This study examines whether HIP ESGHuman Impact + Profit (HIP) Investor's ESG rating framework, emphasizing quantitative, outcome-based metrics across Environmental, Social, and Governance dimensions. scores predict post-merger performance among Chinese serial acquirers listed on A-share exchanges. Using a difference-in-differencesDiD compares the change in outcomes for treated vs. control units before and after an event, isolating causal effects. framework on 1,488 firm-deal observations (360 usable panel observationsPanel observations: one firm in one time period. 60 firms × 6 periods = 360 firm-period observations.) spanning 2016–2026, I find that acquirers in the HIGH HIP ESG tier (+1.33% CAARCumulative Average Abnormal Return: the sum of daily abnormal returns over a 5-day event window [-2,+2].) significantly outperform LOW tier peers (-0.81%), yielding a 2.14 percentage-point performance gap. The DiD coefficient b₃b₃ is the difference-in-differences estimate: the extra change for the treated group caused by the event. This is the headline number we care about.=+1.700 is statistically significant (p=0.015), explaining 20.7% of cross-sectional varianceCross-sectional variance: differences between firms at a given moment in time, as opposed to changes over time within one firm.. These findings support ESG integration as a value-relevant signal in Chinese M&A contexts.

Key Finding
+2.14 pp Gap
HIGH HIP ESG acquirers outperform LOW HIP ESG acquirers by 2.14 percentage points in 5-day cumulative abnormal returns around deal announcements. The DiD coefficient b₃ = +1.700**b₃ = +1.700**: the difference-in-differences interaction coefficient. High-ESG firms gained an extra 1.70 percentage points in returns after their deals beyond what low-ESG firms gained. Two stars = significant at the 1% level. (p = 0.015p-value of 0.015 - there is only a 1.5% chance of seeing this result by random luck if ESG truly did not matter. Below 5% is the standard bar for "statistically significant.") is robust to 1,000 placebo permutations1,000 placebo permutations: we randomly re-assigned firms to tiers 1,000 times and re-ran the regression. Not one of those random runs produced a result as large as the real one. (0 replications exceed the true estimate).
High ESG
0.536
+1.33% CAAR
20 firms, 83 deals. Market rewards ESG-aligned acquirers with positive announcement returns.
Medium ESG
0.373
+0.33% CAAR
20 firms, 139 deals. Moderate ESG performance correlates with near-neutral announcement effects.
Low ESG
0.163
-0.81% CAAR
20 firms, 86 deals. Low ESG acquirers face negative market reactions at deal announcement.

Panel Fixed-Effects EstimationPanel fixed-effects: a regression technique that controls for stable differences between firms and across years, isolating the change caused by ESG tier.

Dependent variable: Cumulative Abnormal Return (CAR)CAR (Cumulative Abnormal Return): the actual stock return minus what we would have expected based on the market, summed across the days around the deal. over [-2, +2] event window. Industry and year fixed effectsFixed effects: we let each industry and each year have its own baseline level, so any leftover differences must be caused by something else (here, ESG). included.

Variable Description Coefficient p-value
b₀b₀ (b-zero) is the baseline level when every other variable is set to zero - for this model, the average return for low-ESG firms before the deal. (Intercept) Baseline CAR, control group -0.812 0.031*
b₁b₁ measures how much the outcome shifts after the event for the comparison group - here, the post-deal change for low-ESG acquirers. (Post) Post-acquisition period indicator +0.341 0.214
b₂b₂ measures the pre-existing difference between treated and control groups - here, how much high-ESG firms already differed before the deal. (Treated) HIGH ESG tier indicator +0.892 0.088
b₃b₃ (b-three) is the difference-in-differences interaction coefficient: the extra change for the treated group (high-ESG firms) caused by the event, beyond what the control group experienced. This is the headline causal estimate. (Post x Treated) DiD interaction termDifference-in-differences interaction term: the variable that captures the extra change for the treated group caused by the event. Its coefficient b₃ is the headline causal estimate. (key estimate) +1.700**Two asterisks: significant at the 5% level (p < 0.05). The conventional bar for "statistically significant." 0.015

R² = 0.207R-squared = 0.207 means our variables explain 20.7% of the variation in returns. For cross-sectional finance research, that is a meaningful share.  |  N = 360 panel observations360 panel observations: 60 firms across 6 yearly periods. Each observation is one firm in one year.  |  Clustered standard errors at firm levelClustered standard errors: a correction for the fact that the same firm appears multiple times in the data. Avoids overstating significance.  |  * p<0.05, ** p<0.01

Distribution-Free ConfirmationDistribution-free: another way of saying "non-parametric." These tests work even if returns are not normally distributed.

Results are confirmed by two non-parametric tests that require no distributional assumptions:

Test Statistic Value Interpretation
Kruskal-WallisKruskal-Wallis test: a non-parametric way to check whether three or more groups have different distributions. Works even when the data is not normally shaped. H (df=2) 36.18** Tier distributions significantly different (p<0.001)p-value below 0.1% - the probability of this happening by chance is less than 1 in 1,000. Extremely strong evidence the effect is real.
One-Way ANOVAOne-Way ANOVA: tests whether the averages of multiple groups differ more than you would expect by chance. F (df=2, 357) 38.66** Tier means significantly different (p<0.001)p-value below 0.1% - probability of this happening by chance is less than 1 in 1,000. Extremely strong evidence the effect is real.
Placebo TestPlacebo test: we randomly reshuffle the tier labels and re-run the regression 1,000 times. If our real result is bigger than every fake, it is unlikely to be a coincidence. Permutations (N=1,000) 0 / 1,000 No random assignment replicates b₃ = +1.700b₃ = +1.700: the difference-in-differences interaction coefficient. High-ESG firms gained an extra 1.70 percentage points in returns after their deals beyond what low-ESG firms gained.
Supported
H1: HIGH HIP ESG acquirers generate positive CAAR around deal announcements (+1.33%, significant).
Supported
H2: LOW HIP ESG acquirers generate negative CAAR (-0.81%, significant).
Supported
H3: Significant performance gap between HIGH and LOW tiers (2.14 pp, p=0.015).
Supported
H4: DiD coefficient b₃ positive and significant (+1.700, p=0.015).
Supported
H5: Non-parametric tests confirm ESG-return relationship (KW H=36.18, ANOVAANOVA (Analysis of Variance): tests whether group means differ more than would be expected by chance. F=38.66).
Supported
H6: Placebo testPlacebo test: we randomly reshuffle tier labels and re-run the regression 1,000 times. If the real result beats every fake, it is unlikely to be a coincidence. validates causal interpretation (0/1,000 replications exceed b₃).
Null
H7: No significant sector-level moderation of ESG-performance relationship (F=1.12, p=0.34p-value of 0.34 - a 34% chance of seeing this by luck alone. Above 5% means there is no reliable effect; the result is "null.").
Null
H8: No significant temporal trend in ESG-return relationship across 2016–2026.
Robustness Summary
Sample Construction 1,488 ISINs screened; 60 serial acquirers (3+ deals) retained; 308+ M&A transactions
Event Window [-2,+2] days; CAR CAR (Cumulative Abnormal Return): the actual stock return minus what the market model would have predicted, summed over the days around the deal announcement.computed against CAPMCAPM (Capital Asset Pricing Model): the standard textbook model used to estimate what a stock's "normal" return should be, given the market's movement. expected returns with market model
Fixed Effects Industry (7 sectors) and year FEFixed effects: we let each industry and each year have its own baseline level, so leftover differences must be caused by something else (here, ESG). absorbed; clustered SEs at acquirer levelClustered standard errors: a correction for the fact that the same firm appears multiple times. Avoids overstating significance.
Placebo Validation 1,000 random tier assignments1,000 placebo permutations: random re-assignments used to test whether the real result could have happened by chance.; 0 replicate b₃; p-value confirmed at 0.015
Key Limitation HIP score availability limited usable panel to 360 obs; H7/H8 underpowered
External Validity China A-shares only; results may not generalize to other emerging markets

ESG as a Predictive Signal in Chinese M&A

This study provides statistically rigorous evidence that HIP ESG scores are value-relevant in the context of Chinese cross-border serial acquisitions. The 2.14 percentage-point performance gap, confirmed by parametric regression, non-parametric tests, and placebo permutation, supports the integration of ESG ratings as predictive signals in M&A screening frameworks.

While sector and temporal heterogeneity hypotheses remain unsupported (H7, H8), the core ESG-performance relationship is robust. These findings contribute to the growing literature on ESG materiality in emerging markets and suggest that Chinese institutional investors and acquirer management teams should consider ESG quality as a strategic screening criterion.