博雅執中或文飾媚外？─ 結合AI評分機制探討北美執行長博雅教育背景對盈餘操縱行為與分析師反應之調節角色

Abstract

本文聚焦三議題，首先探討北美執行長博雅程度與五項盈餘管理行為之關聯性 ; 其二檢視企業面臨一項或多項盈餘操縱誘因時，執行長博雅背景對於盈餘操縱行為的調節效果 ; 其三則探討企業財報資訊透露盈餘操縱訊號時，執行長博雅程度對標的企業分析師每股盈餘預測下調幅度的抑制效果。本研究使用Wharton Research Data Services (WRDS)資料庫數據，囊括共15,974筆觀測值，資料區間為1987年至2023年間1,157家北美企業財務數據與執行長特徵資料，透過AI輔助追溯到樣本初期的大學博雅程度量化評分機制，對執行長取得大學學位前五年間之該校博雅教育發展程度進行評分，此外，亦使用人工方式將執行長畢業大學對博雅教育之深化程度由高至低分為四類：(ㄧ) 位於歐盟會員國內之大學，(二) 位於北美，並具宗教背景、軍事性質或設有博雅學院 (Liberal Arts College) 之大學，(三) 北美非博雅性質大學，與 (四) 位於開發中國家之大學。實證結果顯示，畢業於歐盟會員國內大學 (博雅程度較高) 的執行長，其面對財務壓力、營業虧損等盈餘操縱誘因時，相較於其他大學執行長，其所屬企業盈餘操縱傾向較低。而來自北美博雅與非博雅性質大學的執行長，其所屬企業遭遇營業淨損或盈餘負成長等困境時，其相較於歐盟、位於開發中國家的大學執行長，傾向以實質盈餘操縱手段美化盈餘。此外，當企業報導之盈餘疑似被裁決性上調時，分析師似未依執行長博雅背景深淺調整其預測，但對博雅程度較弱之北美名校執行長所屬企業之裁量性盈餘操縱行為，分析師盈餘預測下調程度較低。

This thesis aims at investigating three interrelated research themes: (1) The relationship between the degree of Liberal Arts Education and the engagement in five major types of Earnings Manipulation schemes among Top Management Team (TMT) executives in North American firms. (2) Whether the extent of Liberal Arts Education moderates TMT’s responses when firms face one or more incentives to manipulate earnings, and whether it reduces the likelihood of engaging in such earnings manipulation behaviors. (3) If the presence of earnings manipulation signals in financial reports affects analysts' Forecast Revision Index to a lesser extent when executives possess stronger Liberal Arts educational backgrounds in their undergraduate. This paper draws on data from the Wharton Research Data Services (WRDS), comprising 15,974 firm-year observations from 1,157 publicly listed North American firms over the period 1987 to 2023. Executives are classified into four educational categories in descending order of Liberal Arts emphasis based on the characteristics of their undergraduate institutions: (1) universities located in EU member states, (2) North American universities with religious affiliations, military backgrounds, or dedicated Liberal Arts Colleges, (3) other Non-Liberal Arts universities in North America, and (4) universities located in developing countries as of the 1990s. In addition to this manual classification, a complementary mechanism is developed using an AI-assisted scoring system that quantitatively assesses the degree of Liberal Arts education at the undergraduate institutions attended by executives during the five years prior to their graduation. This system provides a continuous measure of the university’s commitment to Liberal Arts education, supplementing the categorical approach. The regression analysis findings reveal that executives graduating from EU-based universities are less likely to boost the report earnings of their employing firms when facing earnings manipulation pressures such as financial distress or operational losses, compared to their counterparts from other educational backgrounds. Executives from North American are more inclined to engage in real earnings manipulation (REM) when confronted with net operating losses or earnings decline. Finally, when firms exhibit signs of upward earnings manipulation, analysts do not appear to adjust their estimates of the firm’s future profitability based on the depth of the liberal arts of the college where the executives went to. Instead, they are less inclined to undo the alleged upward bias in their earnings revisions for firms led by executives from prestigious North American universities with relatively weaker liberal arts exposure.