Earnings management and financial distress or soundness determining CEOs' future over- and under-investment decisions

This study investigates how CEOs' cognitive bias due to earnings management practices and financial distress affects their future investment decisions. It positions earnings management as a driver of investment decision-making, noting prior evidence that firms with high earnings management tend to increase investment excessively in the current period. It synthesizes literature on financial stress and future investment decisions, highlighting that financially depressed firms may hoard cash and exhibit non-uniform investment behavior. The study examines whether a high level of earnings management, accompanied by financial distress, leads to the highest probability of excessive future investments. New contributions include: (1) sequentially blended logic from prospect theory, framing, conservatism, and psychological projection to explain CEOs' future investment decisions based on past earnings information; (2) considering financial distress or soundness as an influence on the association between earnings management and future over/under-investment decisions; and (3) investigating CEOs' over-confidence in an international setting, differentiating developed and developing countries. The study argues that CEOs escalate commitment under loss framing and use psychological projection, abandoning ex-ante conservatism, thereby shaping abnormal future investments. It underscores the capital allocation implications for investors and suggests universality of these behaviors across countries.

The literature review focuses on two clusters. First, the intersection of earnings management and financial distress: financial health and growth opportunities are linked to managerial decision-making. Studies show firms in distress engage in earnings manipulation (e.g., Rosner 2003; Saleh and Ahmed 2005; Jaggi and Lee 2002), with financial conditions incentivizing managers to manage earnings to maintain positions and relieve temporary distress (Habib et al., 2013; Du and Lai, 2018). Second, CEOs' cognitive bias and future investment decisions: cognitive biases arise from heuristics, task artefacts, and error management (Haselton et al., 2015). Earnings management and financial condition generate cognitive dissonance, leading CEOs to resolve discomfort via further earnings management and investment decisions. Prospect theory implies risk-seeking under losses and risk-avoidance under gains (Kahneman and Tversky, 1979), with financial distress creating negative framing and soundness positive framing. Psychological projection operates as a self-defense mechanism, channeling pressures into abnormal investments. Hypotheses: H1a: When past income increases, earnings management positively affects future over-investment. H1b: When past income decreases, earnings management positively affects future under-investment. H2a: Financial distress positively influences future overinvestment. H2b: Financial soundness positively influences future underinvestment. H3a: Financial distress strengthens the relationship between earnings management under past income increases and future overinvestment. H3b: Financial soundness strengthens the relationship between earnings management under past income decreases and future underinvestment. H4a: Financial distress first intervenes through earnings management (with past income increases) to affect future overinvestment. H4b: Financial soundness first intervenes through earnings management (with past income decreases) to affect future underinvestment.

Design and data: The study employs an international sample of manufacturing firms from 46 developed and developing countries (UN classification; U.S. News & World Report Best Countries 2020) over 2009–2019. Data on earnings management, financial position, and abnormal investments are sourced from Bureau van Dijk and Refinitiv Thomson Reuters. Inclusion requires manufacturing classification and available balance sheet and income statement data. Initial sample: 240,624 firm-year observations (21,461 firms). After excluding firms lacking investment disclosure (−148,674), incomplete earnings management items (−15,029), and missing financial condition metrics (−6,334), the final sample is 70,587 firm-years (27.08%). Measures: Abnormal investment (ABI) is the residual from actual investment INV minus predicted investment INV_hat. INV is fixed-asset payments (tangible and intangible) scaled by total assets. Predicted investment follows Richardson (2006) and Baker et al. (2016), regressing INV on firm growth (lagged revenue growth), leverage, cash, firm age, size (log assets), prior return, and lagged investment. Earnings management is proxied by discretionary accruals (DA) using a Kothari et al. (2005) performance-matched modified Jones approach over rolling ten-year industry-year regressions, separating discretionary and nondiscretionary accruals (Dechow and Dichev, 2002; Guay et al., 1996). Financial condition (FC) captures distress/soundness without bankruptcy legal consequences (Pindado et al., 2008), constructed by comparing actual liabilities to optimal predicted liabilities (a modified Z-score concept), where FC = DER_optimal(t−1) − DER_actual(t−1); higher FC indicates better financial condition, lower indicates distress. Controls include lagged ABI, size, leverage, profit growth, asset growth, profit margin, market-to-book, and operating cash flows. Empirical strategy: Main and moderating effects are tested via panel regressions. Equation (6) estimates effects of lagged DA and FC on ABI (split into overinvestment and underinvestment subsamples). Equation (7) adds the interaction DA*FC to test moderation (H3). For staged (mediating) effects (H4), a two-stage least squares (2SLS) framework is used: First stage regresses DA on FC and controls (Eq. 9); second stage regresses ABI on predicted DA and controls (Eq. 8). Country subsamples (developed vs developing) are analyzed for heterogeneity. Variables are winsorized at the 1st and 99th percentiles to mitigate outliers. Model diagnostics ensure linearity and address potential covariance between residuals across stages; lag structures help with temporal ordering.

- Sample and descriptives: Final sample includes 70,587 firm-year observations across 46 countries. Most firms exhibited underinvestment and income-increasing discretionary accruals; mean ABI = −0.00041, mean DA = 0.00124, mean FC = 0.00413, indicating overall financial soundness. - H1 (earnings management and abnormal investment): Supported. DA(t−1) is positively associated with overinvestment when past income increases and with underinvestment when past income decreases. Significance at p < 0.01 overall; significance holds in both developing and developed subsamples (Table 2). - H2 (financial condition and abnormal investment): Supported. FC is positively associated with overinvestment under distress and with underinvestment under soundness at p < 0.01 overall and across subsamples (Table 3). - H3 (moderation by financial condition): Not supported. The DA*FC interaction is not statistically significant in predicting ABI for either distress/overinvestment or soundness/underinvestment settings; signs are opposite to hypothesized strengthening effects (Table 4). - H4 (staged/mediated effects via 2SLS): Supported. First stage: FC(t−1) significantly predicts DA at p < 0.01. Second stage: predicted DA(t−1) significantly predicts ABI at p < 0.01, consistent across developing and developed subsamples (Tables 5 and 6). Interpretation: Financial distress leads to income-increasing earnings management, which then increases future overinvestment; financial soundness is associated with income-decreasing earnings management, which then increases underinvestment. - Cross-country comparison: Patterns are broadly consistent between developed and developing countries; behaviors appear universal. - Additional coefficients: Control variables (size, leverage, asset growth) frequently significant with expected signs in many specifications; lagged ABI significant, indicating persistence in investment behavior.

The findings demonstrate that CEOs' earnings management aligns with subsequent abnormal investment choices, consistent with cognitive biases under prospect theory, framing, and psychological projection. Under financial distress (loss framing), CEOs exhibit risk-seeking, engage in income-increasing accruals, and subsequently overinvest to escape losses. Under financial soundness (gain framing), CEOs tend toward caution, income-decreasing accruals, and underinvest. The absence of a concurrent moderating effect (H3) suggests the process is sequential rather than simultaneous: financial condition first shapes earnings management practices, which then influence investment decisions (H4). The international evidence indicates these behavioral patterns are largely universal across developed and developing contexts. Practical implications include caution for investors and creditors regarding firms exhibiting earnings management coupled with distress, as these may precede abnormal investment and heightened risk. Enhanced disclosure about over/underinvestment and investment normality could improve capital allocation and protect stakeholders; dynamic impacts on WACC and firm valuation underscore the importance of transparent reporting.

The study establishes that CEOs' over- and under-investment decisions are systematically linked to earnings management and financial condition. Income-increasing earnings management under financial distress precedes future overinvestment, while income-decreasing earnings management under financial soundness precedes underinvestment. These behaviors reflect cognitive biases—loss framing, heuristics, and psychological projection—and appear consistent across developed and developing markets. Contributions include integrating behavioral theories with investment and reporting outcomes, documenting staged effects from financial condition to earnings management to investment, and providing broad international evidence. Future research could refine behavioral measurements, extend to market outcomes, and incorporate managerial attributes such as CEO tenure.

- Behavioral constructs not directly measured: Heuristics, self-defense mechanisms, and psychological projections were theorized but lacked explicit instrumental variables. Future work should develop measures or experimental designs to capture these constructs. - Market outcomes and dynamic modeling: The study did not link earnings acceleration and investment choices to stock price or return volatility. A 3SLS framework connecting earnings management, over/underinvestment, earnings acceleration, and price/return dynamics is proposed for future research. - Managerial heterogeneity: CEO tenure and related characteristics were not considered; future designs should incorporate tenure effects on the relationships between financial distress/soundness and investment outcomes.