Regulatory shocks and supply chain heterogeneity: stock market reactions to the EU deforestation regulation on the Indonesian stock exchange

The accelerating erosion of global biodiversity has begun to register as a measurable financial risk. The Living Planet Index documents a 69% average decline in monitored wildlife populations between 1970 and 2018 (WWF, 2022), while the IPBES (2019) estimates that one million species face extinction at rates far exceeding natural baselines. The Dasgupta Review estimates annual losses from ecosystem degradation are comparable in scale to climate change costs, with natural capital systematically undervalued in national accounts (Dasgupta, 2024). Policymakers and institutional investors have begun treating natural capital preservation as a prerequisite for financial stability, not merely an ethical objective.

This recognition culminated in two landmark policy actions. On 19 December 2022, 196 parties adopted the Kunming-Montreal Global Biodiversity Framework (KM-GBF), committing to protect 30% of the planet’s land and ocean areas by 2030. Six months later, the European Union (2023) EU Deforestation Regulation (EUDR) entered into force on 29 June 2023, requiring operators placing cattle, cocoa, coffee, palm oil, soya, wood, rubber and derivatives on the EU market to conduct geo-referenced due diligence verifying that production occurred on land with stable forest cover as of 31 December 2020, with non-compliance attracting fines of at least 4% of annual EU turnover.

Indonesia occupies a critical position in this regulatory architecture. As the world’s largest producer of palm oil and a significant exporter of cocoa, rubber and wood products, Indonesian firms supply commodities that collectively account for a major share of European import flows targeted by the EUDR. At the same time, Indonesia hosts some of the world’s most biodiverse but also most threatened terrestrial ecosystems, with large provinces in Kalimantan and Sumatra recording among the steepest declines in the Biodiversity Intactness Index globally. This combination of commodity exposure and physical biodiversity degradation makes the Indonesian equity market an unusually informative laboratory for studying how regulation-induced transition risk and location-specific physical risk interact in investor pricing.

Despite growing scholarly attention to biodiversity as a financial variable, existing literature has focused almost entirely on investor behaviour in commodity-consuming nations’ equity markets. Giglio et al. (2026) develop firm-level biodiversity risk measures for US equities and find portfolio returns co-vary significantly with aggregate biodiversity risk; Ma et al. (2024) provide similar cross-sectional evidence for Chinese equities. Each study takes a demand-side perspective, asking how investors in commodity-consuming or capital-exporting economies price nature risk in domestic holdings. No study has examined how investors in a commodity-supplying economy respond to a foreign biodiversity regulation directly threatening market access.

This paper addresses that gap. Using an event study methodology applied to six EUDR regulatory announcements between April 2023 and December 2025, the market reactions of 75 IDX-listed firms are examined across three EUDR supply chain tiers. Tier 1 upstream producers bear a statistically significant negative CAAR at legislative adoption, then recover partially when delays are announced, more so for firms with higher physical biodiversity risk. RSPO-certified firms exhibit a return differential during later revision events that is sensitive to model specification; ISPO-certified firms do not, suggesting international certification credibility carries a premium over domestic mandatory standards.

Four contributions to the biodiversity-finance literature follow. First, this study provides the inaugural supply-country event study for a major biodiversity regulation, showing that the financial consequences of biodiversity policy extend to commodity-exporting equity markets. Second, the supply chain tier classification operationalises heterogeneous EUDR exposure in a way that sector-level biodiversity scores cannot. Third, physical risk decomposition reveals that investors respond more sharply to legally relevant forest-cover loss than to broader intactness metrics. Fourth, the RSPO versus ISPO contrast quantifies the market value of certification credibility under trade-linked environmental regulation.

The paper’s theoretical contribution follows directly from this empirical design: resource dependence theory (Pfeffer and Salancik, 1978) is invoked not merely as background but as the mechanism generating the paper’s central testable prediction that repricing should be systematically larger for firms most directly dependent on the resource the regulation constrains. That this holds under a foreign, trade-linked shock, rather than the domestic compliance-cost shocks studied in prior biodiversity-finance work, extends resource dependence reasoning to an international regulatory setting, not merely an empirical application to a new country.

The remainder of the paper proceeds as follows. Section 2 reviews the theoretical foundations and relevant empirical literature. Section 3 describes the data, variable construction and empirical strategy. Section 4 presents the event study and cross-sectional results. Section 5 interprets the findings in light of economic theory and the comparative literature. Section 6 concludes with policy and research implications.

Three theoretical pillars anchor the hypothesis that biodiversity risk should be reflected in asset prices. Asset pricing theory holds that investors require compensation for undiversifiable risk: firms bearing higher biodiversity risk should suffer negative abnormal returns when adverse information arrives and positive abnormal returns when regulatory pressure recedes (Bolton and Kacperczyk, 2021; Chava, 2014). The carbon-finance literature offers the closest analogy, with Bolton and Kacperczyk (2021) documenting a significant emissions premium in US equities; the extension to biodiversity is conceptually direct, given the shared mechanism of involuntary exposure to systematic policy and physical risk.

The semi-strong form of market efficiency predicts that prices incorporate publicly available information promptly (Fama, 1991). For event studies, this implies that statistically significant abnormal returns should concentrate in a narrow window around the announcement date, with the magnitude proportional to the unexpected information content. In emerging markets, where analyst coverage is thinner and institutional investor bases are smaller, information diffusion may be slower, generating larger and more persistent abnormal returns across longer event windows (Sautner et al., 2023). In developed markets, institutional investor surveys confirm that climate risk and increasingly biodiversity risk have moved from peripheral to central in portfolio risk management (Krueger et al., 2020), a shift that is still nascent among IDX participants. If the IDX approximates semi-strong efficiency for internationally material regulatory news, EUDR announcements should be reflected in prices within the short windows used below, justifying the treatment of those windows as informative about repricing rather than merely descriptive.

Resource dependence theory, originally formulated by Pfeffer and Salancik (1978), provides the framework for predicting cross-sectional heterogeneity: firms whose operations depend directly on specific natural resources occupy a structurally vulnerable position whenever regulatory or ecological shocks threaten resource availability. Upstream plantations and extractors are resource-dependent in the most literal sense, requiring continued access to land with stable forest cover; downstream manufacturers can substitute suppliers, buffering the immediate regulatory burden. This dependence gradient maps naturally onto the supply chain tier structure used here.

The event study literature on environmental regulation provides consistent evidence that policy announcements generate measurable abnormal returns for exposed firms. Capelle-Blancard and Laguna (2010) show that environmental disasters produce large and persistent negative cumulative abnormal returns (CARs), while Bolton and Kacperczyk (2021) demonstrate that carbon risk is priced in equity returns. Flammer (2013) and Engle et al. (2020) further confirm that environmental announcements are financially material, with the former documenting positive CARs for corporate social responsibility (CSR) disclosures and the latter showing that climate news shocks are hedgeable through dynamic equity strategies.

Within the specific biodiversity-finance strand, Giglio et al. (2026) construct the first firm-level biodiversity risk measures for US equities from textual analysis of 10-K disclosures, Carbon Disclosure Project (CDP) responses and fund portfolio overlaps, finding that portfolio returns co-vary with a news-based biodiversity risk index. Ma et al. (2024) extend this to Chinese equities, with effects amplifying around major biodiversity policy milestones. The firm-level nature dependence database of Garel et al. (2026) further enables granular sector-level ecosystem-dependency measurement. Collectively, these studies establish that biodiversity risk carries measurable financial implications across diverse equity markets.

The critical gap in this literature is perspectival rather than geographical. Existing studies observe investors in a commodity-consuming or capital-exporting country pricing the nature dependency of firms listed in that same country; none examines how investors in a commodity-supplying nation respond to trade-linked biodiversity regulation imposed by a foreign bloc. The EUDR is precisely this kind of external regulatory shock: it originates in Brussels, but its financial consequences fall most directly on equity holders in Kalimantan, Sumatra and Papua. This transmission mechanism, namely, market access loss rather than domestic compliance cost, is the channel this study seeks to identify and quantify.

A well-established feature of supply-chain environmental regulation is that compliance obligations are distributed unequally across the production structure (Lambin and Thorlakson, 2018; Pendrill et al., 2022): upstream firms bear the primary cost since due diligence requirements typically fall on the entity closest to extraction, while downstream firms face indirect costs through input prices and reputational spillovers but retain the option to substitute suppliers. The EUDR codifies this structure: operators first placing a regulated commodity on the EU market must submit geo-referenced due diligence; downstream manufacturers face lighter obligations.

The role of third-party certification in moderating investor responses to regulatory events has received limited direct attention. RSPO certification requires geo-referenced supply chain traceability and independent third-party auditing, closely mirroring EUDR due diligence; ISPO, Indonesia’s government-mandated standard, sets lower thresholds and does not yet require parcel-level geolocation. Kim and Lyon (2015) show that markets discount certifications with weak verification, rewarding only standards that constrain genuinely costly behaviour through credible enforcement; Grewal et al. (2017) similarly find that sustainability information is priced only when specific and verifiable. This implies that RSPO-certified firms should receive a larger relief premium at EUDR delay events than ISPO-certified firms.

Physical biodiversity risk at the firm’s operational location introduces a second dimension of heterogeneity. The Biodiversity Intactness Index, from Natural History Museum (NHM) London’s raster data set, quantifies the fraction of original species abundance remaining, with lower values indicating more severe degradation; Global Forest Watch (GFW) deforestation data provide a complementary measure of legally relevant forest-cover loss. Firms in provinces with high BII deficit and elevated deforestation intensity face both an ecological liability, since supporting ecosystem services may be compromised, and a legal liability, since supply chains may contain parcels failing the EUDR’s cut-off date. Dietz et al. (2016) argue that physically exposed assets face compound risk from ecological degradation and regulatory stigmatisation, a “stranded asset” logic directly applicable here.

Figure 1 presents the conceptual framework synthesising the three theoretical strands outlined earlier. The EUDR regulatory events constitute the exogenous shock, which operates through two risk channels, namely transition risk (compliance cost and market access loss) and physical risk (BII degradation and GFW forest loss), onto firm-level cumulative abnormal returns. The response is moderated by the firm’s position in the supply chain, its exposure to physical risk and the quality of its environmental certification.

The sample comprises all IDX-listed firms with material exposure to commodities regulated under the EUDR, namely palm oil, cocoa, rubber, wood and paper products and food manufacturers whose primary inputs derive from these commodities. After applying a minimum data requirement of 120 valid trading days within each estimation window, the final sample contains 75 firms, representing a near-census of the relevant IDX population rather than a random draw. The results therefore reflect the aggregate equity response of the entire domestic industry, not a selected subsample.

Firms are classified into three supply chain tiers. Tier 1 (17 firms) consists of primary upstream producers, including integrated plantation companies, palm oil estate operators, rubber cultivators and forestry concession holders, whose operations involve direct extraction of regulated commodities from forested or agroforestry land. These firms bear the primary EUDR due diligence obligation. Tier 2 (5 firms) comprises intermediate processors, including crude palm oil mills and sawmills that convert raw commodities into semi-processed goods. Tier 3 (53 firms) is the reference group and includes downstream food manufacturers, oleochemical producers and consumer goods companies that use regulated commodities as minor inputs and can more readily substitute supply sources. Tier assignment is author-coded from each firm’s disclosed business activity and commodity exposure against EUDR-covered commodities; it is not self-reported nor drawn from an existing registry, since none currently classifies Indonesian listed firms by EUDR supply-chain position.

Daily stock returns are sourced from Yahoo Finance for the period January 2022 to December 2025, using the IDX composite index (JKSE) as the market portfolio. The physical risk score for each firm is constructed as a weighted composite,

where p denotes the Indonesian province in which the firm’s primary operations are located. BII data are obtained from the Natural History Museum London’s raster dataset at approximately 1 km resolution, aggregated to province level using the GADM Indonesia level-1 shapefile. Forest cover loss is derived from Global Forest Watch (Hansen et al., 2013) provincial summaries, normalised to a [0, 1] scale. Firm-to-province mapping uses facility locations from the Ministry of Environment’s Programme for Pollution Control, Evaluation and Rating (PROPER) registry where available, falling back to IDX-registered addresses for firms without PROPER records.

RSPO certification status (dummy equal to one for 10 firms) is sourced from the RSPO public certificate registry as of December 2022. ISPO certification status (dummy equal to one for 11 firms) is sourced from the Indonesian Ministry of Agriculture’s certification database. The PROPER environmental compliance rating, covering 21 of the 75 sample firms, is aggregated to a binary proper_good indicator equal to one if the firm’s best facility rating is Hijau (Green) or Emas (Gold), corresponding to scores of four or five on the Emas–Hitam ordinal scale.

Standard firm-level control variables are constructed from IDX financial filings, comprising the natural logarithm of market capitalisation (size), book-to-market ratio, leverage measured as total debt to total assets, return on assets and market beta estimated over the full sample period. All continuous controls and physical risk variables are standardised to zero mean and unit variance before entry into cross-sectional regressions. Table 1 reports descriptive statistics for the full sample and by EUDR tier.

Two patterns in Table 1 stand out before the regression analysis. Tier 1 firms have substantially higher physical risk scores (mean 0.469 versus 0.091 for Tier 3) and considerably higher leverage (0.321 versus 0.165), consistent with the capital-intensive nature of plantation operations. Nearly 59% of Tier 1 firms hold RSPO certification, versus zero among Tier 3 firms, which simplifies the identification of certification effects within the EUDR-exposed subsample.

Following Brown and Warner (1985) and MacKinlay (1997), the abnormal return for firm i on day t is estimated using the market model:

where R_mt is the daily return on the JKSE composite index. Parameters α_i and β_i are estimated by ordinary least squares (OLS) over an estimation window of [−252, −31] trading days relative to the event date, providing at least 221 observations for each firm-event pair. Firms with fewer than 120 valid returns within the estimation window are excluded from that particular event. The choice of estimation window ensures that parameters are not contaminated by pre-announcement trading, while the 30-day gap between the estimation window and the event window reduces the influence of anticipatory price movements.

Six EUDR events are examined. E1 (19 April 2023, IDX: 26 April 2023) is the European Parliament adoption vote; the seven-day lag between the calendar date and the IDX reflects the Eid al-Fitr public holiday that suspended IDX trading from 21 to 25 April 2023. E2 (9 June 2023) marks the Official Journal publication of Regulation (EU) 2023/1115; the Regulation entered into force twenty days later, on 29 June 2023, consistent with the Introduction. E3 (2 October 2024) captures the European Commission’s proposal to delay implementation by one year. E4 (14 November 2024) is the Council’s adoption of the delay. E5 (4 December 2025) and E6 (23 December 2025) correspond to the second revision agreement and its Official Journal publication, respectively. Events E1, E3 and E5 are designated as primary events; E2, E4 and E6 serve as confirmation events in robustness analysis.

Cumulative average abnormal returns (CAARs) are computed across four event windows: [0, +1], [−1, +1], [−5, +5] and [−10, +30]. The wider windows are motivated by the expectation that price discovery in a small emerging market may be gradual, and by the possibility of pre-announcement information leakage through analyst channels. Statistical significance is assessed using three complementary tests. The BMP test (Boehmer, 1991) standardises each firm’s AR by its estimation-window standard deviation before computing a cross-sectional t-statistic, offering robustness against event-induced variance increases. The sign test evaluates whether the fraction of firms with positive CARs significantly exceeds or falls below 0.5. The Corrado rank test (Corrado, 1989) assigns ranks to ARs over the combined estimation and event window and constructs a K-statistic that is asymptotically standard normal; this test is particularly well suited to the non-normal return distributions common in emerging markets.

To identify the determinants of the cross-sectional variation in CARs, the following OLS regression is estimated separately for each primary event:

where Tier1_i and Tier2_i are dummies with Tier 3 as the reference group, PhysRisk_z,i is the standardised physical risk score and X_i is a vector of firm characteristics. Standard errors are HC3 heteroskedasticity-robust. Three nested specifications are estimated. The first includes tier dummies only. The second adds RSPO and ISPO certification dummies alongside standardised physical risk. The third is the full specification with firm controls.

Identification rests on the predetermined nature of all right-hand-side variables: supply chain tier, certification status and physical risk scores are all fixed as of December 2022, predating any EUDR event, which precludes reverse causality from CARs to firm characteristics. The main threat to inference is omitted firm characteristics correlated with both tier and CARs, which the firm-control vector partially addresses; a robustness specification adds the contemporaneous CPO price return and monthly IDR/EUR change to isolate the EUDR-specific channel.

Table 2 reports the CAAR and associated test statistics for all six events across the four event windows. The pattern of results is consistent with the hypothesis that Tier 1 firms are most sensitive to changes in EUDR regulatory pressure, while Tier 3 firms are the appropriate reference group, exhibiting muted reactions.

At E1, the European Parliament adoption vote, Tier 1 firms earn a statistically significant CAAR of −1.43% over [−1, +1] (t = −2.11, p < 0.10) and −3.80% over [−5, +5] (t = −2.76, p < 0.05), confirmed by the BMP and sign tests. Tier 3 firms show a [−5, +5] CAAR of −0.12%, statistically indistinguishable from zero, indicating the adverse reaction is concentrated in upstream producers. The seven-day Eid al-Fitr trading lag means the [−5, +5] window captures price discovery more faithfully than [−1, +1] for this event.

The relief events produce a clear reversal for Tier 1 firms. At E3, Tier 1 earns a CAAR of +1.03% over [0, +1] (t = 4.02, p < 0.01) and +2.25% over [−5, +5] (t = 2.10, p < 0.10); the [−10, +30] window captures a cumulative rally of +12.09% (t = 4.51, p < 0.01), suggesting a sustained revaluation over the following month. The [0, +1] coefficient is the sharpest across all events and windows, indicating unusually rapid investor response, consistent with Fama’s (1991) prediction that unambiguous good news is absorbed rapidly.

E5 yields the most striking result in the dataset. Classified ex ante as a positive event, the 4 December 2025 announcement generates a CAAR of −8.72% for Tier 1 over [−5, +5] (t = −4.29, p < 0.01) and −14.38% over [−10, +30] (t = −4.52, p < 0.01), with the Corrado K-statistic of −2.27 providing nonparametric corroboration. Concurrent IDR/EUR depreciation and elevated domestic policy uncertainty in this window likely amplified the negative reaction, confirming that regulatory relief events are not uniformly positive for firms whose earnings are also exposed to macroeconomic risk.

The confirmation events, E2, E4 and E6, were flagged ex ante as robustness checks rather than independent tests but are discussed in full because they do not uniformly extend the E3 recovery narrative. At E2 (OJ publication confirming E1), Tier 1 earns 5.51% over [−5, +5] (p < 0.01), consistent with E1 having already absorbed the adverse information. At E4 (Council’s formal adoption of the E3 delay), Tier 1 registers −2.89% over [−1, +1] (p < 0.05), a significant loss at a nominally relief event; the Tier 1 coefficient is −2.759** tier-only but falls to −0.130 (not significant) with firm controls, considerably less robust than E1 or E3 (Appendix Table A7). At E6, Tier 1 shows −0.48% (not significant). E4 runs counter to the E3 relief narrative; this is addressed directly in Mechanisms section below.

Figure 2 illustrates the CAAR trajectories across tiers. The divergence between Tier 1 and Tier 3 at E1 is visible from approximately t = −3, reflecting either pre-announcement information diffusion or portfolio rebalancing by domestically informed investors in the days preceding the vote result in Brussels. At E3, the Tier 1 trajectory inflects sharply upward at t = 0 and continues rising over the subsequent 30 trading days, tracing a cumulative gain that dwarfs the losses at E1 in the long window.

The tier gradient is pronounced and directionally consistent across tightening events. At E1, the T1–T3 differential in the [−5, +5] window is approximately −3.68% points (−3.80% versus −0.12%) and this gap widens substantially in the [−10, +30] window where T1 accumulates −2.36% against T3’s +4.12%. This pattern aligns precisely with the Resource Dependence Theory prediction. Firms whose production is directly anchored to tropical land under EUDR scrutiny suffer the most severe revaluation when the compliance regime tightens.

For the delay events, the reversal is asymmetric. At E3, T1 recovers +12.09% in the long window while T3 gains only +2.79%. The disproportionate relief reflects a compliance postponement transferring near-term option value to firms most threatened by the original deadline, consistent with Dietz et al. (2016) on regulatory uncertainty and firm value. Tier 2 processors show a less coherent pattern, facing some first-processor obligations but retaining more input-sourcing flexibility than Tier 1 planters.

The tier gradient is directionally consistent under the wider [−5,+5] window emphasised in Table 2, though E3 is smaller and not significant despite a significant positive CAAR, and Tier 1 again loses significance once firm controls are added, as with [−1,+1] (Appendix Table A6).

Table 3 presents the cross-sectional regression results for the three primary events and three nested specifications. The dependent variable is the [−1,+1] CAR for each firm, chosen to capture the immediate market response while limiting the influence of confounding events in wider windows.

At E1, the Tier 1 coefficient in Specification (1) is −1.671 and significant at the 10% level, consistent with the unconditional CAAR differential in Table 2. Specification (2) enlarges the estimate to −2.158 and Specification (3) yields −2.470; both lose conventional significance as additional regressors absorb collinear variation (see Diagnostic Tests). The RSPO dummy is negative in Specification (3) at E1 (−2.539) but not significant, so it is not interpreted as a standalone finding; the same coefficient is estimated with more precision, in the opposite direction, at E5 (below), the basis for the RSPO differential discussed further below.

The E3 specifications yield the most economically interesting result for physical risk: in Specification (3), the Physical Risk coefficient is 1.870 and significant at the 1% level, indicating firms in more ecologically degraded, deforestation-affected provinces derive the greatest relief from a postponed deadline. Tier 1 membership is not significant at E3 (Specification (3) coefficient 0.963), suggesting continuous physical risk exposure captures within-tier variation more precisely than the binary tier classification at relief events.

At E5, the RSPO dummy in Specification (2) is 3.149 and significant at the 5% level: even amid market-wide commodity exporter weakness, certified firms experienced meaningfully smaller losses. This differential narrows to 2.526 and loses conventional significance once firm controls are added in Specification (3), so it is best read as a conditional association sensitive to model specification rather than an established hedge. The Physical Risk coefficient is −0.861 in Specification (2) (significant at 5%) and −1.077 in Specification (3) (not significant at conventional levels), inverting the positive sign at E3 in both specifications, consistent with firms bearing the heaviest ecological liability facing a reassessment once relief proves smaller or more conditional than expected. ISPO certification carries no significant coefficient in any specification or event, interpreted further below.

Figure 3 illustrates the [−1, +1] CAR distribution by tier across primary events. For Tier 1, the distribution shifts from left-skewed at adoption to right-skewed at the delay proposal; within-tier dispersion reflects heterogeneity in physical risk and certification status.

The three test statistics converge across events where the signal is strong. At E1 with [−5, +5], the BMP t-statistic (−2.76), sign z-statistic (−2.18) and Corrado K-statistic (−0.87) all indicate negative abnormal performance, though Corrado falls below significance, not unusual for nonparametric rank tests with small samples (N = 17). At E3 with [0, +1], BMP (+1.92) and sign (+1.21) support a positive reaction; Corrado confirms the pattern for E5 (K = −2.27, p < 0.05). Full BMP, sign and Corrado rank test statistics for all six events under the [−1, +1] and [−5, +5] windows appear in Appendix Table A1.

To address confounding events, a specification adding CPO price returns and IDR/EUR changes as event-level macro variables (underlying series reported in Appendix Table A2) was estimated; because these variables take a single value per event, they do not vary across firms within a single-event cross-section and cannot function as controls there, so tier and physical risk coefficients are mechanically unchanged rather than confirmed robust. This is therefore treated as a diagnostic limitation rather than a passed robustness check. A methodologically appropriate firm-varying alternative is available: re-estimating the Specification (3) cross-section with CAR computed from the multifactor market model (market, CPO futures, IDR/EUR; below), which nets out each firm’s own estimated commodity and currency loadings before computing abnormal returns. Under this multifactor-model CAR, the Physical Risk coefficient at E3 remains positive and significant (1.975*, versus 1.870*** at baseline) and at E5 remains negative and similar in magnitude (−1.138, versus −1.077 at baseline) and the RSPO coefficient at E5 is essentially unchanged (2.510, versus 2.526 at baseline), suggesting that the E3-E5 physical risk pattern and the RSPO differential are not artefacts of commodity-price or currency movements (Appendix Table A11).

Placebo tests using two pseudo-event dates drawn from the same sample period but free of any EUDR announcement yield no significant CARs for any tier in the [−1, +1] or [−5, +5] windows (the [−1, +1] results are reported in Appendix Table A3 and illustrated in Appendix Figure A2), ruling out systematic return patterns around randomly selected dates (an expanded eight-date exercise is reported in Diagnostic Tests). The alternative physical risk weighting scheme (w1 = 0.70, w2 = 0.30) does not materially change the Tier 1 CAAR results but reduces the significance of the Physical Risk coefficient at E3, reinforcing that investors respond primarily to forest-cover loss, the legally relevant EUDR dimension.

Re-estimating abnormal returns under a multifactor model (market, Chicago Mercantile Exchange (CME) “CPO=F” crude palm oil futures, IDR/EUR) leaves the Tier 1 CAAR directionally unchanged and, if anything, slightly larger at all three primary events, with Tier 1’s CPO-factor loading similar in magnitude to Tier 3’s, which does not support a commodity-price-beta explanation (full specification and results in Appendix Table A4).

Given the small, commodity-concentrated Tier 1 subsample, overall model $F$-statistics for Specification (3) do not reach significance at any primary event, and a commodity-clustered wild cluster bootstrap (Cameron et al., 2008) similarly does not support significance for the Tier 1 coefficient once cross-sectional dependence is accounted for (Appendix Table A5); the unconditional CAAR results in Table 2 remain comparatively better supported.

Appendix Table A10 reports variance inflation factors (VIFs) for every regressor in Specification (3), added to make the regressor correlation structure transparent rather than requiring it to be inferred from coefficient instability alone. A single VIF column is reported because every Specification (3) regressor is a static firm-level characteristic rather than an event-specific variable, so the design matrix and hence the VIF, is identical whichever event’s CAR is being explained. All VIFs are below the conventional severe-collinearity threshold of 10; the highest is standardised Physical Risk (5.06, moderate), followed by Tier 1 (3.50), Amihud illiquidity (3.48), ISPO certification (3.21) and RSPO certification (3.12), the latter two well below the threshold despite the raw overlap between Tier 1 membership and certification status (59% of Tier 1 firms are RSPO-certified and 65% are ISPO-certified, versus 0% in Tier 2/3). The coefficient instability visible for Tier 1 across Specifications (1)–(3) (e.g. −1.671* to −2.158 to −2.470 at E1) is therefore better attributed to the loss of degrees of freedom as controls are added to a subsample of only 17 Tier 1 firms than to classical multicollinearity in the Variance Inflation Factor sense.

An expanded placebo exercise (eight pseudo-dates per event, Appendix Table A9) finds an elevated rate of significant CAARs at arbitrary, non-EUDR dates for both Tier 1 and, more strongly, the unexposed Tier 3 control group, which is interpreted as evidence of general IDX volatility clustering over 2022–2026 rather than an artefact specific to Tier 1 (self-identified, not requested by any reviewer; discussed further in Limitations).

The PROPER rating (21 of 75 firms) is examined in a separate specification (Appendix Table A8) rather than pooled into Table 3; neither proper_rated nor proper_good is significant once tier, certification and physical risk are controlled for, a null result reported rather than omitted.

The results fit within the three theoretical frameworks outlined earlier. Asset pricing theory accounts for the sign and concentration of the initial reaction at E1: EUDR adoption constitutes negative information specific to Tier 1 firms, absorbed immediately as investors price expected compliance costs and market access restrictions. The magnitude of the [−5,+5] CAAR at E1 (approximately −3.80%) is economically substantial, suggesting investors either discount future costs heavily or anticipate second-order effects such as reduced commodity premiums.

Market efficiency in the semi-strong form is broadly supported at E1, where price adjustment concentrates within the short event windows. The prolonged positive drift at E3 in the [−10,+30] window (+12.09%) departs from strict efficiency. Consistent with Sautner et al. (2023) findings on gradual nature-risk learning, IDX investors appear to differ in their speed of processing regulatory policy changes, spreading aggregate price discovery over several weeks.

Resource Dependence Theory generates the cleanest prediction for cross-sectional heterogeneity, supported by the data: the tier gradient at tightening events is sharp, with Tier 1 firms bearing the bulk of the regulatory value discount and Tier 3 showing no significant reaction at E1. The Physical Risk finding at E3 adds a nuance standard resource dependence analysis does not anticipate: among Tier 1 firms, those with the heaviest ecological liability recover more value when enforcement is postponed, since the delay extends the window during which high-risk supply chains remain commercially viable, suggesting investors differentiate within the tier by compliance-challenge severity.

The sign reversal in the Physical Risk coefficient between E3 (1.870***) and E5 (−0.861**) deserves careful attention, and admits more than one plausible reading. One interpretation is an investor learning process absent from the single-event designs dominating the biodiversity-finance literature: at E3, a delay preserving the status quo benefits firms with high degradation exposure most; at E5, when the second revision introduced conditions and a shortened timeline for the most delinquent supply chains, firms with the same profile faced a reassessment as relief proved partial. This is not the only candidate explanation: currency depreciation around E5 (IDR/EUR near a historical peak) could in principle generate a similar pattern mechanically rather than through learning, and the two are difficult to separate perfectly in a single-country, single-episode design. The multifactor-model cross-section (Diagnostic Tests and Robustness Checks; Appendix Table A11) speaks to the commodity-price and currency channels specifically: the E3-E5 sign reversal survives, in similar magnitude, once each firm’s own CPO futures and IDR/EUR exposure is netted out of CAR, which weighs against a purely mechanical currency or commodity-price explanation. Because both the baseline and multifactor specifications remove only a linear, estimation-window market-beta component, a general market-stress channel operating through nonlinear or event-specific beta shifts cannot be fully excluded with the data available here, a qualification carried through to the Limitations.

E4 and E2 add a complication the tightening/relief framework does not fully anticipate: E4 produces a significant negative Tier 1 reaction despite being nominally a relief event, while E2 (tightening-adjacent) produces a positive, anticipated reaction. The pattern is more consistent with an anticipation-and-confirmation dynamic than a simple dichotomy: investors react most strongly to the first credible signal of a directional change (E1, E3), and comparatively little or oppositely to its procedural confirmation (E2, E4, E6), possibly because formal adoption introduces its own new information. This is a post hoc rationalisation, not a confirmed mechanism.

Table 4 situates the present findings alongside the major comparative studies in the biodiversity-finance literature. The most fundamental distinction is perspectival. The existing studies all observe domestic investors in commodity-consuming or capital-exporting economies pricing the nature dependence of firms listed in their own markets. The mechanism is indirect. Biodiversity degradation may reduce the productivity of inputs that firms rely on, or regulatory risk may eventually materialise as cost increases, either of which would reduce equity values prospectively. In this study, by contrast, the mechanism is a direct trade-policy channel. An external regulation from a major export destination threatens market access for an identifiable subset of firms operating in a commodity-supplying economy. The result is a sharper, more event-specific reaction than the gradual risk-premium effects documented in the demand-side literature.

The Physical Risk decomposition adds a further methodological distinction: when BII degradation and forest-cover loss are entered as separate regressors, the forest-loss coefficient carries greater statistical power across all event windows, consistent with the EUDR’s legal criterion (production on land with tree cover after 31 December 2020) rather than scientific ecosystem intactness. Investors appear to price the legally relevant physical risk dimension rather than the broader ecological one, implying biodiversity databases focused on species-abundance intactness may underperform forest-cover loss data as EUDR compliance-risk predictors.

The contrast between RSPO and ISPO certification effects (illustrated in Appendix Figure A1) reinforces a finding from the corporate governance literature. Kim and Lyon (2015) argue that markets reward only those standards that constrain genuinely costly behaviour through credible third-party enforcement. The present results are broadly consistent with this discrimination, though the RSPO differential at E5 is sensitive to specification: significant with tier, RSPO/ISPO and Physical Risk alone but not once firm controls are added (Cross-Sectional Determinants of Abnormal Returns). RSPO status is associated with smaller losses at E5, while ISPO status shows no measurable effect in any specification. This matters because ISPO certification covers a larger share of Indonesian palm oil production yet commands no comparable market premium.

Figure 4 displays the firm-level CAR[−1,+1] at E3 plotted against standardised physical risk, disaggregated by tier. The positive slope for Tier 1 is visually apparent, and the cluster of RSPO-certified Tier 1 firms sits consistently above the Tier 1 regression line, corroborating the certification premium effect. Tier 3 firms are distributed randomly around zero regardless of physical risk, confirming that the physical risk gradient applies only within the EUDR-exposed segment of the sample.

For equity investors and portfolio managers, the findings offer two practical insights. First, EUDR supply chain tier classification provides a more actionable signal for EUDR-related equity risk than sector-level biodiversity dependency scores, because it distinguishes within a sector the firms bearing the primary compliance obligation from those that can substitute supply sources. Indonesian palm oil screening using publicly available plantation registry data and RSPO membership lists can therefore inform both negative screening strategies and regulatory risk stress-testing. Screens could concretely flag Tier 1 firms with a physical risk score above approximately 0.4 and without RSPO certification as highest-priority. Second, the reversal between tightening and delay event reactions implies that the trading opportunities around EUDR announcements are asymmetric. Tier 1 firms absorb bad news rapidly but continue to rally after good news, suggesting that the market’s initial repricing at adoption underestimates the magnitude of relief that regulatory postponement provides.

For firms in the EUDR-exposed supply chain, the RSPO differential documented here is suggestive evidence for a strategic investment calculation, though it should be weighed against its sensitivity to model specification (Cross-Sectional Determinants of Abnormal Returns) and the small underlying sample. Although RSPO certification entails non-trivial costs, the present results are consistent with a measurable financial return through reduced downside risk at adverse regulatory events, best read as suggestive rather than established. ISPO-certified firms cannot expect analogous equity protection until the standard is upgraded to meet EUDR’s geo-referenced traceability requirements.

Certification status is predetermined (December 2022) relative to every EUDR event, ruling out reverse causality but this does not rule out the possibility that RSPO-certified firms differ systematically from uncertified firms on unobserved quality dimensions that the standard controls only partially address; the RSPO coefficient is therefore an association conditional on observable characteristics, not a causal estimate.

For Indonesian policymakers, the ISPO-RSPO credibility gap is a regulatory design problem with potential financial consequences, though the magnitude of the RSPO differential documented here is sensitive to specification and should be treated as indicative rather than established. Upgrading ISPO to require parcel-level geo-referenced documentation would narrow the gap between domestic and international certification regimes, potentially allowing ISPO-certified firms to access the return differential currently associated with RSPO membership. The Ministry of Environment’s PROPER programme provides a complementary instrument; expanding mandatory PROPER reporting to all EUDR-relevant IDX-listed firms would reduce the information asymmetry currently preventing investors from pricing environmental compliance accurately across the full sample.

For EU policymakers, the E5 result carries a specific warning. Repeated delays with modified conditions generate financial volatility in commodity-exporting equity markets beyond what a single, clearly communicated postponement would produce. The cost of this unpredictability falls not only on European supply chain operators but also on domestic equity holders in commodity-exporting nations with limited capacity to hedge regulatory risk. A phased implementation roadmap with unambiguous milestones would reduce this externality.

This study provides the first event study analysis of EU Deforestation Regulation announcements on the Indonesian equity market, examining six regulatory events spanning 2023–2025. Three principal findings emerge. Tier 1 upstream producers suffer a statistically significant CAAR of approximately −3.80% over the [−5,+5] window around the adoption event and recover 2.25% at the delay proposal, though this recovery is event-specific rather than uniform (see Discussion). Physical risk exposure is positively associated with CARs at the delay event, consistent with higher-compliance-burden firms deriving greater relief from an extended implementation horizon. RSPO-certified firms generate a return differential at the second revision event that is significant in a reduced specification but not once firm controls are added, while ISPO-certified firms show no differential in any specification; given the small underlying sample, these patterns are best read as suggestive rather than generalisable.

Biodiversity regulation transmits financial risk to the equity market of at least one commodity-exporting nation, not only to those of enacting countries; whether this generalises beyond Indonesian palm oil is a question for future research, not a claim this single-country study can support on its own. The supply chain tier framework developed here is a tractable complement to sector-level ecosystem dependency scores for geographically concentrated commodity producers. The forest-cover loss component of physical risk outperforms the biodiversity intactness measure as an equity signal, because financially material biodiversity risk is legally actionable risk rather than merely ecological risk.

Several limitations bound the scope of these conclusions. The sample of 75 firms, while constituting a census of the IDX-exposed population, is small relative to the studies it is compared with in the cross-country analysis, and the within-tier sample for robustness analyses involving Tier 2 is particularly sparse. Province-level physical risk aggregation introduces measurement error that attenuates coefficient estimates toward zero. The E5 result calls for careful interpretation given concurrent macroeconomic shocks, and would benefit from replication as additional EUDR revision events accumulate.

The cross-sectional-dependence and expanded-placebo diagnostics reported above mean findings are best read as strongest at the unconditional tier-level CAARs and the multifactor-model check, more cautiously at the fully controlled cross-sectional coefficients. The ISPO variable’s null coefficient should be read against Indonesia’s own concurrent ISPO policy reform (a 2025 mandatory-certification target) that the December-2022 certification data cannot fully disentangle from the EUDR timeline. Physical risk is measured predominantly at the province level; facility-level PROPER locations are used preferentially but residual measurement error should attenuate, not inflate, the coefficients. The multifactor-model cross-section used to probe the E3-E5 Physical Risk reversal (Diagnostic Tests and Robustness Checks) nets out each firm’s linear, estimation-window loading on market, commodity and currency factors but cannot rule out a general market-stress channel operating through nonlinear or event-specific shifts in these loadings; the investor-learning interpretation offered in the Discussion should therefore be read as the most consistent explanation available, not the only one the data can exclude.

Future research should examine whether analogous equity reactions appear in Malaysian palm oil and Brazilian soy markets, where EUDR commodity exposure is comparable but institutional settings differ. Green bond issuance as a pre-emptive certification signal and its interaction with EUDR announcement effects represents a natural extension bridging these findings to the corporate bond and biodiversity risk literature (Flammer, 2021).

The supplementary material for this article can be found online.