2.1. Conceptualising deliberate decline

Within the various literatures engaging with climate policy and decarbonisation, the intentional decline of carbon-intensive systems has emerged as an increasingly salient topic [2–4]. Previous scholarship has characterised the process of deliberate decline as ‘the managed erosion of lock-ins that perpetuate the production and consumption of fossil fuels’ [2]. Such lock-ins span multiple dimensions, including techno-economic (e.g., sunk investments in carbon-intensive infrastructure), political-institutional (e.g., entrenched incumbent interests or market rules that encourage continued production and use of fossil-fuel technologies), and behavioural (e.g., user norms and practices that perpetuate carbon-intensive lifestyles) [44,45].

The decline of carbon-intensive systems has been extensively examined through the lens of socio-technical transitions [12,46–48]. This tradition conceptualises carbon-intensive systems as large functional systems – for instance, transport, electricity and agri-food – that fulfil critical societal needs such as mobility, energy and food [49]. These systems are stabilised and reproduced by socio-technical regimes, made up of dominant configurations of material elements (technologies, materials, infrastructures) and non-material elements (policies and regulations, business models, user practices) [50,51]. A paradigmatic example is the socio-technical regime underpinning personal road-based transport. Centred around the ICE vehicle, this regime is supported by material elements such as road networks, refuelling infrastructure and vehicle manufacturing just as much by non-material factors like regulatory frameworks, business models (e.g., loans and leasing), user behaviour (e.g., car-based commuting to work) and wider cultural norms that valorise automobility by associating it with freedom and personal success [52–54]. This socio-technical perspective suggests that relevant targets for decline-seeking actions are not limited to technologies and materials but also extend to infrastructure, industries, institutions and individual practices.

To elucidate the processes contributing to the destabilisation and decline of regimes in large socio-technical systems, much of the transitions scholarship has relied on cases of historical or ongoing shifts [4,30,55,56]. Predominantly focused on cases across Europe and North America, this body of work has explored the decline of coal extraction [18,29], coal-fired and nuclear power [19,21,31,57,58], fossil-fuel heating [59] and legacy transport systems such as tramways and steamships [32,55,60]. Scholars have also examined unsustainable practices such as cloud seeding [61] and fishing [62]. While the observed declines of these historical socio-technical systems largely occurred due to changed market and societal conditions, these studies underscore the pivotal role of public policy in catalysing, accelerating or cementing the discontinuation of unsustainable socio-technical systems or specific elements. This recognition of the importance of public policy aligns with a growing body of scholarship examining the conditions that can accelerate socio-technical transitions, which, as historical analyses show, typically require decades to unfold [35,56,63,64]. This agency-centred perspective, which emphasises decline as a product of deliberate policy intervention [9,15], stands in contrast to more processual accounts that tend to view decline as an emergent outcome of broader structural shifts, such as deteriorating economic or technological competitiveness, changed user preferences and societal crises [16,29,31,65]. To be sure, agency has an important role to play in these processual accounts, including through legitimacy struggles surrounding established trajectories and the eventual defection of incumbent players away from these trajectories.

Drawing on this literature, we adopt a broad conceptualisation of deliberate decline, focusing on policies designed to reduce or eliminate one or more core elements of a carbon-intensive system. Consistent with prior research [3,66], our empirical focus includes policies targeting elements such as technologies and infrastructures (e.g., ICE vehicles, fossil-fuel boilers, industry equipment), substances and fuels (e.g., plastics, ozone-depleting substances), industry processes (e.g., landfilling, logging, gas flaring) and practices (e.g., flying). Building on earlier conceptualisations of intentional decline [2,67], we define a decline policy as ‘a governance intervention intended or expected to induce the partial reduction or complete discontinuation of the production or use of one or more incumbent elements within a carbon-intensive system.’

2.2. Direct decline and indirect decline

While the above definition captures policies that directly target carbon-intensive system elements for reduction or discontinuation (e.g., legislation to phase out coal-fired power), it is equally important to recognise policies that indirectly induce decline by replacing incumbent elements with cleaner alternatives. Examples of such policies include renewable energy portfolio standards and biofuel mandates, which both set percentage targets for substituting fossil-based fuels with a minimum share of renewable energy, or subsidies supporting the removal and replacement of emissions-intensive technologies (e.g., ICE vehicles, incandescent public lighting and fossil-fuel heating systems) with low-emission alternatives. Assuming one-for-one substitution, such policies result in an inverse reduction of the incumbent element by the same proportion as the diffusion that occurs.

To illustrate the relationship between direct and indirect decline policies, we propose a conceptual framework shown in Fig 1. Informed by prior research [2,12,68] emphasising the dialectical yet complementary relationship between decline and innovation dynamics in sustainability transitions, this framework depicts two contrasting but mutually supporting policy approaches through which decline can be deliberately pursued. Additionally, this dual view of decline stresses the role of agency and the multiple points from which policymakers can induce socio-technical transitions [9,63,69].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Direct and indirect decline policies.

https://doi.org/10.1371/journal.pone.0334512.g001

We conceive direct decline policies as interventions that reduce the use and/or production of a carbon-intensive element by operating on the downward sloping curve reflecting the relative position of the incumbent element. Direct decline policies thus primarily encompass a destructive function and signal to market actors ‘where not to go’. Illustrative examples include bans, restrictions or mandates to curtail logging, landfilling and fossil-fuel extraction [70], as well as phase-out schedules for technologies such as coal-fired power [18], incandescent lightbulbs [13,71] or substances like ozone-depleting compounds [72].

Conversely, we conceive indirect decline policies as interventions that contribute to decline processes by upscaling alternatives in a fashion which displaces the incumbent carbon-intensive element by an inverse proportion [15,31]. Policies falling into this category include the aforementioned renewable energy portfolio and biofuel mandates as well as ZEV sale targets, which all share the feature of aiming for a specific level of substitution by setting diffusion targets or minimum thresholds. Similarly, policies that mandate or subsidise the replacement of diesel trains, buses or ferries with electric ones, or that promote the substitution of fossil-fuel combustion in industry with electrification or hydrogen, are also expected to result in a proportionate reduction in the use and/or production of the incumbent element. So, if direct decline policies chiefly signal ‘where not to go’, indirect decline policies contribute to the decline process by signalling ‘where to go,’ while at the same time spurring substitution of the incumbent system element.

But not every diffusion policy can be considered an indirect decline policy. This is because many leave room for the uninterrupted continuation of incumbent trajectories, especially in expanding markets where relative decarbonisation gains from a new technology’s diffusion can be offset by the absolute growth of the incumbent system. One mechanism that can undermine decline outcomes is layering, when a new technology, substance or practice diffuses without displacing or substituting the incumbent element to a comparable degree [73,74]. The case of consumer subsidies for ZEVs provides a compelling example. In some jurisdictions and contexts, ZEV-promoting policies have encouraged the purchase of a second vehicle, thus failing to reduce the size of the existing fleet of traditional ICE vehicles [75,76]. A related mechanism is rebound, when per-unit efficiency gains are negated by increased overall use of the incumbent system element. This is particularly relevant for energy-efficiency policies, where rebound effects have been observed to erase gains in reducing fossil-fuel consumption in various domains. In the household sector, this occurs when occupants increase indoor temperatures on their thermostats or when larger dwellings are built [77,78]. Similarly, in the transport sector, many hybrid vehicle owners respond to reduced mile-per-gallon costs by driving longer distances [79].

These concerns about layering and rebound suggest that the potential of certain policy types to reduce the relative share of the incumbent element – whether directly or indirectly – is less certain than other policy types, ceteris paribus. Accordingly, as illustrated in Table 1, we limit our empirical scope in this study to policy types where a decline effect can be anticipated with a high degree of certainty. In doing so, we exclude policies such as energy efficiency upgrades and standards, purchase subsidies without replacement requirements, and broad emissions-reduction goals. By narrowing our empirical scope to policies that explicitly perform a destructive or substitutive function, our conception of two decline approaches distinguishes itself from other work [12,28,80] that take interest in all types of innovation and diffusion policies.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Empirical scope of study and illustrative examples.

https://doi.org/10.1371/journal.pone.0334512.t001

While we do not dismiss the possibility that some of the excluded policy types may, in certain contexts, contribute to the destabilisation or reduction of incumbent carbon-intensive elements, their lack of an explicit destructive or substitutive function places them out of scope from our analysis. With respect to our decision to include all forms of carbon pricing, such as carbon taxes and emissions trading, we follow Kivimaa and Kern’s [12] analysis of policy mixes for ‘creative destruction’, which underscores pollution taxes as critical levers for exerting pressure on emissions-intensive regimes.

A key question arising at this point is which decline approach – direct or indirect – is more effective at spurring decarbonisation and socio-technical transitions. Existing transitions research again provides useful guidance here, suggesting that in most cases successful transitions require both the breakdown of incumbent elements and the emergence of new ones [81,82]. So, while direct decline catalyses the reduction or discontinuation of incumbent arrangements through their destructive function, indirect decline policies contribute to decline through displacement and substitution with the competing technology or practice.

Moreover, indirect decline policies may, at times, prove more socio-politically feasible and easier to justify than more confrontational direct decline policies. Past experiences with direct decline policies targeting established technologies or industries reveal how incumbent power can weaken and delay them [19,83,84]. Thus, to mitigate fears about job losses, asset write-offs and political backlash – challenges repeatedly associated with direct decline policies [18,66,85] – governments may choose to embed decline intentions into innovation-focused agendas. For instance, policies within the United States’ Inflation Reduction Act introduced under the Biden administration indirectly advanced the decline of fossil fuels, while prioritising job creation, industrial development and economic growth through green industrial policies. This approach underscores the higher political viability of indirect decline in some circumstances, as promises of green technology development and diffusion can temper resistance from incumbent industry, such as lobbying, litigation and political discourse.

2.3. An index of ‘decline intensity’

A critical determinant of the ability for decline policies – whether direct or indirect – to achieve their intended effect is their design. For instance, policies can differ vastly with respect to their strictness [13,86]. In most cases, mandatory or regulatory instruments offer greater assurance that a particular degree of reduction will occur compared to softer tools like voluntary agreements [87]. Similarly, establishing explicit numerical targets regarding the magnitude of change (e.g., 50% versus 100%) or the speed of change (e.g., immediate, as in the case of some bans, or gradual and multi-decadal, as seen in many phase-out schedules) can play an important role by signalling to the market how much change is required within what timeframe [10,88].

These insights point to the need to consider both quantity and quality in studying policy design [41–43]. Take, for instance, the case of a legally enforced schedule that commits a country’s power sector to phase-out coal-fired power by 2038, as is the case in Germany. Then compare this to a softer ambition to reduce reliance on fossil-fuel heating by offering financial support for conversion to non-fossil heating or heat pumps through a voluntary programme aimed only at public buildings in a specific region, with no explicit guidance on the market share targeted for reduction. While both policies aim to reduce volumes of the incumbent emissions-intensive technology, the former example of regulatory-backed phase-out at the national scale with a specified schedule can be expected to exert a much larger reduction effect. By the same token, the latter policy can be expected to lead to much larger GHG emissions reductions than policies with weaker decline ambitions or stringency.

Thus, because not all decline policies are created equal, merely counting the number of policies in a particular country or sector may not be a good indicator of the potential decline or decarbonisation effect [86]. Focusing only on quantity while ignoring the qualitative design features of a policy hence risks obscuring variation in policy stringency and undermining the validity of comparative assessments [89]. We deal with this challenge by building on the concept of ‘policy intensity’ [41] coined by policy design scholars to develop a novel ‘decline intensity’ index (elaborated in Methods). We apply this index to score policies with respect to key design features we expect to influence their ability to achieve large-scale and rapid decline, which also impact the degree of GHG reductions achieved.

3.1. Step one: Selection of countries

Annex I parties with emissions reporting obligations under the Kyoto Protocol consist of 44 countries. These collectively represent 44.7% of cumulative GHG emissions globally since 1990 (assigning emissions within the European Union to its member countries to avoid double counting; excluding emissions from land use, land-use change and forestry [LULUCF]) [90]. Our analysis focuses on the ten Annex I countries with the largest cumulative GHG emissions since 1990, whose Fifth Biennial Reports were available in English. In descending order of emissions, our sample comprises the United States, Japan, Germany, Canada, United Kingdom, Australia, France, Italy, Poland and Kazakhstan. Together, these countries account for 31.8% of global cumulative GHG emissions since 1990 (ibid). The selection of these ten countries allows us to investigate the extent to which decline policies have been implemented in high-emitting and industrialised economies while maintaining a sample size suited to piloting the approach developed herein.

Biennial reports from the European Union (EU), Russia, Spain and Ukraine – countries that would otherwise rank among the top ten global emitters – were excluded from the sample. This decision was made to avoid double counting the EU with its member countries, to address the unavailability of documents in English from Spain and Russia, and to account for the wartime circumstances in Ukraine, which we expected would hinder the ability to enact decarbonisation policies and collect reliable data on GHG emissions.

3.2. Step two: Extract decline polices

Decline policies were identified by manually examining descriptions in each country’s ‘Policies and Measures’ table, along with corresponding explanations in Chapter 4 of the Fifth Biennial Report. We determined the relevance of policies in accord with our conceptualisation in Table 1 and the identification procedure outlined in Table 2. We also provide specific examples of decisions about relevant and irrelevant policies in Tables S1 and S2 in S3 File.

The first question in Table 2 identifies direct decline policies that explicitly state the objective to force or induce the reduction or discontinuation of a carbon-intensive technology, substance/fuel or process. Some policies, such as a regulation or law to phase out coal-fired power, fix this as their primary purpose and are hence included. Additionally, the second question allows us to include policies that induce decline indirectly, primarily through the substitution of incumbent arrangements with new innovations. For example, policies that stipulate an increased market share in an alternative technology encompass an implicit objective to inversely induce the reduction of the incumbent technology, such as in the case of targets to reach a 100% share of ZEVs in new vehicle sales by 2035. It should be noted that we also included broad strategies comprising multiple tenets or measures if one of their focus areas involved decline goals. An example is Canada’s ‘Green Buildings Strategy’. Although comprising several tenets, such as retrofitting and green construction, we include the strategy due to its explicit ambitions to phase out fossil-fuel heating.

The process of identifying relevant policies was informed by external evidence such as government websites, policy documents and third-party materials. Care was taken to exclude policies that, while described with decline-related terminology (e.g., reduce fossil fuel use) in the Biennial Report, were found upon closer inspection to not correspond with our empirical focus, presented in Tables 1 and 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Criteria used to identify relevant decline policies.

https://doi.org/10.1371/journal.pone.0334512.t002

3.3. Step three: Database construction and coding

We built a database to compile descriptions of decline policies from each country in accord with variables listed in Table 3. This was organised by country and then structured so that each row corresponds to a specific decline policy and decline target (e.g., coal-fired power, deforestation). The database was populated with information taken directly from the respective Biennial Report regarding the decline target, sector, policy instrument and CO2 reduction estimates in addition to our inductively created coding categories for the decline approach (direct or indirect) and ‘decline intensity’ (described next). In cases where one policy described several measures addressed at different decline targets, we coded each target separately, assigning each its own sectoral code and decline intensity score. Consequently, the database contains 233 policies that correspond to 251 targets, all of which were included in the analysis. The full database and codebook have been made available in S1 Data, enabling readers to reproduce the analyses conducted for this pilot study.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Coding variables used in database.

https://doi.org/10.1371/journal.pone.0334512.t003

When making coding decisions, we supplemented the descriptions of each policy in that country’s Biennial Report with external sources such as official government websites policy documents.

3.3.1. Evaluation of decline intensity.

Inspired by the concept of ‘policy intensity’ [41], we developed an original evaluation scheme to compare the varying decline intensity of each policy included in our database (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Decline intensity evaluation framework.

https://doi.org/10.1371/journal.pone.0334512.t004

This index was inductively and iteratively designed by observing the contrasting design features of the decline policies in our database. Concretely, we adapted the numerical scoring approach of [41] and [43] to develop seven indicators that evaluate each policy based on a set of design features that we expect to influence the magnitude and speed of reduction and decarbonisation. If an indicator was met, we awarded 1 to that policy, and 0 if unmet or not clear from available evidence. For two indicators (speed and magnitude), we included interim values (0.5) to account for partial fulfilment.

As well as using individual scores for each indicator, we also computed ‘total intensity scores’ by taking the unweighted average score for all relevant indicators (i.e., seven if a policy is targeted at technologies and six if targeted at fuels/substances or processes). Following policy design research [42,43], our underlying assumption is that policies with a higher intensity score would be more likely to achieve large-scale and rapid decline than lower intensity policies, thereby yielding greater decarbonisation outcomes. Although some scholars [41,91] apply weightings to different indicators when calculating policy intensity, we refrained from doing so due to insufficient empirical knowledge about the design features that matter the most for achieving large-scale and rapid decline.

3.4. Step four: Analysis

Finally, to build understanding of the decarbonisation potential of different policies and enhance our understanding of associations between their design features and projected mitigation potential, we integrated estimates of yearly mitigation impact, expressed as GHG emissions reductions (kt CO₂e) for various years (e.g., 2025, 2030, 2035). Provided in Biennial Reports, these estimates are calculated by governments in each country and were available for 98 (42%) of the 233 policies in the dataset. We excluded three of these policies (ID79, 163 and 207) from the analysis due to implausibly high estimates of mitigation impact that exceed 100% of annual emissions in that sector.

We used regression analysis to model each policy’s estimated GHG emission reduction as a function of (a) the total intensity score and (b) the decline approach (direct or indirect). The dependent variable in the regression is the projected mitigation impact of each policy, normalised as a share of total annual emissions in the corresponding sector in 2019. This year was chosen for two reasons. First, it was the most recent year for which emissions data were available in the Fifth Biennial Reports at the time of publication (late 2022). Second, emissions in 2019 were unaffected by the global COVID-19 pandemic, which disrupted emissions patterns from 2020 onwards.

This normalisation was necessary to address substantial variation in the estimates of each policy’s mitigation potential and to enable more meaningful comparisons of decarbonisation effectiveness across sectors compared to using unnormalised absolute values. Because the distribution of projected mitigation impacts was highly skewed due to a small number of policies with disproportionately large estimates (ranging between 20–65% of sectoral emissions), we log-transformed the normalised values of avoided sectoral emissions. Due to our interest in the peak mitigation potential of policies irrespective of the timing when this occurs, we used the highest value listed for the policies that provided estimates for multiple years. To account for the hierarchical structure of the data, we applied mixed-effects models with random effects specified at the country level. Finally, to illustrate emission reduction trends in individual sectors, we derived marginal means from the regression results.

Data and Stata code used for the regression are provided in S1 Data and S2 Code.

4.1. Overview of decline policies in our database

We identified a total of 233 decline policies with 251 corresponding targets across the ten highest-emitting Annex I countries in the UNFCCC. The majority are concentrated in transport, energy supply and waste (Fig 3). Conversely, we found less decline activity in LULUCF and agriculture. Cross-cutting strategies, mostly consisting of carbon pricing instruments such as emissions trading and pollution taxes, make up only 11% of policies. This indicates that policymakers typically use decline policies to tackle well-defined targets in specific sectors through tailored interventions as opposed to a ‘shot-gun’ style that simultaneously addresses multiple targets and sectors.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Distribution of policies across sectors by decline approach.

Bars show the relative distribution of decline policies by sector for 233 policies corresponding to 251 decline targets. The corresponding number of targets for each sector is indicated in brackets. Several policies target more than one sector and thus appear multiple times in the results.

https://doi.org/10.1371/journal.pone.0334512.g003

With respect to the distribution of direct and indirect decline approaches, our analysis reveals a nearly even split comprising 53% and 47% respectively. Sectors with a higher prevalence of direct decline approaches include LULUCF, waste and agriculture. These sectors stand in contrast to those where emissions predominantly stem from fossil-fuel combustion – namely transport, energy supply and buildings – where indirect approaches are more commonly adopted. The transport sector in particular exhibits a strong reliance on indirect approaches, reflecting a tendency to promote substitution with ZEV diffusion targets, fleet replacement programs and renewable fuel blending mandates rather than directly restricting the production or sale of ICE vehicles and petroleum fuels.

Although our analysis did not explicitly aim to identify country-level trends due to the limited size of the Annex I group selected for analysis (n = 10), our findings show that some countries predominantly employ direct decline policies whereas others lean more towards indirect approaches (Fig 4). The three countries with the highest use of direct policies are Germany, Poland and France. Among these, Germany stands out, with more than 75% of its decline policies directly targeting incumbent elements, especially in transport, energy and industry. In contrast, Italy, Japan and the United States exhibit a strong inclination towards indirect approaches.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Distribution of policies across countries by decline approach.

Bars show the relative distribution of decline policies by country for 233 policies corresponding to 251 decline targets. The corresponding number of policies for each country is indicated in brackets.

https://doi.org/10.1371/journal.pone.0334512.g004

In terms of the specific emissions-intensive elements targeted by policies, across sectors the most common are fossil-fuel and coal-fired power generation, ICE vehicles (comprising mostly light-duty, but also medium and heavy) and petroleum fuels, fossil-fuel heating, ozone-depleting substances and industrial fossil-fuel combustion, synthetic fertilisers and landfilling of organic waste (see S1 Fig in S3 File). While these technologies, substances and practices are addressed by the majority of the ten countries, our analysis also identified less common but emerging decline targets. The supply of fossil gas in heating and cooking is one example, with four countries (Australia, Canada, UK and France) describing various decline policies at the national or sub-national level. Australia’s Capital Territory has taken a particularly active stance, leveraging regulatory instruments to prevent the installation of new gas grids in buildings and urban developments, financial incentives to switch from gas to electric appliances and institutional commitments to using gas-free buildings in the government and private sector.

Tables 5 and 6 provide illustrative examples of direct and indirect decline policies from our dataset. The examples show that policymakers adopt diverging approaches and instruments to accomplish their aims, including phase-down/out schedules, regulations, prescriptive targets and bans. Consider, for example, how Australia’s biennial report lists an indirect decline policy (Electric Vehicle Strategy) from the state of South Australia that seeks to achieve 50% sales of ZEVs by 2030 – which inversely signifies a 50% reduction in ICE vehicle sales – while the UK’s report describes a regulatory policy (Regulation to phase out all new non-zero emission road vehicles) directly banning the sale of ICE vehicles by 2035.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Direct decline: Illustrative policies by sector.

https://doi.org/10.1371/journal.pone.0334512.t005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Indirect decline: Illustrative policies by sector.

https://doi.org/10.1371/journal.pone.0334512.t006

Our analysis reveals distinct differences between the types of policy instruments used by direct and indirect decline approaches. As shown in Fig 5, we find that direct decline policies are more likely to use regulatory instruments than indirect decline policies. Regulatory instruments feature in more than two thirds (69%) of direct decline policies compared to only 36% of indirect decline policies. Indirect decline policies, on the other hand, tend to involve softer strategies such as economic, fiscal and information/education instruments. Many of these instruments involve subsidies or fiscal incentives to encourage the replacement of incumbent technologies like industry equipment and fuels with cleaner alternatives. We also find that the decline policies frequently involve a mix of instruments, with roughly one third (32%) of the policies in our dataset using two or more instruments.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Relative share of instruments used in direct decline and indirect decline.

Figure shows the relative distribution of all instruments (n = 357) used by all policies (n = 233) following categories indicated in Biennial Reports. The former amount is higher than the latter because roughly one third of policies use two or more instruments.

https://doi.org/10.1371/journal.pone.0334512.g005

4.2. Variations in decline intensity

We now move to consider the rich variation of design features across policies, assessing their ‘decline intensity’ in accord with the indicators outlined in our Methods (Section 3).

Observing raw means, we find that direct decline policies tend to exhibit a higher decline intensity compared to indirect approaches (Fig 6). This result reflects a stronger tendency for direct approaches to pursue decline through mandatory instruments such as regulation, legislation or legal frameworks (indicator 1). Direct approaches also tend to set explicit numerical targets (indicator 2) and faster reduction schedules (indicator 3). Moreover, they tend to cover larger shares of the incumbent element in the targeted system (indicator 5) and are considerably more likely to aim for deep reduction or complete discontinuation, rather than partial reduction (indicator 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Decline intensity of direct and indirect approaches.

Jitter plot showing the distribution for seven indicators used to evaluate decline intensity of policies. Circles and triangles represent individual policies, scored according to the criteria elaborated in Methods Table 4. Total decline intensity (last column) represents the average unweighted score of all indicators; a policy fully satisfying all indicators would receive a total score of 1. Diamonds represent arithmetic means for all indicators, distinguishing between direct and indirect decline. Indicator 7 (stock turnover) is only applied to decline policies targeting technologies. Based on 233 policies encompassing 251 decline targets. Each target received its own decline intensity score and appears in the figure.

https://doi.org/10.1371/journal.pone.0334512.g006

Findings also reveal significant variation in the decline intensity of policies across targets (Fig 7). Policies with the highest total decline intensity are typically directed at coal-fired power, deforestation, the production and use of ozone-depleting substances, and the landfilling of organic waste. In contrast, fertiliser and land conversion in agriculture and fossil-fuel based heating in buildings generally exhibit weaker decline intensity scores. This is largely attributable to weak or missing policy design features concerning stringency, explicit numerical targets and articulated ambitions to achieve a complete phase-out.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Evaluation of decline intensity for most common decline targets.

Results for a subset of the most common targets (n = 233) out of a total of 251 in the dataset. For explanation of indicators, see Methods (Section 3). When viewing score trends across different targets and indicators, readers should note the number of corresponding policies (several policies encompass multiple decline targets) shown in the far-right column. *Stock turnover (indicator 7) is only applied to policies targeting technologies.

https://doi.org/10.1371/journal.pone.0334512.g007

Across all decline targets, some indicators exhibit particularly high variation in intensity scores – notably strictness (indicator 1), system coverage (indicator 5) and reduction magnitude (indicator 6) – while others are associated with low scores throughout. Some targets achieve high scores on most or all dimensions. In the case of policies targeting coal-fired power, this reflects more frequent use of mandatory frameworks, system-wide coverage, and ambitions to achieve complete abolishment. By contrast, policies addressing venting and flaring in oil and gas production or fossil-fuel heating in buildings tend to score relatively low on most dimensions. Petroleum transport fuels, fossil-fuel combustion in industry, fertiliser use in agriculture along with venting and flaring are also notable in that most decline policies aim for partial rather than complete reduction or fail to set explicit reduction timelines. For instance, in the agricultural sector, we identified nine policies across six countries aiming to curb sectoral emissions by reducing fertiliser volumes. Yet only half of these policies are supported by regulatory frameworks, and just three fix numerical targets to specify the pace and magnitude of the reduction.

The above variations in decline intensity are likely due to manifold and context specific factors. Existing research on decline policies emphasises factors such as resistance by industry or affected stakeholders [92,93], the cost and difficulty of eliminating incumbent elements in some sectors [9] and the availability of viable alternatives [94]. This once again foregrounds the critical role of indirect decline policies, which facilitate decline through substitution, by supporting the diffusion of alternatives that replace incumbent carbon-intensive elements. In the case of efforts to reduce landfilling waste, policies promoting alternative means of treating or disposing of waste – such as recycling, composting or incineration facilities – are just as pivotal as policies that force the diversion of waste streams away from landfills.

4.3. The decarbonisation impact of decline policies

The preceding analysis highlights substantial variation in the design features of different decline policies. This prompts the question of whether, and to what extent, these design features are associated with differences in decarbonisation effectiveness. To explore this relationship, we draw on estimates of GHG emission reductions (i.e., mitigation impact) anticipated for the implementation of each policy, as reported in each country’s Biennial Report. These estimates enable us to examine the mitigation impact of 95 out of the 233 policies in our dataset. We employ multiple regression analysis following the procedure outlined in Methods 3.4 to sequentially model the relationship between projected mitigation impact (normalised as a share of sectoral emissions) and two variables of interest: (1) the decline approach (direct versus indirect), (2) and decline intensity. Full results of the regression along with correlations of variables are provided in SI Tables S3 and S4 in S3 File.

Our first partial model (n = 95), using decline approach as the focal predictor variable, reveals a statistically significant relationship (p < 0.05) with mitigation impact. This result indicates that direct decline policies tend to be associated with higher estimates of GHG emission reductions. The second partial model, employing decline intensity as the primary predictor, shows a robust association (p < 0.01) with GHG emission reductions, indicating that higher-intensity policies tend to be linked with larger mitigation impact estimates. In the full model, which includes both independent variables along with sectoral dummies, only decline intensity remains significantly associated with mitigation impact (p < 0.01). This is not surprising, given the moderate correlation between approach and intensity (r = 0.41; see Table S4 in S3 File) and the relatively small sample (n = 95). Substantively, this suggests that the strength and ambition of a policy’s design may be more decisive for reducing emissions than the choice of targeting a carbon-intensive element directly or indirectly. Overall, the results provide tentative evidence for our assumption that decline intensity, reflecting the various policy design features discussed in Table 4, matters strongly for decarbonisation outcomes.

As depicted in Fig 8, we next illustrate mitigation potential at the sectoral level by modelling marginal means for both the decline approach (indirect versus direct) and decline intensity (low versus high). To enhance interpretability, we exponentiated the model predictions, allowing the marginal means to be presented on the original scale – that is, as directly interpretable percentage reductions in sectoral emissions. These marginal means reflect the model-predicted average emission reduction associated with a single policy in each sector.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Marginal means of CO₂ reduction potential by sector and policy type.

Bars show marginal means of the share of sectoral emissions avoided for 95 policies that included mitigation impact estimates in their Biennial Report, along with confidence intervals. A further three policies (not shown) were excluded due to implausibly high estimates that exceeded total annual sectoral emissions. Means are based on mixed-effects models, with random effects at the country level and robust standard errors. To calculate the normalised emission reductions, we used the highest mitigation impact estimate provided for any year between 2020 and 2040 in each country’s Biennial Report and divided this by that sector’s total annual emissions in 2019. The ‘energy’ category includes policies from the ‘buildings’ sector. Mitigation impact estimates for policies in the building sector were normalised in accord with the annual emissions of the energy sector because the former’s emissions are not reported separately by countries. The ‘other’ category includes sectors with too few estimates to derive marginal means (agriculture, LULUCF, cross-cutting). ‘All’ represents the marginal mean obtained across all sectors. Decline intensity is categorised as ‘low’ (≤0.5) or ‘high’ (>0.5) based on intensity scores, which reflect the unweighted average of seven indicators (Table 4).

https://doi.org/10.1371/journal.pone.0334512.g008

With respect to the decline approach, results (Fig 8, top) indicate that direct decline policies are associated with higher mitigation impact estimates in most sectors, with the exception of energy & buildings and transport. The higher mitigation potential of direct policies is particularly apparent in the industry and waste sectors. Here the average reduction potential for direct decline nears 8% of sectoral emissions compared to 3.3% and 1.5% for indirect approaches, respectively. However, as indicated by the relatively wide confidence intervals, this result is accompanied by a notable degree of statistical uncertainty.

Turning to the relationship between decline intensity and mitigation impact (Fig 8, bottom), we find that in all sectors except waste, high-intensity policies (reflecting a total intensity score >0.5) are associated with greater GHG reduction estimates than low-intensity policies (total intensity score ≤0.5). When considering results for all sectors combined (displayed as grey bars), high-intensity policies are projected to reduce sectoral emissions by an average of 3.8% – considerably higher than the 1.1% reduction estimated for low-intensity policies. The difference between low-intensity and high-intensity approaches is also particularly pronounced in the energy & building sector, where a decline policy with low intensity is predicted to achieve a 0.1% reduction of sectoral emissions, while a high-intensity policy is associated with a reduction of 1.4%.

Taken together, the results of the regression analysis (Tables S3 and S4 in S3 File) and the examination of sectoral marginal means (Fig 8) indicate that policy characteristics are a critical factor shaping the potential depth of decarbonisation achieved by decline interventions. While evidence regarding the relative influence of direct versus indirect decline policies remains inconclusive, the analysis reveals a robust association between the ‘intensity’ of decline policies and higher emission reduction estimates. These findings therefore suggest that not all decline policies are equally effective. Rather, qualitative variations resulting from design features such as strictness, reduction magnitude, speed and geographic coverage may be more decisive in shaping a policy’s potential mitigation impact than the choice between direct or indirect approaches. This indicates that both decline approaches – embodying contrasting functions of destruction and substitution – can meaningfully contribute to decarbonisation if well designed.

4.4. Limitations

We emphasise the tentative nature of our findings, as the statistical analysis draws on a limited number of observations – particularly in the LULUCF, transport and waste sectors. This limitation underscores the need for future research to validate the observed relationships between decarbonisation effectiveness and decline intensity or the decline approach with a larger sample of policies and mitigation impact estimates.

A further methodological issue concerns the reliability of the mitigation impact estimates themselves, derived exclusively from figures provided by governments in their Biennial Report. Because the full range of assumptions and modelling methods underlying these estimates are not published, uncertainty remains regarding the comparability and accuracy of these forecasts of mitigation outcomes. Extending the analysis with third-party modelling based on standardised and more transparent approaches would offer an important means to validate or challenge these findings.

Additionally, our understanding of the decarbonisation impacts of different kinds of decline policies is hampered by the varying timescales over which GHG emission reductions occur. For example, policies that induce the reduction of coal-fired power typically achieve large and rapid short-term reductions. In contrast, policies aimed at reducing ICE vehicle sales typically result in slower emissions reductions due to the many years required to turn over existing on-road vehicle stock. To address this, our analysis used the highest estimate of annual mitigation impact provided in Biennial Reports regardless of the year in which this is expected to occur. Yet future research is needed to more fully account for the temporal dynamics of decline processes when comparing the relative decarbonisation effectiveness of different policy types.