Policy Research Working Paper 11094 Climate Policies are Path-Dependent Implications for Policy Sequencing and Feasibility Penny Mealy Michael Ganslmeier Stephane Hallegatte Climate Change Group March 2025 Policy Research Working Paper 11094 Abstract Although the feasibility of introducing climate policies countries are significantly more likely to introduce poli- underpins global efforts to curb climate change, there has cies that are related to their prior climate policymaking been limited analysis estimating the likelihood of introduc- experience. Exploiting this finding, the paper constructs ing specific policies in different country contexts. Drawing empirically validated ‘Climate Policy Feasibility Frontiers’ on a dataset of climate policies introduced globally over the which identify policies that are likely to be more feasible and past 50 years, this paper explores patterns in climate policy could also increase the probability of the adoption of other adoption to quantify policy feasibility across countries. In policies. Complementing traditional cost-benefit analysis, constructing a ‘Climate Policy Space’ network based on feasibility frontiers can inform more realistic and strategic the co-occurrence of policies across countries, the paper climate policy prioritization across countries. shows that climate policy adoption is path-dependent: This paper is a product of the Climate Change Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at pmealy@worldbank.org and shallegatte@worldbank.org The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Climate Policies are Path-Dependent: Implications for Policy Sequencing and Feasibility Penny Mealy1, Michael Ganslmeier2 and Stephane Hallegatte3 1 World Bank. Address: World Bank, 1818 H Street NW Washington, DC 20433 United States. Email: pmealy@worldbank.org. Further affiliations: Institute of New Economic Thinking at the University of Oxford; Smith School of Environment and Enterprise at the University of Oxford; Santa Fe Institute; Monash Business School at Monash University 2 University of Exeter. Address: University of Exeter, Centre for Computational Social Science Clayden Building at Streatham Rise Campus, EX4 4PE Exeter, United Kingdom. Email: m.ganslmeier@exeter.ac.uk. Further affiliations: Department of Methodology at the London School of Economics and Political Science; Department of Social Policy and Intervention at the University of Oxford 3 World Bank. Address: World Bank, 1818 H Street NW Washington, DC 20433 United States. Email: shallegatte@worldbank.org MAIN TEXT INTRODUCTION The feasibility of introducing climate policies is one of the most critical constraints on efforts to mitigate climate change (van den Ven et al, 2023; Jewell & Cherp, 2020, 2022; Patterson et al, 2018). While recent work has sought to quantify the effectiveness of climate policies in terms of reducing emissions (Stechemesser et al, 2024; Zheng et al, 2023; Nascimento & Höhne, 2023), there has been surprisingly little empirical or quantitative analysis assessing the likelihood that a given policy could be introduced in a specific country context. Given the overwhelming reliance of future climate outcomes on the capacity of governments to introduce specific policies to reduce or limit emissions, there is a clear need to better understand such likelihoods, and how this differs across country contexts. However, with the feasibility of policy implementation depending on a range of geophysical, technological, economic, socio-cultural and institutional factors – many of which are temporally dynamic and likely to vary between different groups or actors (IPCC, 2022), an appropriate empirical approach to quantitatively assess climate has proven hard to come by. This paper addresses this gap by providing a novel approach to estimate countries’ capacities to introduce different types of climate policies. Rather than seeking to directly measure underlying factors likely to influence policy feasibility in a given country setting, here we instead aim to infer information about such capacities by studying the type of climate policies that countries have previously introduced. Just as ecologists can infer a great deal about an ecosystem by observing species that inhabit it, observing policies that different countries have been able to introduce can reveal a lot about their capacities to introduce further policies in the future. 2 Drawing on a rich dataset of different climate policy instruments introduced by 198 countries over the last 50 years (Nascimento et al, 2022), we find distinct patterns in climate policy adoption. We show that across countries, climate policy adoption exhibits a nested structure suggesting that certain types of policies require a broader ‘ecology’ of prior policy experience and capabilities to be introduced than others. In constructing a ‘Climate Policy Space’ network based on the co- occurrence of climate policies across countries, we confirm that climate policy is path-dependent: policies become easier to adopt following the introduction of other related policies – which, on inspection, appear to involve similar institutional and policy know-how (Rosenbloom et al, 2019). Such path-dependency in climate policy adoption suggests that it is also predictable. By exploiting the pattern of co-occurrence in climate policies across countries, we construct a measure of policy alignment between a given country’s existing set of climate policies and new policies it could introduce in the future. We show that this measure is predictive of future climate policy adoption, even after controlling for relevant factors such as population size, income levels, fiscal space, carbon emission level, vulnerability towards climate change, climate, and governance capacities. We use this measure to construct empirically validated ‘Climate Policy Feasibility Frontiers’ (CPFFs) for each country. In addition to helping identify which new climate policies could be most feasible for different countries to introduce, these CPFFs also include other elements of policy desirability such as emissions reduction and capacity building potential, which can help further guide climate policymaking strategies. Our findings have several important practical implications for both research and policy. For the research community, insights into the feasibility of different climate policies can contribute to improving climate policy models following the priorities highlighted by Peng et al (2021). They argued that to provide more realistic and useful results, models needed to better capture the political economy and the competence of governments. Our approach can help assess the likelihood of given policy scenarios in each country context and could therefore be used to produce ‘constrained’ policy scenarios. For the policy community, our analysis opens the possibility of recommending and designing climate policies in new ways. While specific climate policy recommendations are often justified in terms of cost-benefit analysis, the practicality of their implementation in different country contexts is rarely rigorously considered. Our mapping of climate policy feasibility in different settings provides a new lens through which climate policies can be more realistically and strategically prioritized. RESULTS THE NESTEDNESS OF CLIMATE POLICIES ACROSS COUNTRIES Climate policy mixes – rather than a single instrument such as carbon pricing – are often advocated in the literature given the need for multiple instruments to address multiple market failures, or the recognition that some policies are complementary and create positive synergies towards intended objectives (van den Bergh et al, 2021). But few studies have sought to provide empirical insights on the potential challenges countries could face in introducing multiple policy instruments given their prior policymaking experience. In Figure 1 we draw on the Climate Policy Database (Nascimento et al, 2022) (see Methods) to examine which types of policy instruments have been introduced by different countries. Here we represent countries along the columns and climate policy instruments along the rows. Squares are colored blue if a country has announced the specific policy at any point in time recorded in the Climate Policy Database, and left blank otherwise. Figure 1. Triangular (nested) distribution of climate policies announced by countries By ordering policy instrument rows by degree (i.e. how frequently they have been announced by countries, and country columns based on the number of different instruments it has announced), we observe a distinct triangular (or ‘nested’) distribution in climate policy adoption. (Nestedness of this matrix is measured as being 0.64 based on the NODF measure (Almeida-Neto & Ulrich, 2011), where 1 relates to a perfectly nested structure and 0 indicates a perfectly un-nested matrix of blocks – see Methods). This structural pattern indicates that countries that have introduced few policy types tend to introduce policies that have been more prevalently introduced, while rare policies tend only to be found in countries that have introduced a wide variety of climate policy instruments. Such nested patterns have been documented in a variety of different areas, such as the distribution of species across locations in ecological systems (Nielsen & Baskompte, 2007), and the distribution of exported products and industrial activity in economic systems (Bustos et al, 2012; Tacchella et al, 2012; Mariani et al, 2019). To our knowledge, this is the first documentation of nestedness in the distribution of policies across countries. One interpretation of this empirical pattern is that common or prevalent policy instruments are easier to introduce (e.g. they could require comparably less institutional capacity, policymaking experience or investment of political capital), while rare policies that are found in countries with highly diverse climate policy portfolios could require more sophisticated institutional capacity and policymaking capabilities, in addition to a supportive political environment. Inspecting the types of instruments that are most prevalent (e.g. strategic planning and non-binding climate strategies, GHG reduction targets and renewable energy targets) lends support to this hypothesis: these policies are relatively easy to introduce as they do not require a country or its constituents to make binding climate commitments. In contrast, policies that have been announced by very few countries include the removal of fossil fuel subsidies (which is known to be politically challenging), retirement premiums of emissions-intensive assets that have not reached the end of their useful life (which can require complex contracting arrangements between governments and asset owners) and white certificate schemes (which involve establishing a trading certificate scheme for energy efficiency improvements). If we rank policy instruments in terms of their commonality and assume this provides an initial indication of their ease of introduction, carbon taxes appear in 25th place. This is consistent with previous work on policy sequencing, which has shown that carbon pricing tends to appear relatively late in countries’ policy sequences – often after a series of different types of policy instruments have been introduced (Linsenmeier and Schwerhoff, 2022). Rankings of all policies can be found in the Supplementary Information (SI) section. THE CLIMATE POLICY SPACE AND THE PATH-DEPENDENCE OF CLIMATE POLICYMAKING While certain policies might be easier to introduce later in a sequence, some policies may be more feasible if a country has acquired specific types of related policymaking know-how and experience beforehand. For example, it could be easier to introduce a carbon tax if a country has prior experience in measuring, reporting and verification processes associated with the introduction of other policies (e.g. specific standards or regulations). To explore this possibility, we investigate the co-occurrence of different policy instruments across countries and estimate the ‘relatedness’ of one climate policy instrument to another based on their likelihood of co-occurring in a given country’s policy mix (see Methods). Similar techniques have previously been applied to analyze the path dependence of economic development across countries (Hidalgo et al, 2007; Zaccaria et al, 2014). Note that we test a variety of different functional forms and time-periods over which we define a country’s policy mix and relatedness measures and find similar results (see SI). Panel (a) of Figure 2 visualizes the relatedness between climate policies in a network we call the Climate Policy Space. Here nodes represent specific policy instruments linked to each other based on their estimated relatedness (for visualization purposes, not all links are shown. Our visualization approach follows Hidalgo et al (2007), see Methods). Climate policy instruments are colored by their prevalence across countries (consistent with row orders in Figure 1), with more commonly introduced policies colored in darker purple and rarer policies colored in darker orange. Figure 2. The climate policy space and associated key policy clusters Panel (b) of Figure 2 shows the same network, but instead with key policy clusters identified with a community detection algorithm (see Methods). The pink and purple clusters at the bottom of the network consist of highly prevalent non-binding targets and climate strategies which are likely to be easier to introduce. The turquoise cluster comprises instruments relating to binding targets and institutional creation, while the blue cluster in the middle of the network largely consists of regulatory instruments (such as industry and product standards), market-based instruments (such as carbon taxes, energy and other taxes, and feed-in tariffs), and key enabling policies relating to monitoring, auditing and coordinating bodies for climate strategy. The olive cluster includes many policy instruments relating to technological deployment and innovation, as well as other rare policies that only tend to be found in countries that have introduced a wide variety of different policy types. Interestingly, the network does not show particularly tight clusters (see the SI). This suggests that policymaking does not occur in a series of discrete, well-defined stages, but is a more fluid process of continually evolving a given policy-mix. Having said that, the evolution of countries’ climate policy mixes across the Climate Policy Space does vary by income level (see Panel (a) of Figure 3). Low-income countries tend to have climate policies focused on the bottom clusters of the network corresponding to non-binding targets and strategies. Lower-middle and upper-middle income countries show a greater spread of policy instruments into the turquoise and blue clusters, suggesting that the building out of policy mixes into binding targets, institutional creation, regulatory and market-based instruments may go hand- in-hand with rising levels of economic development. And finally high-income countries tend to have a much wider spread of climate polices across the Climate Policy Space network, particularly in the olive cluster containing many rare and technology-specific policies. Figure 3. Countries in the climate policy pace at different income levels and over time While panel (a) of Figure 3 gives a cross-sectional perspective on how countries income levels may influence the feasibility of introducing specific climate policies, panel (b) and (c) show examples of Türkiye and Vietnam’s policy evolution through the Climate Policy Space over time. Both countries show a tendency to introduce new climate policies that are connected (or nearby) to existing policies in the Climate Policy Space network in each subsequent period. Such trajectories suggest that climate policymaking shows some degree of path-dependence, and it is easier for countries to introduce climate policies that require similar institutional capacity and know-how to policies they have previously introduced. However, as it is difficult to draw conclusions from a few illustrative examples, we take a more systematic approach to analyze this in the next section. PREDICTING FUTURE CLIMATE POLICY ADOPTION To further examine the path-dependence of climate policymaking across countries, we develop a measure of ‘Policy Alignment’ between a country and a new policy it has not yet introduced. This measure draws on the previously defined relatedness measure between two policies and calculates the average relatedness between the new policy and all policies that a country has had prior experience with (see Methods). We firstly perform an out-of-sample prediction by calculating the Policy Alignment between countries and new policies for rolling ten-year periods in the data from 1990 onwards (data training windows), and then testing the extent to which Policy Alignment is predictive of countries’ future policy adoptions in subsequent five-year periods (data test windows). Robustness checks using alternative time periods can be found in the SI. Our hypothesis is that if climate policy adoption is path-dependent, policies that have greater alignment with a countries’ existing policy mix should be more likely to be introduced in subsequent years. Our regression analysis follows the following specification: = + + + + + with , as binary that is equal to 1 if country c has adopted at least one policy of policy instrument j in the five years after the ten-year data training period t, and 0 otherwise; as the Policy Alignment measure (independent variable of interest); as the coefficient of control variable ; as the policy instrument fixed effect; as the training period fixed effect; and as the error term. In all models, the Policy Alignment between countries and new policies is significantly predictive of policy adoption in future years (Figure 4). To account for important correlates of climate policy adoption identified in previous studies based on the level of economic development (Dasgupta et al 2001), fiscal space, institutional quality (von Dulong & Hagen, 2024) considerations (Hawkins et al, 2016; Bolton et al, 2023) and vulnerability arguments (Brody et al, 2008), we control for population size, income levels, public debt, school enrollment rates, carbon emission levels, precipitation, air quality, and various measures of institutional capacity and political context (Yeganeh et al, 2020). The political and institutional variables include measures such as the vote share of the strongest government party, government effectiveness, control of corruption, rule of law, political stability and absence of violence, regulatory quality, voice and accountability, bureaucratic quality, post-election unrest, and democratic governance (e.g., Polity2 scores). Controlling for these factors, in addition to policy type fixed effects and period fixed effects, and holding all else constant, a one standard deviation increase in Policy Alignment (weighted measure) between a country and a new policy increases the likelihood of that policy’s introduction in the next five years by 10.3 percentage points. Compared to the average policy adoption rate of 11.2 percent, this increase would almost double a country’s average probability of introducing a given policy. In Table S3.1 we show the robustness of our results to different model specifications and fixed effects, data training and testing windows and Policy Alignment functional forms. In all cases, our results remain similar. Further robustness test including control sets, estimation methods (e.g. logistic versus linear probability model), rolling-window approach, alternative Policy Alignment measures, and different country subsets (e.g. high-income vs low-income) can be found in the SI. Figure 4. Robustness of regression coefficient to different controls, data test and train time periods and policy alignment functional form Note: The outcome variable is a binary that is equal to 1 if at least one policy of a given instrument in a given country has been adopted in the test period, and 0 otherwise. The point estimates show the association of two different policy alignment measures: policy alignment with unweighted relatedness, and policy alignment with weighted relatedness (x-axis). The panels differ with respect to the estimation approach: the left panel uses a linear probability model (OLS) using the full sample, the panel in the middle uses a linear probability model (OLS) using a sample that consists of countries with good data coverage, and the right panel uses a logistic regression model using the full sample. The coefficients within each panel differ with respect to the control set included in the regressions (differentiated by colors). A full list of variables for each control set can be found in S3.1. Policy type fixed effects and period fixed effects are included in all estimations. The control sets are described in the footnote of table S3.1. Thin and thick vertical lines around point estimates represent 99% and 95% confidence intervals based on robust standard errors clustered at the country level, respectively. CLIMATE POLICY FEASIBILITY FRONTIERS Having shown the strong tendency for future climate policies to build on countries’ previous policymaking experience, we now exploit these relationships to construct ‘Climate Policy Feasibility Frontiers’ (CPFF) that aim to inform more realistic and strategic climate policymaking across countries. Figure 5 shows example feasibility frontiers for Türkiye and Vietnam. Panels (a) and (c) illustrate the countries’ current positions in the Climate Policy Space and panels (b) and (d) show their CPFF. Each dot in the CPFF represents a new policy that countries have not introduced before 2021 and corresponds to policies colored in grey in the Climate Policy Space. The CPFF encompasses three key dimensions. Policy Alignment (discussed above), which indicates the likely ease of a new policy’s introduction based on countries’ prior policy experience is plotted on the x-axis. Future Policy Enablement (plotted on the y-axis) is a measure that aims to capture how the introduction of a new policy enables the future adoption of other new policies. It calculates the additional alignment a given policy creates to other new climate policies that a country has not yet introduced (see Methods). New policies are also colored based on a Policy Emission Association measure, with darker green indicating policies that tend to be present in countries with lower emissions. This measure uses the same approach as Eskander and Fankhauser (2020) and estimates the association between policies and emissions through a regression analysis of past policy adoptions and associated changes in emissions across countries (see Methods). However, these estimates should be considered as indicative only, as they do not account for the interactions between policies or the level of policy stringency (which is not available for many policies in the Climate Policy Database), and some policies have very little data coverage. Policies colored in orange relate to instruments for which no significant effect could be detected or there were not enough observations across countries to derive reliable emission association estimates. Figure 5. Climate policy feasibility frontiers for Türkiye and Vietnam Looking at Türkiye’s CPFF (Figure 5, panel b), some policies that are likely to be the most feasible to introduce include a legally-binding climate strategy or a binding emissions reduction target (we note that Türkiye’s net zero target and associated draft climate law is still in draft form and not yet official policy) (Climate Action Tracker, 2025). These policy instruments are highlighted in Türkiye’s Climate Policy Space in dark bold circles (panel a) and are adjacent (or connected to) many other policies that Türkiye has prior experience with. Based on our regression analysis in Table S3.1 (model 9), Türkiye is around 25 percent more likely to implement a legally binding climate strategy in the next five years compared to a retirement premium policy (which is further away from Türkiye’s existing policies in the Climate Policy Space). The CPFF can also help identify strategic stepping-stones in a long-term policy pathway. For example, the introduction of a legally-binding climate strategy in Türkiye could be a first step toward a binding emissions reduction target. According to the CPFF, the likelihood of introducing a legally-binding emission reduction target would be around 9 percent higher if the country had a legally-binding climate strategy. The CPFF can therefore be used to design policy sequences that can lead to the implementation of a policy that may appear out of reach at one point in time. The CPFF is based on past policies and does not capture all dimensions relevant to the choice of policy instrument. For instance, it identifies an emissions trading scheme (ETS) or a carbon tax as potential next steps for Türkiye. In practice, the choice to introduce either instrument depends on the political context. In Türkiye, in part due to the geographic proximity with the EU and its ETS, the government has recently announced future plans for the introduction of an ETS, confirming the CPFF’s assessment that it is a highly feasible policy intervention in the country (ICAP, 2023). Policies relating to technological deployment and diffusion and technological development (labeled in purple in Türkiye’s CPFF and highlighted in a purples circle in the Climate Policy Space) are less aligned with Türkiye’s prior policy experience and are likely to be more difficult to introduce in the near term. However, these policies could be considered ‘stretch’ policies that Türkiye could work toward: their position in the CPFF shows that they would have a large impact on the country’s capacity to deploy further climate policies in the future, in part through their benefits in terms of green innovation and technological development. In addition to their climate benefits, such capacities could also have important economic advantages for the country. Equivalent plots are shown for Vietnam in panels (c) and (d). As Vietnam has a different set of policymaking capabilities to Türkiye, its CPFF looks different. Vietnam’s most feasible policies include climate-related institutional creation and introducing a legally binding GHG emission reduction target. Policies to support climate-related infrastructure investment could be advantageous in helping Vietnam build further capacity to introduce new policies. Although carbon pricing appears more challenging to implement for Vietnam in the short term, other policies can help pave the way. Such plots illustrate that there is no overarching climate strategy that is likely to fit all countries: as each country’s institutional context, climate objectives and political environment are different, each country’s climate policy pathway should also be unique. DISCUSSION This paper develops a new approach to quantitatively assess the feasibility of introducing different types of climate policies. We demonstrate that climate policy adoption across countries is nested – suggesting that certain policies might only be feasible in countries with a high degree of climate policymaking experience. We also show that policy adoption is path-dependent and therefore predictable: counties are more likely to introduce policies that are more closely aligned to policies they have previously introduced. Drawing on these insights, we introduced an empirical framework to assess climate policies across countries in terms of their feasibility, emission reduction and future policy potential. By considering where new policies fall along each of these dimensions, policymakers can more realistically prioritize policies and strategically map out policy-pathways in terms of feasible policy stepping-stones. Our work is connected to several strands of literature. We contribute to the literature on policy- sequencing (Meckling et al, 2015; 2017; Pahl et al, 2018; Linsenmeier & Schwerhoff, 2022), which highlights advantages of first introducing policies that build a more favorable political economy (e.g. industrial policies that build coalitions of green industries or bring down green technology costs) before introducing more politically challenging policies like carbon pricing or scaling up climate policy stringency. While previous work on sequencing has largely focused on policy feasibility in terms of political palatability, our research also draws attention to feasibility constraints that nations (particularly lower income countries) could face in terms of a lack of relevant policymaking experience and institutional capacity (Meckling & Nahm, 2017). Any policy sequence needs to account for both political and capacity constraints (Dubash et al, 2021; Guy et al, 2023) Our work also provides a data-driven angle to analyze ‘policy feedbacks’ (Pierson, 1993; Jordan & Matt, 2014) and the path-dependent nature of policy and institutional change (Bednar & Page, 2018). Although recent work on sustainability transitions has emphasized the possibility for policies to either reinforce or undermine the rate or direction of future policy making through conceptual theorizing or discussing specific examples (Farmer et al, 2019; Edmondson et al, 2019; Lockwood, 2022), our methodology allows such dynamics to be explored with greater empirical rigor. Of course, our work leaves plenty of avenues for future research. Firstly, while our approach provides insights into the adoption of climate policies, we have not yet explored the stringency or effectiveness in which those policies are applied. Secondly, in drawing on the Climate Policy Database, our work has focused primarily on mitigation policies. However, with many damaging effects of climate change now being experienced around the world, there is an urgent need to also consider policy feasibility in relation to adaptation and resilience. Thirdly, future research could also investigate the extent to which specific climate policies require other non-climate policies to be in place. For example, energy efficiency standards in buildings would be easier to introduce if some sort of building codes are already in place. Finally, as our approach is generalizable, future work could also apply these techniques to policy domains stretching beyond climate change, looking in particular at the implementation of growth-enhancing or poverty-reducing policies. MAIN REFERENCES Almeida-Neto, M., & Ulrich, W. (2011). A straightforward computational approach for measuring nestedness using quantitative matrices. Environmental Modelling & Software, 26(2), 173-178. Bednar, J., & Page, S. E. (2018). When order affects performance: culture, behavioral spillovers, and institutional path dependence. American Political Science Review, 112(1), 82-98. Bolton, P., Buchheit, L., Gulati, M., Panizza, U., di Mauro, B. W., & Zettelmeyer, J. (2023). On Debt and climate. Oxford Open Economics, 2, odad005. Brody, S. D., Zahran, S., Grover, H., & Vedlitz, A. (2008). A spatial analysis of local climate change policy in the United States: Risk, stress, and opportunity. Landscape and urban planning, 87(1), 33-41. Bustos, S., Gomez, C., Hausmann, R., & Hidalgo, C. A. (2012). The dynamics of nestedness predicts the evolution of industrial ecosystems. PloS one, 7(11), e49393. Climate Action Tracker, 2025, Turkiye country profile, https://climateactiontracker.org/countries/turkey/net-zero- targets/#:~:text=T%C3%BCrkiye's%20draft%20climate%20law%20is,and%20Climate% 20Change%2C%202024b). Dasgupta, S., Mody, A., Roy, S., & Wheeler, D. (2001). Environmental regulation and development: A cross-country empirical analysis. Oxford Development Studies, 29(2), 173-187. Dubash, N. K., Pillai, A. V., Flachsland, C., Harrison, K., Hochstetler, K., Lockwood, M., & Tyler, E. (2021). National climate institutions complement targets and policies. Science, 374(6568), 690-693. Edmondson, D. L., Kern, F., & Rogge, K. S. (2019). The co-evolution of policy mixes and socio- technical systems: Towards a conceptual framework of policy mix feedback in sustainability transitions. Research Policy, 48(10), 103555. Farmer, J. D., Hepburn, C., Ives, M. C., Hale, T., Wetzer, T., Mealy, P., & Way, R. (2019). Sensitive intervention points in the post-carbon transition. Science, 364(6436), 132-134. Guy, J., Shears, E., & Meckling, J. (2023). National models of climate governance among major emitters. Nature Climate Change, 13(2), 189-195. Hawkins, C. V., Krause, R. M., Feiock, R. C., & Curley, C. (2016). Making meaningful commitments: Accounting for variation in cities’ investments of staff and fiscal resources to sustainability. Urban Studies, 53(9), 1902-1924. Hidalgo, C. A., Klinger, B., Barabási, A. L., & Hausmann, R. (2007). The product space conditions the development of nations. Science, 317(5837), 482-487. ICAP (2023), International Carbon Action Partnership, Turkiye factsheet, https://icapcarbonaction.com/en/ets/turkiye IPCC. (2022). Climate Change 2022: Mitigation of Climate Change. Contribution of Working Group III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [P.R. Shukla, J. Skea, R. Slade, A. Al Khourdajie, R. van Diemen, D. McCollum, 19 M. Pathak, S. Some, P. Vyas, R. Fradera, M. Belkacemi, A. Hasija, G. Lisboa, S. Luz, J. Malley, (eds.)]. Cambridge University Press, Cambridge, UK and New York, NY, USA. doi: 10.1017/9781009157926 Jewell, J., & Cherp, A. (2020). On the political feasibility of climate change mitigation pathways: is it too late to keep warming below 1.5° C?. Wiley Interdisciplinary Reviews: Climate Change, 11(1), e621. Jewell, J., & Cherp, A. (2023). The feasibility of climate action: Bridging the inside and the outside view through feasibility spaces. Wiley Interdisciplinary Reviews: Climate Change, 14(5), e838. Jordan, A. & Matt, E. (2014) Designing policies that intentionally stick: policy feedback in a changing climate. Policy Sci, 47: 227–247. https://doi.org/10.1007/s11077-014-9201-x Linsenmeier, M., Mohommad, A., & Schwerhoff, G. (2022). Policy sequencing towards carbon pricing among the world’s largest emitters. Nature Climate Change, 12(12), 1107-1110. Lockwood, M. (2022). Policy feedback and institutional context in energy transitions. Policy Sciences, 55(3), 487-507. Mariani, M. S., Ren, Z. M., Bascompte, J., & Tessone, C. J. (2019). Nestedness in complex networks: observation, emergence, and implications. Physics Reports, 813, 1-90. Meckling, J., & Nahm, J. (2018). The power of process: State capacity and climate policy. Governance, 31(4), 741-757. Meckling, J., Kelsey, N., Biber, E., & Zysman, J. (2015). Winning coalitions for climate policy. Science, 349(6253), 1170-1171. Meckling, J., Sterner, T., & Wagner, G. (2017). Policy sequencing toward decarbonization. Nature Energy, 2(12), 918-922. Nascimento, L., & Höhne, N. (2023). Expanding climate policy adoption improves national mitigation efforts. npj Climate Action, 2(1), 12. Nielsen, A., & Bascompte, J. (2007). Ecological networks, nestedness and sampling effort. Journal of Ecology, 1134-1141. Pahle, M., Burtraw, D., Flachsland, C., Kelsey, N., Biber, E., Meckling, J., ... & Zysman, J. (2018). Sequencing to ratchet up climate policy stringency. Nature Climate Change, 8(10), 861-867 Patterson, J. J., Thaler, T., Hoffmann, M., Hughes, S., Oels, A., Chu, E., ... & Jordan, A. (2018). Political feasibility of 1.5 C societal transformations: the role of social justice. Current Opinion in Environmental Sustainability, 31, 1-9. Peng, W., Iyer, G., Bosetti, V., Chaturvedi, V., Edmonds, J., Fawcett, A. A., Hallegatte, S., Victor, D. G., van Vuuren, D., & Weyant, J. (2021). Climate policy models need to get real about people—here’s how. Nature, 594(7862), 174-176. Pierson, P. (1993). When effect becomes cause: Policy feedback and political change. World politics, 45(4), 595-628. Rosenbloom,D., Meadowcroft, J. & Cashore, B. (2019). Stability and climate policy? Harnessing insights on path dependence, policy feedback, and transition pathways. Energy Research & Social Science, 50: 168-178. ISSN 2214-6296. https://doi.org/10.1016/j.erss.2018.12.009. Stechemesser, A., Koch, N., Mark, E., Dilger, E., Klösel, P., Menicacci, L., ... & Wenzel, A. (2024). Climate policies that achieved major emission reductions: Global evidence from two decades. Science, 385(6711), 884-892. Tacchella, A., Cristelli, M., Caldarelli, G., Gabrielli, A., & Pietronero, L. (2012). A new metrics for countries' fitness and products' complexity. Scientific reports, 2(1), 723. Van de Ven, D. J., Mittal, S., Gambhir, A., Lamboll, R. D., Doukas, H., Giarola, S., ... & Nikas, A. (2023). A multimodel analysis of post-Glasgow climate targets and feasibility challenges. Nature Climate Change, 13(6), 570-578. Van den Bergh, J. C. J. M., Castro, J., Drews, S., Exadaktylos, F., Foramitti, J., Klein, F., ... & Savin, I. (2021). Designing an effective climate-policy mix: accounting for instrument synergy. Climate Policy, 21(6), 745-764. von Dulong, A., & Hagen, A. (2024). Institutions make a difference: assessing the predictors of climate policy stringency using machine learning. Environmental Research Letters, 20(1), 014056. Yeganeh, A. J., McCoy, A. P., & Schenk, T. (2020). Determinants of climate change policy adoption: A meta-analysis. Urban Climate, 31, 100547. Zaccaria, A., Cristelli, M., Tacchella, A., & Pietronero, L. (2014). How the taxonomy of products drives the economic development of countries. PloS one, 9(12), e113770. Zheng, S., Pu, Y., Lu, H., Zhang, J. J., Wang, D., & Ma, X. (2023). Global climate policy effectiveness: A panel data analysis. Journal of Cleaner Production, 412, 137321. 1 METHODS 2 DATA 3 To analyze climate policies across countries, we draw on the Climate Policy Database V20211, 4 which is one of the most comprehensive policy datasets in terms of its coverage of countries 5 and specific climate policy instruments. The dataset contains several other climate policy 6 datasets (e.g. Climate Change Laws of the World2 and the OECD Policy Instruments for the 7 Environment Database (2017)3), and also draws on rich repositories of country climate policy 8 reports and related documents. The data is higher quality and considered most comprehensive 9 for G20 countries and 18 other countries that are larger emitters or required to provide in-depth 10 reporting on their policy implementation (listed in Table S2.1 in the SI). The database includes 11 non-exhaustive information on other countries1,4. 12 13 While the Climate Policy Database encompasses 70 policy instrument categories, some serve 14 as broader umbrellas under which disaggregated categories are classified. We restrict our 15 analysis to the set of most detailed policy instrument categories (54 in total). A list of all policy 16 instrument types analyzed in this paper can be found in Table S1.4 in the SI. The Climate Policy 17 Database also includes policies announced by countries and within-country jurisdictions 18 (regions and cities), however here we only consider policies announced at the country level. 19 20 In addition to the climate policy data, we collect further indicators to control for essential 21 confounding factors that are important to account for when estimating the adoption of climate 22 change policies. To be more specific, we use the following control variables (with sources in 23 parenthesis): population size (log) (World Bank), GDP per capita (log) (World Bank), central 24 government debt as percentage of GDP (International Monetary Fund), carbon emission per 25 capita (log) (World Bank), share of population living in areas below 5m of sea level (World 26 Bank), annual precipitation,5 and various measures of institutional capacity6 capturing a 27 country’s level of corruption, rule of law, regulatory quality, government effectiveness, 28 political stability, and political accountability. The selection of these control variables is based 29 on several long-standing theoretical and empirical contributions on (climate change) policy 30 adoption.7,8,9,11 31 32 MEASURING THE NESTEDNESS OF CLIMATE POLICIES, POLICY RELATEDNESS, POLICY 33 ALIGNMENT AND FUTURE POLICY POTENTIAL 34 We measure the nestedness of the climate policy adoption matrix shown in Figure 1 using the 35 Nestedness based on Overlap and Decreasing Fill (NODF) algorithm developed by Almeida- 36 Neito et al (2008).12 In the SI we show a series of robustness tests using alternative approaches 37 to order the rows and columns, such as the Fitness/Complexity algorithm13 and the Economic 38 Complexity Index and Product Complexity Index.14 While the Fitness/Complexity algorithm 39 provides ordering that better maximize the triangular appearance of the matrix, we find broadly 40 similar results. 41 42 To measure the relatedness between two policy instruments and we firstly construct a 43 matrix where the rows correspond to countries and columns correspond to policy 44 instruments. Each element relates to the number of policies that country has introduced 45 in policy instrument . 46 47 We then construct a binary matrix where again rows correspond to countries and columns 48 correspond to different policy instruments. Here though, each element = if ≥ and 49 0 otherwise. The relatedness between two policy instruments and is then calculated as 50 ∑ ∑ 51 = ( ∑ , ∑ ) (1) 52 where 53 = ∑ . (2) 54 55 While is informative of the intensity in which countries have introduced different policy 56 instruments, the measure has some drawbacks when seeking to make comparisons across 57 countries and policy instruments. For example, countries that are larger, more economically 58 advanced or for which data is more comprehensively collected are likely to have higher 59 values. Similarly, the intensity in which policy instruments need to be introduced differs by 60 policy: although it might make sense for a country to have several policies relating to grants or 61 loans, they should only need to introduce a single GHG or renewable energy target. 62 63 We therefore additionally calculate a measure we call Relative Policy Prevalence (), 64 which aims to account for these country and policy-specific factors. RPP for country in policy 65 instrument is defined as 66 / ∑ 67 = ∑ / ∑ (3) 68 69 where an value greater than 1 indicates country has a introduced a higher share of 70 policies in policy instrument than average policy share across countries. As we show in the 71 SI, our results remain similar whether we use the or in constructing the binary matrix 72 . 73 74 To construct the measure of Policy Alignment between a country and a new policy 75 instrument it has not yet introduced, we calculate the weighted average relatedness of that 76 policy to all policies that country has prior experience with. That is, 77 ∑ 78 = ∑ (4) 79 80 Our measure of Future Policy Potential aims to capture the benefits a country derives 81 from the introduction of a new policy in terms of the relative ease of introducing other new 82 policies in the future. It is calculated as the average Policy Relatedness between the new policy 83 and all other policies that the country has no prior experience with. (Again, we consider 84 prior experience based on the two definitions described above). 85 ∑ 86 = ∑ (5) 87 88 CONSTRUCTING THE CLIMATE POLICY SPACE NETWORK AND IDENTIFYING KEY POLICY 89 CLUSTERS 90 To construct the Climate Policy Space network, we follow Hidalgo et al (2007)15 and first 91 construct a maximum spanning tree from the adjacency matrix to create the network ‘back- 92 bone’. We then add additional edges between each node and if is greater than a threshold 93 . For the visualization in Figure 2, we let = . , which corresponds to 0.5 standard 94 deviations greater than the mean of the distribution of values. While lower values result 95 in a denser network and vice versa for higher values, the general structure of the network 96 remains similar for different values. We use the Gephi software package (V 0.9.2) to visualize 97 the final network shown in Figure 2. 98 99 To identify the key policy clusters in the Climate Policy Space network, we experimented with 100 several community detection algorithms. These included k-means and the Louvain method 101 (discussed further in the SI). As shown in the SI, the approaches yielded broadly similar results. 102 The key policy clusters shown in panel b of Figure 2 are derived using the k-means algorithm, 103 where we identify 5 clusters after applying the Elbow Method to find the optimal number of 104 clusters. 105 106 107 ASSESSING EMISSION REDUCTION POTENTIAL 108 Assessing how emission reduction potential differs across policy instruments is complicated 109 by the fact that multiple complementary policies are often operating at any point in time. 110 Moreover, many policies have not been in force long enough for effects to be fully realized, or 111 have not yet reached a level of stringency where they can be effective (e.g. carbon prices in 112 many jurisdictions are still too low to drive meaningful changes in behavior16. Here we do not 113 seek to provide evidence on causal relationships. Instead, we show indicative estimates of the 114 emissions reduction potential based on existing policy adoption patterns and emission 115 reduction trends to date. 116 117 We closely follow the approach taken by Eskander and Fankhauser (2020)17 who adopted a 118 two-way fixed effect panel regression approach using an annual cross-country dataset of 119 climate laws and legislation. They regressed carbon emission levels on the number of climate 120 laws while accounting for several key control variables. As climate laws take time to result in 121 emission outcomes, they use two measures of climate adoption activity: the number of laws 122 adopted in the previous three years (short-term), and the number of laws adopted more than 123 three years ago (long-term). 124 125 We estimate an identical regression model (using the same control variables), with two key 126 differences. First, rather than drawing on the Climate Change Laws of the World dataset 127 (2023),2 we use the Climate Policy Database.1 Second, we run regressions for each specific 128 policy instrument (rather than the total number of policies announced by a country). In this 129 way, we estimate how emission reduction potential varies across policy types. In detail, our 130 estimation approach takes the following form: 131 132 = + , , + + + + + 133 134 with as (log) CO2 emission per unit of economic output (MtCO2e/GDP) of country in 135 year ; , as the number of policies of a given policy type P adopted in the last three years; 136 , as the number of policies that have been adopted more than three years ago; as a 137 vectors of control variables that captures the first lag of the following factors: rule of law 138 indicators, GDP per capita, squared GDP per capita, the cyclical component of GDP based on 139 the Hodrick-Prescott decomposition, import share, size of the service sector, temperature 140 fluctuation, and a binary capturing the presence of a federal system; , and represent 141 the country fixed effect, years fixed effect and random error term, respectively. 142 METHODS REFERENCES 1. Nascimento, L., Kuramochi, T., Iacobuta, G., den Elzen, M., Fekete, H., Weishaupt, M., & Höhne, N. (2022). Twenty years of climate policy: G20 coverage and gaps. Climate policy, 22(2), 158-174. 2. Climate Change Laws of the World. Grantham Research Institute at the London School of Economics, Climate Policy Radar. (2023). 3. OECD. (2017). Policy Instruments for the Environment. Organisation for Economic Co-operation and Development (OECD). 4. Linsenmeier, M., Mohommad, A., & Schwerhoff, G. (2022). Policy sequencing towards carbon pricing among the world’s largest emitters. Nature Climate Change, 12(12), 1107-1110. 5. Food and Agriculture Organization Aquastat. (2018). AQUASTAT-FAO’s Global Information System on Water and Agriculture. 6. Howell, L. D. (2011). International country risk guide methodology. East Syracuse, NY: PRS Group, 7. 7. Dasgupta, S., Mody, A., Roy, S., & Wheeler, D. (2001). Environmental regulation and development: A cross-country empirical analysis. Oxford development studies, 29(2), 173-187. 8. Hawkins, C. V., Krause, R. M., Feiock, R. C., & Curley, C. (2016). Making meaningful commitments: Accounting for variation in cities’ investments of staff and fiscal resources to sustainability. Urban Studies, 53(9), 1902-1924. 9. Bolton, P., Buchheit, L., Gulati, M., Panizza, U., di Mauro, B. W., & Zettelmeyer, J. (2023). On Debt and climate. Oxford Open Economics, 2, odad005. 10. Brody, S. D., Zahran, S., Grover, H., & Vedlitz, A. (2008). A spatial analysis of local climate change policy in the United States: Risk, stress, and opportunity. Landscape and urban planning, 87(1), 33-41. 11. Yeganeh, A. J., McCoy, A. P., & Schenk, T. (2020). Determinants of climate change policy adoption: A meta-analysis. Urban Climate, 31, 100547. 12. Almeida‐Neto, M., Guimaraes, P., Guimaraes Jr, P. R., Loyola, R. D., & Ulrich, W. (2008). A consistent metric for nestedness analysis in ecological systems: reconciling concept and measurement. Oikos, 117(8), 1227-1239. 13. Tacchella, A., Cristelli, M., Caldarelli, G., Gabrielli, A., & Pietronero, L. (2012). A new metrics for countries' fitness and products' complexity. Scientific reports, 2(1), 723. 28 14. Hidalgo, C. A., & Hausmann, R. (2009). The building blocks of economic complexity. Proceedings of the national academy of sciences, 106(26), 10570-10575. 15. Hidalgo, C. A., Klinger, B., Barabási, A. L., & Hausmann, R. (2007). The product space conditions the development of nations. Science, 317(5837), 482-487. 16. World Bank. (2023). State and Trends of Carbon Pricing 2023. http://hdl.handle.net/10986/39796 17. Eskander, S. M., & Fankhauser, S. (2020). Reduction in greenhouse gas emissions from national climate legislation. Nature Climate Change, 10(8), 750-756. 30 ADDITIONAL INFORMATION Author contributions. PM and SH conceived the idea. PM and MG collected the data, performed the analysis and created the figures. PM, MG and SH wrote the draft and approved the final manuscript. Competing interest declaration. The authors declare no competing financial and non-financial interests. Materials and correspondence. Correspondence should be addressed to Penny Mealy (pmealy@worldbank.org). The datasets and codes for the empirical analysis of the present research article will be made available on our GitHub page, see https://github.com/MGanslmeier/ClimatePolicySpace. SUPPLEMENTARY INFORMATION S1: ROBUSTNESS CHECKS BASED ON DIFFERENT APPROACHES FOR ORDERING THE COUNTRY- POLICY MATRIX Figure 1 in the main text showed a nested (or triangular) pattern of climate policy adoption across countries by ordering the rows and columns of the country-policy matrix by row and column sums (i.e. by degree). This ordering provides insights into the diversity (or breadth) of different climate policies countries have introduced. We note that there are other approaches discussed in the literature to order the rows and columns of the matrix. For example, in Figure S1.1 Panel a), we apply the Fitness and Complexity algorithm, described in Tacchella et al (2012)3, to arrive at Fitness rankings for countries and Complexity rankings for policies and order country columns and policy rows on this basis. These rankings can be found in Table S1.4 and S1.5 below. As can be seen, this approach gives a ‘cleaner’ nested pattern. However, as the correlations between degree ranks and Fitness/Complexity ranks are very high (0.95 for countries and 0.85 for policies), overall differences are fairly minimal. We perform a similar exercise using the Economic Complexity Index (ECI) and Product Complexity Index (PCI) algorithm introduced in Hidalgo and Hausmann (2009)4 (see Figure S1.1 Panel b). Countries are ordered by their ECI, while policies are ordered by their PCI. Here the triangular pattern is not quite as clean. While correlations with the original degree ranking are not quite as strong as the Fitness and Complexity algorithm, they are still fairly similar (0.85 for countries and 0.70 for policies). Of course, in all cases the overall nestedness of the matrix is unchanged – only the rankings of countries and policies are altered. Such rankings could be used to assess or infer the sophistication of institutional capacity associated with a country or a policy – we leave this for future research. 31 32 Figure S1.1: Ordering of country policy matrix by the Fitness and Complexity algorithm and ECI and PCI algorithm 33 Table S1.1: Correlations between alternative approaches for ranking ordering countries Country rank Diversity (number of different Fitness ECI ordering policies introduced) Diversity 1.00 0.95 0.85 Fitness 0.95 1.00 0.92 ECI 0.85 0.93 1.00 Table S1.2: Correlations between alternative approaches for ranking ordering Policies Policy rank Prevalence (number of countries Complexity PCI ordering that have introduced the policy) Prevalence 1.00 0.85 0.70 (degree) Complexity 0.85 1.00 0.92 PCI 0.70 0.92 1.00 34 Table S1.3: Rankings of policies by Prevalence, Complexity and PCI Preval. Preval. Complexity PCI Policy instrument Complexity PCI (degree) rank rank rank Political & non-binding 151 -1.17E-207 0.300 1 1 1 GHG reduction target Strategic planning 133 -6.35E-207 0.150 2 2 4 Political & non-binding 121 -1.07E-206 0.210 3 4 3 climate strategy Political & non-binding 103 -7.22E-207 0.250 4 3 2 renewable energy target Institutional creation 71 -1.36E-106 0.030 5 7 6 Formal & legally binding 56 -1.36E-106 0.010 6 6 9 renewable energy target Obligation schemes 54 -0.000451439 -0.070 7 12 17 Political & non-binding 52 -8.31E-178 0.100 8 5 5 energy efficiency target Tax relief 51 -0.004950639 -0.070 9 15 14 Grants and subsidies 46 -0.009742112 -0.090 10.5 19 25 Feed-in tariffs or 46 -0.011676931 -0.070 10.5 20 16 premiums Energy and other taxes 44 -0.00316253 -0.080 12 13 22 Product standards 43 -0.003568179 -0.080 13.5 14 21 Other mandatory 43 -0.009115041 -0.070 13.5 18 19 requirements Building codes and 41 -0.016885455 -0.110 15.5 22 28 standards Monitoring 41 -0.00500982 -0.070 15.5 16 18 Information provision 40 -0.02226129 -0.110 17.5 25 27 Formal & legally binding 40 -6.82E-106 0.030 17.5 8 7 climate strategy Formal & legally binding 39 -1.00E-91 -0.060 19 11 13 GHG reduction target Grid access and priority 38 -0.012372236 -0.070 20 21 15 for renewables Loans 34 -0.067732623 -0.130 21 31 32 Auditing 33 -0.055758354 -0.120 22 30 30 Infrastructure investments 32 -0.039950633 -0.130 23 27 33 Sectoral standards 30 -0.01883775 -0.100 24 23 26 Coordinating body for 29 -3.56E-92 0.010 25 9 8 climate strategy Tendering schemes 27 -0.007722291 -0.090 26.5 17 23 Advice or aid in 27 -0.23332128 -0.160 26.5 38 44 implementation Vehicle fuel-economy 26 -0.121643056 -0.140 28 35 37 and emissions standards CO2 taxes 24 -0.070794466 -0.140 30 32 35 Comparison label 24 -0.295844205 -0.150 30 39 41 35 Professional training and 24 -0.086480773 -0.130 30 33 34 qualification Voluntary approaches 22 -0.4055516 -0.160 33 40 45 RD&D funding 22 -0.147808378 -0.140 33 36 38 Technology deployment 22 -0.785065438 -0.170 33 43 46 and diffusion Net metering 21 -0.046446611 -0.090 35 29 24 Negotiated agreements 20 -0.817303447 -0.160 37.5 44 42 (public-private sector) Barrier removal 20 -0.020533423 -0.040 37.5 24 11 Technology development 20 -1.207826301 -0.180 37.5 45 49 GHG emissions 20 -0.20626666 -0.160 37.5 37 43 allowances GHG emission reduction crediting and offsetting 18 -0.03724937 -0.120 40 26 31 mechanism Demonstration project 17 -1.515658899 -0.180 41.5 46 48 Endorsement label 17 -1.541217952 -0.190 41.5 47 50 Formal & legally binding 16 -1.00E-91 0.010 44.5 10 10 energy efficiency target Funds to sub-national 16 -2.389566875 -0.190 44.5 49 53 governments Green certificates 16 -0.47777131 -0.150 44.5 41 40 Industrial air pollution 16 -0.040022487 -0.080 44.5 28 20 standards Procurement rules 14 -2.065639163 -0.190 47 48 52 Public voluntary schemes 9 -3.914408488 -0.190 48 50 54 User charges 8 -0.683070914 -0.120 49 42 29 Vehicle air pollution 7 -0.11436868 -0.050 50.5 34 12 standards White certificates 7 -6.1152688 -0.180 50.5 52 47 Retirement premium 6 -5.912576105 -0.150 52 51 39 Unilateral commitments 5 -11.25288486 -0.190 53 53 51 (private sector) Removal of fossil fuel 3 -13.20621314 -0.140 54 54 36 subsidies 36 Table S1.4: Rankings of Countries by Diversity, Fitness and ECI Diversity Diversity Fitness ECI Country Fitness ECI (degree) rank rank rank Korea, Rep. 45 9.811951 0.035 1.5 7 8 Japan 45 11.87499 0.032 1.5 4 17 China 44 7.128719 0.032 3 13 14 Germany 43 12.67766 0.036 4 2 6 India 42 14.20822 0.035 5 1 7 Australia 41 7.156223 0.038 6.5 12 4 France 41 9.76765 0.032 6.5 8 13 Italy 40 10.39098 0.033 8.5 6 10 United Kingdom 40 8.007024 0.034 8.5 10 9 Canada 39 9.513215 0.042 11 9 2 United States 39 12.08517 0.041 11 3 3 Brazil 39 6.449236 0.032 11 16 16 Mexico 38 11.05047 0.031 13 5 18 Spain 37 4.587078 0.037 14 17 5 Argentina 35 3.933089 0.032 15 21 15 Türkiye 34 1.766558 0.024 16 26 24 Indonesia 33 1.546163 0.020 17 27 29 South Africa 32 4.015283 0.026 18 20 22 Sweden 31 4.245544 0.046 19 18 1 Denmark 29 2.080262 0.021 20 25 26 Switzerland 28 7.444398 0.028 21 11 20 Russian Federation 27 0.635352 0.021 22 32 27 Chile 25 1.181485 0.020 23.5 28 28 Norway 25 2.103203 0.033 23.5 24 11 Colombia 23 0.764504 0.019 25.5 30 32 Viet Nam 23 0.965221 0.011 25.5 29 40 Ukraine 22 0.734341 0.019 27 31 31 Belgium 20 4.069776 0.030 28.5 19 19 Saudi Arabia 20 6.625173 0.025 28.5 15 23 Bulgaria 18 0.113659 0.010 32 49 41 Pakistan 18 0.496207 0.016 32 33 36 Thailand 18 0.36269 0.005 32 34 46 Kazakhstan 18 0.209228 0.012 32 41 39 Uzbekistan 18 0.113335 0.003 32 50 48 Uruguay 17 0.115006 0.001 35.5 48 51 Slovak Republic 17 0.197404 0.018 35.5 44 34 Iran, Islamic Rep. 16 7.018877 0.022 37 14 25 Malaysia 15 2.601595 -0.008 38 22 61 Greece 14 0.101293 0.007 39.5 51 44 Tunisia 14 0.228035 0.020 39.5 38 30 Georgia 13 0.12673 0.002 41 47 50 New Zealand 12 0.025921 -0.027 44.5 69 81 37 Austria 12 0.08735 0.005 44.5 53 45 Peru 12 0.06842 -0.006 44.5 57 58 Morocco 12 0.097084 -0.019 44.5 52 73 Tajikistan 12 0.079033 -0.023 44.5 55 77 United Arab Emirates 12 0.220862 -0.006 44.5 40 59 Belarus 11 0.027636 -0.028 51.5 68 83 Ecuador 11 0.025753 -0.012 51.5 70 67 Nigeria 11 0.02852 -0.026 51.5 67 79 Ireland 11 0.009089 -0.011 51.5 81 64 Philippines 11 0.009783 -0.024 51.5 80 78 Bhutan 11 0.13048 -0.007 51.5 46 60 Kuwait 11 2.157557 0.032 51.5 23 12 Uganda 11 0.204072 0.010 51.5 43 42 Portugal 10 0.008094 -0.005 59.5 83 57 Finland 10 0.326459 0.017 59.5 36 35 Algeria 10 0.060522 -0.016 59.5 59 72 Czechia 10 0.010564 -0.034 59.5 78 88 Tonga 10 0.082358 -0.021 59.5 54 74 Netherlands 10 0.018784 -0.030 59.5 74 84 Egypt, Arab Rep. 10 0.016433 -0.003 59.5 75 53 Dominican Republic 10 0.019752 -0.016 59.5 73 71 Slovenia 9 0.145733 0.027 67.5 45 21 Tanzania 9 0.061001 -0.021 67.5 58 75 Israel 9 0.037835 -0.023 67.5 64 76 Mozambique 9 0.034374 -0.016 67.5 66 69 Mongolia 9 0.074199 -0.016 67.5 56 70 Singapore 9 0.223807 -0.004 67.5 39 55 Costa Rica 9 0.308248 -0.028 67.5 37 82 Bangladesh 9 0.03686 -0.032 67.5 65 87 Guatemala 8 0.208139 -0.046 74 42 94 Romania 8 0.038 -0.001 74 63 52 Zambia 8 0.004868 -0.031 74 85 86 Syrian Arab Republic 8 0.046144 -0.010 74 61 63 Poland 8 0.016256 -0.011 74 76 65 Cameroon 7 0.03925 -0.013 78.5 62 68 Angola 7 0.352403 -0.051 78.5 35 97 Hungary 7 0.007475 -0.012 78.5 84 66 Suriname 7 0.023356 -0.047 78.5 71 95 Zimbabwe 6 0.003673 -0.041 83 88 92 Senegal 6 0.00393 -0.064 83 87 107 Rwanda 6 0.021187 -0.040 83 72 91 North Macedonia 6 7.62E-92 -0.065 83 102.5 109 Cuba 6 0.001539 -0.064 83 91 106 Seychelles 5 2.45E-92 -0.082 93 107 129 Luxembourg 5 0.000195 -0.010 93 93.5 62 38 Palau 5 7.62E-92 -0.055 93 98 98 Congo, Dem. Rep. 5 0.011759 -0.026 93 77 80 Ethiopia 5 1.07E-106 -0.084 93 119 130.5 Jamaica 5 0.008122411 -0.072 93 82 118 Venezuela, RB 5 0.003498215 -0.049 93 89 96 Micronesia, Fed. Sts, 5 5.34E-106 -0.071 93 113.5 117 Solomon Islands 5 0.002160121 -0.075 93 90 120 Vanuatu 5 9.04E-178 -0.091 93 132 136 Ghana 5 2.13E-106 -0.062 93 116 104 El Salvador 5 0.05192194 -0.005 93 60 56 Montenegro 5 1.07E-106 -0.069 93 124 116 Gabon 5 1.07E-106 -0.084 93 119 130.5 Kenya 5 6.40E-106 -0.064 93 110 105 Nepal 4 3.12E-206 -0.102 113 140.5 157.5 Yemen, Rep 4 2.45E-92 -0.079 113 107 125 Marshall Islands 4 9.04E-178 -0.097 113 132 141 Moldova 4 7.62E-92 -0.058 113 102.5 100.5 Paraguay 4 3.12E-206 -0.102 113 140.5 157.5 Papua New Guinea 4 5.34E-106 -0.088 113 113.5 132.5 Brunei Darussalam 4 9.04E-178 -0.097 113 132 141 Malawi 4 2.45E-92 -0.075 113 107 121 Tuvalu 4 3.12E-206 -0.102 113 140.5 157.5 Bosnia and Herzegovina 4 7.62E-92 -0.058 113 102.5 100.5 Cabo Verde 4 9.04E-178 -0.097 113 132 141 Trinidad and Tobago 4 0.001363611 -0.065 113 92 108 Croatia 4 1.52E-91 0.003 113 95 49 Korea, Dem. People’s Rep. 4 0.010282532 -0.037 113 79 89 Fiji 4 9.04E-178 -0.097 113 132 141 Maldives 4 3.12E-206 -0.102 113 140.5 157.5 Sri Lanka 4 1.07E-106 -0.077 113 119 122.5 Grenada 4 3.12E-206 -0.102 113 140.5 157.5 Honduras 4 5.34E-106 -0.088 113 113.5 132.5 Armenia 4 3.12E-206 -0.102 113 140.5 157.5 Iraq 4 0.004693288 -0.031 113 86 85 Jordan 4 1.07E-106 -0.075 113 124 119 Lebanon 4 9.04E-178 -0.097 113 132 141 Gambia, The 4 1.07E-106 -0.077 113 119 122.5 Iceland 4 6.40E-106 -0.061 113 111 102 South Sudan 3 2.24E-206 -0.099 138 151 148 Panama 3 2.35E-206 -0.113 138 147 169.5 Togo 3 1.07E-106 -0.068 138 124 110 Serbia 3 7.62E-92 -0.046 138 102.5 93 Liberia 3 9.04E-178 -0.097 138 132 145 Guinea-Bissau 3 9.04E-178 -0.097 138 132 145 Guinea 3 1.81E-206 -0.104 138 156 164.5 39 Lao PDR 3 2.35E-206 -0.113 138 147 169.5 Latvia 3 1.52E-91 0.007 138 96 43 Cyprus 3 0.00019465 0.019 138 93.5 33 Cambodia 3 2.45E-92 -0.077 138 107 124 Botswana 3 1.81E-206 -0.104 138 156 164.5 Benin 3 2.45E-92 -0.056 138 107 99 Lithuania 3 7.62E-92 0.004 138 97 47 Madagascar 3 2.24E-206 -0.099 138 151 148 Belize 3 9.04E-178 -0.082 138 132 128 Liechtenstein 3 5.34E-106 -0.080 138 113.5 127 Guyana 3 2.98E-206 -0.092 138 144 137 Niger 3 9.04E-178 -0.097 138 132 145 Monaco 3 2.35E-206 -0.113 138 147 169.5 Nicaragua 3 2.35E-206 -0.113 138 147 169.5 Andorra 3 2.35E-206 -0.113 138 147 167 Namibia 3 2.24E-206 -0.099 138 151 148 Barbados 3 1.81E-206 -0.104 138 156 163 Albania 3 1.07E-106 -0.061 138 124 103 French Polynesia 2 1.46E-206 -0.114 163 162 174 Estonia 2 7.62E-92 0.012 163 99.5 37.5 Bahamas 2 2.20E-206 -0.104 163 153.5 161.5 Djibouti 2 1.03E-206 -0.122 163 169 180.5 Burundi 2 1.46E-206 -0.114 163 162 174 Antigua and Barbuda 2 1.66E-206 -0.090 163 158.5 134 Niue 2 1.66E-206 -0.090 163 158.5 135 Azerbaijan 2 1.03E-206 -0.122 163 169 180.5 Dominica 2 1.46E-206 -0.114 163 162 174 Côte d'Ivoire 2 1.03E-206 -0.122 163 169 180.5 Afghanistan 2 1.03E-206 -0.122 163 169 177 Samoa 2 2.20E-206 -0.104 163 153.5 161.5 Kiribati 2 9.22E-207 -0.100 163 175 152 Nauru 2 9.04E-178 -0.079 163 132 126 Kyrgyzstan 2 1.46E-206 -0.114 163 162 174 Myanmar 2 1.07E-106 -0.040 163 119 90 Sierra Leone 2 9.22E-207 -0.100 163 175 152 Mauritius 2 1.03E-206 -0.122 163 169 180.5 Somalia 2 9.22E-207 -0.100 163 175 152 Malta 2 7.62E-92 0.012 163 99.5 37.5 Saint Vincent and the 2 1.03E-206 -0.122 163 169 180.5 Grenadines São Tomé and Príncipe 2 9.22E-207 -0.100 163 175 152 Qatar 2 9.22E-207 -0.100 163 175 152 Mali 2 1.03E-206 -0.122 163 169 180.5 Saint Lucia 2 1.46E-206 -0.114 163 162 174 Bahrain 1 7.79E-207 -0.068 185 181 113 40 Saint Kitts and Nevis 1 1.43E-207 -0.132 185 189 189 Mauritania 1 1.43E-207 -0.132 185 189 189 Bolivia 1 7.79E-207 -0.068 185 181 113 Burkina Faso 1 1.43E-207 -0.132 185 189 189 Haiti 1 1.43E-207 -0.132 185 189 189 Central African Republic 1 1.43E-207 -0.132 185 189 189 Chad 1 1.43E-207 -0.132 185 189 189 Cook Islands 1 8.86E-207 -0.112 185 178 166 Timor-Leste 1 7.79E-207 -0.068 185 181 113 San Marino 1 1.43E-207 -0.132 185 189 189 Libya 1 1.07E-106 -0.003 185 124 54 Lesotho 1 1.43E-207 -0.132 185 189 189 Equatorial Guinea 1 1.43E-207 -0.132 185 189 189 Eritrea 1 1.43E-207 -0.132 185 189 189 Sudan 1 7.79E-207 -0.068 185 181 113 Greenland 1 7.79E-207 -0.068 185 181 113 Turkmenistan 1 1.32E-206 -0.096 185 165 138 Oman 1 1.43E-207 -0.132 185 189 189 41 S2: ROBUSTNESS CHECKS BASED ON CONSTRUCTING THE COUNTRY-POLICY MATRIX WITH ONLY HIGH DATA QUALITY COUNTRIES The Climate Policy Database (v2021) contains data on climate policies for 198 countries but is most comprehensive for the 38 countries listed below1,2. Table S2.1: Countries in which the Climate Policy Database (v2021) has the most comprehensive coverage Argentina Japan Switzerland Nigeria Australia Korea, Rep. Chile Pakistan Brazil Mexico Colombia Thailand Canada Russian Federation Egypt, Arab Rep. Ukraine China Saudi Arabia Spain Uzbekistan France South Africa Iran, Islamic Rep. Venezuela, RB Germany Türkiye Iraq Viet Nam India United Kingdom Kazakhstan Indonesia United States Kuwait Italy United Arab Emirates Malaysia In Figure S2.1, we show that the nested (triangular) pattern is still present in a country-policy matrix when we only consider data for countries where the Climate Policy Database is most comprehensive (shown in Table S2.1). Nestedness (as measured by NODF) is 0.783. Moreover, the correlation between the policy prevalence (degree rankings) for this dataset and the unrestricted (full) data set is 0.88 suggesting results are highly similar. Country and policy rankings are shown in Table S2.2 and S2.3 below. Figure S2.1: Country policy matrix (ordered by degree) for only high data quality countries. 42 Table S2.2: Rankings of policies by Prevalence, Complexity and PCI when data is restricted to high data quality countries Prevalence Preval. Complexity PCI Policy instrument Complexity PCI (degree) rank rank rank Strategic planning 36 -0.10737 0.163 1 1 4 Energy and other taxes 33 -0.12187 0.143 2 3 6 Political & non-binding 32 -0.12183 0.153 3 2 5 GHG reduction target Tax relief 30 -0.18473 0.097 4 5 9 Institutional creation 29 -0.26689 0.069 5.5 7 13 Political & non-binding 29 -0.16138 0.187 5.5 4 2 renewable energy target Information provision 28 -0.37197 0.007 8.5 15 22 Grants and subsidies 28 -0.31628 0.046 8.5 10 16 Obligation schemes 28 -0.32027 0.048 8.5 11 15 Other mandatory 28 -0.36845 0.021 8.5 14 20 requirements Political & non-binding 27 -0.3643 0.009 12 13 21 climate strategy Feed-in tariffs or premiums 27 -0.30107 0.101 12 8 8 Product standards 27 -0.30781 0.051 12 9 14 Building codes and standards 26 -0.37461 -0.008 15 17 29 Loans 26 -0.37431 0.073 15 16 12 Monitoring 26 -0.39217 0.004 15 18 25 Infrastructure investments 25 -0.41894 0.005 17 19 24 Political & non-binding 23 -0.35085 0.085 18.5 12 10 energy efficiency target Auditing 23 -0.46462 0.006 18.5 20 23 Sectoral standards 22 -0.48601 0.041 21.5 21 18 Tendering schemes 22 -0.20138 0.107 21.5 6 7 Vehicle fuel-economy and 22 -0.5186 0.002 21.5 22 26 emissions standards Advice or aid in 22 -0.56411 -0.104 21.5 24 37 implementation Comparison label 21 -0.54492 -0.069 24 23 32 Technology deployment and 20 -0.57392 -0.071 25 27 33 diffusion RD&D funding 19 -0.56488 -0.056 27 25 31 Grid access and priority for 19 -0.57269 0.027 27 26 19 renewables Voluntary approaches 19 -0.62733 -0.121 27 28 40 Negotiated agreements 18 -0.68543 -0.100 30 31 36 (public-private sector) Technology development 18 -0.70779 -0.137 30 32 43 Professional training and 18 -0.67956 -0.003 30 30 27 qualification 43 Formal & legally binding 17 -0.6494 -0.006 32.5 29 28 renewable energy target CO2 taxes 17 -0.85296 -0.115 32.5 35 39 Coordinating body for 16 -0.90737 -0.085 35 37 34 climate strategy Demonstration project 16 -0.894 -0.207 35 36 48 GHG emissions allowances 16 -0.80853 -0.097 35 34 35 Net metering 15 -0.73411 0.076 37 33 11 Endorsement label 14 -1.32232 -0.225 39 41 52 Funds to sub-national 14 -1.34237 -0.227 39 42 53 governments GHG emission reduction crediting and offsetting 14 -0.95713 -0.135 39 38 42 mechanism Formal & legally binding 13 -1.15632 -0.153 41 39 44 climate strategy Procurement rules 12 -1.47636 -0.217 42.5 44 50 Formal & legally binding 12 -1.37746 -0.189 42.5 43 45 GHG reduction target Green certificates 11 -1.67616 -0.195 44 47 47 Industrial air pollution 10 -1.49134 -0.114 45 45 38 standards Barrier removal 8 -1.20532 0.164 46.5 40 3 Public voluntary schemes 8 -2.2599 -0.241 46.5 49 54 User charges 7 -1.5645 0.043 48.5 46 17 White certificates 7 -2.62549 -0.191 48.5 50 46 Retirement premium 6 -1.98155 -0.051 50 48 30 Unilateral commitments 5 -4.02976 -0.216 51.5 53 49 (private sector) Formal & legally binding 5 -2.84824 -0.129 51.5 52 41 energy efficiency target Removal of fossil fuel 3 -2.64038 0.417 53.5 51 1 subsidies Vehicle air pollution 3 -6.78271 -0.223 53.5 54 51 standards 44 Table S2.3: Rankings of Countries by Diversity, Fitness and PCI when data is restricted to high quality data countries Diversity Diversity Fitness ECI Country Fitness ECI (degree) rank rank rank Korea, Rep. 45 1.830752 0.120 1.5 2 3 Japan 45 2.288883 0.099 1.5 1 6 China 44 1.825228 0.118 3 3 4 Germany 43 1.67648 0.054 4 6 15 India 42 1.812622 0.060 5 4 13 France 41 1.593207 0.063 6.5 10 12 Australia 41 1.501673 0.108 6.5 12 5 United Kingdom 40 1.641539 0.121 8.5 7 2 Italy 40 1.60704 0.073 8.5 9 10 Canada 39 1.547022 0.128 11 11 1 United States 39 1.612698 0.092 11 8 7 Brazil 39 1.480935 0.069 11 13 11 Mexico 38 1.451607 -0.003 13 15 17 Spain 37 1.680989 0.057 14 5 14 Argentina 35 1.109182 0.000 15 16 16 Türkiye 34 0.988869 -0.051 16 17 18 Indonesia 33 0.899604 -0.065 17 19 21 South Africa 32 0.958437 -0.055 18 18 19 Switzerland 28 1.458719 0.084 19 14 9 Russian Federation 27 0.707735 -0.057 20 20 20 Chile 25 0.70405 -0.101 21 21 22 Viet Nam 23 0.517388 -0.128 22.5 27 24 Colombia 23 0.620089 -0.106 22.5 22 23 Ukraine 22 0.549045 -0.144 24 24 25 Saudi Arabia 20 0.553907 -0.213 25 23 32 Kazakhstan 18 0.402964 -0.153 27.5 29 26 Thailand 18 0.334315 -0.187 27.5 31 29 Uzbekistan 18 0.535026 -0.169 27.5 25 27 Pakistan 18 0.483039 -0.201 27.5 28 30 Iran, Islamic Rep. 16 0.521051 -0.234 30 26 33 Malaysia 15 0.370586 -0.206 31 30 31 United Arab Emirates 12 0.214403 -0.170 32 33 28 Kuwait 11 0.305124 0.092 33 32 8 Egypt, Arab Rep. 10 0.144139 -0.298 34 34 34 Venezuela, RB 5 0.039966 -0.465 35 35 36 Iraq 4 0.031689 -0.442 36 36 35 45 S3: ALTERNATIVE FUNCTIONAL FORMS FOR MEASURING POLICY RELATEDNESS AND POLICY ALIGNMENT AND ASSOCIATED ROBUSTNESS CHECKS As discussed in the Methods section, we consider two alternative approaches for establishing whether a country has prior experience with a policy. The first (specified in equation 4 in the main text) considers whether country has announced any policy of type over the course of history covered by the Climate Policy Database (i.e. ≥ 1). The second considers whether country has announced a greater share of policies of type compared to the average across countries (i.e. ≥ 1). Either approach can be used to construct the binary matrix , that we use to calculate the Policy Relatedness measure ( ) between two policies or the Policy Alignment measure ( ) between a country and a new policy. The difference between these two approaches is shown in Table S3.1 (see Measure 1 and Measure 3). A further functional form variant is to consider calculating the Policy Relatedness ( ) between two policies using the same approach introduced by Hidalgo et al (2007)5. In the main text, we presented results based on the following equation, which is increasing in the co-occurrence of policies within countries, but weights this by how many policies a given country has introduced (represented by ): ∑ ∑ = ( ∑ , ∑ ) where = ∑ . However, we can also consider a Policy Relatedness measure that does not include this additional weight. That is: ∑ ∑ ′ = ( , ) ∑ ∑ Table S3.1 also shows results when using this simpler functional form for Policy Relatedness (see Measure 2 and Measure 4). In all cases, we find little difference across our results, suggesting a high degree of robustness to different choices of functional form. Table S3.1 Regression analysis of predictive power of different policy alignment measures (1) (2) (3) (4) (5) (6) (7) Measure 1: Policy alignment measure (weighted) constructed on the basis of N (baseline) 0.568*** 0.522*** 0.516*** 0.504*** 0.509*** 0.513*** 0.490*** (0.018) (0.027) (0.028) (0.028) (0.029) (0.028) (0.032) Measure 2: Policy alignment measure (unweighted) constructed on the basis of N 0.565*** 0.519*** 0.514*** 0.500*** 0.506*** 0.510*** 0.487*** (0.016) (0.024) (0.025) (0.025) (0.026) (0.025) (0.029) Measure 3: Policy alignment measure (weighted) constructed on the basis of Revealed Policy Prevalence (RPP) rather than N 0.780*** 0.600*** 0.579*** 0.569*** 0.569*** 0.573*** 0.527*** (0.041) (0.045) (0.046) (0.045) (0.046) (0.045) (0.049) Measure 4: Policy alignment measure (unweighted) constructed on the basis of Revealed Policy Prevalence (RPP) rather than N 0.801*** 0.647*** 0.628*** 0.615*** 0.617*** 0.622*** 0.576*** (0.037) (0.044) (0.045) (0.045) (0.046) (0.045) (0.050) N 109,787 87,616 85,264 77,591 82,619 85,264 76,042 Policy Yes Yes Yes Yes Yes Yes Yes FE Period Yes Yes Yes Yes Yes Yes Yes FE 0.248, [0.248], 0.25, [0.25], 0.25, [0.25], 0.253, [0.25], 0.251, [0.25], 0.25, [0.25], 0.26, [0.256], R2 {0.22}, (0.23) {0.23}, (0.236) {0.23}, (0.24) {0.238}, (0.241) {0.235}, (0.238) {0.234}, (0.24) {0.242}, (0.24) Control 1 2 3 4 5 6 7 set Note: The outcome variable is a binary that is equal to 1 if at least one policy of a given instrument in a given country has been adopted in the test period, and 0 otherwise. The point estimates show the association of four different policy alignment measures: policy alignment with unweighted relatedness, policy alignment with weighted relatedness (x- axis), and their respective RPP versions. Each coefficient is based on a separate regression model. The estimation approach uses a linear probability model (OLS) using the full sample, Each column differs with respect to the control set included in the regressions. Policy type fixed effects and period fixed effects are included in all estimations. Control Set 1 includes no additional controls (no country fixed effects). Control Set 2 includes measures of population size, income levels, school enrollment rates, and public debt. Control Set 3 adds indicators for carbon emissions per capita, precipitation, and air quality. Control Set 4 incorporates the vote share of the strongest government party. Control Set 5 replaces the vote share of the strongest government party with the total number of votes in the legislature. Control Set 6 substitutes the number of votes with a measure of democratic governance. Control Set 7 includes measures of population size, income levels, school enrollment rates, public debt, carbon emissions per capita, precipitation, and air quality. It also incorporates the vote share of the strongest government party, post-election unrest, a measure of democratic governance, and various indicators of governance quality, including control of corruption, rule of law, regulatory quality, government effectiveness, political stability, and voice and accountability. In the R2 row, the first (no parenthesis), second (brackets), third (curly braces) and fourth (parenthesis) values refer to the model with the estimate of measure 1, measure 2, measure 3 and measure 4, respectively. Robust standard errors clustered at the country level in parenthesis. *p<0.1; **p<0.05; ***p<0.01. 46 S4: FURTHER ROBUSTNESS ANALYSIS BASED ON DIFFERENT REGRESSION MODELS Figure S4.1: The Association between Past Policy Alignment and Future Policy Adoption Across Varying Data Samples and Estimation Methods Note: The outcome variable is a binary that is equal to 1 if at least one policy of a given instrument in a given country has been adopted in the test period, and 0 otherwise. The point estimates show the association of two different policy alignment measures: policy alignment with unweighted relatedness (red), and policy alignment with weighted relatedness (grey). The coefficients within each panel differ with respect to the train and test period (y-axis). The left panel is based on a linear probability model using all countries available in the sample; the panel in the middle is based on linear probability model using only countries with sufficiently high data coverage; the right panel is based on a logistic regression using all countries available in the sample. In addition to policy type fixed effects, each estimation includes measures of population size, income levels, school enrollment rates, public debt, carbon emissions per capita, precipitation, and air quality, the vote share of the strongest government party, post-election unrest, a measure of democratic governance, and various indicators of governance quality, including control of corruption, rule of law, regulatory quality, government effectiveness, political stability, and voice and accountability. Thin and thick horizontal lines around point estimates represent 99% and 95% confidence intervals based on robust standard errors clustered at the country level, respectively. 47 48 Figure S4.2: The Association between Past Policy Alignment and Future Policy Adoption Across Varying Data Samples and Estimation Methods using a rolling-window approach Note: The outcome variable is a binary that is equal to 1 if at least one policy of a given instrument in a given country has been adopted in the test period, and 0 otherwise. The point estimates show the association of two different policy alignment measures: policy alignment with unweighted relatedness (red), and policy alignment with weighted relatedness (grey). The coefficients within each panel differ with respect to the train and test period (y-axis). The left panel is based on a linear probability model using all countries available in the sample; the panel in the middle is based on linear probability model using only countries with sufficiently high data coverage; the right panel is based on a logistic regression using all countries available in the sample. In addition to policy type fixed effects, each estimation includes each estimation includes measures of population size, income levels, school enrollment rates, public debt, carbon emissions per capita, precipitation, and air quality, the vote share of the strongest government party, post-election unrest, a measure of democratic governance, and various indicators of governance quality, including control of corruption, rule of law, regulatory quality, government effectiveness, political stability, and voice and accountability. Thin and thick horizontal lines around point estimates represent 99% and 95% confidence intervals based on robust standard errors clustered at the country level, respectively. 49 Figure S4.3: Robustness of regression coefficient to different controls, data test and train time periods and policy alignment functional form Note: The outcome variable is a binary that is equal to 1 if at least one policy of a given instrument in a given country has been adopted in the test period, and 0 otherwise. The point estimates show the association of two different policy alignment measures: policy alignment with unweighted relatedness (left panel), and policy alignment with weighted relatedness (right panel). The coefficients within each panel differ with respect to the train and test period (x-axis) and the control set included in the regressions (differentiated by colors). Policy type fixed effects are included in all estimations. The control sets are described in the footnote of S4.1. Thin and thick vertical lines around point estimates represent 99% and 95% confidence intervals based on robust standard errors clustered at the country level, respectively. 50 Figure S4.4: Robustness of regression coefficient to different controls, data test and train time periods and policy alignment functional form using a rolling-window approach Note: The outcome variable is a binary that is equal to 1 if at least one policy of a given instrument in a given country has been adopted in the test period, and 0 otherwise. The point estimates show the association of two different policy alignment measures: policy alignment with unweighted relatedness (left panel), and policy alignment with weighted relatedness (right panel). The coefficients within each panel differ with respect to the train and test period (x-axis) and the control set included in the regressions (differentiated by colors). Policy type fixed effects are included in all estimations. The control sets are described in the footnote of S4.1. Thin and thick vertical lines around point estimates represent 99% and 95% confidence intervals based on robust standard errors clustered at the country level, respectively. 51 Figure S4.5: The Association between Past Policy Alignment and Future Policy Adoption on High- and Low-Income Countries (1960-2021) Note: The outcome variable is a binary that is equal to 1 if at least one policy of a given instrument in a given country has been adopted in the test period, and 0 otherwise. The point estimates show the association of two different policy alignment measures: policy alignment with unweighted relatedness (red), and policy alignment with weighted relatedness (grey). The coefficients within each panel differ with respect to country sample they are estimated on: high- income (upper row) and low income (lower row) countries. The train and test period is 1960-2009 and 2010-2021, respectively. The left panel is based on a linear probability model using all countries available in the sample; the right panel is based on a logistic regression using all countries available in the sample. In addition to policy type fixed effects, each estimation includes each estimation includes measures of population size, income levels, school enrollment rates, public debt, carbon emissions per capita, precipitation, and air quality, the vote share of the strongest government party, post-election unrest, a measure of democratic governance, and various indicators of governance quality, including control of corruption, rule of law, regulatory quality, government effectiveness, political stability, and voice and accountability. Thin and thick horizontal lines around point estimates represent 99% and 95% confidence intervals based on robust standard errors clustered at the country level, respectively. 52 S5: IDENTIFYING KEY POLICY CLUSTERS IN THE CLIMATE POLICY SPACE WITH DIFFERENT COMMUNITY DETECTION ALGORITHMS To identify key policy clusters in the Climate Policy Space network, we first apply the k-means algorithm to the policy relatedness adjacency matrix. To determine the number of clusters to identify (k), we use the Elbow method and calculate the within-cluster some of squares (the squared distance between each point and cluster centroids) for various values of k (see Figure S5.1). As there does not appear to be a distinct ‘elbow’ inflection point in the graph, we set k = 3, 5 and 8 and illustrate these clusters in Figure S5.2 below (panel a-c). Figure S5.1: The Elbow method for identifying the optimal number of clusters (k) We also experiment with the Louvain algorithm6 for identifying clusters, which identifies the number of clusters by optimizing modularity, which is the proportion of edges that fall within the given clusters less the expected proportion if edges were distributed at random7. Modularity ranges in value from (–0.5, 1). Here, 3 clusters are identified, although modularity is not particularly high (modularity = 0.437) (see panel d of Figure S5.2). The K means (k = 3) and the Louvain algorithm identify clusters that are broadly similar, although in the k means algorithm policies relating to binding targets and institutional creation are grouped with the middle cluster comprising regulatory instruments and enabling policies, while the Louvain algorithm groups binding and non-binding targets together. For Figure 2 shown in the main text, we show the clusters identified by the k means (k=5) method, as this usefully distinguishes between binding and non-binding targets, and retains the broad groupings of the Louvain and k means (k=3) approaches. K means (k=8) appeared to give clusters that were too granular to be meaningful. 53 Figure S5.2: Different clusters in the Climate Policy Space Network identified with different community detection algorithms 54 S6: EMISSIONS REDUCTION ASSOCIATION ANALYSIS Figure S6.1 and Table S6.1 show a summary of the emissions association analysis for all policy instruments used in this paper. Estimates for the emissions association are shown for policies where there were at least 30 observations and the result was statistically significant. As discussed in the Methods section, our regression approach follows that taken by Eskander and Fankhauser (2020)8 and includes an rule of law indicator, GDP per capita, squared GDP per capita, the cyclical component of GDP based on the Hodrick-Prescott decomposition, import share, size of the service sector, temperature fluctuation, and a binary capturing the presence of a federal system as control indicators. Country and year fixed effects are also included. Figure S6.1: Short and long term emission association estimates for different policy types Note: The figure plots the short- and long-term emission reduction potential of different policy types. The dependent variable refers to (log) CO2 emission per unit of economic output (MtCO2e/GDP). Estimates for the emission reduction potential are shown for policies where there were at least 30 observations. Our approach follows Eskander and Fankhauser (2020), and takes the following econometric form: = + , , + + + + + with as (log) CO2 emission per unit of economic output , (MtCO2e/GDP) of country i in year t ; as the number of policies of a given policy type P adopted in the last three years; , as the number of policies that have been adopted more than three years ago; as a vectors of control variables that captures the first lag of the following factors: rule of law indicator, GDP per capita, squared GDP per capita, the cyclical component of GDP based on the Hodrick-Prescott decomposition, import share, size of the service sector, temperature fluctuation, and a binary capturing the presence of a federal system as control indicators. Country and year fixed effects are included. Robust standard errors clustered at the country level. Standard errors are based on a 95% confidence interval. 55 Table S6.1: Policy instruments and their associated emissions association relationship Policy instrument Number of policies Date of first More than 30 Statistically Emission reduction announced introduction Obs? significant? coefficient Advice or aid in implementation 190 1970 Y Y -0.00373*** Auditing 81 1980 Y Y -0.01745*** Barrier removal 26 2001 N N Building codes and standards 124 1970 Y Y -0.03066*** CO2 taxes 47 1991 Y Y -0.0782*** Comparison label 77 1985 Y Y -0.0154*** Coordinating body for climate strategy 36 1997 Y N Demonstration project 113 1974 Y Y -0.00701*** Endorsement label 56 1982 Y Y -0.02185*** Energy and other taxes 123 1944 Y Y -0.03707*** Feed-in tariffs or premiums 188 1991 Y Y -0.01792*** Formal & legally binding GHG reduction target 50 2003 Y Y -0.22483*** Formal & legally binding climate strategy 62 1998 Y Y -0.05816*** Formal & legally binding energy efficiency target 27 2003 N N Formal & legally binding renewable energy target 82 2000 Y Y -0.143*** Funds to sub-national governments 56 1976 Y Y -0.02545*** GHG emission reduction crediting and offsetting mechanism 43 1980 Y Y -0.01559*** GHG emissions allowances 40 1999 Y Y -0.04677*** Grants and subsidies 464 1976 Y Y -0.00736*** Green certificates 37 1999 Y Y -0.04087*** Grid access and priority for renewables 59 1990 Y N Industrial air pollution standards 20 1992 N N Information provision 313 1970 Y Y -0.00266*** Infrastructure investments 159 1985 Y Y -0.01112*** Institutional creation 228 1964 Y Y -0.00887*** Loans 123 1974 Y Y -0.01482*** Monitoring 155 1970 Y Y -0.00955*** Negotiated agreements (public-private sector) 129 1980 Y Y -0.00247*** Net metering 29 1991 Y N Obligation schemes 130 1976 Y Y -0.05157*** Other mandatory requirements 204 1941 Y Y -0.00767*** Political & non-binding GHG reduction target 238 2004 Y Y -0.03281*** Political & non-binding climate strategy 211 1993 Y N Political & non-binding energy efficiency target 66 1997 Y N Political & non-binding renewable energy target 172 1996 Y N Procurement rules 47 1992 Y Y -0.01981*** Product standards 162 1976 Y N Professional training and qualification 49 1985 Y N Public voluntary schemes 24 1997 N N RD&D funding 78 1974 Y Y -0.01094*** Removal of fossil fuel subsidies 3 2010 N N Retirement premium 13 2001 N N Sectoral standards 130 1970 Y Y -0.00946** Strategic planning 1039 1958 Y Y -0.00366*** Tax relief 282 1975 Y Y -0.00773*** Technology deployment and diffusion 111 1975 Y Y -0.00633*** Technology development 108 1975 Y Y -0.0045*** Tendering schemes 50 2003 Y N Unilateral commitments (private sector) 9 1993 N N User charges 10 1982 N N Vehicle air pollution standards 8 1996 N N Vehicle fuel-economy and emissions standards 111 1970 Y Y -0.00694** Voluntary approaches 108 1982 Y Y -0.00491*** White certificates 15 1994 N N Note: The table shows the number of (announced) policies for each policy instrument. The last column shows the estimate of the emission reduction potential. The dependent variable refers to (log) CO2 emission per unit of economic output (MtCO2e/GDP). Estimates for the emission reduction potential are shown for policies where there were at least 30 observations, and the result was statistically significant. As discussed in the Methods section, our regression approach follows that taken by Eskander and Fankhauser (2020) and includes a rule of law indicator, GDP per capita, squared GDP per capita, the cyclical component of GDP based on the Hodrick-Prescott decomposition, import share, size of the service sector, temperature fluctuation, and a binary capturing the presence of a federal system as control indicators. 56 Country and year fixed effects are also included. Robust standard errors clustered at the country level. *p<0.1; **p<0.05; ***p<0.01. 57 SUPPLEMENTARY INFORMATION REFERENCES 1. Nascimento, L., Kuramochi, T., Iacobuta, G., den Elzen, M., Fekete, H., Weishaupt, M., & Höhne, N. (2022). Twenty years of climate policy: G20 coverage and gaps. Climate policy, 22(2), 158-174. 2. Linsenmeier, M., Mohommad, A., & Schwerhoff, G. (2022). Policy sequencing towards carbon pricing among the world’s largest emitters. Nature Climate Change, 12(12), 1107- 1110. 3. Tacchella A, Cristelli M, Caldarelli G, Gabrielli A, Pietronero L. A new metrics for countries' fitness and products' complexity. Sci Rep. 2012;2(1):723. 4. Hidalgo, C. A., & Hausmann, R. (2009). The building blocks of economic complexity. Proceedings of the national academy of sciences, 106(26), 10570-10575. 5. Hidalgo, C. A., Klinger, B., Barabási, A. L., & Hausmann, R. (2007). The product space conditions the development of nations. Science, 317(5837), 482-487. 6. Blondel VD, Guillaume JL, Lambiotte R, Lefebvre E. Fast unfolding of communities in large networks. J Stat Mech. 2008;(10):P10008. 7. Newman ME. Modularity and community structure in networks. Proc Natl Acad Sci. 2006;103(23):8577-8582. 8. Eskander, S. M., & Fankhauser, S. (2020). Reduction in greenhouse gas emissions from national climate legislation. Nature Climate Change, 10(8), 750-756.