Think and act for entrepreneurship in Africa

In 4 articles for Entreprenante Afrique, the Café Lumière project in Madagascar is presented, and the results of different alternative evaluation methods carried out at an affordable cost are compared to assess the existence of positive impacts of the project on development objectives. This fourth and final article explores the impacts on households by comparing changes in the variables of interest in treated and untreated localities, similar to the ‘localities’ part of our survey. Thanks to the sample size of households surveyed in each locality[1], we can assess the statistical significance of our conclusions and evaluate the risk of errors in our findings regarding the existence of an impact. We studied variables of interest that can logically be considered as measuring potential impacts of the project on households.

Access to electricity

On the sample of surveyed households, 18% are connected  in May 2023to the mini-grid in localities equipped with a Café Lumière. The other forms of access to electricity were mainly individual fixed or mobile solar panels. The rate of access to electricity, defined as including all possible sources of electricity, was relatively similar between the two groups of localities in 2023, with 46% of households having access in localities with a Café Lumière compared to 43% in localities without one.

It is important to note that the rate of access to electricity was higher in localities that were subsequently equipped with Café Lumière (37%) compared to those without (27%), which is a significant difference. While there is no significant improvement in access to electricity from a quantitative point of view, there is a replacement of individual solar panels by connections to the mini-grid, which is considered a qualitative improvement.  According to the Multi Level Framework (MTF) defined by the World Bank, the first type of access is considered level 1, while the connection to the mini-grid is level 2. This is due to two differences between the two types of access: higher available power and lower intermittency. Data on the use of electrical appliances corroborates this finding.

Use of electrical appliances

In order to analyse the use of electrical appliances, we calculate an index, which is the sum of the electrical appliances used by a household. The appliances taken into account are: cooker, refrigerator, fan, radio, television, video recorder, DVD player, computer, tablet, fixed telephone, and mobile phone. Thus, a household owing a fan and a radio would then have a score of 2, and a household possessing all of the mentioned items would have a score of 11.

During the second wave of surveys, we observe that households living in localities with Café Lumière use more electrical appliances than those living in localities without Café Lumière. Households living in localities with Café Lumière have an average of 1.7 electrical appliances compared to only 1.5 in other localities. Although this difference may appear small, it is statistically significant at the 1% threshold.

The increase usage of electrical appliances in the treated localities is clearly due to the improved quality of access to electricity for the 18% of households now connected to the mini-grid.

To corroborate this conclusion, we construct a second index of electrical appliance usage, considering only appliances that require relatively high power to operate, typically unavailable when electricity access is at level 1 of the MTF. This index includes the following appliances: cooker, refrigerator, fan, television, video recorder, DVD player, and computer.

On this new index, we find a very significant positive impact of the Cafés Lumière. On average, households used only 0.20 high-power electrical appliances in 2023 in localities without a Café Lumière, whereas they used 0.37 in localities with Café Lumière (this figure even rises to 1.20 for households using the mini-grid, whereas there was no significant difference in 2017/2018).

Household wealth

To assess any potential impact of Café Lumière on household wealth levels, we construct a series of composite indices based on variables reflecting households’ material possessions. It is important to clarify that this index represents a stock of assets, rather than a flow such as the household’s monthly or annual income. We proceed with Multiple Correspondence Analysis (MCA), which is a data analysis method that synthesises information from multiple qualitative variables into a few complementary axes. The axis containing the most information is generally considered representative of household wealth. The possessions of households used to calculate this composite wealth index take into account elements related to housing quality, such as materials used for the floor, walls, or roof, the number of rooms and floors, etc., as well as electrical appliances (see list above) and all belongings owned by the household, such as a bicycle, a moped, a car, a watch, etc. Each household is then assigned a score representing its level of wealth. This index has a mean of 0 and ranges from a minimum of -2.17 for the poorest household to a maximum of 5.22 for the wealthiest household.

When we analyse the difference in scores in 2023 between localities equipped with Café Lumière and those without, we observe that the average wealth index of households living in localities with Café Lumière is higher, albeit not quite significantly. As illustrated in Figure 1, the average wealth index for localities with Café Lumière is 0.19 compared to only 0.03 for localities without Café Lumière. The absence of disparity in 2017/2018 strengthens the robustness of this conclusion. It is noteworthy that when we calculate the wealth index without taking into account ownership of electrical appliances, what we could call a “non-electricity wealth index”, we find no impact of Café Lumière. This leads us to conclude that while Café Lumière may have had a positive impact on electricity access and its usage by households, this improvement in their material living conditions has not yet spread more widely.

Socio-economic data

Beyond the objectives of improving the overall energy and economic situation in the treated localities, the main objective of the Café Lumière project is to contribute to socio-economic progress. Regarding economic poverty, we observe a slight improvement in the distribution of the wealth index benefiting the poorest 20%, but this cannot be directly attributed to the introduction of Café Lumière.

Looking at access to health and education, which are the areas most often studied in similar evaluations, we find some evidence of impact in the first area, but not in the second. This is consistent with the localities section of our survey, which showed many health centres connected to the mini-grids, but proportionately fewer schools.

Finally, we examined the project’s effects on the insecurity associated with theft of livestock, crops, or other goods experienced by households in rural regions of Madagascar. It is often argued that the introduction of public lighting at night may help improve this situation, but this is not clearly proven at this stage.

In terms of health issues, if we compare reports of symptoms such as respiratory problems (cough/cold, etc.), diarrhoea, fever, headache and eye problems or burns, we find no significant difference between the two groups of localities in 2023. However, it is interesting to note that during the first wave of the survey, i.e., before the installation of the Cafés Lumière, symptoms were much more prevalent in the localities that received the Cafés Lumière, and these localities experienced a significant decrease in symptoms prevalence between the two waves, which could be interpreted as a consequence of the Cafés Lumière project.

Focus on childbirth: we observe that childbirths occurring in a well-lit environment due to electricity have become almost routine in localities with a Café Lumière. Indeed, for women living in a locality with Café Lumière, 78% of them give birth in facilities with electric lighting compared to only 40% in other localities. This difference in average is statistically highly significant.

Conclusions

This fourth and final stage of our series of articles on the evaluation to date of the impacts of the Café Lumière project in Madagascar has allowed us to study the benefits received by the population of the concerned localities in terms of access to electricity and livings standard, as well as its impacts on socio-economic development. Before the implementation of the project, the studied localities were not characterised by absolute energy poverty, which can be defined by the absence of access to any source of electricity. On average, in 2017-2018, one-third of households already had access to electricity through standalone solar panels.

The most visible energy impact of the project on households is that 18% of households in equipped localities now benefit from a connection to the mini-grid. For these households, this energy impact is significant as it results in the use of more numerous and higher-powered electrical appliances. In our previous blog article, we had also shown that some subscribed households used their subscription for both their own consumer needs and for income-generating activities.

We have observed that in the equipped localities, the wealth index of households has increased, but only when taking into account the ownership of electrical appliances. Therefore, the enrichment of households through the arrival of the mini-grid does not seem to have created economic transformation beyond the acquisition by connected households of some electrical appliances requiring the transition to higher power and lower intermittency provided by the mini-grid. Efforts still need to be made for the poorest households, as even though they benefit from improved public services and shop services, they cannot yet trigger genuine economic development. This observation echoes the fact that we had shown in our previous article that, so far, the impact on income-generating activities had been limited to the electrification of their production tools by the mini-grid.

In the social spheres, the arrival of mini-grids has had positive impacts on health. The situation has particularly improved due to the connection of primary health centres (csb2) to mini-grids. The improvement in public health has also benefited from synergy with other national and international programmes in collaboration with the World Bank and UNICEF, notably for vaccination campaigns.

The impacts observed so far appear to be limited, but this observation needs to be put into perspective by the fact that we are currently only seeing short-term impacts. Thus, if all equipped localities were to achieve the same results as the most successful locality in terms of household connections, Talata Dondona (36%), we would double the number of household connections. Furthermore, economic transformations that can be stimulated by purely energy-related impacts necessarily take a long time to materialise. The same applies to socio-economic progress, which may require complementary public policies, as in the case of healthcare.

[1] 599 households were interviewed during the first wave (2017/2018), and 595 during the second wave (2023). 133 households from the first wave were not found during the second. They were replaced by 129 new households. In total, 466 households were surveyed during both waves.

Further reading in the same serie of articles on “Measuring the Impact of Decentralised Electrification Projects”: Cafés Lumière in Madagascar (1/4), Using Remote Sensing: Initial Results on the Impact of Cafés Lumière (2/4), Characterisation of the Impacts of Decentralised Electrification Projects on Access to Electricity Using Locality Data (3/4).

In 4 articles for Entreprenante Afrique, we are presenting the Cafés Lumière project in Madagascar and comparing the results of different alternative evaluation methods carried out at an affordable cost to test the existence of positive impacts of the project on development objectives. In this third part, we draw on the activity reports of the operator Anka, which provide, for each equipped locality, a monthly series of electricity consumption data by major type of use, and associated data on the number of customers. We supplement this information with the initial results of the socio-economic surveys conducted in 2017/18 and 2023, in its locality survey section. These data, some of which  enable comparisons to be made with control (non-equipped) localities, enable identification of the main transmission channels for access to electricity in relation to sustainable development objectives. This analysis will be supplemented in the fourth part of this blog by the household section of the survey, in which the responses from 50 households per locality will be used to refine the diagnosis by considering the downstream effects of electrification on household well-being.

Overall Performance of Electricity Generation

As demonstrated in the previous article, the Cafés Lumière project has positively influenced access to electricity in the equipped localities. However, it is important to note that this impact may differ from one locality to another. Figure 2 illustrates the effect of Cafés Lumière on electricity provision through the mini-grid by comparing the total annual electricity consumption in 2022 with the production potential of the solar panels (converted into annual MWh).

Figure 1: Potential Electricity Generation and Total Electricity Consumption by Locality in 2022

Figure 1 shows that the higher the wattage of the installation, the greater the amount of electricity consumed, excluding street lighting. However, this correlation does not necessarily imply causation, as the power of the installations was determined based on the population to be electrified. The relatively low ratio between daily consumption and potential generation reflect the non-uniform daily load curve. In the 2023 locality survey, it was frequently noted that the mini-grid often reaches its capacity during peak periods, posing a significant constraint for the few three-phase subscribers.

At Ambatonikolahy and Amparaky, the facilities appear to produce below their potential, possibly due to the limited number of household subscribers, and the largest consumers, as observed elsewhere, are constrained by capacity. In the case of Ambohimalaza, insufficient photovoltaic production during peak times (when the husker is in use) is compensated for by the utilization of the emergency diesel generator. A detailed examination of Anka’s activity reports provides clarification of these observations and also offers an opportunity to compare the progress achieved in the main consumption categories.

Household Consumption

Initially, households benefited from the Cafés Lumière project through the establishment of shops. These shops, functioning as energy kiosks, target individuals who may not have the means to afford a subscription to the mini-network. The income generated from these shops suggests that many households were initially customers in the shop before subsequently subscribing to the mini-grid. While the role of the shops was crucial in the early stages, the main quantitative impact of the project now occurs through the mini-grids.

Figure 2: Growth in the Number of Subscriber Households

Figure 2 illustrates the fluctuations in the number of subscribing households. It is a dynamic scenario with a wide range of situations. With the exception of Cafés Lumière in Ambatonikolahy, Ambohimalaza, and Amparaky, there has been substantial growth in household electricity usage, but this growth is broad rather than intensive (with no significant rise in consumption per subscriber, which is approximately 5 kWh per month). For a tangible economic impact, households will need to augment their electricity usage, expanding beyond domestic lighting consumption to activities that could generate income. Further insights from Anka regarding the structure of electricity consumption in 2023 allows us to observe the start of this dynamic in the next section.

Consumption by Income-Generating Activities

Another significant source of economic impact is the development of income-generating activities (IGAs), the establishment or modernisation of which relies on the use of electricity. From this perspective, the outcomes are varied. Currently, Anka’s activity reports mention very few IGAs, with the highest number, 4, in Antanamalaza. Moreover, most of these activities consume modest amounts of electricity, with the only exceptions being huskers and grinders associated with agricultural activities in Ambatonikolahy, Ambohimalaza, Antsampandrano, and Amparaky. Other diversified activities include bars, carpentry, hairdressing, pastry-making, car painting, fishmonger, and video games.

With a few exceptions, the development of IGAs is slow. Additional data provided by Anka on the current situation at the beginning of 2023 indicates a genuine potential for change. Anka has observed that several households are already using their subscription to power electrical appliances for productive activities without officially registering as entrepreneurs. By mid-2023, there were 32 users of electricity produced by the mini-grid engaged in providing services, including 12 officially declared as IGAs. Among the 20 producers identified in households (constituting 5% of the total subscriber households), a variety of activities are pursued, such as bars, grocers, small restaurants, fruit juice sales, telephone recharging, multiservices, and welding. These activities are present in all localities, and their diversity reflects a certain level of dynamism. It is worth noting that the small restaurants and shops identified in the locality section of our socio-economic survey, on average 5 per locality, are only partially listed as service producers by Anka.

It is not possible to separate the consumption of households engaged in service-producing activities from the overall household consumption, but an approximation can be derived using consumption data for officially declared income-generating activities (IGAs). The consumption of IGA households is markedly higher, approximately 40% more than that of other subscribing households. The expansion of these activities could subsequently drive an increased demand for electricity for productive purposes, thereby enhancing the economic impact of the Cafés Lumière.

Data from the locality section of the socio-economic survey enable a more precise interpretation of these consumption figures. There was no noteworthy increase in the number of service-producing activities, both in the localities electrified by the mini-grid and in the control localities. However, where these activities previously operated, and still do in the control localities, using diesel generators and individual solar panels, almost all of them in the equipped localities have transitioned to the electricity provided by the mini-grid. This marks progress for the equipped localities, and for all the activities that have switched from diesel to the solar mini-grid (on average more than 6 per locality), and it contributes to environmental conservation.

Public services

The last category of electricity consumption identified by Anka is public services, it represents a minority share of the electricity consumed on the mini-grids in 2022. Nevertheless, it can have a significant impact on the achievement of sustainable development goals: health (SDG3), education (SDG4), and peace, justice, and effective institutions (SDG16).

The health objective has garnered considerable attention, with all the basic health centres (CSB2) in the electrified localities (5 CSB2s in 6 localities supplied by the mini-grid) having an average monthly consumption of 23 kWh per CSB2. For comparison, only half of these centres were electrified (via solar panels) in 2017. Currently, all these health centres are equipped with lighting and refrigeration, but the use of electricity to operate medical equipment remains infrequent. In contrast, the situation is markedly different in the control localities, where only half of the health centres have electricity access in 2023, generated by solar panels.

Education is also a focal issue but with more modest aspirations. From a scenario where no school was electrified in 2017, half of the schools in the equipped areas are now electrified, albeit with low consumption (1.8 kWh per month per electrified school). In comparison, none of the schools in the other localities have access to electricity.

The collective activities associated with SDG16, including 8 administrative structures and 10 churches, consumed an average of 8 kWh per month per structure in electrified localities in 2022. In these areas, approximately half of the town halls and police stations are electrified in localities equipped with a mini-grid, while all churches have access to electricity.

Public lighting can also impact MDG16 for security. All localities electrified by Cafés Lumière have invested in public lighting, although not always as a priority, with its connection to the mini-grid sometimes being delayed. Outside Ambatonikolahy and Talata Dondona, the nighttime coverage is limited. Moreover, the Amparaky town council decided against such an investment in 2022 due to perceived high costs (this situation will change in 2023). Overall, in 2022, electrified localities consumed only 12 kWh per month for public lighting, which is less than the consumption for administrative structures or churches.

The availability of public lighting does not significantly differ in localities with a mini-grid compared to control localities, with an average of around 5 streetlights per locality. The mini-grid supplies two-thirds of the streetlights in the concerned localities, but this does not seem to enhance the security situation. The 2 localities with a mini-grid providing extended night lighting do not report any reduction in security problems. The main security issues are linked to crop theft, and it appears that the police activities play a more significant role in improving security.

Further reading in the same serie of articles on “Measuring the Impact of Decentralised Electrification Projects”: Cafés Lumière in Madagascar (1/4),  Using Remote Sensing: Initial Results on the Impact of Cafés Lumière (2/4), Characterisation of the Impacts of Decentralised Electrification Projects on Access to Electricity Using Household Data (4/4).

Typically, current assessments of decentralised electrification projects struggle to establish their tangible impact. Achieving this would necessitate expensive and time-intensive efforts, because it would demand extensive data not only from the equipped localities but also from control localities. Nonetheless, substantiating the presence of an impact is crucial to assist public decision-makers, both at the national and international levels, in expanding the adoption of these solutions.

In a series of four articles for “Entreprenante Afrique,” we introduce the Cafés Lumière mini-grids project in Madagascar and compare the outcomes of various affordable evaluation methods applied to these projects. These methods aim to assess the project’s positive effects on development objectives. In this second of four articles, we explore the feasibility of evaluating impacts through the use of remote sensing data, which is generally a low cost approach.

The contribution of remote sensing to impact analysis

Remote sensing uses high-resolution satellite imagery that covers nearly the entire globe, offering near real-time accessibility at a low cost. This technology can effectively capture various measurable terrestrial phenomena relevant for the study of human activities. For the assessment of electrification projects’ impacts, satellite imagery measurements of night-time luminosity can be employed, once sufficient experience has been gained in interpreting this data. It has been consistently demonstrated that an increase in night-time luminosity correlates well with the increase in electrification over time, even at fine levels of granularity (see Berthélemy, 2022, in The Conversation). It has even been argued that rise in night-time luminosity reflects growth in economic activity (Hu and Yao, 2022).

Nonetheless, critics of this new approach, which relies on observations of natural phenomena partially linked to the outcomes of human activities, raise concerns about potential assessment bias. In our context, night-time luminosity specifically measures the light generated at night, primarily from lighting sources, especially public lighting. It does not have a direct connection with total electricity consumption, of which it typically represents a minor portion, and even less so with its impacts on socio-economic development. While public lighting can influence safety, and consequently economic activity, it can only capture a fraction of the consequences of electrification on human activities.

Application: the Cafés Lumière Mini-grids have a Significant Impact from 2021 onwards

Night-time luminosity can serve as a real-time indicator for detecting the effects of electrification. To illustrate this, we have calculated the average annual night-time luminosity for the 6 equipped localities and compared it to that of 6 similar non-equipped localities, which were randomly selected by Electriciens sans frontières to create a control group. The data was analysed taking into consideration the following constraints :

• The different localities were not equipped at the same time ;
• The Cafés Lumière were progressively set up, with priority given to installing the shop before commissioning the mini-grid. Since the shop has little effect on light emission, the effects of Café Lumière can be initially evaluated by examining the influence of the mini-grid ;
• In 2 localities, Ambatonikolahy and Talata Dondona. public lighting operates during specific hours when the satellite passes overhead (between 12pm and 2am in our case). Other localities prefer to have public lighting from 9pm to midnight and from 5am to 6am.

Figure 1: Assessment of night-time brightness (annual averages 2013-2022) in both equipped and non-equipped localities (radiance measured in w/cm2_sr)

Figure 1 shows a similar trend in the average night-time brightness for both the 6 equipped villages and the 6 control villages up to 2020. However, the equipped group exhibit more brightness in 2021 and 2022. This difference is approximately 10% compared to previous data, which, though modest given the initially low level, is statistically significant. Our findings from Figure 1 are corroborated by a more formally rigorous statistical test that utilizes monthly data for each locality, while also controlling for the seasonal effects and fixed effects associated with each treated or untreated locality. The deployment of the mini-grid leads to an increase in night-time luminosity comparable to that depicted in Figure 1, and this increase is statistically significant.

Concerns about potential bias arising from the presence of street lighting are only potentially justified in the case of 2 out of the 6 localities (Ambatonikolahy and Talata Dondona). In principle, these concerns can be examined by considering that street lighting is often installed subsequent to the mini-grid becoming operational. These fears are not borne out.  In fact, when we attempt to account for both the presence of the mini-grid and street lighting simultaneously, the latter shows no significant effect.

This does not imply that public lighting has no effect; rather, it suggests that we are unable to provide demonstrable evidence of its effect. This limitation may be attributed to the statistical power of the test, given the small number of observations with the presence of public lighting.

In the third article, we will employ more precise data, notably on monthly electricity consumption categorized by usage, which will enable us to assess the contribution of public lighting to the increase in night-time luminosity.

Further reading in the same serie of articles on “Measuring the Impact of Decentralised Electrification Projects”: Cafés Lumière in Madagascar (1/4), Characterisation of the Impacts of Decentralised Electrification Projects on Access to Electricity Using Locality Data (3/4), Characterisation of the Impacts of Decentralised Electrification Projects on Access to Electricity Using Household Data (4/4).