Climate policies and adaptive capacities: a cross-analysis of sectoral, intersectoral, and international actions

1 Introduction

The intensification of climate change has become one of the major challenges facing sustainable development. The increasing frequency and severity of droughts, floods, heatwaves, wildfires, and sea-level rise highlight the growing vulnerability of economic, social, and environmental systems worldwide. Although climate-related risks affect all countries, their consequences are often more severe where financial resources, institutional capacities, and public infrastructure remain limited. In this context, adaptation readiness; defined as the economic, institutional, and social capacity to mobilize investment and translate it into adaptation action; is a central component of countries' resilience to future climate risks.

In response to these challenges, governments and international organizations have progressively adopted a broad range of climate policies intended to strengthen the conditions supporting adaptation (Intergovernmental Panel on Climate Change, 2022a). These policies can be classified into three main categories. Sectoral policies target specific areas such as agriculture, water, health, energy, transport, and infrastructure. Cross-sectoral policies seek to coordinate objectives and instruments across several policy domains in order to address the multidimensional nature of climate risks. International policies operate through multilateral agreements, international commitments, climate-finance mechanisms, technical cooperation, and the transfer of knowledge and technology (Saud et al., 2020; Scordato and Gulbrandsen, 2024).

The literature identifies different mechanisms through which these policy categories may be associated with adaptation readiness. Cross-sectoral policies may improve coherence, reduce institutional fragmentation, and account for interdependencies among sectors (England et al., 2018; Schoenefeld et al., 2022; Devine et al., 2024). Sectoral policies may respond more directly to specific vulnerabilities and benefit from specialized technical expertise, although their implementation in isolation may create inconsistencies or transfers of risk across policy domains (Barbera et al., 2021). International mechanisms may facilitate access to finance, technology, expertise, and knowledge, particularly in countries with limited domestic resources, but their contribution depends on national implementation capacity and the effective translation of international commitments into domestic action.

Despite the growing literature on climate governance, relatively few studies directly compare the associations of sectoral, cross-sectoral, and international climate policies with changes in adaptation readiness. This gap is important because these associations may differ according to countries' economic conditions. National income affects governments' ability to finance infrastructure, develop administrative capacity, support innovation, and sustain the implementation of climate policies over time (Acemoglu and Robinson, 2002; Hallegatte and Dumas, 2009; Fallahi et al., 2024). At the same time, countries that already possess relatively advanced institutions and infrastructure may experience smaller marginal gains from the adoption of additional policies. The moderating role of income may therefore differ across policy categories rather than uniformly strengthening their contribution.

Against this background, this study examines the comparative association of sectoral, cross-sectoral, and international climate-policy adoption with annual changes in adaptation readiness across 59 countries over the period 2002–2022. Adaptation readiness is measured using readiness_delta, defined as the annual first difference of the ND-GAIN Readiness Index. The analysis therefore focuses on improvements or deteriorations in the economic, governance, and social conditions supporting adaptation rather than on countries' absolute readiness levels. It also investigates whether the relationships between the three policy categories and readiness_delta vary with national income.

The study tests four main hypotheses. First, cross-sectoral policies are expected to display the most consistent positive association with annual improvements in readiness because they facilitate coordination across interdependent policy domains. Second, sectoral policies are expected to be positively associated with readiness improvements, although their contribution may be weakened by institutional fragmentation. Third, the association of international policies with readiness is expected to be stronger at lower levels of national income, where access to international finance, technology, and knowledge may generate larger marginal gains. Fourth, national income is expected to moderate these relationships, but the direction and magnitude of this moderation may differ across policy categories.

This study makes two main contributions. First, it provides a comparative international analysis of sectoral, cross-sectoral, and international climate policies rather than treating climate-policy adoption as a homogeneous form of intervention. Second, it examines how the association between each policy category and annual changes in readiness varies with national income. The empirical strategy combines joint fixed-effects OLS and seemingly unrelated regression models with several robustness analyses, including a dynamic pooled Mundlak specification, fixed-effects quantile regressions, Difference-GMM estimation, and a sensitivity analysis based on the OECD Climate Actions and Policies Measurement Framework. By identifying which forms of policy are most consistently associated with improvements in adaptation readiness, the study provides practical evidence for governments, development partners, and international organizations seeking to design more coherent and context-specific adaptation strategies.

The remainder of the article is organized as follows. Section 2 presents the theoretical framework and reviews the relevant literature. Section 3 describes the data and empirical methodology. Section 4 reports the main results and robustness analyses. Section 5 discusses the findings and their policy implications. The final section concludes.

2 Literature review

The literature on climate policies and adaptive capacity has expanded considerably over recent decades, driven by the intensification of climate-related risks and the need to identify public interventions capable of strengthening resilience over the long term. However, previous studies have largely examined the effectiveness of climate policies as an undifferentiated whole, without distinguishing between policy categories or accounting for the institutional heterogeneity that conditions their implementation. Existing studies show that a country's adaptive capacity depends not only on its exposure to climate risks, but also on its economic resources, level of human development, and ability to design and implement coherent public policies (Bostrom, 2022; Intergovernmental Panel on Climate Change, 2022b; Sarvestani and Millar, 2024). From an economic perspective, climate adaptation involves externalities, underinvestment, financing constraints, and coordination failures among public and private actors. The benefits of adaptation investments often extend beyond those who bear their costs, affect multiple sectors, and materialize over relatively long time horizons. Public intervention is therefore required to guide investment, reduce vulnerabilities, and coordinate actions whose effects are closely interconnected.

The analysis of these mechanisms rests primarily on two complementary theoretical frameworks. The first is North (1990)'s institutional theory, further developed by Acemoglu and Robinson (2002). This approach holds that institutions shape incentives and determine the ability of public actors to mobilize available resources, enforce adopted decisions, and translate them into economic and social outcomes. Applied to climate adaptation, it explains why the formal existence of a policy does not automatically guarantee its effectiveness: its translation into concrete action depends on the administrative, technical, and financial capacities required for implementation. The second framework is Ostrom (2010)'s theory of polycentric governance, according to which complex and systemic problems are more effectively addressed when several relatively autonomous yet interdependent decision-making centers coordinate their interventions. In the climate context, this approach highlights the need to connect sectors, territorial levels, and categories of actors affected by risks that simultaneously influence agriculture, energy, health, water resources, infrastructure, and local development. Together, these two frameworks provide the general theoretical foundation for examining the mechanisms associated with sectoral, international, and cross-sectoral climate policies.

2.1 Sectoral climate policies and adaptation capacity

Within this broader framework, sectoral policies represent a first form of intervention. Their functioning can be interpreted through the logic of functional specialization derived from the division of labor, according to which concentrating skills and resources within a clearly defined area enhances expertise, reduces certain information costs, and facilitates task implementation. Applied to climate action, this logic explains why policies targeting agriculture, energy, transport, health, water, or buildings may provide responses tailored to the vulnerabilities of each sector. Such policies may, for example, support irrigation, promote drought-resistant crops, strengthen health early-warning systems, or protect infrastructure exposed to extreme events. Their specialized nature may also facilitate the definition of measurable objectives, data collection, and programme monitoring. Schmidt et al. (2008) show that sectoral targeting can simplify policy administration, improve the availability of information required for monitoring, and facilitate the diffusion of appropriate technologies when the number of actors involved remains relatively limited. This logic is consistent with institutional theory, since the effectiveness of such interventions depends on the existence of specialized technical and administrative capacities.

Specialization may nevertheless generate adverse effects when it develops into institutional compartmentalization. Public agencies organized around specific mandates may prioritize their own objectives, resources, and instruments without adequately considering the consequences of their decisions for other sectors. Fragmented public action may therefore lead to duplicated expenditure, regulatory inconsistencies, or transfers of risk across sectors. A policy supporting agricultural production may, for instance, increase pressure on water resources, while an energy intervention may create new territorial vulnerabilities. Schmidt et al. (2008) also identify risks of incomplete coverage, reduced economic efficiency, and the displacement of activities toward unregulated sectors. Such effects may result in maladaptation, understood as responses that temporarily reduce vulnerability while generating new risks over the medium or long term (Abomo et al., 2024; Birchall et al., 2025; Hickel and Kallis, 2020). Moreover, an increase in the number of adopted policies does not necessarily ensure comprehensive coverage or proportionate outcomes. Nascimento et al. (2022) show that substantial gaps persist across several sectors despite the expansion in both the number and coverage of climate policy instruments.

2.2 International climate policies and adaptation capacity

International climate policies, by contrast, respond to the transnational nature of climate change and to the considerable disparities in resources across countries. Their underlying logic can be explained through the theory of global public goods and collective action. Because the climate is a global public good, the benefits generated by one country's efforts cannot be confined to that country alone, whereas the costs of action are largely borne at the national level. This creates an incentive for free-riding, whereby each state may seek to limit its own contribution while benefiting from measures adopted by others. Multilateral agreements, nationally determined contributions, national adaptation plans, climate finance mechanisms, and technical cooperation programmes seek to address this problem by coordinating commitments and organizing transfers of financial resources, technologies, and knowledge. These arrangements may ease the financing constraints faced by countries with limited domestic resources and facilitate the diffusion of climate-related standards, expertise, and innovations.

However, collective-action theory also helps explain why international policies may produce limited effects. In the absence of a supranational authority endowed with effective enforcement powers, agreements frequently rely on voluntary or weakly binding commitments. Countries may therefore adopt unambitious targets, delay implementation, or comply only partially. Proost (2026) analyses this problem through the prisoner's dilemma and shows that international agreements may result in a collective level of effort below the social optimum, since each state has an incentive to benefit from the actions of others while limiting its own costs. In addition to these cooperation difficulties, international policies face significant challenges of domestic implementation. Countries must possess the resources required to prepare projects, access funding, comply with the technical requirements of international partners, and ensure programme continuity. Where such capacities are insufficient, commitments may remain largely declaratory. Abugnaba-Abanga et al. (2026) show that the formal integration of climate objectives into Ghana's national health policies did not produce effective preparedness at the local level because of insufficient funding, limited human resources, and weak cross-sectoral coordination. Institutional theory therefore complements collective-action theory in this context: international cooperation may facilitate access to resources, but the translation of those resources into outcomes depends on domestic implementation capacities.

2.3 Cross-sectoral climate policies and adaptation capacity

In response to the limitations of sectoral compartmentalization and the difficulties involved in translating international commitments into concrete action, a growing body of literature has emphasized cross-sectoral policies. Their functioning can be explained through the theory of policy integration, according to which cross-cutting problems require the objectives, instruments, and decision-making processes of several sectors to be made mutually compatible. This approach directly extends the logic of polycentric governance: the purpose is not to eliminate sectoral administrations, but to organize their interdependencies and establish mechanisms of cooperation among multiple decision-making centers. In the climate context, cross-sectoral policies seek to coordinate interventions relating to agriculture, energy, health, water, infrastructure, and spatial planning. They may reduce inconsistencies across policies, prevent duplicated expenditure, and promote a more coherent allocation of resources. Murthy (2025) argues that cross-sectoral collaboration makes it possible to pool the expertise, resources, and knowledge of public authorities, businesses, non-governmental organizations, and local communities, while fostering innovation and greater coherence in climate action.

Empirical studies suggest that cross-sectoral approaches can improve information flows, collective learning, investment coherence, and the anticipation of systemic risks (Adger, 2000; Schoenefeld et al., 2022; Schuller, 2016). Combining expertise and instruments from several areas of public policy may generate more comprehensive responses than those developed by a single administration acting in isolation. Such integration is particularly relevant when the causes and consequences of climate risks extend beyond the boundaries of any one sector. Botah (2023) shows that policy integration can reduce fragmentation, inconsistencies, and conflicts among sectoral objectives. Similarly, Kalpakchiev et al. (2025) emphasize that cross-sectoral coherence can connect otherwise fragmented public priorities and enable more effective use of existing instruments when an environmental challenge simultaneously affects several policy domains.

These benefits do not, however, imply that cross-sectoral policies are automatically superior. Policy integration theory emphasizes that coordination is both complex and costly. Public administrations may pursue divergent objectives, operate according to different time horizons, and defend their institutional mandates or budgetary resources. Integration therefore requires consultation procedures, a clear allocation of responsibilities, shared objectives, and mechanisms for resolving conflicts. It may slow decision-making, increase administrative costs, or result in overly general compromises. Botah (2023) shows that, despite its potential, policy integration remains constrained by implementation complexity, insufficient political commitment, and operational difficulties. Murthy (2025) similarly identifies conflicting priorities, institutional barriers, resource limitations, and inadequate communication as major obstacles to cross-sectoral collaboration. Kalpakchiev et al. (2025) further show that competing problem definitions, sectoral priorities, and established interests may prevent integrated solutions from reaching the policy agenda. The literature therefore cautions against treating sectoral policies as inherently ineffective and cross-sectoral policies as systematically optimal. Their relevance depends on the nature of the problem, the resources available, and the ability to establish substantive coordination rather than merely formal integration.

2.4 role of economic resources

These mechanisms also highlight the importance of economic resources. High-income countries generally benefit from greater fiscal space, better access to finance, and more developed technical capacities. They are therefore better able to finance specialized sectoral interventions, comply with the requirements of international mechanisms, and bear the administrative costs associated with cross-sectoral coordination. By contrast, low-income countries often face competing immediate needs and budgetary constraints that restrict the implementation of policies even when they have been formally adopted. Several studies consequently report a positive association between income, human development, and adaptive capacity (Fallahi et al., 2024; Sarkodie et al., 2022). Income alone does not, however, guarantee policy effectiveness; rather, it constitutes a resource that facilitates policy financing, coordination, and continuity.

Overall, the literature identifies distinct mechanisms associated with sectoral, cross-sectoral, and international climate policies. Sectoral policies may benefit from specialization and technical expertise but can also reinforce institutional fragmentation, while international policies may facilitate access to finance, technology, and knowledge but remain constrained by collective-action problems and domestic implementation capacities. Cross-sectoral policies may improve policy coherence and coordination across interdependent domains, although their implementation can involve substantial administrative costs and institutional conflicts. Despite these differences, few studies directly compare how these three policy categories are associated with annual changes in adaptation readiness or examine whether these relationships vary with national income. This study addresses this gap using data for 59 countries over the period 2002–2022. It expects cross-sectoral policies to display the most stable positive association with adaptation capacity, while the relationships associated with sectoral and international policies may be more sensitive to implementation conditions. National income may strengthen these relationships by providing greater financial and institutional resources, but it may also reduce marginal policy gains in countries that already possess relatively developed adaptive capacities. The direction and magnitude of this moderating role therefore remain empirical questions.

3 Methodology

3.1 Data and Variables

This study is based on an unbalanced panel of 59 countries (Table 2) observed over the period 2002–2022. This period covers both the consolidation of international climate governance following the Kyoto Protocol and the acceleration of national climate action after the adoption of the Paris Agreement in 2015. The data are drawn from internationally recognized sources covering climate-policy adoption, adaptation readiness, national income, and socioeconomic conditions. The sources and measurement of the variables used in the empirical analysis are described below.

The dependent variable is derived from the Notre Dame Global Adaptation Initiative (ND-GAIN) Readiness Index. The original Readiness Index measures a country's ability to mobilize economic, institutional, and social resources and convert them into effective adaptation action. Unlike vulnerability indicators, which focus primarily on exposure and sensitivity to climate-related risks, readiness captures the conditions that facilitate the mobilization and implementation of adaptation investment. It comprises three equally weighted dimensions: economic readiness, which reflects the capacity of the business environment to support adaptation-related investment; governance readiness, which captures the institutional conditions facilitating the implementation of such investment; and social readiness, which includes social inequality, access to information and communication technologies, educational attainment, and innovation. This multidimensional structure provides a broad and internationally comparable measure of the conditions supporting adaptation (Otto et al., 2021; Sarkodie et al., 2022; Sarvestani and Millar, 2024).

The empirical analysis does not use the absolute level of the Readiness Index as its dependent variable. Instead, it uses readiness_delta, defined as the annual first difference of the index:

readiness_deltait=Readinessit-Readinessi,t-1.(1)

A positive value indicates an annual improvement in the economic, governance, and social conditions supporting adaptation, whereas a negative value indicates a deterioration. A value equal to zero indicates no annual change. Consequently, the dependent variable measures annual changes in adaptation readiness rather than countries' absolute level of adaptive capacity. The values represent changes in index points and should not be interpreted as percentage growth rates.

Climate policies constitute the main explanatory variables. The baseline indicators are constructed from the Climate Change Laws of the World (CCLW) database, which provides detailed information on national climate laws, policies, plans, strategies, and other policy documents. Based on their scope and institutional level, the identified policies are classified into three categories: sectoral, cross-sectoral, and international. Sectoral policies target a clearly identified area, such as agriculture, water, energy, transport, health, forests, or buildings. Cross-sectoral policies apply across several policy domains or establish economy-wide coordination and governance arrangements. International policies refer to national measures adopted in connection with international cooperation, commitments, financing mechanisms, or multilateral climate frameworks.

The three policy variables are coded as binary country-year indicators, with a value of one indicating the adoption or presence of a policy in the corresponding category and a value of zero otherwise. These indicators therefore measure policy adoption rather than policy stringency, implementation quality, or policy effectiveness. Their disaggregation allows the analysis to compare the associations of different forms of climate intervention with annual changes in adaptation readiness. This classification is consistent with the literature emphasizing differences in sectoral coverage, cross-sectoral coordination, and international climate action (Nascimento et al., 2022; Nuta, 2025).

Gross national income (GNI) is introduced as the moderating variable to examine whether national economic resources condition the relationship between climate-policy adoption and changes in readiness. The data are obtained from the World Bank's World Development Indicators. Before the estimation, GNI is centered and standardized. The policy–income interaction terms are then constructed using the standardized GNI variable. This procedure facilitates the interpretation of the direct policy coefficients, which represent their estimated associations at the average level of national income, and reduces non-essential collinearity between the constitutive terms and their interactions. The inclusion of GNI is consistent with evidence showing that economic resources influence countries' capacity to finance adaptation, develop institutions, and mobilize investment (Sarkodie et al., 2022).

Two control variables are included to account for time-varying socioeconomic conditions that may simultaneously influence climate-policy adoption and adaptation readiness. The first is GDP growth (annual percentage), which captures short-term changes in national economic activity. The second is urban population growth (annual percentage), which accounts for demographic and spatial pressures associated with urbanization. Both variables are obtained from the World Bank's World Development Indicators.

As an additional sensitivity analysis, the baseline policy indicators are replaced with the sectoral, cross-sectoral, and international policy stringency measures from the OECD Climate Actions and Policies Measurement Framework (CAPMF). The CAPMF indicators are measured on a common scale ranging from 0 to 10, with higher values indicating greater policy stringency (Nachtigall et al., 2024). Unlike the binary CCLW indicators, they capture the extent to which climate-policy instruments prescribe, encourage, or enable climate action. Because CAPMF data availability is more limited, this sensitivity analysis is conducted on a separate sample of 49 countries and should not be interpreted as using exactly the same sample as the baseline analysis. The CAPMF indicators are employed to assess whether the main conclusions remain consistent when climate policies are measured using an alternative source and a continuous measure of policy stringency.

The descriptive statistics and the correlation matrix of the variables are provided in the Appendix with Tables 8 and 9.

3.2 Model specification

The objective of this study is to examine the associations between sectoral, cross-sectoral, and international climate-policy adoption and annual changes in countries' adaptation readiness. The dependent variable is readiness_delta_*it*, defined as the annual first difference of the ND-GAIN Readiness Index for country i in year t. The empirical analysis therefore focuses on annual improvements or deteriorations in the economic, governance, and social conditions supporting adaptation rather than on countries' absolute readiness levels.

The empirical analysis relies primarily on the Seemingly Unrelated Regression model originally developed by Zellner (1962). The SUR framework jointly estimates several linear equations and allows their disturbances to be contemporaneously correlated. This approach is appropriate for the present study because sectoral, cross-sectoral, and international climate policies operate within the same national institutional and socioeconomic environments. Unobserved shocks associated with one policy category may therefore also influence the relationships estimated for the other policy categories.

The SUR system contains three policy-specific equations. Each equation includes the relevant climate-policy indicator, standardized gross national income, the interaction between the policy indicator and income, the control variables, country effects, and year effects. The system is specified as follows:

readiness_deltait=αS+βSSectoralit+θSGNIit* +δS(Sectoralit×GNIit*) +γS′Xit+μS,i+λS,t+εS,it.(2)

readiness_deltait=αC+βCCrossSectoralit+θCGNIit* +δC(CrossSectoralit×GNIit*) +γC′Xit+μC,i+λC,t+εC,it.(3)

readiness_deltait=αI+βIInternationalit+θIGNIit* +δI(Internationalit×GNIit*) +γI′Xit+μI,i+λI,t+εI,it.(4)

In these equations, Sectoral_*it*, CrossSectoral_*it*, and International_*it* are binary variables equal to one when country i has adopted a policy belonging to the corresponding category in year t, and zero otherwise. These indicators measure policy adoption rather than policy stringency, implementation quality, or policy effectiveness.

The variable GNIit* denotes standardized gross national income. The vector X_*it* contains annual GDP growth and annual urban population growth. The terms μ_{*k, i*} represent country effects, whereas λ_{*k, t*} represent year effects, with k∈{S, C, I}. Country effects account for time-invariant national characteristics, while year effects control for common shocks affecting all countries during the same period.

The terms ε_{*S, it*}, ε_{*C, it*}, and ε_{*I, it*} are the disturbances associated with the three equations. Following the SUR framework developed by Zellner (1962), these disturbances are allowed to be correlated across equations for the same country-year observation:

Cov(εS,itεC,itεI,it) =Σ =(σSSσSCσSIσSCσCCσCIσSIσCIσII).(5)

The diagonal elements of Σ represent the variance of the disturbance in each equation. The off-diagonal elements capture the contemporaneous covariance of unobserved factors across the three policy equations. When these covariances differ from zero, estimating the equations jointly may improve efficiency relative to estimating each equation independently (Zellner, 1962).

3.2.1 SUR estimation and statistical inference

The estimation follows the feasible generalized least-squares procedure proposed by Zellner (1962). In a first stage, the equations are estimated separately to obtain preliminary residuals. These residuals are used to estimate the cross-equation disturbance covariance matrix Σ. In a second stage, the complete system is estimated jointly using the estimated inverse covariance matrix.

The SUR estimator can be represented as:

β^SUR=[X′(Σ^-1⊗I)X]-1 ×X′(Σ^-1⊗I)y,(6)

where y stacks the dependent variables of the three equations, X contains the corresponding explanatory variables, Σ^ is the estimated cross-equation disturbance covariance matrix, I is an identity matrix, and ⊗ denotes the Kronecker product.

Because the observations are repeated over time within countries, statistical inference must account for possible heteroskedasticity and within-country dependence. The reported inference therefore relies on a wild cluster bootstrap implemented at the country level, following Cameron et al. (2008) and MacKinnon and Webb (2017). This procedure provides bootstrap-based p-values that are robust to heteroskedasticity and within-country error dependence. It does not, however, eliminate possible endogeneity, reverse causality, or omitted time-varying factors.

3.2.2 Interaction terms

The interaction terms examine whether the association between each climate-policy category and annual changes in readiness varies according to national income. The use of multiplicative terms follows the interaction framework discussed by Cohen (1978). Their interpretation is based on the recommendations of Brambor et al. (2006), who emphasize that the constitutive variables must remain in the equation and that the conditional association should be evaluated across relevant values of the moderating variable.

Gross national income is centered and standardized before the interaction terms are constructed:

GNIit*=GNIit-GNI¯sGNI,(7)

where GNI¯ represents the sample mean and s_*GNI* represents the sample standard deviation. A value of zero therefore corresponds to the average income level in the estimation sample, while values of −1 and 1 correspond to one standard deviation below and above the mean, respectively.

The three interaction variables are constructed as follows:

SectoralGNIit=Sectoralit×GNIit*,(8)

CrossSectoralGNIit=CrossSectoralit×GNIit*,(9)

InternationalGNIit=Internationalit×GNIit*.(10)

Standardizing GNI facilitates interpretation and places the income variable on a common scale. It does not, by itself, make the interaction term orthogonal to its constituent variables. The policy indicator, GNI, and their product are therefore all retained in each equation, in accordance with the hierarchical specification principle discussed by Brambor et al. (2006).

The coefficients β_*S*, β_*C*, and β_*I* represent the estimated associations of the corresponding policies with readiness_delta when GNIit=0, that is, at the average income level. The interaction coefficients δ_S, δ_C, and δ_I* indicate how these policy associations change when standardized GNI increases by one unit.

The coefficients θ_*S*, θ_*C*, and θ_*I* represent the association between GNI and readiness_delta when the corresponding policy indicator equals zero. Because the policy indicators are binary and are not centered, these coefficients should not be interpreted as the association of GNI at the average level of policy adoption.

3.2.3 Marginal associations

Following Brambor et al. (2006), the conditional association of each policy is calculated by taking the partial derivative of the corresponding equation with respect to its policy indicator. At a given value g of standardized GNI, the marginal associations are:

MESectoral(g)=βS+δSg,(11)

MECrossSectoral(g)=βC+δCg,(12)

MEInternational(g)=βI+δIg.(13)

The marginal associations are evaluated at low income (g = −1), average income (g = 0), and high income (g = 1). This procedure determines whether the association between policy adoption and annual changes in adaptation readiness strengthens, weakens, or changes direction across national income levels.

3.2.4 Cross-equation Wald tests

The SUR framework allows formal comparisons of coefficients across the three equations. These comparisons are conducted using Wald tests, based on the general hypothesis-testing framework developed by Wald (1943).

The pairwise equality hypotheses for the direct policy coefficients are:

H0SC:βS=βC,(14)

H0SI:βS=βI,(15)

H0CI:βC=βI.(16)

The joint equality hypothesis is:

H0Joint:βS=βC=βI.(17)

The equality of the interaction coefficients is also examined:

H0Interaction:δS=δC=δI.(18)

Rejecting a pairwise null hypothesis indicates that two policy categories display statistically different conditional associations with annual changes in adaptation readiness. Rejecting the joint hypothesis indicates that the three direct policy coefficients are not all equal. These tests compare conditional associations and do not establish that one policy category has a stronger causal effect than another.

4 Empirical results

4.1 Basic results

The main results are presented in Tables 1 and 2. Table 1 examines, across 13 columns, the relationship between sectoral, cross-sectoral, and international climate policies and adaptive capacity. Columns 1, 4, 7, and 10 report the OLS estimates, which simultaneously include the three climate policy indicators, their interactions with national income, and the control variables. Columns 2, 5, 8, and 11 correspond to the SUR A model, in which each policy is treated as the main explanatory variable in a separate equation. Columns 3, 6, 9, and 12 present the SUR B model, which additionally includes the other two policy categories as cross-controls (Zellner, 1962). The first three columns report the baseline policy coefficients and their interactions with income, whereas Columns 4 to 12 present the marginal effects at low, average, and high income levels. Finally, Column 13 reports the variance inflation factors. This structure makes it possible to assess the robustness of the results across estimation methods, the simultaneous inclusion of the other policy categories, and countries' income levels.

Table 1. Climate policies and adaptive capacity: baseline and marginal effects.

Columns 1, 4, 7, and 10 report OLS estimates. Columns 2, 5, 8, and 11 report SUR A estimates, in which each policy category is successively treated as the main explanatory variable in a separate equation. Columns 3, 6, 9, and 12 report SUR B estimates, which additionally include the other two policy categories as cross-controls. Columns 4–6, 7–9, and 10–12 report marginal effects at low, average, and high GNI levels, respectively. Column 13 reports variance inflation factors (VIF). The diagnostic tests in the lower panel refer to the joint OLS specification. Significance levels are based on robust inference clustered at the country level. ***p < 0.01, **p < 0.05, and *p < 0.10. Source: Authors' calculations.

Table 2. Wald tests of coefficient equality and joint significance.

Entries are Wald test statistics. “Overall” reports tests on the baseline coefficients or interaction terms, while the Low, Average, and High GNI columns report pairwise tests of the corresponding marginal effects. ***p < 0.01, **p < 0.05, and *p < 0.10. Source: Authors' calculations.

The Breusch–Pagan and Wooldridge tests indicate the presence of heteroskedasticity and within-country serial correlation, respectively (Breusch and Pagan, 1979; Drukker, 2003), while the Jarque–Bera test reveals substantial non-normality of the residuals (Jarque and Bera, 1987). To account for these violations, statistical inference relies on a wild cluster bootstrap, which resamples entire countries in order to preserve within-country dependence in the simulated distribution of the test statistics (Cameron et al., 2008). The Pesaran CD test does not indicate significant cross-sectional dependence (Pesaran, 2004). Finally, several variance inflation factors associated with the interaction terms exceed the conventional threshold of 10. The variables were therefore mean-centered before constructing the interaction terms in order to reduce collinearity between the main effects and their interactions (Cohen, 1978; Brambor et al., 2006).

The main finding from Table 1 is that cross-sectoral climate policies display the most stable positive association with adaptive capacity. Their coefficients are positive and statistically significant across all three specifications, with values close to 0.030. A one-unit increase in the cross-sectoral policy index is therefore associated with an increase of approximately 0.030 units in readiness_delta, all else being equal. The stability of this coefficient across the OLS, SUR A, and SUR B models indicates that the result is not driven by a particular econometric specification.

The interaction between cross-sectoral policies and national income is, however, negative and statistically significant in all three models, with coefficients ranging from −0.009 to −0.010. The positive association of these policies therefore decreases as income rises, without disappearing. Indeed, the marginal effects remain positive and statistically significant at all three income levels. At the high-income level, they still amount to 0.022 in the OLS model, 0.023 in SUR A, and 0.022 in SUR B. This finding points to diminishing marginal gains rather than the disappearance of the effect, which may reflect the fact that higher-income countries already possess more developed institutional and infrastructural capacities.

Sectoral climate policies also exhibit an overall positive association with adaptive capacity, although the result is more sensitive to the estimation method. In the OLS model, the coefficient is positive but statistically insignificant, at 0.016. It becomes positive and significant in SUR A and SUR B, with coefficients of 0.023 and 0.020, respectively. Its continued significance in SUR B indicates that sectoral policies retain an independent contribution even after controlling for cross-sectoral and international policies.

The interaction between sectoral policies and national income is not statistically significant in the OLS specification. It becomes negative in SUR A and SUR B, with coefficients of −0.011 and −0.010, but is significant only at the 10% level. The moderating role of income is therefore less robust for sectoral policies than for cross-sectoral policies.

International climate policies display a different pattern. Their baseline coefficients are small and statistically insignificant across the three specifications, ranging from 0.006 to 0.007. This finding does not provide evidence of a systematic association between international policies and readiness_delta across the full sample. Their interactions with income are, by contrast, negative and strongly significant in all three models, with coefficients ranging from −0.022 to −0.026.

The marginal effects show that international climate policies are positively associated with adaptive capacity in low-income countries, with a coefficient of 0.022 that is significant at the 5% level in both SUR specifications. This association weakens and becomes statistically insignificant at the average income level. At the high-income level, the estimated effects become negative but remain statistically insignificant. The appropriate conclusion is therefore not that international policies reduce adaptive capacity in richer countries, but rather that their additional contribution gradually diminishes as income rises.

Table 2 complements these findings by testing the equality of the coefficients across the three policy categories, as well as their joint statistical significance (Wald, 1943). The joint nullity test of the three interaction terms with lnGNI is not significant in the OLS model, with a test statistic of 7.56, but is strongly significant in SUR A and SUR B, with values of 20.77 and 17.62, respectively. The moderating role of income therefore becomes more apparent when interdependence across the equations is taken into account (Zellner, 1962).

The pairwise comparisons reveal no statistically significant difference between sectoral and cross-sectoral policies, whether their baseline coefficients or their marginal effects at different income levels are considered (Wald, 1943; Brambor et al., 2006). Although cross-sectoral policies display larger and more stable coefficients, their statistical superiority over sectoral policies cannot therefore be established.

The difference between sectoral and international policies becomes significant at the high-income level in the SUR A model. The differences between cross-sectoral and international policies are more pronounced: they are not significant at the low-income level but become significant at the average and high-income levels in both SUR specifications (Wald, 1943). As income rises, cross-sectoral policies thus retain a positive and statistically significant association with adaptive capacity, whereas the contribution of international policies becomes statistically indistinguishable from zero.

The joint equality test of the three baseline policy coefficients is not statistically significant in the OLS model or in the overall SUR B specification, but it is significant in SUR A (Wald, 1943; Zellner, 1962). When the effects are examined by income level, the equality hypothesis is also rejected at the average and high-income levels in the SUR models. These results indicate that the three categories of climate policies are not fully interchangeable, particularly in countries with higher income levels.

Finally, the joint nullity test of the direct effects and interaction terms is strongly significant across all three models, with test statistics of 21.21 for OLS, 82.41 for SUR A, and 83.51 for SUR B (Wald, 1943). The hypothesis that climate policies and their interactions with income are collectively unrelated to adaptive capacity is therefore rejected.

Overall, Tables 1 and 2 show that the three categories of climate policies are not uniformly associated with adaptive capacity. Cross-sectoral policies produce the most robust results: their association with readiness_delta remains positive and statistically significant across all specifications and income levels. Sectoral policies are also positively associated with adaptive capacity, but this result is more dependent on the estimation method. International policies, by contrast, appear to be primarily associated with improvements in adaptive capacity in low-income countries. These findings highlight the importance of cross-sectoral coordination and suggest that international climate policies may be especially relevant for countries with more limited financial and institutional resources.

4.2 Test sensitivity

As a sensitivity analysis, the climate policy indicators constructed from the CCLW database are replaced with the sectoral, cross-sectoral, and international policy indicators from the OECD Climate Actions and Policies Measurement Framework (Climate Actions and Policies Measurement Framework [CAPMF], 2024), covering a sample of 49 countries (Nachtigall et al., 2024). The equations are re-estimated using a Mundlak specification, which accounts for the potential correlation between the explanatory variables and unobserved country-level heterogeneity by including the country-specific means of the time-varying regressors (Mundlak, 1978). The model reports the simple effects of each climate policy category, as well as their marginal effects evaluated at low, average, and high levels of GNI (Brambor et al., 2006). A one-period lag of the dependent variable, readiness_delta_*t*−1, is also included to account for the persistence of adaptive capacity over time (Wooldridge, 2010). This analysis examines whether the main findings remain valid when both the measurement of climate policies and the estimation method are modified. To ensure comparability, the sensitivity analysis retains the same dependent variable, control variables, income interactions, and standardization procedure and, as far as possible, uses a common sample of countries and years.

The CAPMF–Mundlak sensitivity analysis (Table 3) provides partial support for the baseline findings. First, the positive association between cross-sectoral climate policies and adaptive capacity is confirmed in the between-country component of the Mundlak model (Mundlak, 1978). Countries with higher average levels of cross-sectoral policy stringency exhibit significantly higher readiness_delta, and this result remains stable in the lagged specification. The Wald tests (Table 4) also confirm that cross-sectoral and international policies have statistically different effects, particularly in the within-country dimension and at average and high levels of GNI (Wald, 1943). This supports the baseline conclusion that cross-sectoral policies display a more favorable and robust association with adaptive capacity than international policies.

Table 3. CAPMF Mundlak results.

The table reports CAPMF estimates obtained using the Mundlak specification. “Between” denotes the effects associated with the country-specific means of the explanatory variables, whereas “Within” denotes the effects associated with deviations from these means. Low, Average, and High GNI report marginal effects evaluated at the corresponding income levels. Lag 1 reports estimates obtained using the climate policy indicators lagged by one period. VIF denotes the variance inflation factor. The table reports coefficients only. ***p < 0.01, **p < 0.05, and *p < 0.10.

Table 4. CAPMF–Mundlak Wald Tests for between and within effects.

Entries are Wald test statistics. The “Between” column reports tests based on the country-specific means of the explanatory variables, whereas the “Within” column reports tests based on within-country deviations from those means. Low, average, and high GNI correspond respectively to one standard deviation below the mean, the sample mean, and one standard deviation above the mean. The null hypothesis in each pairwise test is that the corresponding policy effects are equal. The joint tests examine the simultaneous equality or nullity of the three policy coefficients or interaction terms.

However, the sensitivity analysis does not fully reproduce the baseline income-moderation result for cross-sectoral policies. Whereas the baseline OLS and SUR estimates indicated that the positive effect of cross-sectoral policies declined significantly with income, the corresponding Mundlak interaction is not robust and is weakly positive in the within-country component. Moreover, the marginal effects are not statistically significant at all three income levels (Brambor et al., 2006). The positive role of cross-sectoral policies is therefore confirmed, but the hypothesis of systematically diminishing marginal gains as income rises is not.

The positive contribution of sectoral policies is also not confirmed by the Mundlak specification (Mundlak, 1978). Although the baseline SUR estimates produced positive and significant coefficients, neither the between-country nor the within-country sectoral effect is statistically significant in the CAPMF model. This result is nevertheless consistent with the sensitivity already observed in the baseline analysis, where the sectoral coefficient was insignificant under OLS and depended more strongly on the estimation method. The sensitivity analysis therefore suggests that the sectoral result is less robust than the cross-sectoral result rather than providing clear evidence of an opposite effect.

For international climate policies, the Mundlak results confirm the absence of a positive systematic association across the full sample. Their between-country coefficient remains small and statistically insignificant, consistently with the baseline estimates. However, the positive marginal effect previously identified for low-income countries is not reproduced. Instead, the within-country international policy coefficient is negative and significant, including at average and high GNI levels. This finding is broadly consistent with the baseline observation that the contribution of international policies weakens as income rises, but it provides stronger evidence of a negative within-country association than the original OLS and SUR estimates. Since the policy–GNI interactions are not jointly significant in the Mundlak model, this pattern cannot be interpreted as robust evidence of systematic income moderation (Brambor et al., 2006; Wald, 1943).

Overall, the sensitivity analysis confirms the central distinction between cross-sectoral and international climate policies, as well as the greater robustness of the cross-sectoral policy result. It does not, however, fully confirm the positive effect of sectoral policies or the systematic moderating role of national income. The baseline conclusions are therefore robust with respect to the relative importance of cross-sectoral policies, but more sensitive regarding sectoral policies and income-dependent effects.

4.3 Robustness tests

Three robustness tests are presented in this subsection: a dynamic estimator with Mundlak, quantile regressions (QR) and the generalized method of moments (GMM). The additional robustness tests are motivated by the fact that the results presented above have some shortcomings which we investigate below.

4.3.1 Dynamic pooled Mundlak estimator

The first robustness check relies on a dynamic pooled Mundlak specification, which extends the baseline model by including the one-period lag of the dependent variable, readiness_delta_{*i, t*−1}, to account for the persistence of adaptive capacity over time (Wooldridge, 2010). The time-varying explanatory variables are decomposed into their country-specific means and their deviations from these means, thereby distinguishing between-country differences from within-country changes and allowing the regressors to be correlated with unobserved time-invariant country characteristics (Mundlak, 1978). The model is estimated on pooled country-year observations with common slope coefficients across countries and year fixed effects to control for common temporal shocks (Wooldridge, 2010). Unlike a dynamic fixed-effects error-correction model, this specification does not estimate an error-correction term or directly separate short-run from long-run effects. The coefficient on readiness_delta_{*i, t*−1} measures the persistence of adaptive capacity, while a coefficient below one in absolute value indicates a dynamically stable process (Wooldridge, 2010). The results provide partial support for the baseline findings: the positive between-country association of cross-sectoral climate policies remains statistically significant, whereas the negative within-country association of international policies is also preserved after controlling for the lagged dependent variable. By contrast, the effects of sectoral policies and most interactions with national income remain statistically insignificant, indicating that the comparatively favorable association of cross-sectoral policies is more robust than the sectoral and income-moderation results.

4.3.2 Quantile regressions

The second robustness check relies on quantile regressions (QR), initially developed by Koenker and Bassett (1978). While the OLS and SUR estimators describe the average conditional association between climate policies and adaptive capacity, they may conceal substantial heterogeneity across the distribution of the dependent variable. Quantile regression makes it possible to examine whether sectoral, cross-sectoral, and international climate policies are associated differently with countries located in the lower, median, and upper parts of the conditional distribution of readiness_delta. This distinction is particularly relevant in the present study because the effects of climate policies may differ between countries experiencing relatively weak improvements in adaptive capacity and those displaying stronger improvements. Quantile regression is also less sensitive than conditional-mean estimators to extreme observations and does not require normally distributed error terms, which is useful given the non-normality of the residuals identified in the baseline estimations. More generally, when the dependent variable displays substantial heterogeneity, quantile regression provides a more complete description of its conditional distribution than an estimator focusing exclusively on the conditional mean (Cade and Noon, 2003). To preserve comparability with the baseline models, the specification includes the three climate policy indicators, their interactions with national income, the control variables, and country and year effects. Following Binder and Coad (2011), the conditional quantile model can be written as:

Qτ(readiness_deltait∣Xit,αi,λt)=αi+λt+Xit′βτ,(19)

where readiness_delta_*it* denotes the change in adaptive readiness for country i in year t, X_*it* is the vector containing the sectoral, cross-sectoral, and international climate policy indicators, their interactions with GNI, and the control variables, α_*i* represents country-specific effects, and λ_*t* denotes year effects capturing common shocks. The vector β_τ contains the coefficients associated with the τ-th conditional quantile of readiness_delta. The estimator is obtained by solving the following asymmetric absolute-deviation minimization problem:

β^τ=argminβ∈ℝK∑i=1N∑t=1Tiρτ(readiness_deltait-αi-λt-Xit′β),(20)

where the quantile loss function is defined as:

ρτ(u)=u[τ-1(u<0)].(21)

Estimating the model at different quantiles therefore makes it possible to determine whether the magnitude, sign, and statistical significance of the climate-policy coefficients vary across the conditional distribution of adaptive-capacity changes.

4.3.3 Estimation by the generalized method of moments

The Threeth robustness check relies on dynamic panel estimators based on the generalized method of moments (GMM). This approach is used to address the potential endogeneity of climate policy indicators, reverse causality between climate policies and adaptive capacity, and the correlation between the lagged dependent variable and unobserved country-specific effects. The dynamic specification includes the one-period lag of readiness_delta, the sectoral, cross-sectoral, and international climate policy indicators, their interactions with national income, the control variables, and year effects. The baseline dynamic equation is written as follows:

readiness_deltait=α+ρreadiness_deltai,t-1+Pit′β +(Pit×GNIit)′γ+Zit′δ+μi+λt+εit,(22)

where P_*it* is the vector containing the sectoral, cross-sectoral, and international climate policy indicators, P_*it*×GNI_*it* denotes their interactions with national income, Z_*it* contains the control variables, μ_*i* represents unobserved time-invariant country-specific heterogeneity, and λ_*t* captures common year effects. The Difference-GMM estimator, introduced by Holtz-Eakin et al. (1988) and developed by Arellano and Bond (1991), removes the country-specific effect by transforming the model into first differences:

Δreadiness_deltait=ρΔreadiness_deltai,t-1+ΔPit′β +Δ(Pit×GNIit)′γ+ΔZit′δ+Δλt+Δεit.(23)

However, the lagged dependent variable remains correlated with the transformed error term because readiness_delta_{*i, t*−1} contains ε_{*i, t*−1}, implying that Cov(Δreadiness_delta_{*i, t*−1}, Δε_*it*)≠0. Difference GMM addresses this problem by using sufficiently lagged levels of the endogenous and predetermined variables as instruments for their first differences. Nevertheless, when the dependent variable and the climate policy indicators are highly persistent, their lagged levels may constitute weak instruments. For this reason, the analysis also considers the System-GMM estimator developed by Arellano and Bover (1995) and Blundell and Bond (1998), which jointly estimates an equation in levels and an equation in first differences:

System-GMM→{readiness_deltait=α+ρreadiness_deltai,t−1 +Pit′β+Zit′γ+μi+λt+εit,Δreadiness_deltait=ρΔreadiness_deltai,t−1 +ΔPit′β+ΔZit′γ+Δλt+Δεit.(24)

In the differenced equation, lagged levels are used as instruments, whereas lagged differences instrument the corresponding variables in the level equation. The validity of the GMM estimates depends on the relevance and exogeneity of these instruments. Hansen's J test is therefore reported to assess the joint validity of the instrument set. The Arellano–Bond tests for first- and second-order serial correlation are also presented. First-order correlation in the differenced residuals is expected, whereas the absence of second-order serial correlation is required for the moment conditions to remain valid. Because the sample contains only 49 countries, particular attention is paid to instrument proliferation. The lag depth is restricted, the instrument matrix is collapsed, and the number of instruments is kept below the number of countries, following the recommendations of Roodman (2009b,a). The preferred specification is estimated using the two-step System-GMM estimator with finite-sample-corrected robust standard errors. The resulting estimates make it possible to assess whether the main conclusions regarding sectoral, cross-sectoral, and international climate policies remain robust after accounting for persistence, potential reverse causality, and the endogeneity of the policy indicators.

The robustness estimations (Table 5) provide mixed but informative support for the baseline OLS and SUR results. The quantile regressions strongly confirm the positive association of cross-sectoral policies with adaptive capacity, as their coefficients remain positive and statistically significant throughout the conditional distribution of readiness_delta. They also support the positive sectoral-policy coefficients obtained in the SUR models, although this relationship was not significant in the joint OLS specification. The negative interaction between international policies and national income is particularly robust, remaining large and statistically significant at all estimated quantiles. The negative income moderation of cross-sectoral policies is only partially confirmed, since it is concentrated between the 25th and 75th quantiles, while the negative moderation previously associated with sectoral policies is not reproduced. By contrast, neither the dynamic Mundlak nor the Difference-GMM specification produces statistically significant contemporaneous policy coefficients or income interactions. Both dynamic estimators reveal a very high degree of persistence in readiness_delta, suggesting that part of the associations identified by the static OLS and SUR models may reflect the strong temporal inertia of adaptive capacity. Overall, the positive cross-sectoral result and the negative income moderation of international policies receive the strongest support, whereas the sectoral effect and the income moderation of cross-sectoral policies remain more sensitive to the estimation strategy.

Table 5. Robustness tests: dynamic Mundlak, Quantile regressions, and GMM.

Coefficient cells report estimated coefficients only. Dynamic Mundlak reports the within-country coefficients obtained from a dynamic pooled correlated-random-effects specification including country-specific means of the time-varying regressors, the one-period lag of readiness_delta, and year effects. Q10, Q25, Q50, Q75, and Q95 denote joint fixed-effects quantile regressions estimated at the 10th, 25th, 50th, 75th, and 95th conditional quantiles, respectively, with robust asymptotic inference. GMM reports estimates from a two-step Difference-GMM model using collapsed lag instruments. The AR(1) and AR(2) rows report the Arellano–Bond serial-correlation test statistics. First-order serial correlation in the differenced residuals is expected, whereas the absence of statistically significant second-order serial correlation supports the validity of the moment conditions. ***p < 0.01, **p < 0.05, and *p < 0.10.

These results are illustrated in Figure 1, which highlights substantial heterogeneity in the association between climate policy adoption and readiness_delta across the conditional distribution of adaptation readiness. At the average level of GNI, cross-sectoral policies display a positive and statistically significant association at all quantiles, with the largest coefficients observed at Q25 and Q50 before declining at Q75 and Q95. Sectoral policies are also positively associated with readiness_delta, with stronger effects around the middle of the distribution, particularly at Q50 and Q75, although the association becomes weaker at Q95. By contrast, international policies exhibit a positive and significant association only at Q10, while their coefficients are small and statistically insignificant across the remaining quantiles. Overall, the figure confirms that the positive association of cross-sectoral policies is more consistent across the readiness distribution than those of sectoral and international policies.

5 Discussion and recommendations

The results obtained for 59 countries over the period 2002–2022 indicate that cross-sectoral climate policies display the most robust and consistent positive association with annual improvements in adaptation readiness. However, the Wald tests do not reject the equality of the coefficients associated with cross-sectoral and sectoral policies. The results therefore do not establish that the magnitude of the cross-sectoral coefficient is statistically greater than that of the sectoral coefficient. The main advantage of cross-sectoral policies lies instead in the stability of their positive association across models and income levels. This finding supports Ostrom (2010)'s theory of polycentric governance, according to which complex risks are more effectively addressed when several decision-making centers coordinate their actions. It is also consistent with Botah (2023), Murthy (2025), and Kalpakchiev et al. (2025), who argue that policy integration improves information sharing, the joint mobilization of resources, and coherence across sectors. Governments should therefore strengthen cross-sectoral coordination by establishing permanent mechanisms connecting, in particular, the ministries responsible for agriculture, water, health, energy, infrastructure, finance, and spatial planning. These mechanisms should rely on shared objectives, clearly defined responsibilities, common information systems, and joint budgetary, monitoring, and evaluation procedures.

The hypothesis that sectoral climate policies are positively associated with readiness is also supported by the baseline results. Given that the Wald tests identify no statistically significant difference between the sectoral and cross-sectoral coefficients, these two policy approaches should be regarded as complementary rather than substitutable. This result is consistent with Schmidt et al. (2008), who shows that specialization can strengthen technical expertise, improve the definition of policy objectives, and facilitate programme monitoring. However, sectoral policies may also generate inconsistencies, duplicated expenditure, and transfers of risk when they are implemented in isolation. The fragmentation and maladaptation risks identified by Nascimento et al. (2022), Abomo et al. (2024), and Birchall et al. (2025) therefore justify combining sector-specific expertise with broader coordination mechanisms. Governments should strengthen cross-sectoral coordination without weakening the specialized administrative, technical, and implementation capacities required within individual sectors. In practical terms, each new sectoral policy should be assessed for its potential consequences for other areas, particularly water, agriculture, energy, health, infrastructure, and spatial planning.

By contrast, the hypothesis that international climate policies systematically improve readiness across all countries is not confirmed. Their association is mainly favorable at low levels of GNI and tends to weaken as national income rises. This finding is consistent with the theory of global public goods and collective action: international agreements can facilitate access to finance, technology, expertise, and knowledge, but their contribution depends on the ability of domestic institutions to translate these resources into concrete adaptation measures. Proost (2026) highlights the constraints created by free-riding and weak enforcement mechanisms, while Abugnaba-Abanga et al. (2026) shows that formal commitments may remain largely declaratory in the absence of adequate financial, human, and administrative capacities. International partners should therefore complement climate finance with technical assistance, support for project preparation, simplified access procedures, and long-term institutional capacity building.

The hypothesis that higher national income systematically strengthens the benefits associated with climate policies is also not confirmed. Instead, the baseline results indicate that the marginal associations of cross-sectoral and international policies tend to decline as GNI rises. This conclusion does not contradict Fallahi et al. (2024) and Sarkodie et al. (2022), who show that economic resources facilitate the financing and implementation of adaptation. Rather, it suggests that countries already possessing relatively developed institutions and infrastructure may obtain smaller additional gains from the adoption of another policy. Climate strategies should therefore be differentiated according to national circumstances. Lower-income countries primarily require financial resources, technology transfers, and stronger administrative capacity, whereas higher-income countries should give greater priority to implementation quality, policy coherence, innovation, and the evaluation of existing instruments rather than merely increasing the number of policies adopted.

The sensitivity analysis provides additional evidence regarding the robustness of these conclusions, although some results differ from the baseline estimates. These divergences may reflect three main factors. First, the CCLW and CAPMF samples differ in their geographical coverage, with the baseline analysis including 59 countries and the CAPMF sensitivity analysis covering 49 countries. Second, the two databases measure different dimensions of climate policy: the CCLW indicators capture policy adoption, whereas the CAPMF indicators reflect policy stringency. Third, the econometric specifications differ, since the baseline results are primarily derived from the SUR framework, whereas the sensitivity analysis relies on a Mundlak specification that separates within-country changes from between-country differences. The sensitivity analysis should therefore not be interpreted as an exact replication of the baseline model, but as an assessment of whether the main conclusions remain broadly valid when the sample, policy measurement, and estimation method are modified. In particular, it confirms the comparatively robust association of cross-sectoral policies, while the sectoral and income-moderation results appear more sensitive to these alternative choices.

Finally, the strong persistence of readiness_delta indicates that changes in countries' economic, institutional, and social adaptation conditions occur progressively. This finding supports the institutional theory of North (1990) and Acemoglu and Robinson (2002), according to which the formal adoption of a policy produces meaningful outcomes only when it is supported by durable administrative, technical, and financial capacities. Governments should therefore avoid evaluating climate policies solely on the basis of their immediate outcomes. Adaptation strategies should be designed over several years, supported by stable funding, and assessed through intermediate implementation indicators as well as long-term changes in readiness.

Overall, the study identifies cross-sectoral policies as displaying the most robust and consistent positive association with improvements in adaptation readiness. Nevertheless, the Wald tests show that their coefficient is not statistically different from that of sectoral policies. The empirical evidence therefore supports the complementarity of cross-sectoral coordination and sector-specific expertise rather than the superiority of one approach over the other. The study does not confirm either a general positive association of international policies across all income levels or the assumption that national income systematically increases the benefits associated with climate-policy adoption. The main policy implication is that governments should strengthen coordination across sectors while preserving specialized sectoral capacities. More generally, improvements in adaptation readiness depend less on the number of policies adopted than on their effective coordination, implementation quality, alignment with national capacities, and continuity over time.

6 Conclusion

This study examined the comparative association of sectoral, cross-sectoral, and international climate policies with annual changes in countries' adaptation readiness across 59 countries over the period 2002–2022. The results show that cross-sectoral policies display the most robust and consistent positive association with improvements in readiness_delta, supporting the importance of institutional coordination highlighted by England et al. (2018), Schoenefeld et al. (2022), and le Polain de Waroux et al. (2024). However, the Wald tests do not indicate a statistically significant difference between the coefficients of cross-sectoral and sectoral policies. The findings therefore support the complementarity of cross-sectoral coordination and sector-specific expertise rather than the superiority of one approach over the other. Sectoral policies also contribute positively, particularly when governments are able to coordinate specialized interventions and prevent policy fragmentation.

By contrast, the hypothesis of a systematic positive contribution from international policies is not confirmed across all countries and income levels. Their association with readiness is more favorable at lower levels of national income and gradually weakens as income rises, indicating that access to international finance, technology, and expertise must be accompanied by sufficient domestic implementation capacity. More generally, the findings show that economic resources facilitate climate action but do not automatically increase the marginal gains from additional policy adoption, in line with Di Gregorio et al. (2019), Fallahi et al. (2024), and Scordato and Gulbrandsen (2024). The strong persistence of readiness_delta further indicates that improvements in economic, institutional, and social readiness are cumulative and require sustained implementation over time.

The sensitivity analysis broadly confirms the comparatively robust association of cross-sectoral policies, although some results differ from the baseline estimates. These divergences may reflect differences in country coverage between the CCLW and CAPMF samples, differences between policy adoption and policy stringency measures, and differences between the SUR and Mundlak specifications. Overall, the study shows that strengthening adaptation readiness depends less on the number of policies adopted than on their coordination, effective implementation, consistency with national capacities, and continuity. Governments should therefore strengthen cross-sectoral coordination while preserving sectoral expertise and implementation capacity, and tailor international support to countries' institutional and economic conditions.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/frevc.2026.1914208/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Lory Barile, University of Warwick, United Kingdom

Reviewed by: Manzir Abbas, Okan University, Türkiye

Mengzhi Xu, Yunnan Minzu University, China