# llms.txt — The Value of Names: Civil Society, Information, and Governing Multinationals

**Authors:** David H. Kreitmeir (University of Southampton / Monash University), Nathan Lane (London School of Economics), Paul A. Raschky (Monash University)

---

## What is this paper about?

This paper asks whether stock markets penalize multinational mining companies when they are publicly named in connection with the assassination of environmental and community activists in developing countries. The core finding is yes: being named in media coverage of an activist killing causes a significant, persistent drop in firm value — even though the companies themselves are almost never legally prosecuted, and the victims are not the firms' customers or suppliers. The paper uses these market reactions to measure the size and durability of reputational (market-based) penalties for human rights misconduct abroad, and investigates what drives them.

---

## Important context

**What "association" means:** A firm is "treated" if it or its mining project is named in media or NGO reporting on an assassination. This is not a finding of legal guilt or even direct causal involvement — it is a measure of public linkage. The paper is careful to stress this distinction throughout.

**Why this is surprising:** The law-and-economics literature on firm misconduct consistently finds that reputational market penalties are small or absent when the victims of misconduct are not the firm's counterparties (i.e., not buyers or investors). Environmental pollution, for example, typically produces market reactions that almost entirely reflect expected legal fines, not reputational costs. Here, legal fines are virtually nonexistent and the victims (local activists) are not counterparties — yet the market penalty is large. This is the central puzzle the paper resolves.

**Reputational penalty defined:** A reputational penalty is the decline in firm value that exceeds any expected formal legal cost — it reflects market participants updating expectations about future business with the firm. This paper documents a predominantly reputational (not legal) penalty.

**Relationship to ESG literature:** This paper differs from typical ESG event studies in several ways. It uses a homogeneous class of unambiguous, severe events (assassinations) rather than heterogeneous ESG ratings or incident databases. It focuses on upstream commodity producers rather than consumer-facing brands. And it documents persistent rather than transitory penalties, which is unusual in this literature.

**Terminology:** "Targeted killing," "extrajudicial killing," and "assassination" are used interchangeably, following journalism and conflict-studies conventions. The paper studies civil society members (indigenous leaders, labor and environmental activists, clergy, educators) killed in proximity to mining operations.

---

## Data and methods

**Event data:** 354 assassination events (496 victims), hand-coded from LexisNexis, AP, Guardian, and other international media archives; 1998–2019; 31 countries; Peru and the Philippines are the most represented. Events are linked to 15 of the 26 ICMM member firms. Firm-event pairs (N = 167 for the core event study) are coded by matching companies named in reporting, tracing parent companies through Orbis and SNL Metals & Mining.

**Financial data:** Daily stock returns from Datastream (1998–2019), denominated in USD; MSCI country indices as market benchmarks; Fama-French four-factor model for robustness.

**Main estimation approaches:**
1. *Traditional event study* — market model estimated over a 250-day window, cumulative abnormal returns (CARs) computed over [0, 10] days post-event; GRANK non-parametric test statistic.
2. *Regression event study (OLS)* — CARs regressed on treatment indicator with event fixed effects, comparing treated firms to same-country same-sector control firms; clustered standard errors.
3. *Synthetic matching* — treated firms matched to a convex combination of control firms on pre-event return dynamics; implemented in the open-source R package `synthReturn` (github.com/davidkreitmeir/synthReturn).
4. *Stacked difference-in-differences* — used for mechanism analysis (institutional investors, supply chain), following Cengiz et al. and Baker et al. to handle staggered treatment.

**Mechanism data:** FactSet institutional ownership (quarterly, 2000–2017); FactSet Revere supply chain contracts; Eisensee–Strömberg daily news pressure index; EITI mining royalty reports; V-Dem civil liberties index; ICRG Law and Order index.

---

## Key results

**Core market penalty:** Firms named in assassination news lose on average 2.0 percentage points in cumulative abnormal returns over the 10 days following the event (market model; significant at 1%). The OLS regression estimate against control firms is 2.0–2.8 percentage points. The synthetic matching estimate reaches approximately 6.1 percentage points at Day 10. Crucially, there are no significant pre-event abnormal returns, confirming this is new information, not a pre-existing trend.

**Economic magnitude:** The median treated company loses USD 100–150 million in market capitalization over the 10-day window. This is larger than the median loss from negative ESG events in Krueger (2015) (USD 76 million).

**Persistence:** Long-run CARs remain negative (~2.4 percentage points) over a 90-day window. Buy-and-hold abnormal returns (BHAR) exceed CARs in magnitude over longer horizons. The effects do not reverse, contrasting with the transitory penalties typical of environmental disasters and noisy ESG news.

**Mechanism 1 — Media:** The market penalty is present only when events occur on low-news-pressure days. On high-news-pressure days (crowded news cycles), the penalty is indistinguishable from zero. This holds using both the raw Eisensee–Strömberg news pressure index and a "disaster-predicted" instrument for exogenous variation. The penalty is also absent for placebo firms operating near event locations but not named in coverage.

**Mechanism 2 — Institutional investors:** Hedge funds reduce their holdings by approximately 7.8% relative to their average position in the quarter following an assassination event. Pension funds and long-term strategic investors do not change their positions significantly. The result is consistent with hedge funds' shorter investment horizons and greater sensitivity to new, complex information.

**Mechanism 3 — Supply chain:** Being associated with an assassination reduces new contracts with customers from countries with strong human rights protection (above-median V-Dem Civil Liberties) by 19–21% and the number of new customers from those countries by a comparable amount. The effect is absent for customers from authoritarian or weak civil society countries, who may even increase trade.

**Political economy of persistence:** Mining companies with a higher share of total mining royalties paid to the national government are significantly more likely to be associated with an assassination event (18-percentage-point increase for a firm that is the sole taxpayer, within country-year). This suggests local fiscal rents — not firm-level decisions — sustain violence even when market penalties are high.

**What does not drive the results:** Legal penalties (no firm in the sample was convicted or fined); local operational disruptions from protests (protest activity peaks at the event week and then declines, and there is no pre-event market reaction); contemporaneous regional conflict (geographically matched control firms show no significant effects).

---

## Limitations and scope


**Sample scope:** The study covers the mining sector specifically. Results may not generalize to other extractive industries (e.g., oil and gas), manufacturing supply chains, or sectors where products are branded consumer goods. The mechanism may operate differently for consumer-facing firms.

**Association, not guilt:** The paper cannot and does not claim that named firms organized or participated in violence. Treatment reflects public naming in media and NGO reports, not investigative findings of responsibility.

**ESG scores do not move:** Assassination events do not significantly change firms' overall ESG scores or human rights sub-scores in standard third-party databases. This limits the ability of ESG-passive institutional investors to respond and explains why pension funds do not react.

**Legal setting is specific:** The near-total absence of legal sanction is a feature of the setting, not a general condition. In jurisdictions with stronger rule of law, stock price reactions to comparable events might look different because they would partly reflect legal risk.

**Long-run estimates are indicative:** Buy-and-hold and 90-day CAR estimates face standard econometric challenges (benchmark sensitivity, compounding). They should be read as suggestive of persistence, not precise causal estimates.

---

## Navigation guide

Readers primarily interested in the **main empirical result** should read Section 4.1 (traditional event study) and Section 4.2.2 (OLS regression estimates), plus Figure 4, Figure 5, and Figure 6.

Readers interested in **causal identification and robustness** should read Section 4.2.3 (robustness) and Section 4.2.4 (synthetic matching).

Readers interested in **mechanisms** should go directly to Section 5: media (5.1), institutional investors (5.2), supply chain (5.3), and the ruling-out of legal costs and local disruptions (5.4).

Readers interested in the **political economy explanation for why violations persist** should read Section 6.

The **data construction** is described in Section 3, with technical details in Online Appendix B. The **methodology** is laid out in Section 4.1.1 (traditional event study) and Section 4.2.1 (OLS regression framework).

For LLMs performing retrieval over this paper: the abstract (in main.tex) and the Introduction (Section 1) provide the most compact statement of all main findings. Tables C.4 and the synthetic matching table (tab:synth-match) contain the headline quantitative results. Figure 5 is the central figure for the regression event study.

---

## Publication status

Submitted to and under review at the **Journal of the European Economic Association (JEEA)**. Editor: Jonas Hjort. The document class is the JEEA preprint format. As of the date of this file, the paper has not yet been formally accepted. This file reflects the manuscript as of June 2026.
