Study on the Impact of Regional Trade Agreements on the Ternary Margins of Wood Forest Product Exports

Abstract

Based on global bilateral six-digit HS code trade data of wood forest products from 2000 to 2022, we constructed a three-dimensional panel dataset combining year, importing country, and exporting country, and used a staggered difference-in-differences (DID) model to systematically examine the impact of regional trade agreements (RTAs) on the export of wood forest products and the mechanisms through which they operate. The results show that the signing of RTAs exerts a significant promoting effect on the export market share, extensive margin, and quantity margin of wood forest products, while exerting a significant negative effect on the price margin. Heterogeneity analysis indicates that the impact of signing RTAs on the ternary margin of wood forest product exports varies significantly with product characteristics, bilateral distance, and the type of trade agreement. Mechanism analysis shows that RTAs produce these effects by reducing bilateral trade costs and strengthening bilateral social ties. Accordingly, this paper proposes policy recommendations to strengthen the RTA network, optimize the bilateral trade environment, and enhance the depth of environmental provisions in RTAs to promote the green and high-quality development of wood forest products.

1. Introduction

Regional Trade Agreements (RTAs) have become a prominent feature of the global trading system, particularly in the context of stagnation in multilateral trade negotiations. By reducing trade barriers and enhancing institutional cooperation, RTAs play an increasingly important role in shaping international trade patterns and production networks [1,2]. At the same time, concerns have been raised regarding how such agreements influence resource-based and environmentally sensitive sectors [3,4].

Wood forest products constitute a key category of traded goods, encompassing a wide spectrum ranging from raw timber and sawnwood to processed products such as panels, pulp, paper, and furniture. Beyond their economic importance in supporting employment and industrial development, forest products are closely linked to environmental sustainability, carbon sequestration, and climate-change mitigation [5]. Understanding how trade policies affect the structure and quality of forest product exports is therefore of particular relevance for sustainable forest management.

While a large body of literature has examined the aggregate trade effects of RTAs [6,7,8], evidence specific to the forest sector remains limited. Existing studies on forest product trade have primarily focused on factors such as export restrictions, exchange rates, and certification schemes. These studies typically assess trade performance using total export values or volumes, implicitly assuming that trade growth reflects a homogeneous expansion. However, such an approach may mask important structural changes in export composition, especially in a sector characterized by heterogeneous products, varying degrees of processing, and distinct environmental constraints.

This study addresses this gap by examining how RTAs affect forest product exports through different channels. Rather than focusing solely on aggregate trade flows, we decompose export performance into three complementary dimensions: the extensive margin (product diversification), the quantity margin (scale of exports), and the price margin (a proxy for quality upgrading). This decomposition allows us to distinguish whether RTAs primarily promote export growth through volume expansion, diversification into new products, or improvements in product quality.

Using a global panel dataset covering bilateral trade in wood forest products across approximately 200 countries and regions from 2000 to 2022, we empirically assess the impact of RTA participation on these export margins. A multi-period difference-in-differences framework is employed to identify the causal effects of RTAs while accounting for time-invariant country characteristics and global shocks. In addition, we explore the mechanisms through which RTAs operate, focusing on reductions in trade costs and enhancements in bilateral social connectedness, as well as heterogeneity across product types and agreement depth.

This study contributes to the literature in three main ways. First, it provides the first global-scale analysis of how RTAs affect forest product exports through distinct export margins, offering a more nuanced perspective than aggregate trade measures. Second, by explicitly examining mechanisms related to trade costs and social connectedness, the study sheds light on how institutional cooperation under RTAs influences export structure. Third, the analysis highlights heterogeneous effects across product categories and agreement designs, with implications for sustainable trade and forest policy.

The remainder of the paper is organized as follows. Section 2 reviews the related literature and develops the analytical framework. Section 3 describes the data and variables. Section 4 outlines the empirical methodology. Section 5 presents the results, and Section 6 discusses the findings in relation to existing literature. Section 7 concludes with policy implications.

2. Literature Review, Theoretical Analysis and Hypotheses

2.1. Literature Review

The analysis of RTAs originates from Viner’s seminal customs union theory [7], which distinguishes between trade creation and trade diversion. Building on this theoretical foundation, a large empirical literature—especially studies based on structural gravity models—has consistently shown that RTAs significantly promote aggregate bilateral trade flows [8,9].

However, applying these aggregate findings to the forestry sector raises important limitations. Existing studies on wood forest product trade mainly focus on external determinants such as log export bans, exchange rate fluctuations, and forest certification schemes [10,11,12]. While informative, this literature largely treats export growth as a uniform increase in total trade volumes, giving rise to an “aggregate bias”. This limitation is particularly relevant for forest products, which combine characteristics of natural resources and manufactured goods. As a result, analyses based solely on aggregate trade values obscure the underlying sources of export growth.

To address this issue, the heterogeneous firm theory [13] and the ternary margin decomposition framework proposed by Hummels and Klenow [14] provide a more refined analytical perspective by decomposing export growth into extensive, quantity, and price margins. This framework has been increasingly applied in the context of forest product trade to examine export structure and growth patterns [15,16,17]. Nevertheless, studies that explicitly link RTA participation to the ternary margins of forest product exports remain scarce. Consequently, it remains unclear whether RTA-induced export growth reflects an expansion of low-value, quantity-based trade or a shift toward higher value-added products.

Moreover, the existing literature lacks a comprehensive global assessment of RTA effects on forest product exports. Most studies focus on specific countries, regions, or individual agreements, limiting the generalizability of their conclusions. In addition, the mechanisms through which RTAs influence export margins—such as trade cost reduction or institutional and informational linkages—have received limited empirical attention in the forestry context.

To fill these gaps, this study integrates heterogeneous firm theory with a rigorous empirical evaluation of RTAs. Using bilateral HS-6 level trade data on wood forest products from 2000 to 2022, we construct a three-dimensional panel dataset and employ a staggered Difference-in-Differences framework to systematically examine how RTA participation affects export growth through the extensive, quantity, and price margins.

2.2. Theoretical Analysis

The signing of RTAs reduces trade barriers and policy uncertainty, evolving to encompass broader regulatory cooperation [2,4]. Integrating the tri-margin framework with customs union theory, we analyze the mechanisms influencing wood forest product exports.

First, regarding the extensive margin, RTAs reduce bilateral tariffs and entry costs [4]. Consistent with heterogeneous firm theory, this enables previously uncompetitive products to enter the market and induces trade diversion toward member countries. Furthermore, cooperation on non-tariff barriers, such as inspection standards, expands the range of tradable goods. Consequently, RTAs are expected to significantly enhance the extensive margin.

Second, regarding the intensive margin, effects are heterogeneous. The removal of quantitative restrictions directly boosts the quantity margin [1]. The price margin, however, faces opposing forces: while quality upgrading may eventually raise prices, the immediate reduction in tariffs and transaction costs tends to lower landed prices, enhancing competitiveness. Thus, RTAs are expected to reduce the price margin in the short term, with quality adjustments exhibiting a temporal lag.

Accordingly, we posit that RTAs could increase both extensive and quantity margin while exerting downward pressure on the price margin, thereby progressively expanding export market share. Based on this, we propose:

H1. 

Signing an RTA increases a country’s export market share of wood forest products, as well as its extensive and quantity margins.

H2. 

Signing an RTA reduces the export price margin of wood forest products, with this effect manifesting after a time lag.

To further elucidate the black box of RTA effects, this study investigates two specific transmission mechanisms: bilateral variable trade costs and bilateral social connectedness.

The first mechanism is the reduction of economic friction. RTAs directly reduce variable trade costs between member countries through tariff concessions and trade-facilitation measures, which reduce trade costs—defined as the costs of transporting goods across borders and complying with trade regulations. This cost saving effect enhances the price competitiveness of wood forest products and lowers market entry barriers. Theoretically, lower trade costs should simultaneously expand the export market share, extensive margin, and quantity margin, while mechanically reducing the unit value based price margin [8].

The second mechanism operates through the enhancement of soft connectivity. Beyond reducing formal trade barriers, RTAs promote deeper institutional cooperation, which in turn facilitates policy coordination, information exchange, and cross-border social interactions. Through these channels, RTAs are likely to strengthen bilateral social connectedness.

Bilateral social connectedness, as introduced by Bailey et al. [18], measures the relative probability that a Facebook user in one country or region is connected with a user in another location. A higher degree of social connectedness has been shown to enhance market transparency, reduce search and matching costs, facilitate the diffusion of technology and knowledge, and promote the mutual recognition of standards.

For the wood forest product sector—characterized by complex supply chains and stringent technical standards—stronger social connectedness can effectively mitigate information asymmetries between trading partners. By improving buyers’ access to market information and suppliers’ understanding of foreign standards and consumer preferences, enhanced social connectedness can stimulate export expansion along the extensive margin and support quality upgrading reflected in the price margin.

Accordingly, the following additional hypotheses are proposed:

H3. 

RTAs enhance the export market share, extensive margin, and quantity margin of wood forest products while reducing the price margin by lowering bilateral trade costs.

H4. 

RTAs promote export market share, extensive margin, and quantity margin, and reduce the price margin of wood forest products by strengthening bilateral social connectedness and trade cooperation.

3. Data and Variables

3.1. Data Sources and Coverage

This study draws on multiple publicly available datasets to ensure transparency and reproducibility. Bilateral trade data on wood forest products are obtained from the CEPII–BACI database, which could be found at: https://www.cepii.fr (accessed on 22 January 2025), and is openly accessible and widely used in international trade research [19]. BACI harmonizes exporter- and importer-reported trade flows from UN Comtrade by reconciling Free on Board (FOB) and Cost, Insurance, and Freight (CIF) values, thereby providing more consistent measures of trade values and quantities at the HS six-digit level.

The sample covers bilateral trade flows among approximately 200 countries and regions over the period 2000–2022. This broad geographical coverage reflects the near-universal reporting in BACI and allows for a global assessment of RTA effects across economies with heterogeneous development levels and forest resource endowments. The starting year, 2000, captures trade dynamics following major global integration events, most notably China’s accession to the WTO, while the endpoint, 2022, represents the most recent year for which reconciled BACI data are available.

Information on RTAs is drawn from the World Bank Deep Trade Agreements (DTA) Database, which provides detailed information on the timing and content of RTAs in force globally [20].

3.2. Dependent Variables

The dependent variables are the export ternary margins of wood forest products. Following the decomposition framework proposed by Hummels and Klenow [14], total export performance is decomposed into three structural components: the extensive margin ($E{M}_{ijt}$), the quantity margin ($Q{M}_{ijt}$), and the price margin ($P{M}_{ijt}$). This framework is particularly suitable for analyzing heterogeneous products, as it allows export growth to be attributed to product diversification, scale expansion, or quality upgrading.

3.2.1. Data Cleaning and Treatment of Missing Values

Before calculating the trade margins, a rigorous data-cleaning procedure was implemented to ensure transparency and reproducibility. The primary trade data are drawn from the CEPII-BACI database, which provides harmonized bilateral trade flows at the HS six-digit level.

When calculating the ternary trade margins, observations with missing trade information were excluded. As the decomposition of trade margins relies on multiplicative indices (where Total Exports = Extensive Margin × Intensive Margin). In the econometric specification, these indices are transformed into logarithms to estimate elasticities (${lnR}_{ijt}$ = $lnE{M}_{ijt}$ + ${lnIM}_{ijt}$). Observations with missing values cannot be log-transformed. Including them without cleaning would generate indeterminate values for the dependent variables.

To avoid overstating the effects of RTAs on margin decomposition, zero-trade observations were removed, and the analysis was restricted to country–product pairs with positive trade flows. Following Hummels and Klenow [14] and Shi [21], a benchmark country was then constructed for each importer–year combination. Specifically, the benchmark value is defined as the total exports of HS six-digit product k from all exporting countries to a given importing country j in year t, which serves as the reference point for decomposing the trade margins.

3.2.2. Construction of the World Benchmark

Consistent with Hummels and Klenow [14] and Shi [21], a world benchmark is constructed for each importer–year combination. Specifically, for a given importing country $j$ and year $t$, the benchmark export value $V_{Wjt}$ is defined as the total exports of HS six-digit product $k$ from all exporting countries to importer $j$. The corresponding product set $I_{Wjt}$ is the union of all products imported by $j$ from the world in year $t$.

This benchmark represents the aggregate import demand faced by all exporters in a given market and serves as a common reference point for normalizing country-specific export performance across importing countries of different market sizes.

3.2.3. Formal Decomposition

The export market share of country i to importer j in year t, denoted by $R_{ijt}$, can be expressed as the product of the extensive margin ($E{M}_{ijt}$) and the intensive margin ($I{M}_{ijt}$). Following the established trade literature, the intensive margin can be further decomposed into the quantity margin and the price margin, yielding a three-margin decomposition of export performance. This decomposition approach is widely used to distinguish between changes in export participation, export quantities, and unit values, and has been extensively applied in empirical studies on trade patterns and policy effects [13,14,15,16,17]. Accordingly, the export market share can be written as the product of the extensive margin, the quantity margin, and the price margin.

$$\begin{gathered} R_{ijt}={\displaystyle \frac{{\sum }_{k\in I_{ijt}}{V}_{ijt}}{{\sum }_{k\in I_{Wjt}}{V}_{Wjt}}}={\displaystyle \frac{{\sum }_{k\in I_{ijt}}{V}_{Wjt}}{{\sum }_{k\in I_{Wjt}}{V}_{Wjt}}}\times {\displaystyle \frac{{\sum }_{k\in I_{ijt}}{V}_{ijt}}{{\sum }_{k\in I_{ijt}}{V}_{Wjt}}} \\ {EM}_{ijt}={\displaystyle \frac{{\sum }_{k\in I_{ijt}}{V}_{Wjt}}{{\sum }_{k\in I_{Wjt}}{V}_{Wjt}}} , {IM}_{ijt}={\displaystyle \frac{{\sum }_{k\in I_{ijt}}{V}_{ijt}}{{\sum }_{k\in I_{ijt}}{V}_{Wjt}}} \end{gathered}$$

(1)

where $I_{ijt}$ denotes the set of products exported by country $i$ to country $j$, $V_{ijt}$ is the export value of product $k$ from $i$ to $j$, and $V_{Wjt}$ is the corresponding world reference value.

The extensive margin $E{M}_{ijt}$ measures the weighted breadth of the product set exported by country $i$ relative to the world benchmark. A higher value of $E{M}_{ijt}$ indicates greater product diversification in the destination market.

The intensive margin is further decomposed into price and quantity components:

$$\begin{gathered} {IM}_{ijt}={PM}_{ijt}\times {QM}_{ijt} \\ {PM}_{ijt}={\prod _{k\in I_{ijt}}\left({\displaystyle \frac{p_{ijt}}{p_{Wjt}}}\right)}^{w_{ijt}}, {QM}_{ijt}={\prod _{i\in I_{ijt}}\left({\displaystyle \frac{q_{ijt}}{q_{Wjt}}}\right)}^{w_{ijt}} \end{gathered}$$

(2)

where $p$ and $q$ denote unit values and quantities, respectively, and $w_{ijt}$ is the logarithmic mean of product-level export shares.

The natural logarithms of these indices—$lnE{M}_{ijt}$, $lnP{M}_{ijt}$, and $lnQ{M}_{ijt}$—are used as the dependent variables in the regression analysis.

3.3. Independent Variable

The core explanatory variable is RTA participation ($DI{D}_{ijt}$). This binary indicator takes the value of 1 if a Regional Trade Agreement between country $i$ and country $j$ is in force in year $t$, and 0 otherwise. Constructed using global RTA data from 1958 to 2022, this variable captures both the contemporaneous and sustained effects of trade liberalization associated with RTAs [4].

3.4. Control Variables

To control for country-level heterogeneity, we incorporate standard covariates derived from the structural gravity model [22,23]. Economic controls include differences in GDP ($lnGDP$) between exporting and importing countries. Bilateral trade costs are proxied using variables from the CEPII Gravity Database [23], including geodesic distance ($lnDIST$), contiguity (CONTIG), common language (COMLANG), colonial ties (COMCOL), shared legal origins (COMLEG), and shared religious beliefs (COMRELIG).

3.5. Mediation Variables

To examine the transmission mechanisms of RTAs, two mediating variables are introduced. Bilateral variable trade costs capture the direct economic channel through which RTAs reduce transaction barriers and policy frictions [24]. Bilateral social connectedness captures an indirect relational channel, reflecting how RTAs foster institutional cooperation and social linkages that reduce information asymmetries, facilitate technology diffusion, and lower search and matching costs [19].

Descriptive statistics for all variables are reported in Table 1.

Table 1. Descriptive statistics.

Variable	N	Meaning	Mean	SD	Min	Max
lnR	481,882	Log of Export Market Share	−8.334	3.514	−23.93	−0.012
lnEM	481,882	Log of Extensive Margin	−2.956	2.079	−16.349	0
lnPM	481,882	Log of Price Margin	0.317	0.689	−7.694	10.241
lnQM	481,882	Log of Quantity Margin	−5.696	2.499	−24.32	0
DID	481,882	Regional Trade Agreement	0.225	0.417	0	1
lncosts	96,424	Log of Social Connectedness Index	5.701	0.497	2.124	7.555
lnSocialConnect	404,437	Log of Bilateral Trade Costs	7.774	1.792	4.043	17.452
lnGDP	481,882	Diff. in Log GDP	7.945	1.645	0.473	12.927
lnDIST	481,882	Log of Bilateral Distance	8.625	0.838	2.079	9.894
CONTIG	481,882	shared land border	0.024	0.153	0	1
COMLANG	481,882	common official language	0.178	0.383	0	1
COMRELIG	460,887	Shared Religious Beliefs	0.174	0.251	0	0.994
COMCOL	481,882	colonial relationship	0.015	0.122	0	1
COMLEG	481,880	shared legal origins	0.408	0.491	0	1

Note. ln denotes the natural logarithm. The dependent variables (lnR, lnEM, lnPM, lnQM) represent the decomposition of export growth into market share, extensive margin, price margin, and quantity margin, respectively. DID is a binary variable taking the value of 1 if a trade agreement is in force. lnGDP denote the natural logarithms of the GDP ratios between exporting and importing countries. Gravity variables lnDist represents the natural logarithm of the geodesic distance between capital cities. CONTIG, COMLANG, COMRELIG, COMCOL, COMLEG are binary dummy variables.

4. Methodology

4.1. Estimation Framework and Baseline Specification

Based on the panel dataset described in Section 3, this study employs a fixed-effects panel estimation framework to examine the impact of RTA participation on the export margins of wood forest products. A fixed-effects approach is preferred over a random-effects specification because unobserved country-specific and country-pair-specific characteristics—such as institutional quality, historical trade relations, and long-term policy orientation—are likely to be correlated with RTA participation, thereby violating the exogeneity assumption required for random-effects estimation.

To identify the causal effect of RTAs, we adopt a staggered DID model, specified as follows:

$${\mathit{ln}Y}_{ijt}={\beta }_0+{\beta }_1{DID}_{ijt}+\sum {\beta }_n{X}_n+{\mu }_{year}+{\mu }_{\mathrm{exporter}}+{\mu }_{importer}+{\epsilon }_{ijt}$$

(3)

where $i$, $j$, and $t$ denote the exporter, importer, and year, respectively. The dependent variable $Y_{ijt}$ represents one of the export margin measures: export market share ($\mathit{ln}{R}_{ijt}$), extensive margin ($lnE{M}_{ijt}$), price margin ($lnP{M}_{ijt}$), or quantity margin ($lnQ{M}_{ijt}$). The key explanatory variable $DI{D}_{ijt}$ is a binary indicator that equals one if an RTA between exporter $i$ and importer $j$ is in force in year $t$, and zero otherwise.

The vector ${\mathrm{X}}_{ijt}$ includes standard gravity-type control variables. Year fixed effects (${\mu }_t$) control for global shocks affecting all trade flows, while exporter fixed effects (${\mu }_i$) and importer fixed effects (${\mu }_j$) absorb time-invariant country-specific heterogeneity. The error term ${\epsilon }_{ijt}$ captures idiosyncratic shocks.

4.2. Nature of Variables and Logarithmic Transformations

The dependent variables in this study are continuous indices derived from the export ternary margin decomposition framework. Because export performance is multiplicative in nature—where total exports can be expressed as the product of extensive, price, and quantity margins—the dependent variables are transformed using natural logarithms. This transformation allows the estimated coefficients to be interpreted as elasticities and helps mitigate heteroskedasticity commonly observed in trade data.

The core policy variable $DI{D}_{ijt}$ is binary and therefore enters the model in levels rather than logarithms. Similarly, other categorical variables—such as contiguity, common language, colonial ties, shared religion, and common legal origin—are included as dummy variables. Continuous trade-cost variables, including geographic distance and bilateral trade costs, are log-transformed following standard practice in the gravity literature. This consistent treatment of variables ensures transparency and facilitates economic interpretation of the estimated coefficients.

4.3. Rationale for Multiple Dependent Variables

Rather than relying on aggregate export values alone, this study employs multiple dependent variables corresponding to different export margins. These margins capture distinct but complementary dimensions of export performance. The extensive margin reflects product variety and market diversification, the quantity margin captures scale expansion of existing products, and the price margin proxies quality upgrading and value enhancement.

Using multiple dependent variables allows us to disentangle the channels through which RTAs affect trade outcomes. Specifically, this approach enables an assessment of whether RTAs primarily promote export growth through diversification, volume expansion, or quality upgrading. Such differentiation is particularly important for wood forest products, where trade expansion may be constrained by resource endowments, sustainability regulations, and technological requirements. The use of margin-level outcomes therefore provides a more nuanced understanding of RTA effects than aggregate trade measures.

4.4. Dynamic Effects and Event-Study Design

The validity of the DID approach relies on the parallel trends assumption—that, in the absence of RTA implementation, treated and untreated country pairs would have followed similar trajectories in export margins. To assess this assumption and to examine the dynamic evolution of RTA effects over time, we implement an event-study specification:

$${\mathit{ln}Y}_{ijt}=\alpha +\sum _{k=-5,k\ne -1}^7{\beta }_k{DT}_{ijt}^k+\sum {\beta }_n^{\prime }{X}_n+{\mu }_{year}+{\mu }_{\mathrm{exporter}}+{\mu }_{importer}+{\epsilon }_{ijt}$$

(4)

where $D{T}_{ijt}^k$ is a set of relative-time dummy variables indicating the number of years before or after the RTA enters into force. The coefficients on the pre-treatment dummies test for the presence of anticipatory effects, while the post-treatment coefficients capture the dynamic adjustment path of export margins following RTA implementation.

4.5. Mediation Analysis

To investigate the mechanisms through which RTAs influence export margins, we employ a causal mediation framework. This framework explicitly models a temporal sequence in which RTA implementation first affects intermediate economic conditions and subsequently alters export outcomes.

The mediation analysis is conducted using the following stepwise regressions:

$${lnM}_{\mathrm{ijt}}={\gamma }_0+{\gamma }_1{DID}_{\mathrm{ijt}}+\sum {\gamma }_n{lnX}_n+{\mu }_{\mathrm{year}}+{\mu }_{\mathrm{exporter}}+{\mu }_{\mathrm{importer}}+{\epsilon }_{\mathrm{ijt}}$$

(5)

$${lnY}_{\mathrm{i}\mathrm{j}\mathrm{t}}={\theta }_0+{\theta }_1{DID}_{\mathrm{ijt}}+{\theta }_2{lnM}_{\mathrm{ijt}}+\sum {\theta }_n{X}_n+{\mu }_{\mathrm{year}}+{\mu }_{\mathrm{exporter}}+{\mu }_{\mathrm{importer}}+{\epsilon }_{\mathrm{ijt}}$$

(6)

where $M_{ijt}$ denotes the mediating variables, namely bilateral trade costs and bilateral social connectedness. This approach allows us to assess whether RTAs affect export margins indirectly through changes in trading frictions and relational connectivity, while controlling for confounding factors using high-dimensional fixed effects.

4.6. Addressing Endogeneity and Heterogeneity

A key concern in evaluating RTA effects is potential endogeneity arising from non-random agreement formation. To address this issue, we adopt multiple complementary strategies.

First, an instrumental variable approach is implemented using the Diplomatic Divergence Index and the lagged RTA indicator as instruments for current RTA participation. These instruments affect the likelihood of RTA formation through political alignment and institutional persistence but are plausibly exogenous to contemporaneous bilateral trade outcomes.

Second, to control for time-varying multilateral resistance and country-specific shocks, we re-estimate the baseline model using exporter–year and importer–year interaction fixed effects. This specification absorbs all exporter- and importer-specific factors that vary over time, providing a more stringent identification strategy.

In addition, heterogeneity analyses are conducted along two dimensions. Product-level heterogeneity is examined by estimating the model separately for resource-intensive, labor-intensive, and capital-intensive wood forest products. Agreement-level heterogeneity is explored by distinguishing RTAs according to their institutional depth, including WTO-plus provisions and environmental clauses. These analyses allow us to assess whether RTA effects differ systematically across product characteristics and agreement design.

4.7. Placebo Test Explanation

To further validate the causal interpretation of the estimated effects, we conduct placebo tests by randomly assigning RTA treatment status across country-pair–year observations. This procedure generates a distribution of placebo coefficients under the null hypothesis of no treatment effect. Comparing the baseline estimates with this distribution helps rule out the possibility that the observed effects are driven by random chance, serial correlation, or spurious correlations.

4.8. Estimation Software

All data processing and econometric estimations are performed using Stata (V18.0). High-dimensional fixed-effects regressions are estimated using the reghdfe package, which efficiently handles multiple fixed effects and provides appropriate degrees-of-freedom adjustments.

5. Results

5.1. Baseline Regression Results

Table 2 reports the baseline estimates of the impact of RTAs on the export ternary margins of wood forest products. The dependent variables include export market share, extensive margin, quantity margin, and price margin. A positive coefficient indicates that RTA participation increases the corresponding export margin, while statistical significance reflects the robustness of the estimated effect.

Table 2. Benchmark regression.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	lnR	lnEM	lnPM	lnQM	lnR	lnEM	lnPM	lnQM
DID	1.2503 ***	0.6560 ***	−0.0620 ***	0.6564 ***	0.3094 ***	0.1445 ***	−0.0083 *	0.1733 ***
	(0.0324)	(0.0195)	(0.0042)	(0.0200)	(0.0235)	(0.0145)	(0.0043)	(0.0173)
ln_GDP					−0.1806 ***	0.1085 ***	−0.0116 ***	−0.2775 ***
					(0.0156)	(0.0109)	(0.0030)	(0.0111)
lnDIST					−1.4293 ***	−0.8519 ***	0.0849 ***	−0.6623 ***
					(0.0161)	(0.0109)	(0.0030)	(0.0113)
CONTIG					0.7458 ***	0.1596 **	−0.0454 ***	0.6316 ***
					(0.0879)	(0.0661)	(0.0108)	(0.0506)
COMLANG					0.6093 ***	0.4928 ***	−0.0405 ***	0.1571 ***
					(0.0332)	(0.0207)	(0.0063)	(0.0248)
comrelig					0.4180 ***	0.2058 ***	−0.0381 ***	0.2504 ***
					(0.0448)	(0.0275)	(0.0088)	(0.0346)
COMCOL					0.7812 ***	0.3414 ***	−0.0195	0.4593 ***
					(0.0893)	(0.0615)	(0.0134)	(0.0607)
COMLEG					0.2410 ***	0.1326 ***	−0.0194 ***	0.1279 ***
					(0.0216)	(0.0134)	(0.0042)	(0.0170)
Constant	−8.6153 ***	−3.1030 ***	0.3313 ***	−5.8436 ***	5.1034 ***	3.3398 ***	−0.2993 ***	2.0629 ***
	(0.0121)	(0.0073)	(0.0020)	(0.0085)	(0.1864)	(0.1334)	(0.0333)	(0.1237)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Exporter FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Importer FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	481,882	481,882	481,882	481,882	460,887	460,887	460,887	460,887
R2	0.5998	0.5252	0.1161	0.4632	0.7129	0.6344	0.1282	0.5131

Note. ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively. The data in brackets represent the robust standard errors for the clusters.

The results show that RTA participation is significantly associated with changes in export structure rather than uniform expansion. Specifically, RTAs are positively related to the extensive margin, indicating that trade agreements promote product diversification in forest product exports. In contrast, the effects on the quantity margin are weaker, suggesting that RTAs do not necessarily lead to proportional increases in export volumes. The price margin exhibits a positive and statistically significant response, implying that RTAs may facilitate quality upgrading or higher value-added exports.

Overall, these findings suggest that the trade effects of RTAs in the forest sector operate primarily through structural and qualitative adjustments rather than simple scale expansion.

5.2. Robustness Tests

5.2.1. Parallel Trend Test and Dynamic Effect Analysis

The validity of using a DID model for policy effect evaluation depends on the parallel trends assumption [4,16], which requires that, prior to policy implementation, the outcome variables of the treatment and control groups exhibit similar trends.

Figure 1 presents the event-study estimates of RTA effects on export margins. The coefficients in the pre-treatment period are statistically insignificant, supporting the validity of the parallel trends assumption. Following the entry into force of RTAs, the effects on export margins emerge gradually and persist over time, indicating that the adjustment process is not instantaneous but unfolds in the medium to long run.

Figure 1. Parallel trend test. Note: Figure plots the estimated coefficients from the event-study specification to test the parallel trends assumption and analyze the dynamic effects of RTAs. The horizontal axis represents the time relative to the entry into force of the agreement, where t = 0 denotes the year of implementation. The vertical axis indicates the estimated impact on the logarithmic values of (a) Export Market Share (lnR), (b) Extensive Margin (lnEM), (c) Price Margin (lnPM), and (d) Quantity Margin (lnQM). The markers represent point estimates, and the vertical dashed lines indicate the 95% confidence intervals. The fact that the coefficients prior to t = 0 are statistically indistinguishable from zero supports the validity of the parallel trends assumption.

5.2.2. Placebo Test

To verify that the observed RTA impacts are not driven by random factors, we employed a placebo test following Moser [25] by randomly reassigning trade pairs and policy timing 1000 times. As shown in Figure 2, the resulting pseudo-coefficients follow a normal distribution centered near zero, with most p-values exceeding 0.1. The significant divergence between these insignificant pseudo-estimates and the baseline coefficients confirms that our results are robust and not artifacts of random noise.

Figure 2. Illustrates the distribution of the estimated coefficients based on 1000 random samples. Note: Figure reports the placebo test results based on 1000 random permutations of the policy variable. Panels (a–d) correspond to lnR, lnEM, lnPM, and lnQM, respectively. The curves denote the kernel density estimates of the placebo coefficients, and the scattered dots represent their corresponding p-values. The vertical dashed lines indicate the actual coefficients from the baseline regression. The results show that the true coefficients are significantly distinguishable from the placebo distribution centered near zero, confirming the robustness of the findings.

5.2.3. Heterogeneous Robust Estimator of the Multi-Period DID Model

To address potential biases associated with staggered treatment adoption and heterogeneous treatment effects, we re-estimated the baseline model using the Extended Two-Way Fixed Effects (ETWFE) estimator proposed by Wooldridge [26]. This method explicitly accounts for cohort-specific time trends, providing a consistent estimator of the Average Treatment Effect on the Treated. The results are presented in Table 3.

Table 3. The Wooldridge ETWFE estimator.

Variable	(1)	(2)	(3)	(4)
	lnR	lnEM	lnPM	lnQM
ETWFE estimator	0.4004 *** (0.0272)	0.1718 *** (0.0173)	−0.0163 *** (0.0055)	0.2450 *** (0.0199)
Country FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Obs.	456,865	456,865	456,865	456,865

Note. *** denotes significance at the 1% levels. The data in brackets represent the robust standard errors for the clusters.

The results in Table 3 confirm that RTA participation significantly boosts the overall export market share of wood forest products. Decomposing this aggregate growth reveals that the expansion is primarily driven by the quantity margin and the extensive margin, indicating that trade liberalization effectively promotes both the scale of existing exports and the diversification of product varieties. Conversely, the price margin exhibits a slight but significant decline, suggesting that RTAs enhance export competitiveness through cost reductions. Notably, the sum of the three margin coefficients aligns precisely with the total market share coefficient, validating the internal consistency and robustness of our conclusions. Therefore, by re-estimating the heterogeneous robust estimator, we can conclude that the multi-period DID model in this paper is relatively robust.

5.2.4. Alternative Core Explanatory Variables

To ensure the reliability of the baseline estimates and address potential biases arising from the staggered adoption of trade agreements, we conducted a series of robustness checks and endogeneity tests.

First, to mitigate bias from cross-country heterogeneity and the lack of pre-treatment observations for early RTAs, we restricted the sample to agreements formed after 2000. As shown in Table 4 (Columns 1–4), the coefficients remain consistent with the baseline, confirming that the results are not driven by legacy agreements.

Table 4. Robustness Tests with Alternative Core Explanatory Variables.

Variable	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	lnR	lnEM	lnPM	lnQM	lnR	lnEM	lnPM	lnQM
DID	0.2529 *** (0.0098)	0.0619 *** (0.0064)	−0.0092 *** (0.0035)	0.2002 *** (0.0093)
DID_pre3					0.4408 *** (0.0074)	0.2253 *** (0.0048)	−0.0004 (0.0024)	0.2159 *** (0.0066)
Constant	0.2017 (0.1353)	0.9433 *** (0.0915)	−0.5474 *** (0.0466)	−0.1942 (0.1250)	0.5617 *** (0.1204)	0.8947 *** (0.0814)	−0.5437 *** (0.0399)	0.2107 * (0.1084)
Control Variable	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Country FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	413,043	413,043	413,043	413,043	481,799	481,799	481,799	481,799
R2	0.7200	0.6458	0.1278	0.5149	0.7148	0.6343	0.1273	0.5175

Note. ***, * denote significance at the 1%, and 10% levels, respectively. The data in brackets represent the robust standard errors for the clusters.

Second, addressing the “anticipatory effect” often observed in institutional economics, we tested for behavioral adjustments during the negotiation phase. By constructing a lead variable (DID_pre3) shifting the effective date 36 months earlier, we found significant positive effects on the extensive and quantity margins but insignificant effects on prices (Table 4, Columns 5–8). This confirms that while firms may expand variety in anticipation of market opening, price adjustments typically exhibit a lag, aligning with Hypothesis 2.

Third, to address concerns regarding staggered treatment effects common in multi-period DID models, we relied on the Event Study specification (see Parallel Trend Test). The absence of significant pre-trends supports the validity of the parallel trend assumption, suggesting that the dynamic effects captured are causal rather than artifacts of differential timing.

5.2.5. Endogeneity Test Results

We employed two strategies to address potential endogeneity. First, we use the Diplomatic Divergence Index [24] and the lagged RTA variable (L.DID) as instrumental variables. The Diplomatic Divergence Index as instrumental variable is grounded in the political economy literature on trade agreement formation. Measures of bilateral diplomatic distance based on UN voting behavior have been widely used to capture political alignment between countries and to explain variation in the likelihood and design of regional trade agreements, as political divergence reduces incentives for international cooperation [27,28]. Following this literature, the Diplomatic Divergence Index provides a plausible instrument for RTA participation, as it is strongly correlated with agreement formation while remaining plausibly exogenous to sector-specific trade outcomes. In addition, the use of lagged RTA variables as instruments is consistent with established approaches to addressing the endogeneity of trade agreements, which emphasize the persistence of agreement status and the implausibility of reverse causality from current trade outcomes to past policy decisions [6,29]. Using a two-stage least squares (2SLS) estimation, the results reported in Columns (1)–(4) of Table 5 show that the estimated coefficients on DID remain consistent in both direction and magnitude with the baseline results, confirming the robustness of the core findings.

Table 5. Endogeneity Test.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	lnR	lnEM	lnPM	lnQM	lnR	lnEM	lnPM	lnQM
DID	0.4757 *** (0.0084)	0.1932 *** (0.0054)	−0.0101 *** (0.0027)	0.2926 *** (0.0075)	0.4709 *** (0.0079)	0.2209 *** (0.0051)	−0.0052 ** (0.0025)	0.2552 *** (0.0070)
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Country FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Country-Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	393,950	393,950	393,950	393,950	481,799	481,799	481,799	481,799
R2	0.3355	0.2664	0.0163	0.1290	0.7271	0.6541	0.1752	0.5410

Note. ***, ** denote significance at the 1%, 5% levels, respectively. The data in brackets represent the robust standard errors for the clusters.

Second, to further reinforce robustness, the fixed-effects structure is redefined to include exporter–year and importer–year interactions. As shown in Columns (5)–(8) of Table 5, the estimated DID coefficients remain both statistically and economically stable, suggesting that the main conclusions are not sensitive to alternative fixed-effects specifications.

5.3. Heterogeneity Analysis

This section explores heterogeneity in the effects of RTAs on the three export margins of wood forest products, accounting for variations in product type and agreement depth.

5.3.1. Product Type

As the existing literature has shown, the effects of RTAs vary systematically across products and sectors according to factor intensity and technological characteristics [9,30]. This approach has been already used in the trade literature to address heterogeneity in the impacts of trade agreements across industries and product types [15,16,17], highlighting that differences in factor intensity and production technology lead to heterogeneous responses along various trade margins.

Following this literature, wood forest products are classified into resource-intensive, labor-intensive, and capital-intensive categories. Resource-intensive products include raw logs and sawn timber (HS 4401–4407); labor-intensive products comprise wood panels (HS 4408–4413), wooden articles (HS 4414–4421), and wooden furniture (HS 940161, 940169, 940330–940360); and capital-intensive products include pulp (HS 4701–4706) and paper or paperboard (HS 4707, HS 48–49). Table 6 reports the corresponding regression results, with Columns (1)–(4) referring to labor-intensive products, Columns (5)–(8) to resource-intensive products, and Columns (9)–(12) to capital-intensive products.

Table 6. Heterogeneity Test by Product Type.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
	lnR	lnEM	lnPM	lnQM	lnR	lnEM	lnPM	lnQM	lnR	lnEM	lnPM	lnQM
DID	0.0183 (0.0161)	0.1148 *** (0.0098)	−0.0034 (0.0072)	−0.0931 *** (0.0177)	0.3268 *** (0.0099)	0.0657 *** (0.0042)	0.0193 *** (0.0042)	0.2418 *** (0.0096)	0.3131 *** (0.0098)	0.0476 *** (0.0039)	−0.0231 *** (0.0039)	0.2885 *** (0.0092)
Constant	2.9947 *** (0.0884)	2.3151 *** (0.0528)	−0.5355 *** (0.0400)	1.2151 *** (0.0961)	3.9628 *** (0.0576)	2.3112 *** (0.0243)	−0.7384 *** (0.0235)	2.3900 *** (0.0549)	6.3010 *** (0.0550)	2.8722 *** (0.0219)	−0.7800 *** (0.0209)	4.2087 *** (0.0503)
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Country FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	107,512	107,512	107,512	107,512	265,950	265,950	265,950	265,950	347,762	347,762	347,762	347,762
R2	0.5390	0.2735	0.2111	0.4579	0.6709	0.4219	0.1886	0.6163	0.6846	0.5434	0.1587	0.6217

Note. *** denotes significance at the 1% levels. The data in brackets represent the robust standard errors for the clusters.

The regression results indicate that RTAs significantly increase market share across all product categories, with the strongest effect observed for capital-intensive products. Notably, the mechanisms of export growth vary considerably across product types:

For resource-intensive products, RTAs significantly expand the extensive margin but suppress the quantity margin, while their effects on price and market share are statistically insignificant. This can be attributed to (a) the low value-added and high price elasticity of primary wood products, and (b) natural resource constraints and sustainable harvesting regulations that limit volume expansion despite improved market access.

For labor-intensive products, RTAs significantly enhance market share as well as extensive, price, and quantity margins. The positive effects on the extensive and quantity margins far outweigh the negative impact on prices. This suggests that RTAs primarily stimulate export growth through diversification and volume expansion. However, standardized production processes and low technological barriers lead to intense product homogeneity and narrow price margins, resulting in a “more volume, less value” development dilemma.

For capital-intensive products, RTAs significantly boost market share, extensive, and quantity margins while exerting a negative effect on the price margin. These results suggest that capital-intensive, high-tech, and high-value-added products benefit from technology spillovers and intellectual property protection clauses embedded in deep RTAs. Enhanced dynamic efficiency reduces marginal costs, enabling exporters to adopt price-competitive strategies that expand export scale and promote overall growth.

5.3.2. Agreement Depth

Given variations in liberalization depth, RTAs may exert differentiated impacts on export performance. Using the DTA Database [20], this study examines how agreement depth affects the three export margins of wood products.

Results presented in Table 7 show that greater agreement depth, measured by the WTO-Plus Index, has positive effects on the extensive and price margins but negative effects on market share and quantity margins. This suggests that deeper RTAs expand product diversity while intensifying market competition, prompting firms to pursue quality upgrading and differentiation. These dynamics raise production costs and prices, thereby constraining volume expansion and market share growth.

Table 7. Heterogeneity by Agreement Depth.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	lnR	lnEM	lnPM	lnQM	lnR	lnEM	lnPM	lnQM
lnWTO_Plus	−0.0307 *** (0.0057)	0.0156 *** (0.0039)	0.0130 *** (0.0019)	−0.0593 *** (0.0051)
ln_EnvDepth					0.0194 *** (0.0040)	0.0521 *** (0.0027)	0.0120 *** (0.0013)	−0.0448 *** (0.0036)
Constant	4.9811 *** (0.0553)	4.0784 *** (0.0367)	−0.5069 *** (0.0175)	1.4096 *** (0.0469)	4.9527 *** (0.0552)	4.2648 *** (0.0369)	−0.4387 *** (0.0173)	1.1265 *** (0.0462)
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Country FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	185,433	185,433	185,433	185,433	185,433	185,433	185,433	185,433
R2	0.7383	0.6714	0.1591	0.5399	0.7383	0.6720	0.1593	0.5400

Note. *** denotes significance at the 1% levels. The data in brackets represent the robust standard errors for the clusters.

While deeper liberalization enhances export diversity, it simultaneously pressures firms to undertake structural upgrading and quality improvement.

Further analysis reveals that environmental provisions within RTAs significantly promote market share, extensive, and price margins, while reducing the quantity margin. This indicates that stricter environmental clauses—by curbing pollution-intensive exports and fostering clean technology cooperation—encourage diversification and quality enhancement. Although output volume may decline, export structure optimization ultimately strengthens overall market competitiveness and sustainability.

5.4. Mechanism Analysis

5.4.1. Trade Cost

Regression results are reported in Columns (1)–(5) of Table 8. Column (1) indicates that the signing of RTAs exerts a significantly negative effect on variable trade costs among member countries, confirming the cost-reducing role of such agreements. To further explore the mechanism through which trade cost reduction affects the ternary margins of wood product exports, bilateral variable trade cost data for wood products are incorporated into the baseline regression model.

Table 8. Mechanism Analysis of Trade Cost.

Variables	(1)	(2)	(3)	(4)	(5)
	lnCosts	lnR	lnEM	lnPM	lnQM
DID	−0.1270 *** (0.0029)	0.2397 *** (0.0113)	0.0587 *** (0.0075)	−0.0069 * (0.0035)	0.1879 *** (0.0103)
lnCosts		−1.1806 *** (0.0137)	−0.3433 *** (0.0088)	0.0667 *** (0.0042)	−0.9040 *** (0.0126)
Constant	3.5280 *** (0.0156)	7.1272 *** (0.0767)	4.1606 *** (0.0509)	−0.5897 *** (0.0233)	3.5564 *** (0.0688)
Controls	Yes	Yes	Yes	Yes	Yes
Country FE	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Observations	96,422	96,422	96,422	96,422	96,422
R2	0.5934	0.8016	0.6889	0.2704	0.6582

Note. ***, * denote significance at the 1%, and 10% levels, respectively. The data in brackets represent the robust standard errors for the clusters.

The results in Columns (2)–(5) reveal two key findings. First, RTAs continue to exert significantly positive effects on the extensive margin, quantity margin, and market share of wood product exports. Second, after including trade costs, the absolute values of the core explanatory variable coefficients decrease, indicating that variable trade costs partially mediate the relationship between RTAs and wood export performance.

Based on these results, Hypothesis 3 is supported: RTAs promote wood product exports by reducing bilateral variable trade costs, thereby lowering the price margin and increasing the extensive, quantity, and market share margins.

5.4.2. Bilateral Social Connectedness

Theoretically, RTAs can promote cultural exchange, technological cooperation, and labor mobility among member countries, thereby enhancing bilateral social connectedness. Stronger social ties reduce information asymmetry and transaction-related trust costs, which in turn can stimulate export activities.

To empirically test this mechanism, this study employs the Social Connectedness Index from the CEPII database as a mediating variable and constructs the corresponding econometric models.

Results reported in Table 9 show that RTAs significantly enhance bilateral social connectedness, indicating that such agreements foster cultural and human exchanges and create favorable conditions for deeper economic cooperation. After incorporating the social connectedness index into the baseline model, RTAs remain positively associated with export market share, extensive margin, and quantity margin, while continuing to exert a negative effect on the price margin—consistent with the baseline regression results.

Table 9. Mechanism Analysis of Bilateral Social Connectedness.

Variable	(1)	(2)	(3)	(4)	(5)
	lnSocialConnected	lnR	lnEM	lnPM	lnQM
DID	0.4192 *** (0.0041)	0.2810 *** (0.0078)	0.0732 *** (0.0051)	−0.0069 *** (0.0027)	0.2147 *** (0.0073)
lnSocialConnected		0.3767 *** (0.0032)	0.2880 *** (0.0022)	−0.0054 *** (0.0011)	0.0940 *** (0.0029)
Constant	18.4823 *** (0.0262)	−3.1562 *** (0.0726)	−1.2489 *** (0.0497)	−0.3193 *** (0.0241)	−1.5880 *** (0.0655)
Controls	Yes	Yes	Yes	Yes	Yes
Country FE	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Observations	404,437	404,437	404,437	404,437	404,437
R2	0.7344	0.7287	0.6615	0.1279	0.5043

Note. *** denotes significance at the 1% levels. The data in brackets represent the robust standard errors for the clusters.

Furthermore, after controlling for social connectedness, the absolute values of the RTA coefficients decline across models, suggesting that social connectedness partially mediates the relationship between RTAs and the ternary margins of wood product exports.

Therefore, Hypothesis 4 is verified: RTAs promote wood product export growth by strengthening bilateral social connectedness, which increases the extensive, quantity, and market share margins while reducing the price margin.

6. Discussion

This study provides robust evidence that RTAs significantly expand the export market share of wood forest products, corroborating the trade creation effects widely documented in the literature [6,7,8]. By applying a ternary margin decomposition framework, however, our analysis shows that export growth is driven predominantly by expansions along the extensive margin and the quantity margin, while the price margin exerts a statistically significant negative effect. This pattern is broadly consistent with heterogeneous-firm trade theory [13,14], which predicts that reductions in trade costs lower productivity cutoffs and enable a larger set of firms and product varieties to access foreign markets. Similar margin-based evidence has been reported by Bernard et al. [31] and Helpman et al. [32], who emphasize the importance of firm entry and product expansion following trade liberalization. Our findings reinforce this mechanism in the context of wood forest products, a sector characterized by relatively standardized goods and strong cost competition.

A key contribution of this study lies in the identification of a negative price margin effect associated with RTAs. This result contrasts with the quality-upgrading hypothesis advanced by Hummels and Klenow [14], according to which trade liberalization allows exporters to charge higher unit values through quality differentiation. Instead, our findings align more closely with the “pro-competitive effect” highlighted by Feenstra and Weinstein [33], whereby intensified competition compresses markups and transfers efficiency gains to importing markets. In the wood forest product sector, high product substitutability and limited scope for vertical differentiation appear to constrain quality-based price increases, leading exporters to compete primarily on cost. The observed temporal lag in the price margin adjustment further suggests that price competition intensifies gradually as market entry expands and competitive pressure accumulates.

The heterogeneity analysis further reveals important sectoral implications. Labor-intensive wood product exports exhibit a “volume-driven” growth pattern accompanied by declining prices, raising concerns about potential lock-in effects in low–value-added segments of global value chains [34]. In contrast, resource-intensive products experience expansion in product variety but stagnation in export volumes. This divergence lends support to recent studies arguing that environmental provisions embedded in modern RTAs impose constraints on raw material extraction [3], thereby encouraging diversification rather than scale-based growth.

These limitations also point to promising avenues for future research. Micro-level customs or firm-level datasets could be employed to disentangle the respective roles of firm entry, product switching, and scale expansion in driving export growth. Moreover, integrating carbon accounting frameworks with trade models would allow researchers to assess whether RTA-induced export expansion in wood forest products is environmentally sustainable, thereby linking trade policy evaluation with climate and resource governance. Finally, future studies could explore how the design of environmental clauses in RTAs shapes the balance between quantity expansion, product diversification, and price competition across different resource-based sectors.

7. Conclusions and Implications

7.1. Conclusions

Using bilateral HS-6 level trade data from 2000 to 2022 and a multi-period DID framework, this study examines the impact of Regional Trade Agreements on the ternary margins of wood forest product exports. The results show that RTAs significantly increase export market share, primarily through the expansion of product variety and export volumes. At the same time, RTAs exert a modest but significant negative effect on export prices, indicating that competitive pressures may outweigh quality-upgrading effects in this sector.

The effects of RTAs are heterogeneous across product types. Labor- and capital-intensive products benefit mainly from scale expansion, while resource-intensive products face quantity constraints, particularly under agreements with stricter environmental provisions. Mechanism analysis confirms that RTAs influence export margins by reducing bilateral trade costs and strengthening bilateral social connectedness.

7.2. Implications

Based on these findings, this study proposes policy recommendations that move beyond simple trade liberalization toward the construction of sustainable and efficient forestry trade networks.

First, regarding the structure of agreements, policymakers should prioritize “quality-oriented” provisions over simple tariff reductions. Since our results indicate that standard RTAs can lead to a “low price, high volume” trap, future negotiations should incorporate “behind-the-border” measures—such as the mutual recognition of timber legality verification and sustainability certifications (e.g., FSC/PEFC). Establishing “Green Lanes” for certified products within RTAs can reduce compliance costs for high-quality exports, thereby countering the negative price margin effect observed in the data.

Second, a nuanced approach is required to manage the trade-off between environmental regulation and export volume. Our heterogeneity analysis shows that deeper environmental clauses restrict the quantity margin for resource-intensive products. To mitigate the economic shock of this transition, RTAs should include specific capacity-building clauses. Developed nations should provide technical assistance to developing trading partners to help them meet strict environmental standards. This ensures that environmental provisions function as instruments for upgrading industrial structure rather than merely acting as non-tariff barriers that block market access.

Third, governments should strengthen institutional coordination to maximize the social connectedness channel. Since social ties significantly mediate trade growth, policy should actively support industry associations and trade fairs that foster direct buyer–seller relationships. By reducing informational frictions, these platforms allow exporters to move away from standardized, price-sensitive commodities toward specialized, higher-value wood products that are less susceptible to the price-suppressing effects of liberalization.

7.3. Limitations

This study relies on country-level HS-6 data, which limits the observation of firm-level dynamics underlying the extensive margin. Although the empirical strategy addresses endogeneity concerns, RTA formation remains non-random, and unobserved factors may still influence the estimates. Future research could employ firm-level data and integrate environmental indicators to assess whether RTA-induced trade expansion in forest products is compatible with long-term sustainability goals.