taiwan， and that such effects are more evident when inequality is measured using market-oriented income concepts.
