Indoor air pollution as an issue of nonattention.
in Nairobi’ s informal settlements
Nid Zimmenmanrt*, Kaveh Dianati*, James Milner, Mwangi Chegé, Balazs Csuvar’,
* UCL Institute for Environmental Design and Engineering
> London School of Hygiene and Tropical Medicine
° Africa Population Health Research Center
4 BuroHappold Engineering
* Comesponding author (n.zimmermann@ucl.ac.uk)
Proceedings of the 35" Intemational Conference of the System Dynamics Society,
Cambridge, MA, July 2017
Abstract
58 percent of Nairobi’s population live in slums under extremely poor and unhealthy condi-
tions. In these settlements, pneumonia is one of the top causes of health issues and deaths among
children and adults, for which indoor air pollution is a known contributor. Yet, the topic of
indoor air pollution receives no attention. Regulatory frameworks and budgets for indoor air
pollution do not exist and humanitarian organisations neglect the topic despite its drastic health
effects. This paper addresses the dynamics of indoor air pollution from two sides: It investigates
the underlying structural mechanisms of organisational and govemmental attention dynamics.
It also analyses how govemment policies interact with these mechanisms and create indoor air
pollution and health outcomes. We employed participatory system dynamics to investigate at-
tention to indoor air pollution, to develop the model structure, policies and to discuss wider
consequences. Participants included community members, local and central policy-makers,
parastatals, NGOs and academics. Modelling suggests possible avenues to improve indoor air
pollution in Nairobi’ s informal settlements and the participatory process also gave insights into
their feasibility. The participatory work also showed some potential of awakening and mohilis-
ing community attention.
Keywords
Attention, indoor air pollution, health, Nairobi, informal settlements
etal. (2017) Indoorair ion as an issue of nonattention in Nairohi' s informal settlements
1 Introduction and objectives
Nairobi city, according to the most recent national census was home to 3.14 million inhabitants
(Kanya National Bureau of Statistics, 2010), having grown fromjust under half a million at the
country’s independence in 1963. The city’s population growth is fuelled both by natural in-
crease and migration from rural and other urban areas. For a long time, Nairobi has remained.
the country’s principal city and it remains an attractive destination for people looking for live-
lihood opportunities that are lacking in the mostly agricultural rural areas.
The rapid growth of the city population has not been accompanied by sufficient provision of
affordable housing and other social amenities leading to the mushrooming of slum settlements.
It is estimated that Nairobi alone has over 150 slumsettlements, scattered across the city. These
settlements which occupy less than five percent of the city’ s land mass are home to an estimated.
60-70 percent of the city population (Beguy et al., 2015). Numerous studies have reported the
challenges that slum residents face, including the near absence of the public sector and poor
access to public goods and services; with negative implications for various health outcomes as
well as other outcomes such as educational attainment (African Population Health Research
Center, 2002, 2014; Kyobutungi, Ziraba, Ezeh, & Yé, 2008; Mugisha, 2006).
Typical housing units in slums have tin/corrugated iron roofing and mud or tin/comugated iron
sheet walls. Most households rent one room measuring about 10ft by 10ft and these rooms serve
as the kitchen, bedroom and living room The rooms usually have one door and one window
although in some cases there are no windows at all. Most households rely on kerosene for cook-
ing and lighting as well as charcoal or wood for cooking. In the poorest of households, the use
of plastic waste, cloth rags and other unconventional materials as fuels has been reported
(Muindi, Egondi, Kimani-Murage, Rocklov, & Ng, 2014) with grim implications for indoor air
pollution and associated toxic fumes. Further, a separate study found that about seven percent
of the respondents are exposed to cigarette smoke while in their homes, where one or more
members of the household are reported to smoke. The same study recorded high levels of par-
ticulate matter with aerodynamic diameter of 2.5 microns and less (PM25), especially in the
evenings and in households buming charcoal/wood and kerosene (Muindi, Kimani-Murage,
Egondi, Rocklov, & Ng, 2016). Other than housing features and behaviours that impact on the
air quality, slums are located in areas close to primary sources of air pollutants. For example,
many slums are built near busy highways, within industrial zones or in close proximity to open
dumpsites (though in most cases it is the dumpsite that is sited close to the communities and
not vice versa). Indeed, outdoor measurements of PM2.5 concentrations in slum areas found that
there were spatial and temporal variations with slum villages close to major outdoor sources
such as dumpsites having higher concentrations, while momings and evenings were also noted
to have elevated levels (Egondi, Muindi, Kyobutungi, Gatari, & Rocklév, 2016). In a context
of weak or non-existent policies to minimise emissions from various sources, slums experience
high exposure to outdoor air pollution compared to non-slum areas of the city.
It is against this backdrop that the project, Housing in Nairobi’ s Informal Settlements (HINIS),
was launched in June 2016. The project is a collaborative effort. HINIS brought together mul-
tiple stakeholders including goverment officials, researchers and academics, NGO represent-
atives as well as community members fromtwo of Nairobi’ s informal settlements - Korogocho
and Viwandani - where APHRC has been working for over a decade. We worked closely with
these different stakeholders to understand the primary needs and constraints around housing
and health in Nairobi’ s informal settlements. The overarching goal of HINIS was to maximise
the opportunity for transformative change in the urban environment to achieve health and sus-
tainahility objectives.
etal. (2017) Indoorair ion as an issue of nonattention in Nairohi' s informal settlements
We quickly observed that indoor air quality significantly contributes to the health burden of
slum residents. For example pneumonia, for which household air pollution is a known contrib-
utor, is the fourth most common cause of death among adults (Egondi, Kyobutungi, Kovats,
Muindi, & Ettarh, 2012) and the leading cause among under five year old children, more fre-
quent than diarrhoea, malnutrition or birth-related causes (Ye et al., 2009). At the same time,
however, govemments and humanitarian organisations focus on issues such as infrastructure,
matemal care and malaria, but they pay little attention to pneumonia and indoor air pollution.
This paper therefore investigates the issue of indoor air pollution in two Nairobi slums. It as-
sesses the drivers of poor air quality and develops a system dynamics model showing interac-
tions between drivers of poor air quality and policies aimed at addressing them. It also estimates
the potential contribution to health and well-being arising from implementing appropriate pol-
icies. At the same time, it reports on a pilot of using participatory system dynamics to under-
stand the challenges of poor air quality, explore policy options and develop potential solutions.
2 Air pollution and healthcare context
The underlying govemance and financial structures affect the indoor air pollution within infor-
mal settlements. Understanding which stakeholders are the key decision makers who allocate
budgets is useful when moving forward with the project and trying to influence change.
The 2015/16 national and county level health budget data suggests that the Ministry of Health
and local counties distribute their funds (KES 144bn) to recurrent expenses (KES 90bn), such
as personnel, drugs and medical supplies, as well as to development expenditures (KES 54nb),
ie. construction of buildings, federal and local health programmes (ex. free matemity care), as
well as initiatives funded by intemational donors. The Ministry of Health accounts for a yearly
KES 19bn of funds arriving from donors, most of which is earmarked to HIV, reproductive
health, immunisation and other health system support programmes.
The 2015 total planned health expenditures in Nairobi specifically account for KES 6.5bn. Over
80 percent of this is to pay for recurrent expenses, the remainder is to development, primarily
building hospitals and health centres. Programmes focusing on preventive and promotive health
services account for a total budget of KES 43 million, divided into five categories: TB, Malaria,
Family planning and Environmental Health. There are no specific mentions of any programme
focusing on air pollution within the Nairobi country council’s health accounts. Looking at it
more broadly, pollution control appears under Environment, Water & Energy sector, focusing
on environmental monitoring and enforcement, but is not targeted towards air pollution specif-
ically.
Yet slum residents hardly participate in existing schemes such as the National Hospital Insur-
ance Fund (NHIF) that all working Kenyans contribute to. With monthly contributions between
KES 150 and KES 1700, KES 500 for the self-employed, the scheme is costly compared to the
estimated average monthly income in a Nairobi slum of KES 4000 per person (Desgroppes &
Taupin, 2011). It does not match the informal work that slum dwellers often pursue either.
Reports suggest that only seven percent of the poorest 40 percent of Kenyans are covered under
this scheme (FSD Kenya, 2016), applicable to most slum dwellers studied during this project.
Therefore these poorest 40 percent frequently use savings, rely on social networks or sell assets
to pay for health-related costs (Gubbins & Ravishankar, 2016; Zollmann & Ravishankar, 2016).
Tt has been established that large benefits can be gained fromyreducing air pollution, e.g. through
switching to cleaner fuel sources such as LPG (Morgan, 2015). Despite apparent benefits, there
are currently no programmes focusing on air pollution from a health point of view, neither at
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
federal nor at county level. This might be due to lack of specific intemal or donor funding, but
could also be the result of a lack of understanding at govemmental level of the importance of
the issue.
3 Methodology: participatory system dynamics
To address indoor air pollution, this study combines different methods: participatory system.
dynamics modelling and engagement in workshops as well as further model refinement and.
modelling off-site. We also conducted health impact modelling using a life table model, but do
not detail it is this paper. The participatory system dynamics approach allowed us to make use
of the different kinds of expertise available in our interdisciplinary stakeholder groups and at
the same time to inform them and our workshop through state of the art modelling.
SD modelling is used to organise the knowledge in a visual structure that promotes leaming
and allows policy design through simulation. This was done by asking participants to identify
the chains of causality within the system, by asking questions such as: What are the main
sources of indoor air pollution in the slums? What are the main drivers for these sources?
Following this process, the causal structure of the system can be captured within the model and.
important causal feedback loops can be identified.
As we were addressing a complex issue involving multiple perspectives and groups as well as
implementation challenges, we used a participatory approach. Allowing policymakers to rely
on their own thinking process to model the situation and subsequently evaluate altemative pol-
icies offers several advantages. For instance, using Participatory System Dynamics engenders
asense of ownership and commitment to the outcome of the modelling process and in this way
increases the chances of successful implementation of resulting policies (Vennix, 1996).
We brought in real world data back in the office in the period of October 2016 to Decent+
ber 2016 to develop a quantified simulation model from the collaborative map generated
through stakeholder workshops. To make concrete policy recommendations, we moved as far
as possible towards formal stock and flow modelling and simulation. Y et, in several areas of
this research, quantifying variables and obtaining data for them has posed some challenges.
The modelling relies on recent monitoring studies that estimated average exposure to particulate
matter (PMp5) in Korogocho at 108.9 g/m? in homes (Muindi et al., 2016) and 166.4 yg/m? in
ambient air (Egondi et al., 2016). Although these studies involved relatively limited sampling,
we use those estimates here as the basis for illustrative calculations of the health burden asso-
ciated with PMb.s in Korogocho. Under the crude assumption that local residents spend 50 per-
cent of their time indoors at home and 50 percent of their time outdoors, the current time-
weighted average PM25 exposure is 137.7 ig/m’. Our health life table modelling suggests that
among the 180,000 people living in Korogocho, particulate matter (PM25) is responsible for
about 250 deaths annually, of which about 75 are children aged under 5. These deaths result in
almost 12,500 years of life lost and more than 10 years of lost life expectancy at birth.
4 Our participatory process and its intermediate results
The primary means used to implement the participatory process was facilitated discussions dur-
ing two rounds of workshops in Nairobi. We held the first round of workshops in September
2016 and the second one in January 2017. We reached out to a diverse group of stakeholders to
ensure that different aspects of housing in informal settlements, air pollution and its effects on
health would be discussed extensively which would contribute to the development of a system.
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
dynamics model. Initial discussions during the first round of workshops involved a broad range
of stakeholders that included both experts as well as non-experts (lay-people). Later, the dis-
cussions mostly involved individuals with expertise on air quality and its impacts on health, as
well as those working on policy development and implementation. These discussions with ex-
perts served to refine the input received during the workshops with broader participation.
Besides workshops, we also used follow-up surveys and focus group discussions as part of the
participatory processes. Following the first workshop, we used a survey tool to solicit opinions
on the relative importance of different policies that aim to address air quality issues. This pri-
oritisation by the stakeholders themselves informed the selection of policies used in the model-
ling. During the second round of workshops, we held focus group discussions (FGDs) with
community members from Korogocho and Viwandani, right within the informal settlements
where they live. The discussions tackled indoor air quality and housing. These participants pro-
vided rich information on their experiences, their agency and the challenges they face in trying
to improve the quality of indoor air within their homes. The FGDs were of mixed gender and.
age because the discussion did not cover sensitive issues that would require disaggregation by
gender or age.
Following each round of workshops, the discussions were transcribed to capture details of the
issues raised by the different stakeholders. These transcriptions formed an important part of the
qualitative data that was included in the system dynamics model.
41 Project partidpants
The participants of the projects who consented to participate were drawn from the following sectors:
e Commumity members from Korogocho and Viwandani, two infonmel settlements in
Nairobi where APHRC has been working for over a decade. The community members
were represented by tenants, landlords and commumity leaders. These representatives
offered their opinions based on their lived experiences while local leadership offered
insights into current developments that could have a bearing on housing and conse-
quently indoor air quality. For instance, the ongoing land allocation and issuance of title
deeds to Korogocho residents is expected to have a positive impact on quality of housing
and consequently on indoor air quality.
e Academia and researchers from local institutions including the University of Nairobi
and Jomo Kenyatta University of Agriculture and Technology. They possess expertise
inair quality monitoring, altemative fuel development as well as design and urban plan-
ning. These individuals provided technical input into the discussions.
e Govemment departments working at the national and sub-national levels. The sub-
national representatives were drawn from two counties - Nairobi and Kisumu. Partici-
pants from Nairobi County represented the health and environment departments as well
as the slum upgrading programme. Kisumu County was represented by officials from
the health, environmental and economic planning departments. These goverment de-
partments are the ‘owners’ of the problem identified in this study given their mandate
as implementers of programmes in their respective counties.
e National parastatals that work in housing, health and environmental regulation. These
were the National Housing Corporation (NHC), National Hospital Insurance Fund.
(NHIF) and the National Environmental Agency (NEMA). These stakeholders were in-
strumental in providing insights into policy and practice as well as pathways into oper
ationalising some of the outputs of the two workshops.
‘Zimmenmamn et al. (2017) Indoor air i ‘issue of ion in Nairohi’ s informal
e Non-govemmental organisations (NGOs) and civil society organisations with inter
ests in housing or air quality such as the Stockholm Environmental Institute, Global
Alliance for Clean Cookstoves (GACC) and Kenya Alliance of Residents Associations
(KARA). They brought into the discussions perspectives from other contexts facing
similar challenges.
e UNagencis focused on environment, housing and urban planning i.e. UN Environment
and UN Hahitat. Representatives from these organisations provided technical input into
the discussions based on their previous and ongoing projects as well as insights from
ongoing global efforts to improve urban housing and air quality.
42 Workshop I: Hopes for the fubure of Nairobi stums
While the HINIS project most closely focuses on air pollution, on Day I we started with a
broader scoping/visioning exercise to explore pressing issues for potential future projects. After
initial opening introductions and presentations, we rana participatory activity in which we elic-
ited participants’ hopes for the future of informal settlements. We then sorted and clustered
Figure 1: Hopes
Subsequently, we used a voting system to prioritise the identified clusters or themes, as viewed.
by participating stakeholders. Table 1 lists the identified clusters of issues in order of im-
portance according to participants:
‘Table 1: Ranked clusters of hopes
Rank —_ Cluster of Hopes (Issues) | Number af votes
Integrated Govemance and Management
Land Tenure and Ownership
Services Impn
Implementation of Integrated Waste Management
Socio-Economic Empowennent
Nw Oo Py)
auhwne
During the aftemoon, we selected the three highest ranked themes, together with the fourth
theme of indoor air quality as the central topic of the HINIS project, to continue working on
and define more closely. According to their interests and expertise, participants formed four
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
workgroups around these four topics according to their interests and expertise and spent several
hours discussing each of these topics in more depth at their table. The purpose of the aftemoon
exercise was to develop preliminary project proposals around each of the themes. These pro-
posals included elements such as project title, background, problem definition, problem type,
primary audience and potential policies. At the end of the group activity, participants recon-
vened to the plenary to present their work and get feedback on it from other groups. The pre-
liminary project proposals developed on this day has the potential to be used in order to define
other snall or large projects within the context of Nairobi’ s informal settlements.
The agenda for Workshop I is presented in Table 2.
Table 2: Workshop I agenda
Time Activity Objective
0830-9.00 Coffee and registration
09.00- 10.00 Weloone, introductions, opening Background
10.00- 11.00 Hopes elicitation activity Exploring future scenarios for Nairobi informal
settlements
11.00- 11.30 Coffee break
11.30- 1200 _—_Priaritisation of hopes Ranking themes/clusters according to perceived
importance/uryency
12.00- 1300 = Lamchbreak
13.00- 15.00 Specifying the most pressingissues Developing useful problem definitions and pro-
ject proposals
15.00- 1530 Coffee break
15.30- 17.00 Group presentations
17.00- 17.30 — Closing
43 Workshop II: A causal map of indoor air pollution
The second workshop day focused on the issue of indoor air quality in Nairobi’ s informal set-
tlements. We started the day by identifying the most central variables conceming indoor air
quality and related policies in addition to the ones identified on the previous day. We paired up
participating stakeholders and asked each pair to spend a few minutes to list as many key vari-
ables in this context as they can. Facilitators then elicited these variables one by one, ina round-
robin fashion and listed on the board. The listing was done under three categories: (a) policy
variables, (b) drivers and (c) indicators. Where there was disagreement as to the variable name
or under which category a certain variable should be listed, we discussed it until we reached.
consensus. The list of elicited variables sorted under three categories can be seen in Table 3.
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
e Public publication and scnr
tiny of data
¢ Enforcement
¢ Zoning
Land subdivision
Table 3: List of identified variables
Prevalence of clean cook-stoves
Prevalence of clean lighting sys-
tems
Implementation of emission stand-
ards
Inplenetbon of building public
th standards —
pealarinrewetl
Prevalence of clean ST eaiaans
Population density
Intensity’ of nearby industrial pollu-
Literacy rate
Availability of clean cooking
stoves
Prevalence of optimal ventilation
behaviour
: ae
Buming of rubbish
Price of clean fuels
Availability of clean fuels
Hours stayed at home
Perceived security and use
e Effect of house type
| Indicators
¢ Indoor air pollution (PM2.5,
CO, black smoke)
Outdoor air pollution
e Prevalence of respiratory dis-
eases
¢ Health outcomes
o Days missed at work and
other secondary indicators
¢ Pollutant-attributable burdens
(childrespiratory illness, IHD,
COPD)
Subsequently, the facilitators briefly introduced the System Dynamics methodology, which is
the predominant modelling approach in this project. In this project we built a formal System
Dynamics model, starting with a participatory causal loop diagramming workshop with stake-
holders with various affiliations and with different areas of expertise. Figure 2 depicts (part of)
the causal loop diagram (CLD) developed by the participants in Workshop II. Facilitators with
expertise in System
Dynamics modelling guided this process. As can be seen, the level of in-
door air pollution is at the heart of this diagram, with various driving variables surrounding it.
We later transferred this picture into the Vensim System Dynamic
3), to be used as basis for a formal quantified model useful for policy simulation.
ics simulation software (Figure
etal, (2017) Indoorair polluti issue of ion in Nairobi’ s informal
Figure 2: Causal Loop Diagram developed during Workshop II
‘ationto rise perceived goverment
+ fail for indbor
Sr Peto
Figure 3: Causal Loop Diagram transferred into Vensim.
The resulting qualitative diagram then needed to be refined, quantified and parameterised to
obtain a formal model that can be simulated and used for policy analysis. For this purpose and
within the context of this pilot project, we needed to focus on a limited number of policies in
order to delineate the boundaries of the model and be able to delve deeper into those particular
focus areas with a more developed version of the model. Therefore, we circulated a question-
naire to our stakeholders asking them to rank the importance of policies identified in the course
of the workshop. Table 4 shows the final ranking of policies based on stakeholder opinions.
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
Table 4: Indoor air related policies ranked by stakeholders according to importance
No Policy Average Importance
1 Monitoring of air quality 450
2 Fuel quality standards 433
3 Enforcement of regulations 417
4 Health impact assessment 414
5 Emission standards 4.00
6 Promotion of clean fuels 4.00
7 Appliance performance standards 4.00
8 Subsidising or provision of clean appliances 3.83
9 Building standards 3.83
10. Investment in economic empowerment 3.67
11. Industrial environmental audit and assessment 3.60
12. Public publication and scrutiny of data 3.33
13 Zoning regulations 3.00
14 Landsubdivision 2.17
On the basis of this ranking, the project team chose the following final list of policies to be
included in the formal model:
1. monitoring of air quality
6. price of kerosene vs. price of LPG
6. price and coverage of electricity
8. prices of clean appliances
ventilation and outdoor air pollution.
With a clearer focus for the project and better defined boundaries for the model, the project
team then refined, quantified and parameterised the above qualitative model into one that can
be simulated and used for evaluating the potential short and long-term effects of the above
policies. In the following section, we will look at a simplified schematic of this model.
The agenda for Workshop II can be seen in Table 5.
Table 5: Workshop II agenda
Time Adiivity Objective
830- 9.30 Coffee and registration
9.30 - 10.00 Intro and recap Background, orientation, goal clarification
10.00- 11.00 Variable elictation Preparation for model structure elicitation
11.00- 11.30 Coffee break
11.30- 1200 —_ Cancept model presentation Familiarisation with modelling
12.00- 1300 Model structure elicitation Coming up with a causal model of the problems
and potential improvements
13.00- 14.00 Lunch break
14.00- 15.30 Model structure elicitation Divided into suk-topics: impacts on air quality,
health effects and policies
1530-1600 = Model review Identifying the most important insights
16.00 Closing
44 Workshop III: Health impacts and policy effects
The third workshop held in January 2017 aimed at discussing and refining with the stakeholders
the model that emerged from the September workshop and discussing the feasibility of policies.
10
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
We reviewed the work done on the model to-date showing its evolution from the initial causal
loop diagram developed (Figure 2) to that shown in Figure 4. This involved a systematic walk-
through of the different elements forming the model and their interactions. We sought to verify
with the participants whether these elements and the interactions depicted resonated with them.
and reflected their understanding of the many inter linked issues around indoor air pollution in
Nairohi’s slums. Subsequently, we simulated different scenarios based on the policies selected.
in the survey conducted following the first workshop. Participants discussed the outcomes of
different scenarios modelled, providing perspectives on what resonated with them and. what
surprised them. The scenarios are discussed in more detail in section 6.2.1 of this paper.
enforcement, political
will, and good
public awareness about
Public concem about air pollution from
indoor air pollution +
governance healthcare costs }
Ne A
effective expenditure to Bi: public awareness about B2: HIA health issues
reduce indoor air Me = indoor air pollution from ” attributed to air
2 lonitoring f
pollution (frpritoring pollution
indoor air_ x x
monitoring’
coverage health impact’
w assessment
share of indoor air coverage health issues due
quality expenditure for Pr to air pollution
price of kerosene. monitoring ‘share oF als quitly
price of proportion of expenditure for health
housholds using impact assessment
to
price of clean stoves ean staves, 4 polition _
share of air quality proportion of = household ME an (.
expenditure for households using <_ pollution
appliance subsidies * cleaniighting ae Bs
74 f-
price ofclean™ electricity electricity ‘outdoor air
lighting coverage price ventilation pollution
Ret i G ‘i ‘ Blue: rei i Brown: balancing loops
Figure 4: ified causal loop di discussed during Workshop III, i.e. day one of the January workshop
The principal participatory activity for the day involved prioritisation of the policies selected in
the survey. This was done by forming five groups, each of which represented the different clas-
ses of stakeholders present. We placed stakeholders in these groups so that they could articulate
a unified view based on their similar backgrounds. The five groups formed were:
County officials
National parastatals
Academics and researchers
UN agencies, NGOs and civil society
Community members
We asked stakeholders in each group to rank the most important policies that would be critical
to ensure achievement of improved air quality in Nairohi’s informal settlements. Each group
then presented the results of their discussions and the rationale behind their ranking. Stake-
holder groups then proceeded to define their role and interests in the improvement of indoor air
quality in Nairobi’s slums. They further discussed implementation of the selected policies, in
tems of who should be the primary actor as well as obstacles and unintended consequences
arising from their implementation. While there was no clear consensus on the priority policies,
there were rich discussions across the different groups, especially as they elaborated on their
ownzoles and perceptions of other stakeholder roles. Results fromthe discussions in the plenary
are shown in Table 6.
11
‘Zimmenmamn et al. (2017) Indoor air i issue of ion in Nairohi’ s informal
Table 6: Results fromthe plenary discussions on policy implementation.
Who benefits? Implementation Barriers / supporting factors
1/3 Moni-— ¢ Publichealth « All citizens ¢ Barely ¢ Lack of overarching air quality
toring of air departments = « Publichealth e UNEP did strategy and budget at govem-
qualityand = Academia service (re- studies > No ment level
health ine and universi- duced costs) policiesfor ¢ Always beeping device >
Pact assess- ties ¢ Insurance com- air pollution, what exactly is the problem?
ment Private sector panies only regula’ = People may tamper with moni-
¢ Govemment tions! > toring systems
(better data Hard to en- e Education can generate de-
available > force mand for this
better deci- ¢ Inabsence of
sions) govemment
action, aca-
demia is fill-
ing the gap,
but typically
only for
short-term
studies
‘Unintended
i consequences
. |e might feel ‘Oh, we've solved this problem’ > reduce investment.
¢ Itmay demoralise people if levels are always high
¢ Unfair use of information, e.g. by insurance companies
e Potentially higher costs of health insurance for polluted areas
4 Enforce- « Producers ¢ Citizens e Barely ¢ Costs of enforcement (checking,
ment izing, etc.)
¢ Combine education and enforce-
ment. Sensitising the public may
bea way around enforcement
e Tuming regulatory body to cash.
cow (corruption)
Long prosecution times
e Low fines and punitive measures
¢ High incentive to not conform.
‘Unintended consequences
¢ Too much enforcement > cause corruption?
6 Lower e Central gov- « Newoonsum « Existing e Distribution system.
Prices of emment as e Human waste gas systems
dean fuels ¢ Businessus) Residents Clubs can reduce costs
Yaise price qs ¢ Businesses ¢ Community investments
of kerosene = « Taxpayer ¢ Distributors
© Matatu driv-
as
Unintended
Consequences
e Does not work for poor because of cash flow - poor now pay more for kerosene
e More LPG canisters results in more explosions
¢ LPG dist. very profitable > fire incidents more common
12
‘Zimmenmamn et al. (2017) Indoor air i issue of ion in Nairohi’ s informal
Who benefits? Implementation Barriers / supporting factors
& Provide © Central gov- « Households « No Initial investment ($40)
dean stove emment . ¢ Failure to sensitise oommuni-
subsidies ¢ Corporations ties on proper use
e Businesses ¢ Costs for establishing regula-
selling subsi- tory system. Regulation re-
dized appli- ied!!!
ances and al- e Availability of clean fuels (in-
temative fuds ).
¢ Enforcement
¢ Inconsistency of policy
¢ Government intemal commu
nications
e Must be made profitable for
businesses
Lentered Oxenrers
2 Smee efficient’ stoves are even more polluting (CO)
Some commercial organisations pocket carbon credits instead of passing them on to end
‘USES
¢ May lead to disemployment in manufacturing organisations if they are unable to manage
market competition
The agenda for Workshop IIL, i.e. the first day of the January workshop can be seen in Table 7.
Table 7: Workshop III agenda
Time Activity Objective
08.30 Coffee
09.00 ‘Welcone and presentation of Participants have spoken, feel welcomed
09.10 Recap and CUSSH Background, orientation, trust-building and goal clari-
fication
09.20 Health impacts Presentation and discussion
09.50 Presentation of simulation. Familiarisation of participants with the system dynam-
model ics model
10.25 Break
10.45 Model faniliarisationinple- Simulating a number of preconceived scenarios and
nary looking at results
Reviewing generated insights in one slide
1145 Discussion on model and Discussing simulations, whether the results seem real-
results istic, and what modifications do people suggest to the
model
Rb Lunch break
B35 Plenary debate Prionitisation of policies, elicitation of obstacles and.
‘unintended consequences
15.30 Summary of insights Relation of this work to general concems
16.00 Closing Ideas about project future, hopefully feeling of time
well-spent.
45 Focus groups: Real impediments
On the day following Workshop III, the facilitation team met with the two slum communities
where the APHRC runs the demographic surveillance. This allowed us to follow up on some
emerging issues from the workshops. The meetings were convened with the respective leaders
of the two slums (the chiefs who represent the national goverment at the administrative loca-
tion in which the two study sites fall). Figure 5 shows one of these sites.
13
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
Figure 5: View of Korogocho
After meeting with the chiefs individually in the moming, we facilitated two focus group dis-
cussions with residents of the two communities. The groups consisted of about 15 participants
each in Korogocho and in Vivandani. We discussed further on indoor air, the barriers residents
see in their adoption of clean cook stoves and other issues touching on housing, outdoor air and.
community/individual agency to agitate for action against known polluters. In Vivandani, only
one participant owned a clean cook stove.
“So we end up using kerosene or even charcoal.’ “There are some of us who cannot afford that
gas or electricity and use the stove therefore it depends on the income.’
We leamed that a $35 investment for a clean cook stove is too high for the households and that
the initial costs of canisters are too high as well, but we discussed the opportunity of forming
self-help groups to afford the initial investments.
Participants alluded to outside air pollution as well.
“Like around 6 am when you wake up at 5 a.m you feel like it is difficult to breathe, because
the air is very bad even outside. Because that is the time the factories open, the water is now
let to the river, all the waste now flows because of those people who [...].’
They attributed some the root causes of this problem to the dumpsites that persist due to cor
Tuption.
“I was saying you can’t open the window and these outdoor pollutions they are contributed by
the government mostly because we have NEMA, isn’t it supposed to deal with those environ-
mental pollutions? But even like this dumpsite already they were told to stop dumping but cor-
ruption makes it continue. Because the person will pay money and continue dumping. Like these
industries, the government cannot ban thembecause they benefit. So we'll just continue to suffer
because we must open them [the windows].’
14
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
“When NEEMA comes and gets the self-help group what do you do? You bribe them and the
day continues.’
Participants mentioned that job opportunities and slum upgrading are their most pressing needs.
Yet, they were aware of air pollution and that it has health consequences and stressed that fur-
ther education on air pollution is strongly required. Y et, especially the residents of Korogocho
expressed that no organisation has ever educated them on the issue of air pollution. But they
emphasised their lack of influence for triggering any change.
These discussions were recorded for later transcription. Table 8 shows the adapted agenda that
we followed on the day.
Table 8: Focus groups agenda
Time Activity Lead
08.30- Travel to field sites -Viwandani and Korogocho APHRC
10.00
10.00- | Courtesy call on Chiefs APHRC
11.00
11.00- Focus Group Disassions APHRC/
13.00 e Short oral introduction (What we want to achieve in the project and today and why is | UCL
it important)
¢ What affects your decisions to buy, to use clean stoves?
¢ Isitrelative price of LPG to kerosene that affects the decisions?
¢ What are your perceptions of outdoor air? Have there been any changes in time?
In what area and to what extent do NGOs not follow through and to what extent
don’t we know it?
¢ What needs to be done to really transform your life positively?
© What do you suggest to deal with the problem?
13.00- Household visits APHRC
14.00
14.00 Close and departure fromfield APHRC
5 A formal model of indoor air pollution in Nairobi sums
Given the potentially high benefits of a reduction of air pollution, we will now investigate the
causal structure of the simulation model we created in order to examine how different policies
affect indoor air pollution and resident health.
5.1 Model structure
The process resulted in a ~180 variable model that can be accessed in the supplementary files.
Due to its complexity, we only describe a simplified causal loop diagram and the main feedback
loops in the subsequent sections.
5.1.1 Simplified causal loop diagram
Figure 6 portrays the simplified causal loop diagram arrived at through distilling the key causal
skeleton of the formal System Dynamics model. The legend on the bottom explains the colour
coding. Starting with the central variable of household air pollution highlighted in red as our
main indicator, let us move backwards (against the direction of the arrows) along the chains of
causality to investigate the key dynamics of the system as modelled here. The key drivers of
the average level of household air pollution in Nairobi’s slums are the levels of outdoor air
15
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
pollution and ventilation extemally and the proportion of households using clean stoves/light-
ing intemally. Within this project we were mainly keen on exploring the intemal factors, ie.
the prevalence of clean appliances. These prevalence levels are mainly driven by the levels of
expenditure for each type of appliance, their prices and also clean fuel prices relative to the
price of brown fuel (in this case kerosene). The lower the prices of clean appliances and/or
clean fuels, the higher the take-up and usage of these by residents of the informal settlements.
In the case of clean lighting, this prevalence is also affected by electricity grid coverage.
The expenditure for subsidising clean appliances comes from the total funds spent for combat-
ting indoor air pollution, effective expenditure to reduce indoor air pollution. This expenditure
is determined not only by public concern about indoor air pollution, but also by the extent of
enforcement, political will and good governance, defined as an index moving within a range of
zero to one and set by the user as a scenario variable. Note the perpendicular delay sign on the
arrow linking public concern about indoor air pollution to effective expenditure to reduce in-
door air pollution that represents the time it takes from becoming increasingly concemed about
indoor air pollution to taking effective action against it. As can be seen on various causal links
inthe diagram, these delays are accounted for in the formulation of the System Dynamics model
and have notable implications on the behaviour of the model, as they do in the real world. Public
concer about indoor air pollution comes from two sources: either direct monitoring of indoor
air pollution levels or through awareness about healthcare costs associated with indoor air pol-
lution, which can be estimated on the basis of information generated through health impact
assessment (HIA) studies. The extent to which either monitoring or HIA are systematically
carried out in the informal settlements depends on the levels of funds available for each, which
are in tum determined by multiplying the total effective expenditure to reduce indoor air pollu-
tion by the share of this expenditure going to either of these initiatives. The levels of awareness
and concem generated through each channel are also driven by the actual levels of indoor air
pollution; directly so in the case of monitoring and in the case of health impact assessment,
travelling through exposure to air pollution, health issues due to air pollution and health issues
attributed to air pollution. Exposure itself is a consequence of either indoor or outdoor air
pollution. Outdoor air pollution lies outside the scope of this project and it is fed as exogenous
data to the model. Its past levels are set according to the limited available real-world data and.
its future levels is decided upon by the user as a scenario variable.
public awareness about
air pollution from
enforcement, political
aaa public concem about
indoor air pollution +
governance healthcare costs }
Ne fk
effective expenditure to B1: public awareness about B2: HIA health issues
reduce indoor air Mo! i indoor air pollution from " attributed to air
ynitoring #
pollution girpnitoring pollution
i #
monitoring
coverage health impact’
. assessment
share of indoor air coverage health issues due
quality expenditure for ie ‘b air pollution
price of kerosene. itoring share of air quality
ack proportion of expenditure for health
Prec housholds using impact assessment
to
price of clean stoves’ oe. a plion
share of air quality + proportion of = household ar f.
expenditure for households using _ pollution
appliance subsidies % cleaniighting Sh t
es:
price ofclean™ electricity electricity ‘outdoor air
lighting coverage price ventilation pollution
Red: indicators Green: policy/scenario vari Blue: reinfarci Brown: balancing loops
Figure 6: Simplified causal loop diagramof the model
16
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
5.1.2 Main feedback loops
Tt becomes immediately evident while looking at the causal structure in Figure 6 that a number
of feedback loops exist within the system that might drive change or counter it in the real world.
Four noteworthy feedback loops are identified and numbered in the diagram. R1: Monitoring
and R2: HIA belong to the class of feedback loops known as ‘reinforcing’, while B1: Clean
Stoves and B2: Clean Lighting are known as ‘balancing’ loops. This inherently different nature
of these feedback loops can be decisive while investigating various policies. It is therefore
worthwhile to look at this distinction more closely.
Let us start with the two balancing feedback loops. A potential increase in expenditure for clean
appliance would, ceteris paribus, help bring down household air pollution, thus making the
public slightly less concemed about this issue. A less worrisome public (be it the goverment,
the communities, or NGOs) would then perhaps think that the issue has to some extent been
contained and perhaps no longer warrants the previously increased level of allocated funds and.
decide to divest those additional funds to other more pressing problems, bringing the level of
expenditure back close to its initial level; hence the use of the label ‘balancing’ .
Yet, if the expenditure for monitoring or health impact assessment studies is boosted, once the
results of such studies are published, this new information could make the public more anxious
about indoor air pollution, leading to a potentially higher budget allocated to this issue for the
next year. Therefore, an increase in the share of indoor air quality for monitoring/health impact
assessment has the potential to enlarge the size of the pie of available resources the next time
round. This means that while an increase in fund available for monitoring or HIA studies may
initially happen at the expense of fewer clean appliances provisioned for the population, with
the passage of time the result of these studies could be a powerful case for demanding a poten-
tially much higher total budget for fighting indoor air pollution, making it possible to spend
more not only on similar studies but also, importantly, on a more expansive provision of clean
appliances than would otherwise have been feasible. This argument makes a theoretical case
for allotting a share of whatever available budget to monitoring and health impact assessment,
a policy that we are going to test in the Modelling results section 6 on page 18.
5.2 Model validation.
The System Dynamics model has undergone extensive validation, both structurally and behav-
iourally. The structure has been validated against expert opinion in the course of the multi-
stakeholder workshops as well as ongoing collaboration with local experts at APHRC. The
model has been parametrised using the limited numerical data available from various sources,
in particular the NUHDSS database.
The behaviour of the model has also been validated against time-series data from this database.
For instance, looking at Figure 7 and Figure 8, it can be seen that the model’s Base Run (blue
curve) captures the general long-term trend in historical data fairly well. Since the focus of this
project is long term policymaking, the fact that short-term oscillations are not captured is not
considered a handicap of the model for our purpose.
17
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
Number of households owning clean stoves
o
a a a a ee
Timo (yeas)
Base Run (b)
Data(a)
Figure 7: Number of households owning clean stoves. Black curve (d): historical data. Blue curve (b): model simulation.
The scarcity of available time-series data for important variables in the model, including our
very central indicator household air pollution, posed a challenge to the behavioural validation
of the model. This is a limitation that entails a degree of caution regarding the use of the model
as the only input to policymaking. Nevertheless, whilst taking such limitations into account, the
model still offers valuable insights for policy, as we will further observe in the next sections.
Number of households owning clean lighting
30,000
2,500
Data (@)
Figure 8: Number of households owning clean lighting. Black curve: historical data. Blue curve: model simulation.
6 Modelling results
In this section, we will start by examining the Base Run, which is the model’s projection of
Current trends under business-as-usual. However, as we will see, these projections are not
merely extrapolations of current trends. Instead, variables can undergo changes in behaviour
mode, as the behaviour of the model is driven by its structure, and not by inputs from outside.
This will be made clearer once we look at the Base Run for our key variable of household air
pollution. Afterwards, we will explore four different scenarios and look into potential implica-
tions and relative merits of various combinations of policies for improving indoor air quality.
18
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
61 Baserun
Firstly, let us take a look at fubure developments of our main indicators, under a ‘business-as-
usual’ scenario, according the model projection. Allowing the model to run up to 2030, we will
see that household air pollution continues to fall slowly, before reaching a plateau before 2020
(Figure 9). As mentioned earlier, we have but one data point relating to past values of average
household air pollution and therefore we cannot be positive about the actual historical trend in
this key indicator. Mean indoor pollution levels were at 108.9 yg/nt in homes in 2014 (Muindi
et al., 2016). We do, however, have quite good data series of the prevalence of clean appliances
and its growth path since 2003, which enables us to postulate the implied behaviour of house-
hold air pollution. Pollution levels have been falling in the past due to the take-up of clean
appliances, particularly the relative precipitous take-up of electric lighting (Figure 10). How-
ever, we will soon reach a point where almost all households in the informal settlements under
study have access to electricity and electric clean lighting. Therefore, from that point onwards
the only way to achieve further improvements in air quality (as captured in our model) is
through more extensive take-up of clean stoves. However, as depicted in Figure 10, growth in
the prevalence of clean stoves is completely dwarfed by that of clean lighting. In other words,
the fonrer is so sluggish that the resulting improvements from clean stoves are almost imper-
ceptible once the prevalence of clean lighting reaches saturation around 2018.
‘Household Air Pollution
425 oe
m5
o
Dore a mat coy 2
WHO standard Bese Run
Figure 9: Household air pollution: Base Run (thick blue) vs. WHO standard (thin black)
Insights fi li
Under business-as-usual, we will soon reach a point where improvements in indoor air quality
will cone to a halt.
19
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
Coverage of Clean Appliances
2008 2008 202 2015, 2018 2a cory 207
number of
number of Base Run
Figure 10: Coverage of clean lighting (Black[1]: Data, Blue [2]: Base Rum) and clean stoves (Dark green [3]: Data, Light
green [4]: Base Run)
G2 Scenario analyses
6.2.1 Description of scenarios
Inthis section, we will examine potential outcomes of four sets of policies. These four scenarios
are summarised in Table 9. A detailed characterisation of these scenarios with regards to pa-
rameter values in the model can be found in the appendix.
Inshort, Scenario I involves a redirection of subsidies from kerosene to LPG and to supporting
local manufacturers of clean stoves. In the model, these assumption are proxied by a step-wise
50 percent increase in kerosene prices by 2023 and a step-wise 25 percent decrease in prices of
LPG and clean stoves by the same year. In the more integrated Scenario II, we complement the
above policies by gradually redirecting the funds previously spent for the provision and subsi-
dising of clean lighting to clean stoves, as the prevalence of clean lighting will soon reach full
coverage regardless. In addition, in Scenario II we also aim to double enforcement, political
will and good governance with regards to the regulations and policymaking related to indoor
air pollution by 2030. This can be achieved, for instance, via more actively fighting comuption.
There are no investments in monitoring or health impact assessment in this scenario.
Table 9: Summarised description of scenarios
Summarised description Notes
paee fuel ¢ LowerLPG prices Adjusting prices of fuels can be attained.
and stove « Lower prices of clean stoves by lowering/increasing subsidies. Lower
Prices Higher kerosene prices stove prices could be a result of support-
ing local manufacturers. Funds for in-
creasing LPG subsidies or supporting
stove manufacturers can be sourced from
savings on kerosene subsidies.
Scenario II: ¢ All of the above in addition to: In the Base Rum, atiny share of available
+stove subsi- = «Higher share of appliance subsidies spent for budget (only 2%) goes towards monitor-
dies and en- stoves, rather than lighting ing. In this scenario, it is assumed that
forcement Higher enforcement, political will and good even that level of funding is stopped.
govemance
¢ No investment in monitoring or health impact
assessment
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
Scenario Summarised description Notes
Scenario III: e All of the above in addition to a higher share of
+monitoring available budget spent for monitoring and
and HIA. health impact assessment.
Scenario IV: ¢ All of the above in addition to adrastic fallin This is the most comprehensive scenario.
+outdoor and outdoor air pollution, along with a drastic rise
ventilation in ventilation
In Scenario III, we accompany all the above changes in policy with gradually ratcheting up the
share of the available budget going towards monitoring and health impact assessment, up to
15 percent for each by 2023. This will gradually bring down the share of the available budget
going to the provision and/or subsidising of clean appliances to 70 percent by 2023. It is worth
noting the size of the available ‘pie’ is not fixed and is endogenously determined under the
influence of public concern about indoor air pollution. The more concemed the public is, the
higher the budget available for fighting it. The effective amount of fund is also mediated by
enforcement, political will and good governance.
Finally, in the most comprehensive Scenario IV, we complement the above indoor-air related.
policies witha drastic (50 percent) reduction in outdoor air pollution, and a drastic (50 percent)
increase in ventilation, in order to demonstrate the potential of improving household air via
improvements in outdoor air.
6.2.2 Results of Scenarios
We now compare these scenarios against each other and against the Base Run. In Figure 11, the
project future path of household air pollution under various assumptions is shown. Blue is Base
Run, red is Scenario I, dark green is Scenario II, light green is Scenario III and brown is Sce-
nario IV. The graph shows how each scenario performs slightly better than the previous one,
thanks to a more comprehensive package of policies implemented, but that the largest leverage
comes from a combination of policies with a reduction of outdoor air pollution.
Household Air Pollution: Scenarios
2008 72006 2008 212 2015 2018 a1 204 O27 i
Time (year)
BaseRun
Scenario I: fuel Pe 2 =. =
Scenario I: ext
Soemutio II: 2 2 3 3 3 3 3 3
Scenario IV: + outdoor & enileion
WHO Standard
Figure 11: Comparing household air pollution under different scenarios
21
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
Figure 12 gives a clearer picture of how the four scenarios fare against each other. This bar
graph captures the improvement that each portfolio of policies generates over and above busi-
ness-as-usual (Base Run), by 2030. This improvement in household air pollution results in a
comparable improvement in average life expectancy, as seen in Figure 13.
Results show that manipulating fuel subsidies and appliance prices alone (Scenario I) is hardly
enough to give inspiring improvements by the end of our simulation period. If, however, we
complement this by redirecting more funds towards clean stoves and by better enforcement
(Scenario IT), by investing in monitoring and health impact assessment (Scenario III), we can.
hope for a more significant betterment of indoor air quality, that is likely to extend the average
life expectancy (at birth) of the residents by one third of year (Figure 13). We need to treat these
mumbers with caution because of the uncertainty of some inputs. Y et it is clear, however, that
the best results by far are only made possible via combining the above policies with a drastic
reduction in outdoor air pollution (Scenario IV).
Household Air Pollution
Improvement over BAU @ 2030
pi @ S iol fuel and stove
38% paces
. +stove subsidies
Scenario II ad aif
Scenario +monitoring and
a” smo
(4% Scenario + outdoor and
IV ventilation.
Scenario | Scenario Il Scenario Ill Scenario IV
Figure 12 - Household air pollution: A comparison of the outcomes of scenarios by 2030
Life Expectancy
Improvement over BAU @ 2030
S iol fuel and stove
3.61 prices
. +stove subsidies
sommnio l and enforcement
Scenario +monitoring and
Il HIA
0.34
0.10 26, Td Scenario | +outdoor and
. . . . IV ventilation
Scenario | Scenario Il Scenario Ill Scenario IV
Figure 13 - Life expectancy: A comparison of the outcomes of scenarios by 2030
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
Insights frommodelling
Merely redirecting fuel subsidies to cleaner fuels does not go far in the way of better indoor
air. However, by implementing a concerted set of policies on various fronts, an improvement
in the order of several months in average life expectancy is made possible by 2030.
So far, these results were probably to be expected and unsurprising. What is perhaps more strik-
ing is that even the best results attained through solely indoor air related policies (Scenario III),
are still far above the WHO guidelines for acceptable levels of exposure to air pollution (the
thin purple line at 10 pg/m? in Figure 11).
Insights frommodelling
Results from combining merely indoor air related policies are underwhelming. This points to
the fact that without tackling the sources of outdoor air pollution, it would not be possible to
get indoor air pollution closer to acceptable levels.
6.2.2.1 Short-termvs. long-term policies
Figure 14. zooms in on the projected behaviour of the number of households owning clean stoves
between now and 2025 for Scenarios II & III, still using the earlier colour coding (dark green
for Scenario II and light green for Scenario III). Note that the only difference between these
two scenarios was that the share of the budget allocated to monitoring and health impact as-
sessment is zero in the former and as high as 15 percent (by 2023) in the latter.
Under close inspection, one can see that the curve for the more comprehensive scenario starts
slightly lower, but then crosses over and above Scenario IT around year 2022, increasingly out-
performing it as we move forward in time. This somewhat counterintuitive behaviour has to do
with the inherently opposite nature of the feedback loops (section 5.1.2) involving the two types
of policies: those targeted at appliance subsidies and those targeted at data collection, research.
and awareness creation.
‘We can explain this particular behaviour with reference to the structure of the model: When we
start to allocate a fraction of available funds to monitoring and health impact assessment (start
ing at 5 percent and going up to 15 percent for each), this inevitably comes at the expense of
being slightly smaller subsidies for clean stoves to the population; hence Scenario III curve
lying a touch below Scenario IT. Overtime, however, the results from the monitoring and health
impact assessment studies start to generate concem over the quality of the air slum residents
breathe, its consequences on their health and the subsequent burden of healthcare costs on the
public purse. Local politicians and activists would then be in a position to leverage this in-
creased awareness to demand and access more funds in order to combat indoor air pollution.
This would enlarge the size of the ‘pie’, which means that even though an increasing share of
these funds is redirected from providing subsidised appliances to funding monitoring and health
impact assessment, in due course there will be more and more money available for providing
subsidised appliances. Due to the reinforcing nature of the dynamics involving monitoring and
health impact assessments, the gap between the two scenarios grows increasingly wider. How
wide it will grow and how much the pie grows depends on how intensely public awareness will
translate into concem and expenditures. We deliberately kept our estimates conservative, thus
there might be higher leverage.
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
Such worse-before-better situation is common in policymaking, where oftentimes short and
long term goals do not necessarily align. In such circumstances, it is not unusual for policy-
maker to aim for the best-possible short term results, inevitably sacrificing the system's long-
term capacity to improve itself.
Households with clean stoves: Worse before better
ai
Time (year)
Figure 14: Worse-before-better behaviour in the prevalence of clean stoves
Insights frommodelling
Redirecting a share of available funds from provision of subsidised appliances to monitoring
and health impact assessment generates a ‘worse-before-better’ behaviour, with increasingly
better long-term results in terms of indoor air quality.
6.2.2.2. Synergies among policies
We saw earlier that our most comprehensive portfolio of exclusively indoor related policies,
Scenario III, yields a 15 percent improvement in household air pollution and an additional
0.34 year of average life expectancy by 2030. But what is the contribution of each single policy
in this total progress? Figure 15 outlines these contributions for our two indicators of interest.
From this graph, it becomes clear that the two policies with the highest contributions in this
scenario are (i) redirecting funds from clean lighting to clean stoves (blue bars at the bottom)
and (ii) redistributing subsidies from kerosene to LPG while lowering the price of clean stoves
(yellow bars in the middle). More interestingly, there is no direct contribution from investments
in monitoring or health impact assessment (HIA): There are no orange or grey bars to be seen.
This is probably the reason why the hypothetical policymaker sees no clear reason to invest in
these. What the policymaker is likely to overlook, however, is the existence of the green bars
of synergy; the ‘invisible’ but substantial contributor to improvement and the fact that synergies
are activated and invigorated as a result of monitoring and health impact assessment. In our
scenario with no funds allocated to these, the contributions from synergies would go down from
5.3 percent to 1.9 percent for household air pollution and from 0.12 to 0.05 years for life ex-
pectancy.
etal. (2017) Indoor air it issue of in Nairobi’ s informal
Policy Contribution Graph
Improvements fromsingle policies by 2030
20% 0.4
m= Synergy
15% 0.3
™ Enforcement
© Fuel and Stove
10% Prices 0.2
= HIA
s
5% Monitoring 0.1
@ Stove Subsidies
0% 0.0
Household Air Pollution Life Expectancy
Figure 15: Improvement over Base Run by the end of the simulation; Household Air Pollution and Life Expectancy
Once again, this has to do with the reinforcing nature of the R1 and R2 feedback loops described
in Section 5.1.2. Generally, such loops have the potential to create synergies in systems; and
were it not for the existence of other stabilising forces and limits to growth within the system,
these synergies could generate exponential growth. Reinforcing cycles do not always tum in
our favour, but when they do, as in this case, it is wise to invest in setting them in motion.
Insights from modelling
Investments in monitoring and health impact assessment, although not directly contributing to
improvements in indoor air quality, have the potential to trigger reinforcing mechanisms that
create synergies among existing policies and elevate our retum on investment.
7 Discussion and condusions
7.1 Recommendations
While Housing in Nairohi’s Informal Settlements (HINIS) was a relatively short pilot, and alt-
hough there was not an abundance of numerical data for parameterisation, calibration and val-
idation of the quantitative System Dynamics model, which makes the model lean in an explor-
atory direction, still a number of well-founded, non-trivial and useful insights emerged as a
results of this practice. We saw, for instance, that under business-as-usual, the current trend of
slowly improving indoor air quality would soon come to a halt due to the saturation of the take-
up of electric lighting and the extremely sluggish rate of take-up of clean stoves. This should
be taken as a waming sign that if we are to aim for reaching WHO standards in terms of ac-
ceptable PM2.5 levels, a drastic acceleration in the take-up of clean stoves will be needed.
25
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
We saw how single policies hardly make any difference and we observed how a concerted
portfolio of programmes promises considerably more visible improvements. However, our pro-
jections - without investing unfounded faith in their point-accuracy - showed that even with a
comprehensive package of policies, there is little hope to close the gap between status quo and.
WHO guidelines for indoor air pollution by 2030. This is because even with zero indoor sources
of pollution, high outdoor pollution would still degrade the indoor environment. Our engage-
ment with the community led us to believe that arriving anywhere near WHO guidelines re-
quires addressing sources of outdoor air pollution, such as neighbouring dumpsites, but also
industrial and road pollution in parallel to indoor air pollution.
Our conceptualisation of the problem, which was a result of a participatory multi-stakeholder
approach, points to the potentially high impact of working towards raising the public’s and the
govemment’s awareness and concem about indoor air pollution and its consequences for resi-
dents’ health and in particular its role in causing serious health issues in children. In order to
achieve this, our study suggests diverting some of the available budget (however big or small
it is) to indoor air quality monitoring and health impact assessment studies, in order to ‘close
the loop’ and bring the issue of indoor air quality higher up on the list of public/govemment
Priorities. Such investments, due to the self-reinforcing nature of the dynamics involved, can
entail high retum on investment, as the policymaker would be able to leverage the results of
such studies in order to enlarge ‘the size of the pie’ of available money and resources.
However, one must recognise that redirecting investments towards monitoring and health im-
pact assessments is likely to give slightly-worse-before-better results due to the time it takes
before these policies pay off, as we observed in the Results section (see Figure 14). This delay
may persist fora politician’ s term. It is important to recognise this in ordernot to be discouraged
and not to prensturely rule out such longer-term policies, but also to be realistic about the
difficulties of their implementation. After a while, once the mechanisms triggered by monitor-
ing and health impact assessment studies take off, they have the potential to create substantial
synergies among other policies, multiplying their effectiveness in bringing down indoor air
pollution.
It is also important to note that the large majority of slum residents are not covered by the
national health insurance system (NHIF) and primarily pay for their health expenses out-of:
pocket. Therefore, preventive healthcare, subsidies for cookstoves and health impact assess-
ments paid out of the healthcare budget might not actually reduce long-term healthcare costs
for the same department. Economic benefits of improved health would need to be linked with
sectors such as economic planning in order to identify synergies and better distribute preventive
healthcare costs.
7.2. Indoor air pollution as a topic of non-attention.
In order for indoor air pollution to be addressed, it has to move into the centre of the public’s
and decision-makers’ attention. But when we asked workshop participants to articulate their
hopes for Nairobi slums, only one participant mentioned the reduction of indoor air pollution:
‘To provide every single household with free cooking stoves and the removal of indoor air
pollution.’ The informal nature of slums, land ownership, services and waste management were
higher on people's agenda. Nevertheless, for a sub-group of stakeholders indoor air pollution
was sufficiently important to attend two full workshop days. In addition, the residents who we
met in focus groups recognised it as an important topic and stated that more education on the
drastic consequences of indoor air pollution is required. Thus, there exists a puzzle between the
importance that some attribute to the issue of indoor air pollution and the little attention it gen-
erally receives.
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
This bears similarity with how the issue of air pollution was dealt with in highly industrial
countries several decades ago. For example, US city and county administrators were perceived.
to support air pollution control, but ethnic and civil groups as well as organisations were per-
ceived to not pay any attention (Crenson, 1971). Today’s situation in Kenya and Nairobi is
similar because no NGO directly addresses air pollution. Y et, some academics and goverment
representatives are aware of the importance of the topic, but find it difficult to put the topic on
the agenda because of lacking public support.
This paper has extensively discussed the underlying causal mechanism (Figure 6 on page 16).
The reinforcing nature of loops R1: Monitoring and R2: HIA locks the system ina vicious spiral
of a lack of knowledge about pollution and its health consequences, non-attention, lack of re-
sources for the issue and for further investigation. The modelling of the scenarios showed that
triggering these reinforcing mechanisms can create leverage in the system. They re-allocate
resources to indoor air pollution and additionally enlarge the size of the pie available. Y et, this
means that indoor air pollution will then also fight with all the other issues for resources, acting
as a serious constraint. In addition, the success depends on the strength of the relationship be-
tween attention, concem and expenditure. Work on reducing the limiting factors is needed so
that these reinforcing feedback loops can fully work and trigger repetitive momentum towards
a higher focus on the issue.
The residents reported the difficulties of getting heard on this and other issues. This creates a
further vicious cycle of public attention and concem. Stakeholders raised socio-economic en
powenment with regards to raising the eaming power of slum residents as a means to break this
cycle. It would improve their incomes and subsequently their ability to live in better environ-
ments and probably spur them to invest in cleaner cooking and lighting devices. The hopes
expressed by stakeholders challenge government agencies and other stakeholders to have equity
as a central theme in their policies and programmes. For example, govemance issues would.
require an equity lens to ensure no section of the city is left behind in programmes and strate-
gies.
The limited attention the indoor air pollution issue has received so far directly influenced this
project and modelling work. As data collection had not been considered important, we lack
reliable data in many areas that would be required for more detailed recommendations to
emerge fromthe modelling. Thus increased investments in (indoor) air pollution would increase
the precision of scientific evidence, and this evidence would hopefully make it easier to agree
on needed policies.
7.3 Impadt, limitations and fubure work
The participatory process used in the HINIS project brought together stakeholders from differ-
ent sectors such as cook stove manufacturers and the policy sector that all have direct impact
on households. For instance while residents spoke of prohibitive cost of cleaner fuels and stoves
as being behind their persistent use of dirty fuels, it was clear that there were other relevant
players suchas the current energy regulation policy that has introduced pricing controls of fuels
such as kerosene but not that of liquefied petroleum gas (LPG). It thus helped draw a multi-
faceted picture of indoor air pollution and brought home to stakeholders the rigorous process
of developing a dynamic model. Its gradual development also demystified the complex inter
linkages of sectors. One of the impacts of these workshops is the raising of participants’ aware-
ness on indoor air pollution and the inter-related factors. This may help awakening and mobi-
lising the public’s attention attention.
27
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
Yet, itis important to recognise the limitations imposed on this study due to a shortage of avail-
able time-series data on such key variables such as indoor and outdoor air pollution and past
expenditures on related policies, among other variables. This points to the crucial importance
of investing in the collection of data over time, as it is hard to manage what you have not meas-
ured. Such data would then allow more rigorous testing of uncertainties, e.g. through sensitivity
simulations, so that more operational recommendations can be given beyond the creation of
behaviour modes.
Another key limitation of this study is the completely exogenous treatment of the level of out-
door air pollution, which we included as a scenario variable whose future value was to be set
by the user. Y et, the effects of policies such as the fuel taxes and subsidies also affect outdoor
air pollution through transport, etc. For a more realistic treatment of the problem of indoor air
pollution, it would thus be desired to endogenously include how indoor pollutants and our sin+
ulated policies interact with both indoor and outdoor air pollution.
Finally, we suggest to more closely focus on implementation. This includes examining the in-
terrelationships between commumities, the public, organisations and goverment on the issue
of indoor air pollution. Our workshop discussions hint to their quite different goals. While we
spent an aftemoon during the final workshop to discuss implementation issues and potential
unintended consequences of policies, such practical aspects warrant a more thorough examina-
tion and inclusion in the model.
Acknowledgements
The authors gratefully acknowledge the financial support from the ‘Housing in Nairobi’s In-
formal Settlements - A Complex Urban System (HINIS)’ project, funded under the EPSRC
Global Challenges Award (project code: 535444, award code: 172313), as well as bridge fund-
ing from the Wellcome Tnust for the ‘Complex Urban Systems for Sustainability and Health
(CUSSH)’ project (award code: 205207/Z/16/Z).
References
African Population Health Research Center. 2002. Population and health dynamics in Nairohi's
informal settlements: report of the Nairobi cross-sectional slums survey (NCSS) 2000.
Nairobi: African Population and Health Research Center.
African Population Health Research Center. 2014. Population and Health Dynamics in
Nairobi’s Informal Settlements: Report of the Nairobi Cross-sectional Slums Survey
(NCSS) 2012. Nairobi: African Population and Health Research Center.
Beguy, D., Elung’ata, P., Mberu, B., Oduor, C., Wamukoya, M., Nganyi, B., & Ezeh, A. 2015.
HDSS profile: the Nairobi urban health and demographic surveillance system
(NUHDSS). International journal of epidemiology: dyu251.
Crenson, M. A. 1971. The Un-Politics of Air Pollution: A Study of Non-Decisionmaking in the
Cities. Baltimore, ML: The John Hopkins University Press.
Desgroppes, A., & Taupin, S. 2011. Kibera: The Biggest Slum in Africa? Les Cahiers de
l'Afrique de!’ Est 44(halshs-00751833): 23-34.
Egondi, T., Kyobutungi, C., Kovats, S., Muindi, K., & Ettarh, R. 2012. Time-series analysis of
weather and mortality pattems in Nairobi’ s informal settlements. Global Health Action
5(19065): 23-32.
‘Zimmenmamn et al. (2017) Indoor air pollution as an issue of nonattention in Nairohi' s informal settlements
Egondi, T., Muindi, K., Kyobutungi, C., Gatari, M., & Rocklov, J. 2016. Measuring exposure
levels of inhalable airbome particles (PM2.5) in two socially deprived areas of Nairobi,
Kenya. Environmental Research 148: 500-506.
FSD Kenya. 2016. 2016 FinAccess Household Survey. Nairobi: FSD Kenya.
Gubhbins, P., & Ravishankar, N. 2016. A look at Kenya’s struggle to purchase healthcare: FSD
Kenya.
Kanya National Bureau of Statistics. 2010. Population Distribution by Age, Sex and
Administrative Units. In KNBS (Ed.), The 2009 Kenya Population and Housing Census,
Vol. 1C. Nairobi.
Kyobutungi, C., Ziraba, A. K., Ezeh, A., & Y6, Y. 2008. The burden of disease profile of
residents of Nairobi's slums: Results from a Demographic Surveillance System
Population Health Metrics 6(1): 8.
Morgan, T. 2015. The socioeconomic impact of switching to LP Gas for cooking: A report of
the World LP Gas Association. In W. L. G. Association (Ed.). Neuilly-sur-Seine,
France.
Mugisha, F. 2006. School enrollment among urban non-slum, slumand rural children in Kenya:
Is the urban advantage eroding? International Journal of Educational Development
26(5): 471-482.
Muindi, K., Egondi, T., Kimani-Murage, E., Rocklov, J., & Ng, N. 2014. “We are used to this”:
a qualitative assessment of the perceptions of and attitudes towards air pollution
amongst slum residents in Nairobi. BMC Public Health 14(1): 226.
Muindi, K., Kimani-Murage, E., Egondi, T., Rocklov, J., & Ng, N. 2016. Household Air
Pollution: Sources and Exposure Levels to Fine Particulate Matter in Nairobi Slums.
Toxics 4(3): 12.
Vennix, J. A. M. 1996. Group Model Building: Facilitating Team Learning Using System
Dynamics. Chichester, NY: Wiley.
Ye Y., Zulu, E., Mutisya, M., Orindi, B., Emina, J., & Kyobutungi, C. 2009. Seasonal Pattem.
of Pneumonia Mortality among Under Five Children in Nairobi’ s Informal Settlements.
The Arrerican J ournal of Tropical Medicine and Hygiene 81(5): 770-775.
Zollmann, J., & Ravishankar, N. 2016. Struggling to thrive: How Kenya’ s low-income families
(try to) pay for healthcare. In FSD Kenya (Ed.). Nairobi, Kenya.
‘Zimmenmamn et al. (2017) Indoor air issue of in Nairohi’s informal
Unit Baserun Scenario Scenario II
future shareaf dimensionless 98% 98% 100%
air quality ex-
penditure for
appliance subsi-
dies
fubureshareof dimensionless 2% 2% 0
indoor air qual-
ity
tenes DEE
futureshareof dimensionless 0 0 0
air quality ex-
penditure for
health impact
futureshareof dimensionless 5% 5% @2017: 50%
i subsi- @2020: 75%
dies for dean. @2023: 100%
stoves
futurepriceof KSHperlite 634 @2017: 70 @2017: 70
@2020: 80 @2020: 80
@2023: 90 @2023: 90
future priceof = §KSHper6kg 1246 @2017: 1160 @2017: 1160
cylinder @2020: 1080 @2020: 1080
@2023: 1000 @2023: 1000
futurepriceof KSHperunit 5000 by 2030 linearly by 2030 linearly
ean stoves down to 4000 down to 4000
futureenforce- dimensionless 0.25 0.25 by 2030 linearly
ment, political = (conceptual- up to0.5
will, and good. ised as a0 to
governanoe 1 index)
Scenario II
(@2017: 90%
@2017: 5%
@2020: 10%
@2023: 15%
- @2017: 5%
@2020: 10%
@2023: 15%
@2017: 50%
@2020: 75%
@2023: 100%
@2017: 70
@2020: 80
@2023: 90
@2017: 1160
@2020: 1080
@2023: 1000
by 2030 linearly
down to 4000
by 2030 linearly
upto0.5
