KIM JUNGSOO
BOOKS
The Dynamic Causal Effects of Corporate AI Adoption — cover
KCI

The Dynamic Causal Effects of Corporate AI Adoption on Profitability and Market Value: An Empirical Analysis of KOSDAQ Panel Data

Jungsoo Kim (first author), Bongkee Baek (corresponding author)

  • INDEXEDKCI — Korea Citation Index
  • PUBLISHEDMarch 2026 · Vol. 6, No. 2 · pp. 135–154
  • AFFILIATIONMidwest University · Soongsil University
  • DATA922 KOSDAQ-listed firms · 7,376 firm-year observations · 2018–2025

Abstract

This study analyzes the causal effects of AI adoption on the corporate performance of KOSDAQ-listed companies from 2018 to 2025. The timing of adoption was identified through a multi-step, contextually validated text analysis of DART business reports, and endogeneity was controlled using a two-way fixed effects (TWFE) model and Propensity Score Matching (PSM). The analysis results reveal that AI adoption had a significantly negative (−) impact on the operating profit margin (OPM), consistent with short-term "J-curve" transition costs including process redesign and capability buildup. No statistically significant change was observed in ROA, reflecting the offsetting dynamics of income compression and asset expansion during the early adoption phase. While the average effect on market value (Tobin's Q) was not significant across all firms, a heterogeneous effect was observed in the ICT industry, where Tobin's Q significantly increased following AI adoption. This result indicates that the capital market evaluates AI investment as a future "growth option" selectively within industrial contexts characterized by strong data infrastructure, digital workforce readiness, and technological absorptive capacity. By demonstrating that the value creation process of AI adoption varies according to industry-specific complementary asset structures and ecosystem conditions, this study provides significant implications for digital transformation strategies, investment decision-making, and AI diffusion policy.

Key Findings

  1. 01

    Adopting AI costs money before it makes money.
    Operating profit margin falls in the short run — the signature of process redesign, retraining, and integration work that lands before any return does.

  2. 02

    The market pays for it anyway — but only where it fits.
    Among ICT firms, Tobin's Q rises significantly. Investors price AI adoption as a growth option, not as this year's earnings.

  3. 03

    Industry context decides the outcome, not the technology.
    The same adoption produces different value depending on the complementary assets a firm already holds. AI is not a uniform treatment.

  4. 04

    Measured on 922 firms, 7,376 firm-year observations, 2018–2025.
    Adoption identified from DART filings through a multi-step verification that excludes AI-washing — mentions without externally corroborated deployment.

Tables

Table 1 — Variable Definitions

CategoryVariableDefinition (Formula)Expected Sign
Dependent VariablesROANet Income / Total Assets
OPMOperating Profit / Revenue
Tobin's Q(Market Cap + Total Debt) / Total Assets
Independent VariableAI_dummyFirm-year AI adoption dummy (Adoption = 1)+
Complementary Assets (Moderators)ZRNDR&D Intensity (Standardized)Reinforce AI effect
ZIntangIntangible Asset Ratio (Standardized)Reinforce AI effect
Industrial Context (Moderator)ICT_dummyICT Industry Status (1/0)Reinforce AI effect
ZDigitalStandardized industry digital maturity index: three equally-weighted dimensions — AI-related exposure (ICT concentration), evidence (depth of AI deployment outcomes), and assets (infrastructure level, HHI) — robustly normalized and Z-scoredReinforce AI effect
Control VariableslnTALog of Total Assets
LeverageTotal Debt / Total Assets
LiquidityCurrent Assets / Current Liabilities
GrowthYear-over-Year Revenue Growth Rate
CAPEXCapital Expenditure / Total Assets

Table 2 — Summary of Descriptive Statistics and Preliminary Analysis

VariableMeanStd. Dev.MinMax
ROA0.0050.112−0.5970.338
OPM0.0260.181−1.5650.525
Tobin's Q1.3730.8870.3617.297
AI_w0.1310.33801
Size(lnA)18.7650.85016.11622.591
Leverage0.3410.1920.0270.997
Liquidity0.1080.1070.0000.831

Table 3 — Summary of Baseline Model (Firm/Year Fixed Effects) Results

Dependent Var.AI Coeff. (β)Std. Errorp-valueSummary Interpretation
ROA−0.0090.0040.231Not Significant: Offsetting transition-cost and asset-expansion effects
OPM−0.0170.0070.008**Significantly Negative (−): J-curve / Transition Cost dominance
Tobin's Q0.0010.0260.975Not Significant (Average): ICT heterogeneity confirmed in Sect. 4.3

* p < .05, ** p < .01, *** p < .001

Table 4 — Summary of Moderating Effects on ROA, OPM, and Tobin's Q

Moderator (Interaction)ROA (β, p)OPM (β, p)Tobin's Q (β, p)Summary Interpretation
ZRND0.000, p=0.9570.005, p=0.609−0.050, p=0.227Not Significant: accounting proxy limitation
ZIntang−0.001, p=0.7130.008, p=0.144−0.028, p=0.178Not Significant: accounting proxy limitation
ZDigital0.003, p=0.5410.010, p=0.123−0.052, p=0.061Negative sign, marginal p: contrary to H3 direction; inconclusive
ICT_dummy0.004, p=0.598−0.003, p=0.8330.192, p=0.000***Significant in ICT: Core Heterogeneous Finding (RBV, GPT theory)

* p < .05, ** p < .01, *** p < .001

Model

A two-way fixed effects (TWFE) panel regression identifies the causal effect of AI adoption. Firm fixed effects control for unobserved time-invariant firm characteristics; year fixed effects control for common annual shocks. Standard errors are clustered at the firm level.

Yit = β1AIit + γ′Xit + μi + λt + εit

Yit = β1AIit + β2Mit + β3(AIit × Mit) + δ′Xit + μi + λt + εit

Yit is ROA, OPM, or Tobin's Q; AIit is the adoption dummy; Xit the control set; Mit complementary assets (ZRND, ZIntang) or industrial context (ICT_dummy); μi and λt the firm and year fixed effects.

Full Text

The full text is available through the Korea Citation Index.