Pay Restrictions and Labor Investment: Cao, Hasan, Huang & Zhao (2026)

What this is. A distilled skeleton of Cao, Hasan, Huang & Zhao (2026), “Pay restrictions and labor investment,” Journal of Corporate Finance 99, 102990. Read the original to replicate or extend. This summary is LLM-distilled, not human-verified, and not reproduced.

TL;DR

Using China’s 2014 SOE executive compensation reform as a quasi-natural experiment, the paper estimates a 3.91% to 4.82% post-reform reduction in abnormal labor investment (ALI) in SOEs relative to non-SOEs. The reduction is concentrated in over-investment rather than under-investment. Results are consistent with internal governance and reduced social comparison as channels: the reform narrows executive pay gaps, and ALI declines more in firms with higher internal governance (subsample difference p = 0.076) and lower executive-to-worker pay disparity. The reform also reduces over-hiring, improves labor quality among firms with lower ALI, and increases employee well-being (employee shares, work safety, vocational training, and director communication channels).

Core results

#	Result	Locator	Magnitude as reported
R1	Pay restrictions reduce SOE abnormal labor investment	Table 4 cols 1-4, p. 15	SOE x Restriction = -0.046***, -0.048***, -0.062***, and -0.039***; paper reports a 3.91% to 4.82% reduction in ALI
R2	Reduction concentrated in over-investment, not under-investment	Table 4 cols 5-6, p. 15	SOE x Restriction = -0.075*** (over-invest); 0.003 n.s. (under-invest); F-test p < 0.001
R3	Reform cuts average SOE executive compensation	Table 3 Panel A col 1, p. 14	SOE x Restriction = -0.096***; 10.08% reduction (~49,000 CNY/year)
R4	Reform increases internal governance effectiveness by 14.05%	Table 5 col 1, p. 16	SOE x Restriction = 0.141** on internal governance score
R5	Higher internal governance firms see 5.57% reduction in ALI	Table 5 cols 2-3, p. 16	SOE x Restriction = -0.056** (high IG), -0.032 n.s. (low IG); F-test p = 0.076
R6	Reform reduces CEO-to-rank pay disparity in SOEs	Table 6 Panel A col 1, p. 17	SOE x Restriction = -1.180*** (SE 0.374)
R7	Reform reduces over-hiring by 8.73%	Table 8 col 1, p. 23	SOE x Restriction = -0.087***; under-hiring, over-firing, under-firing insignificant
R8	Reform improves labor quality and employee well-being	Table 9, p. 23	Labor quality +1.33% (col 1); stock ownership +7.63%, work safety +6.22%, training +3.41%, communication +3.73%
R9	Compensation falls more in SOEs in high-corruption regions	Table 3 Panel A cols 2-3, p. 14	SOE x Restriction = -0.124*** in high-corruption regions, -0.036 n.s. in low-corruption regions; F-test p = 0.008; reported pay reduction 13.20% (about CNY 64,000/year) in high-corruption regions; subsample covers 2011-2017
R10	The reform reduces the pay-for-performance sensitivity of SOE executives	Table 3 Panel A col 4, p. 14	SOE x Restriction x ROA = -1.135***; SOE x Restriction = -0.049**; F-test p = 0.000
R11	The reform reduces SOE executive perks	Table 3 Panel A col 5, p. 14	SOE x Restriction = -0.114***; reported reduction = 12.08%
R12	The reform reduces excess SOE executive compensation	Table 3 Panel A col 6, p. 14	SOE x Restriction = -0.119***; reported reduction = 12.64%
R13	The reform reduces the likelihood of executive misconduct	Table 3 Panel B col 1, p. 14	SOE x Restriction = -0.031*** (SE 0.010); reported reduction = 3.10%
R14	The reform reduces the number of executive misconduct cases	Table 3 Panel B col 3, p. 14	SOE x Restriction = -0.161*** (SE 0.043); reported reduction = 16.09%
R15	Misconduct likelihood falls more for high-compensation executives	Table 3 Panel B cols 2, p. 14	High-compensation interaction = -0.041*** (SE 0.011); low-compensation interaction = -0.014 n.s.; F-test p = 0.026; reported high-compensation reduction = 4.14%
R16	Misconduct case counts fall more for high-compensation executives	Table 3 Panel B col 4, p. 14	High-compensation interaction = -0.213*** (SE 0.045); low-compensation interaction = -0.074 n.s.; F-test p = 0.012; reported high-compensation reduction = 21.32%
R17	The reform narrows top-three-executive to rank-and-file pay disparity	Table 6 Panel B col 1, p. 17	SOE x Restriction = -1.004*** (SE 0.334)
R18	The reform narrows all-top-executive to rank-and-file pay disparity	Table 6 Panel C col 1, p. 17	SOE x Restriction = -0.335*** (SE 0.078)
R19	ALI falls in low-disparity firms for the top-three-executive pay comparison	Table 6 Panel B cols 2-3, p. 17	High-disparity coefficient = -0.020 n.s.; low-disparity coefficient = -0.057***; F-test p = 0.042; reported ALI reduction = 5.72%
R20	ALI falls in low-disparity firms for the all-executive pay comparison	Table 6 Panel C cols 2-3, p. 17	High-disparity coefficient = 0.003 n.s.; low-disparity coefficient = -0.062***; F-test p = 0.011; reported ALI reduction = 6.24%
R21	Pre-reform event-study coefficients do not reject parallel trends	Table 7 Panel A, p. 19	Joint pre-coefficient F-test p = 0.828 (full sample), 0.400 (over-investment), and 0.820 (under-investment); individual pre coefficients are insignificant
R22	Event-study estimates turn negative after reform for ALI and over-investment	Table 7 Panel A, p. 19; Figure 1, p. 20	Full-sample SOE x Post_1..Post_5 = -0.015, -0.043**, -0.047**, -0.081***, -0.037**; over-investment = -0.077*, -0.104**, -0.118***, -0.197***, -0.102***; under-investment post estimates are insignificant
R23	The baseline ALI effect persists in entropy-balanced samples	Table 7 Panel B, p. 19	Full-sample SOE x Restriction estimates = -0.043**, -0.051***, -0.040**, -0.034* across four weighted specifications; over-investment = -0.092***; under-investment = 0.022 n.s.; F-test p < 0.001
R24	The result persists under the first alternative ALI measure	Table 7 Panel C cols 1-3, p. 19	Abnormal Labor Investment2 SOE x Restriction = -0.042***, -0.054***, -0.033** (N = 12,741 in each specification)
R25	The result persists under the second alternative ALI measure	Table 7 Panel C cols 4-6, p. 19	Abnormal Labor Investment3 SOE x Restriction = -0.023***, -0.024**, -0.017* (N = 11,558 in each specification)
R26	The ALI effect is robust to alternative fixed-effect structures	Table 7 Panel D, pp. 19-20	SOE x Restriction = -0.053***, -0.045***, -0.049***, -0.030**, -0.068** (over-investment), and 0.012 n.s. (under-investment) across the reported specifications
R27	Oster bounds indicate the baseline estimate is unlikely to be driven by omitted variables	Table 7 Panel E, p. 20	Full sample: beta* = -0.044 within [-0.063, -0.015], delta = 3.432; over-investment: beta* = -0.086 within [-0.121, -0.029], delta = 3.156; both classified “Unlikely” to reflect omitted-variable bias
R28	The reform does not significantly change rank-and-file employee compensation	§4.5 footnote 23, p. 16; Appendix E, p. 30	SOE x Restriction = -0.034 (SE 0.043) for current compensation and -0.055 (SE 0.046) for next-year compensation; both n.s.
R29	Randomized-treatment placebo estimates center near zero, unlike the actual estimates	Figure 2, p. 22	Across 1,000 placebo draws, the true ALI coefficient is -0.039 and the true over-investment coefficient is -0.075; both appear in the left tail of their placebo distributions
R30	The reform has no significant effect on the other three labor-investment types	Table 8 cols 2-4, p. 23	SOE x Restriction = -0.011 for under-firing, 0.002 for under-hiring, and 0.004 for over-firing; all n.s.
R31	The CEO pay-disparity split shows a larger ALI reduction in low-disparity firms	Table 6 Panel A cols 2-3, p. 17	High-disparity coefficient = -0.015 n.s.; low-disparity coefficient = -0.047***; F-test p = 0.048; reported low-disparity ALI reduction = 4.68%
R32	Labor quality rises in low-ALI firms, with no significant increase in high-ALI firms	Table 9 Panel A, p. 23	SOE x Restriction = 0.013** (SE 0.007) in low-ALI firms and -0.005 n.s. (SE 0.009) in high-ALI firms; F-test p = 0.018
R33	The reform increases employee stock ownership	Table 9 Panel B col 1, p. 23	SOE x Restriction = 0.076*** (SE 0.016); reported increase = 7.63%
R34	The reform increases work-safety management systems	Table 9 Panel B col 2, p. 23	SOE x Restriction = 0.062*** (SE 0.016); reported increase = 6.22%
R35	The reform increases vocational training for employees	Table 9 Panel B col 3, p. 23	SOE x Restriction = 0.034** (SE 0.015); reported increase = 3.41%
R36	The reform increases employee-director communication channels	Table 9 Panel B col 4, p. 23	SOE x Restriction = 0.037** (SE 0.016); reported increase = 3.73%

Overall. China’s 2014 SOE executive compensation reform is associated with lower abnormal labor investment, primarily over-investment. The mechanism tests are consistent with stronger internal governance and reduced social comparison: the reform narrows executive pay gaps, and ALI declines more in higher-governance and lower executive-to-worker disparity subsamples. These patterns are not equally strong: the governance subsample difference has p = 0.076, while the three social-comparison splits have p-values from 0.011 to 0.048. Additional results show that the reform cuts over-hiring specifically, improves labor quality in lower-ALI firms, and increases employee well-being (employee shares, work safety, vocational training, and director communication channels).

Theory / model

The paper has no formal theoretical model. It tests two competing predictions: pay caps may worsen labor-investment efficiency if they weaken managerial effort and retention, or improve it by limiting managerial rent extraction and strengthening oversight. The proposed channels are (i) internal governance, where closer compensation and horizon alignment lets subordinate executives monitor CEOs, and (ii) social comparison, where smaller executive-to-worker pay gaps improve fairness and employee cooperation. The 2014 central-government SOE pay reform is the quasi-natural shock: the treatment is SOE status interacted with the post-reform period, with 2014 excluded from the sample. The authors argue that firms cannot influence the Political Bureau directive (PDF §3.2.2, p. 8).

Method

The design is a firm-panel difference-in-differences study. Abnormal labor investment is the absolute residual from an industry-by-year model of expected net hiring. Net Hire is the percentage change in employees. The paper’s full prediction model (equation 1, PDF §3.2.1, p. 7) is:

$$\begin{aligned} \text{Net Hire}_{it} ={}& \alpha + \beta_1 \text{Sales Growth}_{i,t-1} + \beta_2 \text{Sales Growth}_{it} + \beta_3 \Delta\text{ROA}_{i,t-1} + \beta_4 \Delta\text{ROA}_{it} + \beta_5 \text{ROA}_{it} + \beta_6 \text{Size Rank}_{i,t-1} \\ &+ \beta_7 \Delta\text{Quick Ratio}_{i,t-1} + \beta_8 \Delta\text{Quick Ratio}_{it} + \beta_9 \text{Quick Ratio}_{i,t-1} + \beta_{10} \text{Leverage}_{i,t-1} + \beta_{11} \text{LossBin}_{1,i,t-1} + \beta_{12} \text{LossBin}_{2,i,t-1} + \beta_{13} \text{LossBin}_{3,i,t-1} + \beta_{14} \text{LossBin}_{4,i,t-1} + \beta_{15} \text{LossBin}_{5,i,t-1} + \varepsilon_{it} \tag{1} \end{aligned}$$

The residual’s absolute value is Abnormal Labor Investment (ALI); a positive residual is over-investment and a negative residual is under-investment. Internal governance combines standardized subordinate-executive horizon and ability measures (equations 2-3, PDF §3.2.3, p. 8):

$$\text{Executive Horizon}_{it} = 65 - \text{Average Age of Key Subordinate Executives}_{it} \tag{2}$$ $$\text{Executive Ability}_{it} = \frac{\text{Average Annual Compensation of Key Subordinate Executives}_{it}}{\text{Annual Compensation of CEO}_{it}} \tag{3}$$

The excess-compensation residual comes from the following firm-level model with industry and year fixed effects (equation 4, PDF §3.2.5, p. 9):

$$\begin{aligned} \text{Executive Compensation}_{it} ={}& \alpha + \beta_1 \text{Firm Size}_{it} + \beta_2 \text{Leverage}_{it} + \beta_3 \text{ROA}_{it} + \beta_4 \text{ROA}_{i,t-1} + \beta_5 \text{Sales Growth}_{it} \\ &+ \beta_6 \text{Dual}_{it} + \beta_7 \text{SOE}_{it} + \beta_8 \text{Independent}_{it} + \text{Industry FE} + \text{Year FE} + \varepsilon_{it} \tag{4} \end{aligned}$$

The mechanism measures are the subordinate executives’ monitoring horizon and compensation influence, and three executive-to-rank-and-file compensation ratios (CEO, top three paid executives, and all top executives; PDF §§3.2.3-3.2.4, pp. 8-9). The abnormal labor investment measure follows Jung, Lee & Weber (2014); the paper extends the employee-friendly treatment and labor-investment link in Cao and Rees (2020), and the CEO-director tie work in Khedmati, Sualihu & Yawson (2020). Its over-hiring prediction tests the power theory of Bebchuk and Fried (2003), and its internal-governance measure adopts Cheng, Lee & Shevlin (2016). The panel regressions use firm and year effects and firm-clustered standard errors unless a robustness table states a different fixed-effect structure. Continuous variables are winsorized at the 1st and 99th percentiles. The main sample contains 14,988 firm-year observations for 2,889 firms, 2009-2019 excluding 2014 (PDF §3.1, p. 7; Table 1, p. 8).

Empirical specifications

The policy-effect regressions first test compensation outcomes (equation 5), then whether the reform changes pay-for-performance sensitivity (equation 6). The misconduct specifications are equations 7-8 (PDF §3.2.7, pp. 10-11):

$$\text{Compensation Variables}_{it} = \alpha + \beta_1(\text{SOE}\times\text{Restriction})_{it} + \beta_2\text{SOE}_{it} + \gamma\text{Control Variables}_{i,t-1} + \text{Fixed Effects} + \varepsilon_{it} \tag{5}$$ $$\begin{aligned} \text{Executive Compensation}_{it} ={}& \alpha + \beta_1(\text{SOE}\times\text{Restriction}\times\text{ROA})_{it} + \beta_2(\text{SOE}\times\text{Restriction})_{it} + \beta_3(\text{Restriction}\times\text{ROA})_{it} \\ &+ \beta_4(\text{SOE}\times\text{ROA})_{it} + \beta_5\text{SOE}_{it} + \beta_6\text{ROA}_{it} + \gamma\text{Control Variables}_{i,t-1} + \text{Fixed Effects} + \varepsilon_{it} \tag{6} \end{aligned}$$ $$\text{Executive Misconduct}_{it} = \alpha + \beta_1(\text{SOE}\times\text{Restriction})_{it} + \beta_2\text{SOE}_{it} + \beta_3\text{Restriction}_{it} + \gamma\text{Control Variables}_{it} + \text{Fixed Effects} + \varepsilon_{it} \tag{7}$$ $$\begin{aligned} \text{Executive Misconduct}_{it} ={}& \alpha + \beta_2(\text{SOE}\times\text{Restriction}\times\text{High Compensation})_{it} + \beta_3(\text{SOE}\times\text{Restriction}\times\text{Low Compensation})_{it} \\ &+ \beta_4\text{Executive Compensation}_{it} + \beta_5\text{SOE}_{it} + \gamma\text{Control Variables}_{it} + \text{Fixed Effects} + \varepsilon_{it} \tag{8} \end{aligned}$$

In equations 5-8, the outcomes are executive compensation, perks, excess compensation, pay-for-performance, or misconduct (dummy or case count), as applicable. Models use firm and year fixed effects and firm-clustered robust standard errors. Equation 5 and equation 6 use lagged controls; equations 7-8 use contemporaneous controls. The compensation/misconduct sample sizes vary by outcome and table: Table 3 reports 9,701-14,939 observations (PDF Table 3, p. 14).

The baseline ALI difference-in-differences specification is equation 9 (PDF §3.2.7, p. 11):

$$\text{Abnormal Labor Investment}_{it} = \alpha + \beta_1(\text{SOE}\times\text{Restriction})_{it} + \beta_2\text{SOE}_{it} + \beta_3\text{Restriction}_{it} + \gamma\text{Control Variables}_{it} + \text{Fixed Effects} + \varepsilon_{it} \tag{9}$$

Equation 9 includes firm, industry, and year fixed effects in the main specification; standard errors are robust and clustered by firm. Controls include leverage, book-to-market, ROA, quick ratio, firm age and size, fixed assets, Herfindahl index, institutional share, SA index, and CEO gender, age, and education. The main sample is 14,988 firm-years (Table 4, p. 15); the reported columns vary fixed effects and sample splits.

Equations 10-12 estimate the governance channel, with equation 10 using internal governance as the outcome and equations 11-12 estimating ALI separately above and below the median governance score (PDF §3.2.7, p. 11):

$$\text{Internal Governance}_{it} = \alpha + \beta_1(\text{SOE}\times\text{Restriction})_{it} + \beta_2\text{SOE}_{it} + \beta_3\text{Restriction}_{it} + \gamma\text{Control Variables}_{it} + \text{Fixed Effects} + \varepsilon_{it} \tag{10}$$ $$\begin{aligned} \text{ALI}_{(i,t)_{\text{High I.G.}}} ={}& \alpha + \beta_1(\text{SOE}\times\text{Restriction})_{(i,t)_{\text{High I.G.}}} + \beta_2\text{SOE}_{(i,t)_{\text{High I.G.}}} \\ &+ \gamma\text{Controls}_{(i,t)_{\text{High I.G.}}} + \text{Fixed Effects} + \varepsilon_{(i,t)_{\text{High I.G.}}} \tag{11} \end{aligned}$$ $$\begin{aligned} \text{ALI}_{(i,t)_{\text{Low I.G.}}} ={}& \alpha + \beta_1(\text{SOE}\times\text{Restriction})_{(i,t)_{\text{Low I.G.}}} + \beta_2\text{SOE}_{(i,t)_{\text{Low I.G.}}} \\ &+ \gamma\text{Controls}_{(i,t)_{\text{Low I.G.}}} + \text{Fixed Effects} + \varepsilon_{(i,t)_{\text{Low I.G.}}} \tag{12} \end{aligned}$$

Equations 10-12 include firm and year fixed effects and firm-clustered robust standard errors. The governance outcome sample is 13,088 firm-years; each median-split ALI sample has 6,544 observations (Table 5, p. 16).

Equations 13-15 test social comparison, where equation 13 uses each executive-to-worker disparity measure and equations 14-15 estimate ALI in high- and low-disparity subsamples (PDF §3.2.7, p. 11):

$$\text{Social Comparison Variables}_{it} = \alpha + \beta_1(\text{SOE}\times\text{Restriction})_{it} + \beta_2\text{SOE}_{it} + \beta_3\text{Restriction}_{it} + \gamma\text{Control Variables}_{it} + \text{Fixed Effects} + \varepsilon_{it} \tag{13}$$ $$\begin{aligned} \text{ALI}_{(i,t)_{\text{High S.C.}}} ={}& \alpha + \beta_1(\text{SOE}\times\text{Restriction})_{(i,t)_{\text{High S.C.}}} + \beta_2\text{SOE}_{(i,t)_{\text{High S.C.}}} \\ &+ \gamma\text{Controls}_{(i,t)_{\text{High S.C.}}} + \text{Fixed Effects} + \varepsilon_{(i,t)_{\text{High S.C.}}} \tag{14} \end{aligned}$$ $$\begin{aligned} \text{ALI}_{(i,t)_{\text{Low S.C.}}} ={}& \alpha + \beta_1(\text{SOE}\times\text{Restriction})_{(i,t)_{\text{Low S.C.}}} + \beta_2\text{SOE}_{(i,t)_{\text{Low S.C.}}} \\ &+ \gamma\text{Controls}_{(i,t)_{\text{Low S.C.}}} + \text{Fixed Effects} + \varepsilon_{(i,t)_{\text{Low S.C.}}} \tag{15} \end{aligned}$$

Equations 13-15 also use firm and year fixed effects and firm-clustered robust standard errors. Equation 13 has 14,316 observations for each of the three disparity measures; each high- and low-disparity subsample has 7,158 observations (Table 6, p. 17). The high/low splits test whether the predicted social-comparison channel is concentrated among firms with initially smaller pay gaps.

The robustness checks include a dynamic event study, entropy balancing, alternative ALI residual constructions, additional fixed-effect structures, and Oster bounds (Table 7, pp. 19-20). The parallel-trend event-study estimates show insignificant pre-period coefficients; the placebo exercise randomizes treatment events 1,000 times (Figure 2, p. 22). The paper also reports that using 2015 rather than 2014 as the event year leaves the main results consistent (PDF §3.2.2, p. 8).

Datasets used

Dataset	Role in paper	Wiki page
CSMAR (China Stock Market & Accounting Research)	Financial variables, stock market data, firm characteristics, executive compensation, abnormal labor investment construction, executive misconduct data	/wiki/commercial/csmar/
CNRDS (Chinese Research Data Services)	Firm ownership data, CEO characteristics, employee well-being data (stock grants, work safety, training, communication channels), labor quality	no page yet
Procuratorial Yearbook of China	Provincial-level embezzlement and bribery cases (political corruption measure)	no page yet
China Statistical Yearbook	Provincial-level civil servant counts (denominator for political corruption measure)	no page yet

Sample: 14,988 firm-year observations representing 2,889 unique A-share listed firms in China across 71 CSRC 2012 industries, spanning 2009-2019 (event year 2014 excluded; financial firms and ST-designated firms removed). The main sample is annual.

When to read the full paper

Read Cao et al. (2026) if you are studying: (1) the labor investment consequences of executive compensation regulation in emerging markets; (2) the corporate governance of state-owned enterprises, particularly how internal pay structures affect investment efficiency; (3) causal identification of executive compensation effects using quasi-natural experiments; or (4) the mechanisms through which pay inequality between executives and workers affects organizational outcomes. Key tables are Table 4 (main DiD), Table 5 (internal governance channel), Table 6 (social comparison channel), Table 8 (over/under-hiring decomposition), and Table 9 (labor quality and employee well-being).

Attribution and rights

Cao, J., Hasan, I., Huang, Z., & Zhao, J. (2026). Pay restrictions and labor investment. Journal of Corporate Finance, 99, 102990. https://doi.org/10.1016/j.jcorpfin.2026.102990

Copyright 2026 Elsevier B.V. All rights reserved. This page reproduces no substantial text from the article; results, magnitudes, and table references are extracted for scientific commentary under fair-use principles (extract-only). The article is paywalled; access requires a subscription to the Journal of Corporate Finance or institutional Elsevier access. This summary is LLM-distilled by paper-distiller (gpt-6-luna) and has not been human-verified or independently reproduced.