家庭收支調查與稅收資料之比較分析

Abstract

本文使用2004至2020年間台灣家庭收支調查與財稅資料，分析兩資料中不同所得項目在估計所得分配上的差異，並探討不同資料來源在所得分配估計上的準確性和限制。過去文獻指出，高所得族群在調查資料中有低報傾向，導致無法準確掌握真實所得分佈情形。本研究針對各所得項目，分別計算百分位臨界值、平均值比率及所得占比，進行調查資料與稅收資料的比較。結果顯示，在受僱人員報酬、投資收入和總收入方面，高所得群體在調查資料中的收入低於稅收資料，呈現高所得者低報，或頂端所得者缺漏的現象。利息收入和財產所得則在各收入群體皆呈現調查資料低於稅收資料的情形。另一方面，產業主所得和租金收入在稅收資料中整體低於調查資料，推測因部分所得免稅、查定課稅或逃漏稅情形，造成稅收資料出現低估。整體而言，不同資料來源和所得項目皆具有各自特性與限制，建議未來在進行台灣所得分配研究或應用在相關政策時，應同時參考家庭收支調查和稅收資料，並且留意兩種資料的限制，以更完整掌握所得分配的情況。

The study uses Taiwan survey of family income and expenditure with taxation data from 2004 to 2020 to examine differences in income distribution estimates across data sources. We evaluate the reliability and limitations of each data source. Previous studies suggest that high income individuals tend to underreport their earnings in survey data, making it difficult to capture the true income distribution. To compare the two data sources, we calculate percentile thresholds, average ratios, and income shares for different income categories. Results show that for compensation, investment income, and total income, survey data show lower income levels for high income individuals compared to tax data, which may reflect underreporting or missing rich in survey data. For interest income and property income, survey data are lower than tax data across all income groups. On the other hand, entrepreneur income and rental income are generally lower in tax data than in survey data. This may result from tax-exempt income, presumptive taxation, or tax evasion, which may cause the tax data to underestimate actual income. Overall, both data sources have their own features and limitations. Future research on income distribution in Taiwan or related policies should consider both survey and tax data to more clearly capture the overall income distribution.