Transforming Catalonia’s Mobility System for Net Zero

Annex E. Methodological note: policy categorisation from a systems lens

This annex describes a qualitative methodology to assess policies’ transformative potential and allow for discussions on policy prioritisation in the light of such potential. The methodology is part of the “Understand” step of the OECD “Systems Innovation for Net Zero” process (see Chapter 1) and builds on a key insight of systems thinking: the results and patterns of behaviour observed are not independent from, but rather caused by, the structure of a system (Sterman, 2000[30]; Meadows, 2008[31]; Systems Innovation, 2020[32]). Systems thinking tools are used to shed light on the system structure from which unsustainable behaviours and lifestyles emerge, and to identify the extent to which different policies are able to transform or redesign such structure.

Transformative policies are defined as those able to modify the system structure, so that systems foster – by design – patterns of behaviour aligned with desired results. In systems jargon, transformative policies “push” on high leverage points, namely places in which a small intervention may lead to a system transformation capable of triggering large behavioural changes (Meadows, 2008[33]).

The notions of transformative policies, systems structure and behavioural change (Box A E.1) are intrinsically linked. A policy is considered as transformative when, after its implementation, the “transformed” system structure leads to patterns of behaviour that were not observed in the system “of the past”. For example, when a city design encourages most people to walk, cycle or use public transport for the bulk of the trips; while, in the past, people used to drive to meet their needs.

Box A E.1. Behavioural change from a systems approach

The term “behavioural change” is often associated to individual preferences, which are treated as mostly exogenous, while ignoring the role of the system, its structure, and prevailing rules in making such choices feasible and attractive. When this is the case, policies tend to focus on raising awareness, based on the assumption that this is the better information can lead to people making more sustainable choices.

In this framework, the term “behavioural change” is intrinsically linked to the (complexity science) insight that individual preferences (and the resulting patterns of behaviour) are largely determined by the system structure in which they are embedded and the mental models that have shaped such structure. For example, more people “choose” to drive in territories organised around car driving, where driving is the safest, fastest, and sometimes only option.

From a systemic perspective, solely raising awareness is neither sufficient nor the most effective way to trigger large-scale behavioural change. Raising awareness can be important, but it will only trigger large-scale change in a system where sustainable choices become a real and attractive option.

Policies have a higher change to trigger behavioural change if they focus on redesigning systems. For example, contributing to redesigning territories so that these are organised around sustainable modes instead of the private car; and thus, walking and cycling are the safest and fastest option by design.

Policies are classified according to their intent and transformative potential given the local context and system’s characteristics (Figure A E.1). The policy’s intent refers to the aim with which the policy was designed. Policies are categorized based on whether they react to an event (reactive intent), anticipate the impact of a pattern expected to take place (anticipatory intent), or aim at transforming the system structure and/or engrained mental models that lock in the existing system (transformative intent). Policies categorized as reactive and anticipatory can also be seen as policies aiming to reduce the harm of the results or patterns of behaviours the system produces. The policies’ transformative potential is linked to the actual impact the policy has on the systems’ structure; and is characterised as low, medium, or high. The policies’ transformative potential is assessed in terms of the policy effect on: i) systems’ dynamics, ii) the systems’ physical stocks (infrastructure and space allocated per mode), and iii) the mindsets affecting policy decision-making and its public acceptability.

Figure A E.1. Assessing policies’ intent and transformative potential

Source: Authors.

The next sub-sections describe the assessment of policies transformative intent and potential. Note that while this methodology assesses individual policy intent and transformative potential, no single policy can transform a complex system. An assessment of each policy can, however, help policy makers identify those with high transformative power and design policy packages that prioritise them. Furthermore, if the policy package is designed with a transformative intent, the potential for change of policies with a reactive or anticipatory intent may increase (e.g. carbon prices: see section 3.2.2 in (OECD, 2022[34])).

Assessing policies’ intent

The policy’s intent refers to the aim with which the policy was designed. Policies are categorised as having a reactive, anticipatory, or transformative intent. The categorisation is visualised with the analogy of an iceberg (Figure A E.2).

The iceberg analogy aims to illustrate that observed outcomes or events (e.g. traffic jams, pollution peaks, road fatalities, growing emissions) are the “tip of the iceberg”. These events (patterns, when observed over time) are the result of systems that have been designed in a certain way, built on dominant mental models. Both the system design or structure and the mental models are “below water", invisible to the naked eye. The systemic tools described in the sub-section below (and briefly in Box 1.3 in Chapter 1) shed light on the system structure or design. An analysis of mental models allows the identification of the ideas underlying such structure (see Chapter 2).

Figure A E.2. The iceberg model for categorising policies by their intent

Source: Authors.

A focus on what is visible and immediate (the “tip of the iceberg”) has led to decades of policies and actions that react to events or anticipate patterns. Often, these policies take past and/or current patterns of behaviours as given and attempt to deaccelerate and/or reduce the harm of such behaviours.

Reactive policies are designed to minimise the harm of observed events and do not target root causes. For example, on days of peak air pollution in Paris, cars with certain registration numbers are not allowed to circulate. While this is a necessary action, the policy does not decrease the likelihood or frequency of pollution peaks in the future.

Anticipatory policies are designed to reduce the harms of predicted trends (often based on historical information). For example, a trend towards more frequent pollution peaks coupled with evidence of the effect of air pollution on children’s health (e.g. asthma) may lead city officials to install air purifiers in schools to reduce children’s exposure to polluted air. This action anticipates negative impacts on children’s health and “gets the city ready” to reduce them but does not decrease the likelihood or frequency of pollution peaks in the future.

While reacting to events and anticipating patterns can be fundamental, policy packages mainly focused on reacting or anticipating have a small chance to change unsustainable patterns of behaviour. As the examples above show, this is because the structure of the system, which lies at the source of such patterns, remains intact (Sterman, 2000[30]; Meadows, 2008[31]; Systems Innovation, 2020[32]).

Policies with a transformative intent are designed to shift away from unsustainable systems dynamics and mindsets at the roots of past and current behaviours. They focus on shaping systems from which patterns of behaviour aligned with envisioned results emerge by design. For example, space reallocation in Paris in favour of sustainable transport modes has led to reduced traffic and improved air quality in the city.

Assessing policies’ transformative potential

Identifying the intent underlying policy design is a necessary but insufficient step towards understanding the policy transformative potential. This potential is highly dependent on the level of physical (e.g. road infrastructure) and non-physical (e.g. mindsets) lock-in of the system the policy is trying to influence, and the policy’s level of ambition vis-à-vis such lock-in.

An in-depth understanding of the system’s feedback loop structure is necessary to identify a policy’s transformative potential, i.e. the actual effect a policy can have on changing the dominance of the system’s feedback structure.

Three systemic tools are used to identify policies’ transformative potential: causal loop diagrams, stock and flow analysis, and Meadows’ leverage points framework. When used in combination, the tools trigger questions such as whether a policy strengthens or weakens feedback loops, can change a loop’s dominance, or lead to the creation of new loops. The rest of the section describes these tools in detail.

Causal-loop diagrams

Causal loop diagrams (CLDs) help in the identification of transformative policies by shedding light on the system’s structure. CLDs can be seen as a deep dive into the iceberg model’s “system structure” level (Figure A E.2), enabling the policy maker to better understand the interconnections or causal relationships that produce the results at the tip of the iceberg; and the ways in which a given policy can change (or not) such relations.

Figure A E.3 provides an example of a CLD summarising one of the key dynamics underlying high private car use in Catalonia: induced car demand. Induced car demand refers to the phenomenon by which investment in road expansion aimed at reducing congestion has the opposite effect: they induce car use and increase congestion. The coloured arrows in the Figure show the relationship between variables. A pink arrow between variables means that they vary in the same direction: an increase in a variable leads to an increase in the variable it points at; a decrease in a variable leads to a decrease in the variable it points at. A blue arrow means that variables vary in the opposite direction: an increase in a variable leads to a decrease in the variable it points at; a decrease in a variable leads to an increase in the variable it points at. Each loop label (e.g. B1) denotes a feedback loop, which is either reinforcing (R),1 or balancing (B).

Figure A E.3 can be read as follows. The more people choose to drive, the more congestion and travel time by car (1 in Figure A E.3) increases. As congestion and travel time by car (1) increase, so does the pressure to reduce congestion (2), as no one likes to be stuck in traffic jams. The conventional policy response to this pressure has been to increase public investment in roads for cars (3). Public investment in roads for cars (3) increases road capacity for cars (4), which, all else being equal, reduces congestion and travel time by car (1). However, as congestion and travel time by car are reduced (1), the attractiveness of driving (5) compared to other modes increases. This results in fewer users of shared, micro and active modes, a higher number of cars in the region, and longer average distance driven by car per day; all of which increases traffic volume (6), congestion and travel time by car (1). As explained above, note that this is the opposite effect that increased public investment in road for cars (3) intended to obtain.

Figure A E.3. System dynamics underlying induced car demand.

Note: The coloured arrows show the relationship between variables. A pink arrow between variables means that they vary in the same direction: an increase in a variable leads to an increase in the variable it points to; a decrease in a variable leads to a decrease in the variable it points to. A blue arrow means that the variables vary in the opposite direction: an increase in a variable leads to a decrease in the variable it points to; a decrease in a variable leads to an increase in the variable it points tp. Each loop label (e.g. B1) denotes a feedback loop, which is either reinforcing (R) or balancing (B).

Source: Authors, adapted from (Sterman, 2000[27]), Business Dynamics: System Thinking and Modeling for a Complex World, https://www.researchgate.net/publication/44827001_Business_Dynamics_System_Thinking_and_Modeling_for_a_Complex_World.

Stock and flow analyses

Stock and flow analyses complement CLDs in the study of policies’ transformative potential by helping policy makers understand the system’s physical lock-in.2 Stocks and flows are the elements of a system. Stocks – such as vehicle fleets, public transport infrastructure, or public perceptions - change over time due to inflows and outflows. Stocks are the accumulation of net flows (inflows-outflows) over time, they measure the state of the system and can be seen as the “system memory”. Flows correspond to the activity that has led to stocks. For example, the yearly investment in road infrastructure in the last decades (a flow) has increased road capacity for cars (a stock).

The analogy of a bathtub helps visualise stocks and (in/out) flows. Inflows can be seen as the water pouring from a tap. Stocks are the water that accumulates in the bathtub. Outflows are the water that leaves the bathtub when the plug is off. Figure A E.4 applies the analogy to investments in road infrastructure. Road infrastructure investments (a flow) lead to road construction (not shown in Figure) which increases road capacity for cars (a stock).

Figure A E.4. Stocks and flows: an illustration

Source: Authors.

Understanding the magnitude of stocks and flows can shed light on policies’ transformative potential in several ways. First, it can help identify potential mismatches between the level of ambition or policy stringency and the magnitude of the stocks the policy is trying to influence. For example, while efforts to improve public transport infrastructure may have a transformative intent, their transformative power is limited in systems characterised by decades of investment leading to an accumulation of road capacity for cars. Note that the same efforts made decades ago, when the gap between car and public and active transport infrastructure would have been smaller, could have had a higher transformative power, as the case of the Netherlands illustrates regarding bicycle infrastructure (see section 3.2.2. in (OECD, 2022[34])). Second, stock and flow analyses can remind policy makers that policy action is not limited to opening and closing taps (parameters, in Meadows’ terminology (Meadows, 2008[31])- see below), but can also, for example, pull the plug and reduce stocks3 that lock systems into undesired equilibrium states (e.g. road space reallocation reduces the over dimensioned stock of road capacity for cars. Third, stock and flow diagrams are also a fundamental tool to deepen the understanding of the system structure. Such diagrams are the building blocks of quantitative system dynamic modelling. So far, the system structure analysis in this methodology has been qualitative, based on CLDs (visually easier to interpret than qualitative stock and flow diagrams). Authors see the inclusion of quantitative system dynamic modelling as a next step to improve this framework.

Meadow’s leverage points framework

Meadows’ leverage points framework combines insights from CLDs and stock and flow analysis to identify twelve places to intervene in complex systems, referred to as “leverage points” (Meadows, 1999[35]) Figure A E.5. High-leverage points are places in which a small intervention may lead to a system transformation capable of triggering large behavioural changes. Low-leverage points are places in which small interventions lead to (proportionally) small changes.

The notion of leverage points guides the assessment of policies’ transformative potential and can be mapped onto the four levels of the iceberg model. The leverage point with the lowest potential to trigger change (the 12th) is positioned at the tip of the iceberg. The leverage point with the highest potential is positioned “deep under the surface”. More information on the framework is available at (Meadows, 1999[35]).

Figure A E.5. Leverage points on the iceberg model

Source: Extracted from (Systems Innovation, 2021[36]).

Notes

← 1.  Figure 2.1 only includes balancing loops.

← 2.  The system’s physical lock-in is linked to the notion of hysteresis. The notion of hysteresis emphasises the dependence of a system state on its memory or history and the importance that past events have in shaping the system’s results today (Systems Innovation, 2022[80]).

← 3.  Cognitive science demonstrates that our brain prefers to add rather than subtract (Adams et al., 2021[79]). This biological tendency may help explain why most policies focus on parameters (opening or closing faucets) rather than on reducing undesired stocks (pulling the plug).