Decomposition Analysis of the Evolution of the Local Energy System as a Tool to Assess the Effect of Local Actions: Methodology and Example of Malmö, Sweden

This paper proposes the use of decomposition analysis to assess the effect of local energy-related actions towards climate change mitigation, and thus improve policy evaluation and planning at the local level. The assessment of the impact of local actions has been a challenge, even from a strictly technical perspective. This happens because the total change observed is the result of multiple factors influencing local energy-related greenhouse gas (GHG) emissions, many of them not even influenced by local authorities. A methodology was developed, based on a recently developed decomposition model, that disaggregates the total observed changes in the local energy system into multiple causes/effects (including local socio-economic evolution, technology evolution, higher-level governance frame and local actions). The proposed methodology, including the quantification of the specific effect associated with local actions, is demonstrated with the case study of the municipality of Malmö (Sweden) in the timeframe between 1990 and 2015.


Introduction
The role of local actions towards the achievement of global targets on climate change mitigation is widely recognized by academics and policy makers. The physical characteristics of local energy systems added to the regulatory competences of local authorities, prompt the local level as an appropriate level of action [1]. The call for a paradigm shift towards more decentralized energy management and use (production/consumption), mentioned in several EU publications (including the Clean Energy Package [2]) and in worldwide discussions, reinforces this relevance of local actions. Local authorities, as the regulatory bodies closest to the citizens, are in a strategic position to promote this transition towards decentralization, including lifestyle changes and local sustainable energy investments.
The recognition of local actions' contribution towards climate change mitigation has been accompanied by an increasingly active role of local actors-including local authorities as well as other local stakeholders, such as local businesses, municipal utilities, citizens associations and individual citizens [3]. For instance, more than 10,000 European municipalities have committed to reduce their GHG emissions level by at least 20%, becoming signatories of the Covenant of Mayors initiative [4,5]. There are also several initiatives promoted at national and EU level with the objective of levering sustainable energy planning and climate change mitigation at the local level, such as the European Energy Award, CIVITAS, Managenergy and others.
While local actions become more and more common, it is necessary to properly assess their impact on greenhouse gas (GHG) emissions, i.e., their contribution towards climate change mitigation. Nonetheless, a thorough assessment of local actions and respective effects is still challenging, given the technical difficulty in properly quantifying the effects of The importance of assessing effectiveness, efficiency and relevance of different policies is globally recognized. Policy evaluation is key for the identification and transfer of good practices, as well as for the adjustment of ongoing and development of new policies. There have been several studies focusing on the assessment of energy, climate and environmental policies at different levels, which resulted on the development of a vast set of methodologies to quantify the effects of different policy actions.
Nonetheless, the specific methods for evaluation have not been a recurrent subject for study, at least not as much as would be expected. This might be due to the difficulty in properly assessing their effects. Given the unavoidable interaction with other policies, their contribution to climate change mitigation depends on factors that are not fully controlled by local authorities, and whose disaggregation is neither obvious nor immediate.
Current practice on the evaluation of local actions also reflects this difficulty. The most common assessment of the effects of the Sustainable Energy Action Plans (SEAPs) in the context of the Covenant of Mayors initiative corresponds to the comparison of the energy and emissions inventories before and after the actions' implementation [9,10]. This comparison is relevant, as it provides an insight on the overall evolution of the local energy system, both in terms of energy use and in terms of GHG emissions [11]. However, this is not sufficient to assess the specific impacts or merits of local actions for climate change mitigation.
There are already some studies that go beyond the analysis of observed changes, which provide more detailed analyses of specific actions [12,13]. Such studies use bottomup models, where the effects are estimated according to the changes caused to the system by the specific policy actions that were implemented at the local level. This alternative approach requires a deep knowledge of the energy system and of the implemented actions, including the changes incurred in the system in physical units (e.g., number of photovoltaic (PV) panels installed, number of persons per kilometer shifted from diesel to electric car, number of appliances/lamps replaced)). Here, the major shortcoming is the difficulty in having the necessary data to establish a reliable and detailed baseline, and to perform a robust bottom-up analysis, especially when assessing the effect of a set of actions covering different sectors.
Thus, in practice, there is still a gap in the assessment of local policy actions, as stateof-the-art practices do not provide methodology for the evaluation of local actions and/or plans which: (1) can be used in a systematic manner with commonly available data and (2) provide an estimate of the real impact of local actions (disaggregated from the effects associated with external factors).
Therefore, taking into account existing limitations from current evaluation practices and considering recent advances in the understanding and representation of the local energy system and respective evolution over time, this paper proposes a new methodology for the assessment of local climate change related actions built upon the top-down decomposition model recently developed and published in [14].

Assessing Local Actions Using a Decomposition Analysis of the Evolution of the Energy System
This section describes how the disaggregation of observed changes of a specific local energy system into the individual effects of the different factors of change can contribute to an improved assessment of local actions. First, there is a short description of the model published in [14] and used in the quantification of the individual effects (including the methodology and the main inputs and outputs). Then, an explanation on how this model may be used as an assessment tool for local actions is presented, highlighting innovation compared to current practice.

Model for the Quantification of Local Actions Effects-Brief Description
The factors of change that are considered include the natural evolution of the local energy system, the higher-level governance framework and local climate change mitigation actions. The natural evolution of the system corresponds to the changes in local population and local economy which are not directly associated with energy and climate policies. Factors associated with higher-level governance framework refer to the energy-related policies implemented at international, national and regional level within the studied timeframe. Finally, local actions comprehend all local energy-related policy actions implemented with the aim of reducing local GHG emissions, including both actions taken under a formal plan and isolated actions. In total, the model disaggregates the observed changes into the effects of 18 individual factors, as identified in Table 1. The model itself corresponds to a decomposition analysis that, starting from the energy matrices from the base year, sequentially constructs new energy matrices that result from introducing the changes caused by the different factors of change into the local energy system (Equation (1)). Local actions effects are assumed to be the remaining gap between observed changes and the sum of all the effects from the external factors of change (Equation (2)).
where ∆GHG is the variation in local GHG emissions between the base year (t 0 ) and the reporting year (t 1 ) for the municipality α, and E i is the effect of factor of change i in local GHG emissions between the base year (t 0 ) and the reporting year (t 1 ) for the municipality α.
The individual effects for the different factors of change are estimated by using already known relations (theoretical or empirically proven), taken from the literature, between those factors and the local energy system. These known relations are translated into mathematical formulas that reflect the changes caused by each individual factor of change to the local energy system. Assumptions have been made to adjust the formulas to the existing data and to allow for their applicability in different contexts. Figure 1 presents a schematic representation of the decomposition process and the quantification of the different effects. The model output consists of the quantification of the individual effects associated with each of the 17 external factors and the overall effect of local climate change mitigation actions in terms of final energy and GHG emissions.
Acknowledging the impossibility of guaranteeing the consideration of all the relevant factors (given the diversity of local contexts) and the risks of assuming the remaining effect as an estimation of the effect of local actions, the methodology includes a set of pro-  The model output consists of the quantification of the individual effects associated with each of the 17 external factors and the overall effect of local climate change mitigation actions in terms of final energy and GHG emissions.
Acknowledging the impossibility of guaranteeing the consideration of all the relevant factors (given the diversity of local contexts) and the risks of assuming the remaining effect as an estimation of the effect of local actions, the methodology includes a set of procedures that improve the robustness of the model results. These correspond to (1) the implementation of the model to municipalities that are within the same (or in a similar) context than the municipality being studied and (2) the comparison between obtained results and reported actions.
A more detailed description of the model, the governing equations and accompanying validation procedures can be found in [14].

How to Use the Model as a Tool for the Assessment of Local Actions for Climate Change Mitigation
The disaggregation model described in the previous section can be a useful tool for the assessment of the local path towards a sustainable energy system. It provides an overview of the observed changes in the local energy system for a specific timeframe, and also a robust estimation of the effect of local actions per sector.
In what concerns the overall path of the local energy system, the model outputs can be useful in two distinct ways: first, on the assessment of the evolution of the local energy system over time and second, by assisting on the identification of external synergies and/or barriers regarding the achievement of the established sustainability goals. Concerning the first, the comparison between the energy matrices from the base and the reporting years may provide important insights regarding the evolution of the local energy system, regardless of the factors that have led to the observed changes. This analysis, done per sector and per energy carrier, gives a first overview of the changes that occurred in the local energy system and may result in a set of indicators regarding the unitary energy use and respective GHG emissions and their evolution over time. Concerning the second, as a specific feature of this model, the disaggregation of the observed changes into the individual effects of each factor of change can be used to have a better understanding what contributed more significantly to the observed changes. It can provide relevant insights on the influence of each factor upon the evolution of local GHG emissions, including the identification of which factors are hampering and which are contributing the local path towards a sustainable energy system.
In what concerns the quantification of the effects associated with the local actions towards climate change mitigation, their assessment can be used to evaluate the actual contribution of local actions (overall and per sector) for the achievement of a sustainable energy system. The comparison between the results of the disaggregation model and the targets established by local authorities can be seen as a tool to compare what was achieved with what was intended. This analysis may inform decision makers on the need to review, accelerate or simply maintain their actions towards climate change mitigation. On the other hand, a comparison between the ex-ante projections of the implemented actions with the estimated effects provides an assessment of the effectiveness of those actions.

Application to the Case Study of Malmö (Sweden)
This section is dedicated to the implementation of the evaluation methodology to the municipality of Malmö (Sweden). This choice was based on the following cumulative criteria: (1) to be a signatory of the Covenant of Mayors, with an approved SEAP (submitted in 2011) and with the first monitoring report already published in 2015; (2) being in the forefront on the availability of data from the local energy system and its evolution over time and (3) having a long history regarding climate change mitigation.

Characterization of the Case-Study
Malmö is a municipality located in the southern region of Sweden, in Skåne County. In 2015, it had a population of about 323 thousand inhabitants distributed in an area of 157 km 2 . The municipality has grown over the last years due to its strategic location-15 min away from Copenhagen (Danish capital) and from Lund (with one of the largest Swedish universities). This growth has been accompanied by a continuous increase in local population, which increased almost 50% over the last 30 years. Malmö is mostly an urban municipality, and the provision of services corresponds to its main economic activity. A brief characterization of the municipality and its evolution between 1990 and 2015 is presented in Table 2. In what concerns energy and climate actions, Malmö has been a pioneer, with concerted actions since the 1990s. Its first interventions were done in the context of the Local Investments Program (LIP), a national investment program that promoted the development of local environmental plans and initiatives. Currently, Malmö local administration is committed to become climate-neutral by 2020. Moreover, in the Sustainable Energy Action Plan submitted to the Covenant of Mayors, the city set as a target the supply of local energy use to be exclusively from renewable sources by no later than 2030 [15].
The timeframe considered for this case study is between 1990 and 2015. The choice of the base year is associated with the fact that 1990 was the year considered for the first inventory of local GHG emissions for the municipality of Malmö, being also the base year of the municipality's SEAP. The reporting year corresponded to the most recent year with available statistical data for local energy use. The studied timeframe covers the period when concerted actions taken by the local authority towards climate change mitigation have been more significant.

Main Input Data
An important stage of the application of the model refers to the collection of input data. Regarding the construction of the energy matrices, the main sources of input data were the official statistics on energy and end-use consumption published yearly by the Statistics Central Bureau (SCB) [16,17]. Moreover, as there are no empirical data regarding the enduse efficiency for the different energy uses for the municipality of Malmö, the estimations provided by the energy matrices for the Metropolitan Area of Porto [18] (complete and readily available) were used as reference and adapted to the local context. The level of detail from the information available for the Metropolitan Area of Porto, along with the limited impact that the assumptions on end-use efficiency could have on the model results, justify this choice. The values for end-use efficiency were adapted to local context, taking into account existing data on technology options and average efficiency of different appliances in Portugal and Sweden. Table 3 lists the main input data required to build both base and reporting years' energy matrices with the respective data sources. Table 3. Input data to build the energy and emissions matrices for the municipality of Malmö for 1990 and 2015, with the identification of main sources and data geographical level.

Input Data Main Source Geographical Level
Final energy use per sector SCB (statistics) [ In what concerns the disaggregation of the individual effects per factor of change, an effort was made to collect data as accurate as possible (i.e., specific to the municipality and to the studied timeframe) in order to reduce uncertainty. When data was not available for the specific geographical context, the collected data corresponded to the closest territorial level to which the data was available and, if necessary, the values were adapted to the municipality level using downscaling indicators. Moreover, when specific values for base and/or reporting years were not found, these were estimated from the existing values using trend analysis. Table 4 lists the collected input data along with the respective sources and geographical level.  Recognizing the relevance that assumptions and input data may have on the model results, the methodology includes a fine-tuning process that, as much as possible, ensures the adequacy of the assumptions and the reliability of the input data, in order to guarantee the robustness of the results. The decomposition model was applied to six surrounding municipalities (Helsingborg, Kristianstad, Landskrona, Lund, Trelleborg and Ängelholm) and the obtained results were assessed individually, and in comparison, with each other. This exercise provided additional validation of the results obtained for the municipality of Malmö, by assessing whether plausible results can be obtained for the other municipalities when using the same inputs' data sources and assumptions. In what concerns final energy, the overall energy use remained fairly constant over the considered timeframe. However, changes were observed at the sectorial level, with the exception of the transport sector (both passenger and freight) where the use of energy was kept almost constant. The services sector is the only sector with a significant increase in energy use (30%). The remaining sectors show a decrease in energy use over the studied timeframe. The sector with the highest absolute decrease is the residential sector, reducing about 290 GWh/year, what corresponds to 13% of the sector's energy use in the base year. Nonetheless, the agriculture and fisheries sector is the sector with the highest relative decrease, reducing to 55% of the base year's energy use by 2015.  The evolution of local GHG emissions between 1990 and 2015 shows a downward trend over all the sectors, as shown in Figure 3. GHG emissions refer to CO2, CH4 and N2O; and the level of emissions in tCO2 eq. is estimated based on the carbon content of the different energy carriers and the respective global warming potential, as reported in the IPCC Guidelines for National Greenhouse Gas Inventories [22]. Overall, the emissions in the reporting year correspond to 78% of the emissions in the base year. This reduction seems to be a combination of the shift to less carbon intensive energy carriers with the reduction of the carbon content of fossil heat distributed through district heating. Indeed, it is noticeable that, while final energy use in the services sector has increased, the equivalent GHG emissions decreased (almost 30%), due to the significant weight of district heating on the sectorial energy use. In addition, the decrease in GHG emissions in the residential and industry sectors (39% and 30%) is more significant than the decrease in the equivalent final energy use (13% and 22%). The evolution of local GHG emissions between 1990 and 2015 shows a downward trend over all the sectors, as shown in Figure 3. GHG emissions refer to CO 2 , CH 4 and N 2 O; and the level of emissions in tCO 2 eq. is estimated based on the carbon content of the different energy carriers and the respective global warming potential, as reported in the IPCC Guidelines for National Greenhouse Gas Inventories [22]. Overall, the emissions in the reporting year correspond to 78% of the emissions in the base year. This reduction seems to be a combination of the shift to less carbon intensive energy carriers with the reduction of the carbon content of fossil heat distributed through district heating. Indeed, it is noticeable that, while final energy use in the services sector has increased, the equivalent GHG emissions decreased (almost 30%), due to the significant weight of district heating on the sectorial energy use. In addition, the decrease in GHG emissions in the residential and industry sectors (39% and 30%) is more significant than the decrease in the equivalent final energy use (13% and 22%). seems to be a combination of the shift to less carbon intensive energy carriers with the reduction of the carbon content of fossil heat distributed through district heating. Indeed, it is noticeable that, while final energy use in the services sector has increased, the equivalent GHG emissions decreased (almost 30%), due to the significant weight of district heating on the sectorial energy use. In addition, the decrease in GHG emissions in the residential and industry sectors (39% and 30%) is more significant than the decrease in the equivalent final energy use (13% and 22%). Summarizing, the observed changes in the local energy system between 1990 and 2015 correspond to an almost negligible change in final energy use and a considerable decrease in the respective GHG emissions. The reduction of GHG emissions (19%) may be justified by the decarbonization of the heat production sector and by slight shifts of energy carriers. Despite the significant increase in the number of local inhabitants within the studied timeframe, the level of final energy use was kept constant and the equivalent Summarizing, the observed changes in the local energy system between 1990 and 2015 correspond to an almost negligible change in final energy use and a considerable decrease in the respective GHG emissions. The reduction of GHG emissions (19%) may be justified by the decarbonization of the heat production sector and by slight shifts of energy carriers. Despite the significant increase in the number of local inhabitants within the studied timeframe, the level of final energy use was kept constant and the equivalent GHG emissions decreased, which implies a decrease in the level of final energy use and GHG emissions per capita. The average use of final energy per capita decreased almost 30% between 1990 and 2015, while the observed reduction in GHG emissions per capita (41%) was considerably higher than the overall reduction (19%). The main energy indicators for the municipality of Malmö are presented in Table 5, for both base year and reporting year, in order to provide an overview of the observed evolution of the local energy system.

Disaggregated Effects Associated with the Several Factors of Change
When applying the decomposition analysis to the case study of Malmö, the changes that occurred in the local energy system between 1990 and 2015 were disaggregated into 17 external factors plus the effects associated with local climate change mitigation actions [14]. Figure 4 presents the estimated effects, in terms of GHG emissions, per factor of change. From the analysis of the graphic, it is clear that the largest contributions towards the reduction of local GHG emissions are: (1) Energy taxes, which are all the taxes applicable to energy products, which are associated with energy and climate policies. In Malmö, these include the energy tax (with the goal of improving energy efficiency), the CO2 tax (implemented to promote the shift to less carbon intensive energy carriers) and a special tax (associated with GHG emissions, other than CO2)).
(2) Transport regulations, which in the specific case of Malmö include the incorporation of biofuels in the traditional fuel mixtures and the incentives implemented to promote the substitution of traditional fuels light-weight vehicles with vehicles running with biofuels).
(3) Local climate change mitigation actions. Nonetheless, the contribution of local actions is significantly higher than the ones from the remaining factors. This could be explained by the long history of Swedish municipalities, in general, and Malmö, specifically, in having an active role towards climate change mitigation, a role that has been highly supported by national initiatives.
It was also estimated by the model that a few factors of change had a counteracting effect towards the reduction of GHG emissions. Here, the demographic evolution and the changes in activity level and structure of the local economy were the ones with a more significant effect, corresponding to an increase of 24.1% and 24.5%, respectively, of local GHG emissions compared to the base year's level.
Moreover, when looking at Figure 4, it is worth noticing that the first insights on the overall evolution of the local energy system did not reveal the changes that occurred in the demographic and socio-economic dynamics of the municipality. Even if the energy matrices from the base and reporting years did not reveal abrupt changes in the local energy system, the disaggregation into individual effects showed that there were, in fact, significant changes in the activity level and structure of the local economy as well as in the population size, and that these were counteracted by the effects of other factors of change. From the analysis of the graphic, it is clear that the largest contributions towards the reduction of local GHG emissions are: (1) Energy taxes, which are all the taxes applicable to energy products, which are associated with energy and climate policies. In Malmö, these include the energy tax (with the goal of improving energy efficiency), the CO 2 tax (implemented to promote the shift to less carbon intensive energy carriers) and a special tax (associated with GHG emissions, other than CO 2 )).
(2) Transport regulations, which in the specific case of Malmö include the incorporation of biofuels in the traditional fuel mixtures and the incentives implemented to promote the substitution of traditional fuels light-weight vehicles with vehicles running with biofuels).
(3) Local climate change mitigation actions. Nonetheless, the contribution of local actions is significantly higher than the ones from the remaining factors. This could be explained by the long history of Swedish municipalities, in general, and Malmö, specifically, in having an active role towards climate change mitigation, a role that has been highly supported by national initiatives.
It was also estimated by the model that a few factors of change had a counteracting effect towards the reduction of GHG emissions. Here, the demographic evolution and the changes in activity level and structure of the local economy were the ones with a more significant effect, corresponding to an increase of 24.1% and 24.5%, respectively, of local GHG emissions compared to the base year's level.
Moreover, when looking at Figure 4, it is worth noticing that the first insights on the overall evolution of the local energy system did not reveal the changes that occurred in the demographic and socio-economic dynamics of the municipality. Even if the energy matrices from the base and reporting years did not reveal abrupt changes in the local energy system, the disaggregation into individual effects showed that there were, in fact, significant changes in the activity level and structure of the local economy as well as in the population size, and that these were counteracted by the effects of other factors of change.

SEAP Malmö-Proposed Actions and Implementation Status
The municipality of Malmö was a pioneer in recognizing the importance of local actions towards climate change mitigation and environmental sustainability. Since the 1990s, the local authority has had an active role on the promotion of climate change mitigation by acting on their own infrastructure and by prompting others to act. Since then, there different plans and strategies were developed towards a sustainable future in terms of energy (but not only).
On the vision and goals established for Malmö's energy system, the most recent publication refers to the Covenant of Mayors adhesion, in 2008, where the municipality established the target to reduce local energy use per capita by 20% (compared to average annual use during the period 2001 to 2005) and GHG emissions by 40% (compared to the 1990 level) by 2020. In the Sustainable Energy Action Plan submitted to the Covenant (herein, SEAP Malmö), Malmö's city authority also committed to at least 50% of the local authority's energy use to be from renewable energy sources, with a large proportion being produced locally [15]. The proposed actions covered the different sectors and focused on achieving a more effective energy use and switching to renewable energy sources. Table 6 presents a summary of the technical measures proposed by the local authority.
From 1997 up to today, the city authority publishes an annual report on the local actions towards local environmental sustainability [64][65][66][67][68][69][70][71][72][73][74][75][76][77][78][79][80]. These documents allowed for an appropriate understanding of the efforts being done to achieve the vision described above. Appendix A ( Figure A1) presents a summarized timeline of the actions taken by the local authority towards a more sustainable energy system. Between 2013 and 2015, a survey on the status of implementation of the actions proposed by the city authority was performed. The respective activity reports present an overview of the survey outcomes, concluding that there is still a large number of actions ongoing or not yet implemented that should have been completed by the end of 2015.  Table 7 presents a comparison between the observed changes and the goals established for 2020 by the local authority in the SEAP Malmö, for the local energy system and specifically for the local public sector. This comparison shows that the municipality is indeed moving towards the established goals, especially in what concerns the reduction of final energy use per capita. The evolution of the final energy use associated with the infrastructure owned by the municipality is the exception, given that an increase of 13% in the respective energy use was observed while the established goal was a reduction of 30%. However, this evolution can be justified by the increase in local energy services, due to the municipality's population growth. The comparison between observed changes (as assessed by the model) and the projected evolution of local GHG emissions provides an insight on the alignment of the actual evolution of the local energy system with the goals established for the municipality of Malmö (also considering national and international initiatives). This corresponds to the assessment of the overall distance to the projected pathway, regardless of the factors of change that have caused such evolution. Table 7. Comparison of the overall GHG emissions reduction observed  and the goals established by the municipality of Malmö in the SEAP Malmö (1990-2020).

SEAP Malmö
By 2020 Observed Changes By 2015 Regarding the reduction of local GHG emissions, the observed reduction by 2015 corresponds to about half of the reduction target established for 2020. Moreover, as visible in the plot presented in Figure 5, if a linear evolution over time is considered (dotted line), it is possible to say that the observed reductions in local GHG emissions are behind what was projected by the SEAP. However, considering a trajectory where the decrease in GHG emissions would be smaller up to 2008 (when Malmö adhered to the Covenant of Mayors), the observed reductions would be in line with the established goal for 2020.  Table 7 presents a comparison between the observed changes and the goals established for 2020 by the local authority in the SEAP Malmö, for the local energy system and specifically for the local public sector. This comparison shows that the municipality is indeed moving towards the established goals, especially in what concerns the reduction of final energy use per capita. The evolution of the final energy use associated with the infrastructure owned by the municipality is the exception, given that an increase of 13% in the respective energy use was observed while the established goal was a reduction of 30%. However, this evolution can be justified by the increase in local energy services, due to the municipality's population growth. The comparison between observed changes (as assessed by the model) and the projected evolution of local GHG emissions provides an insight on the alignment of the actual evolution of the local energy system with the goals established for the municipality of Malmö (also considering national and international initiatives). This corresponds to the assessment of the overall distance to the projected pathway, regardless of the factors of change that have caused such evolution. Table 7. Comparison of the overall GHG emissions reduction observed  and the goals established by the municipality of Malmö in the SEAP Malmö (1990-2020).

SEAP Malmö
By 2020 Regarding the reduction of local GHG emissions, the observed reduction by 2015 corresponds to about half of the reduction target established for 2020. Moreover, as visible in the plot presented in Figure 5, if a linear evolution over time is considered (dotted line), it is possible to say that the observed reductions in local GHG emissions are behind what was projected by the SEAP. However, considering a trajectory where the decrease in GHG emissions would be smaller up to 2008 (when Malmö adhered to the Covenant of Mayors), the observed reductions would be in line with the established goal for 2020.

Local Actions' Contribution
The expected contribution of each specific measure (or group of measures) towards the goals established by the local authority is not specified in the Environmental Program of Malmö. Thus, as referred previously, it is not possible to compare the projected outcomes with the estimated effects of local actions. Being so, a comparative analysis between the estimated effects associated with local actions and the overall goals established by the municipality is performed instead. This analysis is also useful to understand the level of contribution of local actions towards the overall targets.
On one hand, it was estimated that the effects associated with local actions between 1990 and 2015 corresponded to a reduction in GHG emissions of over 470 thousand tons of CO 2 eq. , which is equivalent to 41% of the base year's level of emissions. On the other hand, the municipality established the goal of reducing GHG emissions by 40% compared to 1990 level, by 2020. Thus, it could be said that the contribution of local actions already exceeds the reduction targets established by the local authority, even if the overall reduction of GHG emissions is still behind.
Indeed, as visible in Figure 6, in the case of Malmö, the sum of the effects associated with all the external factors (including natural evolution of the local energy system and higher-level policies) corresponds to an increase in the municipality's GHG emissions. This may be justified by the high growth in terms of local demography and economy, which neutralizes the positive effects of national and international policy actions. Moreover, the overall effect of external factors partially counteracts the contribution of local actions towards climate change mitigation. Thus, the mere assessment of the overall change in GHG emissions would lead to an underestimation of the contribution of local actions towards climate change mitigation.

Local Actions' Contribution
The expected contribution of each specific measure (or group of measures) towards the goals established by the local authority is not specified in the Environmental Program of Malmö. Thus, as referred previously, it is not possible to compare the projected outcomes with the estimated effects of local actions. Being so, a comparative analysis between the estimated effects associated with local actions and the overall goals established by the municipality is performed instead. This analysis is also useful to understand the level of contribution of local actions towards the overall targets.
On one hand, it was estimated that the effects associated with local actions between 1990 and 2015 corresponded to a reduction in GHG emissions of over 470 thousand tons of CO2 eq., which is equivalent to 41% of the base year's level of emissions. On the other hand, the municipality established the goal of reducing GHG emissions by 40% compared to 1990 level, by 2020. Thus, it could be said that the contribution of local actions already exceeds the reduction targets established by the local authority, even if the overall reduction of GHG emissions is still behind.
Indeed, as visible in Figure 6, in the case of Malmö, the sum of the effects associated with all the external factors (including natural evolution of the local energy system and higher-level policies) corresponds to an increase in the municipality's GHG emissions. This may be justified by the high growth in terms of local demography and economy, which neutralizes the positive effects of national and international policy actions. Moreover, the overall effect of external factors partially counteracts the contribution of local actions towards climate change mitigation. Thus, the mere assessment of the overall change in GHG emissions would lead to an underestimation of the contribution of local actions towards climate change mitigation. This analysis illustrates the importance of disaggregating the observed changes at the local level in order to properly understand and assess the effect of local actions and the respective contribution on the overall reduction of local GHG emissions. In this specific case, the assessment of the overall reduction of GHG emissions could be misleading, as it would hint that the actions promoted by the municipality of Malmö did not contribute towards climate change mitigation, while the actual contribution of local actions, even if behind what was planned and proposed in the SEAP, was still significant.

Main Findings
From the comparison between the estimated results and the goals defined by local authority, it was observed that the level of local GHG emissions is still far from what is This analysis illustrates the importance of disaggregating the observed changes at the local level in order to properly understand and assess the effect of local actions and the respective contribution on the overall reduction of local GHG emissions. In this specific case, the assessment of the overall reduction of GHG emissions could be misleading, as it would hint that the actions promoted by the municipality of Malmö did not contribute towards climate change mitigation, while the actual contribution of local actions, even if behind what was planned and proposed in the SEAP, was still significant.

Main Findings
From the comparison between the estimated results and the goals defined by local authority, it was observed that the level of local GHG emissions is still far from what is targeted by the municipality. Nonetheless, the values per capita seem to be more satisfactory. The total level of emissions is not justified by ineffectiveness on action but rather by the actual population growth, which is higher than what was considered when developing the SEAP Malmö.
Moreover, when disaggregating the observed changes into the effects of local actions and the ones of external factors of change, the analysis shows that the local contribution to the reduction of GHG emissions nearly reached the proposed reduction goal. The gap between the accomplished and desired GHG emissions reduction is due to the effects associated with external factors, which counteracted the effect associated with local climate change mitigation actions.
Thus, this comparative analysis showed the importance of properly quantifying the actual effects of local actions. Indeed, the difference between the two comparative analyses revealed how the mere assessment of observed total changes can be misleading.

Conclusions
This work proposes a methodology for the assessment of the local path towards climate change mitigation, and of the contribution of local actions towards this path, based on the disaggregation model presented by Azevedo [14]. The model provides a decomposition of the observed total changes of a local energy system into the individual effects of the several factors of change, including external factors and local actions for climate change mitigation. This may be an important asset for the evaluation of local energy systems and their evolution over time. The deeper understanding of what is behind the observed total change in local GHG emissions can help local authorities, as well as higher-level policy makers, to assess current policies and develop future ones.
The model was applied to the municipality of Malmö, for the timeframe 1990-2015.
On the overall change, an almost negligible change in final energy use and a reduction of 22% in the level of GHG emissions was observed. Given the large increase in the number of local inhabitants in the studied timeframe, the changes in energy use and associated GHG emissions per capita were considerably different, with a reduction of 28% and 41% respectively.
When disaggregating the observed change into the effects of the different factors of change, it was estimated that demographic evolution and changes in the local economy were the main drivers for increase of the GHG emissions. On the opposite trend, local climate change mitigation actions were the main driver for decrease in GHG emissions between 1990 and 2015 (41%). Initiatives promoted at the national level, as new energy taxes and transport regulations, also contributed to the reduction in the local GHG emissions, by 12% and 11% respectively.
The fact that the estimated impact of local actions on GHG emissions was a decrease of 41%, while the total observed decrease was only 22%, clearly shows the importance of disaggregating the contribution of local actions from the effects associated with external factors. The assessment of the overall change alone can be misleading, not being representative of the real contribution of local actions. Moreover, the disaggregation by sector and energy carrier estimated by the model can also be used to assess the contribution of specific local actions towards the reduction of local GHG emissions. The comparison between projected outcomes and the estimated effects may be seen as a tool to assess whether the effects of already implemented actions are in line with the respective projections, identifying which mechanisms are effective and which need to be improved or even replaced.
Overall, this paper, with the implementation of the model to the municipality of Malmö, shows the importance of going beyond the quantification of the changes in the local level of GHG emissions over time. The assessment of the contribution of local policy actions towards climate change mitigation could be significantly improved with the disaggregation of the overall change in GHG emissions into the individual effects of the different factors of change.
Finally, considering that the proposed methodology only provides an estimate of the overall effect of local actions, further research on the individual contribution of specific policy instruments is still needed. Indeed, it would be beneficial to estimate the actual contribution of the different policy mechanisms implemented at a more disaggregated level to better assess their effectiveness and efficiency. In order to do so, it would be interesting to develop a bottom-up methodology that would use the overall effects of local actions as well as detailed information on the actions' implementation to estimate these individual effects and to assess their contribution in terms of effectiveness and efficiency. Another area where more work may be useful is the assessment of the precision of input data, and of the typical impacts of data lack of precision upon the model results.

Conflicts of Interest:
The authors declare no conflict of interest. Finally, considering that the proposed methodology only provides an estimate of the overall effect of local actions, further research on the individual contribution of specific policy instruments is still needed. Indeed, it would be beneficial to estimate the actual contribution of the different policy mechanisms implemented at a more disaggregated level to better assess their effectiveness and efficiency. In order to do so, it would be interesting to develop a bottom-up methodology that would use the overall effects of local actions as well as detailed information on the actions' implementation to estimate these individual effects and to assess their contribution in terms of effectiveness and efficiency. Another area where more work may be useful is the assessment of the precision of input data, and of the typical impacts of data lack of precision upon the model results.

Education and training
Creation of organization to promote sustainable transportation (network for companies, municipalities, universities and other stakeholders + gathering and dissemination of knowledge) Release of periodical newspapers and magazines dedicated to environmental issues and targeting local population Green pages; green package for students; etc.
Events to promote environmentally friendly commuting