Resilience 1.0

Low proliferation of climate-resilient housing in slum areas is a problem of massive scale. The United Nations estimates that over 1 billion people live in informal settlements, and within our country of focus, the Philippines, there are about 3.6 million households living in slums (Department of Human Settlements and Urban Development). Our goal is to support informal communities in upgrading to more resilient housing and to help these communities better incorporate climate risks into infrastructure design and planning. We specifically accomplish this by helping families retrofit their homes in a way that boosts a house’s probability of surviving a natural disaster.

The Philippines experiences thousands of earthquakes and an average of 8 to 9 typhoons each year that cause damage to existing homes and property (Government of the Philippines). A 2018 study conducted by the Harvard Humanitarian Initiative finds that only 31% of Filipinos surveyed reported that they felt even slightly prepared for a natural disaster, and many reported that they were financially unable to make disaster-resistant fortifications to their homes.

The lack of widespread climate-resilient housing within informal communities in the Philippines leads to repeated individual and public costs each year. Collapsing homes cause injury and death, and the vulnerability to having a house collapse impacts a family’s ability to make other personal decisions and collect household savings. Collapsing homes also are a large burden to local governments and local economies, as reconstruction bills and temporary rehoming of citizens can cost billions (of dollars) per disaster, and a family’s ability to generate income is temporarily put on hold.

There is no way of disaster-proofing a house. Homes can only have their resilience to natural disasters boosted. This requires two things: (1) the use of more climate-resilient materials and (2) complementing such materials with climate-resilient retrofitting recommendations. Frequent natural disasters are such a problem for low SES residents in the Philippines that many on-the-ground efforts have been successful in producing low-cost, climate-resilient housing materials. Proliferation of climate-resilient housing for the poor has largely been stunted by costs surrounding (2), as retrofitting to bolster climate-resilience of homes is highly personalized. An engineer needs to walk through each housing unit, inspect the existing structure, and then deliver a recommendation on how to retrofit.

Low proliferation of climate-resilient housing within informal communities can be characterized by a classic market failure: prices are high for retrofitting services due to costs, and families are unable to pay market prices due to poverty. 

Our solution, Resilience 1.0, addresses the market failure identified in the problem statement by substantially lowering the per-house cost of generating a personalized climate-resilient retrofitting recommendation. Its ultimate goal is to lower prices for climate-resilient retrofitting services so that households can directly pay for such services to private construction firms.

Resilience 1.0 uses machine learning and unique data collection on existing housing structures to generate personalized, climate-resilient retrofitting recommendations for entire communities at a time. Instead of relying on an individual civil engineer to walk through each housing unit to recommend climate-resilient retrofitting, our solution relies on use of cameras and drone footage to capture the baseline structure of the buildings. It then utilizes machine learning to optimize over the best roof angle (and any other structural component of choice) to maximize the probability of the home surviving a magnitude 6 earthquake or typhoon winds of 110mph. Once the personalized recommendation is generated per house, we can deliver that information directly to the families or to a construction professional who has been contracted to implement disaster-resilient retrofitting. 

Our solution specifically serves the “poor”, which is the lowest SES class within the Philippines. While poor Filipinos can be characterized by the types of communities and homes they live in, our solution explicitly addresses the subsection that live in slums. Our target population mainly live in huts that are hand-built using bamboo and nipa leaves, which are easily salvaged from the environment. These families typically do not have running water or electricity within the home, but they do have smart phones.

Despite being the most vulnerable population to natural disasters, the housing needs of this population are currently underserved by (1) private construction firms, (2) the government, and (3) non-profits. Private construction firms are unwilling to offer services that target poor SES homes because these families are the least likely to seek their services at market prices. As a result, many local construction firms do not have any experience or knowledge on how to work with slum homes. The main reason why governments and non-profits have both not adequately addressed the housing needs of the informal communities within the Philippines is because they are often caught between the desire to help these families improve their homes within the slums and the desire to directly move all families away from slum areas.

Resilience 1.0 will address the housing needs of families living within informal settlements in two ways. First, we will directly implement climate-resilient retrofitting recommendations generated by our solution during the piloting stage for 1,000 to 5,000 homes. This directly addresses the climate-resilient housing needs of those within our pilot. Second, we intend on learning how to take this solution to scale by quantifying the exact willingness to pay from families, the costs of implementation from the firm side, and the savings to local governments post-natural disaster. This will help us determine if our solution can be supported entirely within the free market, or if it will need to be subsidized by the government. We intend to find a long-term, market-based solution to increase climate resilience for this population.

We intend for our solution to substantially boost the financial and physical security of families living in slums. We hope that these families will no longer need to stop working each time there is a natural disaster because they need to spend time rebuilding their home and reinvesting in all of the items that were destroyed.

Our team is well positioned to deliver our solution for two reasons. First, our team itself is comprised of three researchers who have diversity in field, expertise and backgrounds. Dr. Jun Limondo Mata, who pioneered the technology that is being adapted into Resilience 1.0 for informal communities has a PhD in Civil Engineering from Mindanao State University and is a construction professional operating in Davao City. He has firsthand experience working in and growing up within informal communities. Earl Aleluya is a professor of Computer Engineering at Mindanao State University who specializes in applied uses of machine learning for structural design.  I am a doctoral candidate in economics at the University of Chicago specializing in scalable solutions for vulnerable communities and I am dedicated to the communities within and around Davao City through family. Our team’s diverse background and expertise is united by personal connections to the greater Davao region.

Second, our product itself relies heavily on the needs of the informal communities, as Resilience 1.0 is a personalized design recommendation. A house needs to be able to withstand natural disaster, but it also needs to meet the needs of the family that uses it. The latter will vary heavily from family to family, and will vary by religion, location, and source of livelihood. The demands of the community appear can be directly integrated into the design generation process.

During the prototyping process, we gather community needs and family demands qualitatively through interviews and observation of how the family implements a retrofitting recommendation we deliver to them. During the pilot process, which will focus on scalability, we gather this data via large quantitative survey. To collect this data, we partner with Innovations for Poverty Action Philippines, which has over 20 years of data collection experience in informal communities across the Philippines. 

The machine learning optimization method for climate resilient building design that underlies Resilience 1.0 was proposed by a team member, Dr. Jun Limondo Mata as an academic contribution. The technique has been prototyped and implemented for climate-resilient designs on multiple individual buildings that Dr. Mata has been contracted to design in Mindanao.

Our team is in the process of adapting Resilience 1.0 on slum-type homes in Davao City. As of May 2023, we have secured permission from the local government authorities to operate and are in the process of collecting and combining data on the existing building structures in two slum areas. Our timeline of activities is to have our engineering team generate the genetic algorithm design recommendations for slum housing in June. This is to ensure convergence of the algorithm. Then, in July we will conduct random back-checks on 25% of the homes with a structural engineer to make sure the design recommendations are implementable and make sense from a structural standpoint. Finally, we will selectively implement climate-resilient retrofitting with 10 families and gather qualitative data from the families on how the design and implementation process can be enhanced to better serve their needs.  

None at the moment. At least 10 by the time we will be ready for the piloting stage. 

We are applying to MIT Solve for support in 3 domains. First and most importantly, there are huge legal and regulatory hurdles that are constantly arising surrounding any operation within informal settlements. Issues around housing within slum populations are also extremely political and we are interested in seeking the expertise from Solve on best practices. Second, we are seeking assistance around financial matters, as we are technically not a registered organization and our team is based in different countries. Also, no one on the team has extensive experience pitching to business investors. Finally, we are interested in Solve’s expertise in public relations to help boost awareness of our goal to provide a market based solution to improving climate-resilience for informal communities around the world. 

To properly discuss why our solution is innovative, I will first describe the two other methods of generating climate-resilient retrofitting recommendations:

The “standard retrofitting method” is to have a structural engineer walk through a house to generate recommendations. This standard method is likely most effective for higher income households because: (i) these households have higher willingness to pay, and (ii) the housing structures are more complicated. Its weakness is the high variable costs to generate recommendations, meaning the highest prices.

The second method is to send trained individuals to take photos and collect basic data from each home. This data is sent back to a structural engineer, and some generic recommendations can be generated automatically. This method will henceforth be referred to as the “partially automated” method.

Resilience 1.0 has four advantages on the “standard” and the “partially automated” methods: (1) lowest variable costs hence the lowest market price, (2) generates recommendations for an entire community in one session, (3) takes the neighborhood structure of homes into account, and (4) relies on prediction to boost climate-resilience. (3) matters because the angling of the neighbors’ roofs effects how typhoon winds impact homes. (4) matters because it is impossible to earthquake or typhoon-proof a home, as the way a natural disaster impacts a home depends on factors such as the angle of the roof, distance from the fault-line and direction of wind. Resilience 1.0 is more cost effective than the standard and partially automated approaches due to (1) and (2). It is particularly innovative because of (3) and (4), as we believe these may help it outperform recommendations generated by a structural engineer.

Other aspects of Resilience 1.0 that make it particularly innovative are (a) its inherent scalability due to the data collection and design generation process, (b) the fact that it is particularly well suited to slum-type homes due to the simple housing structures, and (c) its ability to incorporate community level feedback into the design generation process.

As mentioned in the problem statement, boosting climate resilience for houses requires both climate-resilient materials and climate-resilient retrofitting. Resilience 1.0 directly catalyzes the positive impacts from on-the-ground work that has generated low-cost climate-resilient materials that were designed specifically for poor and low-income SES house types in the Philippines.

Finally, our team has specifically targeted Resilience 1.0 as a solution to address the market failure identified in the problem statement. By lowering the costs per house of generating a climate-resilient retrofitting recommendation, our solution can directly lower market prices for climate-resilient retrofitting services from private construction firms.

Increasing a house’s ability to survive the frequent natural disasters that hit the Philippines will have a massive and transformative impact on the families living within informal settlements. The impact on their savings, consumption and well being can be captured via the frequent surveys that are conducted by various government bodies and research agencies.

Our impact goal as an organization is to scale our solution. Within the next year, we intend on finishing our prototyping work and pilot our solution on 1,000 houses in Davao City. Within the next 5 years, we intend for our solution to be entirely supported by the free market or partially supported via subsidies by local governments. We intend for over 100,000 slum homes to be retrofit to better resist natural disasters in that time. Our plan for measurement and how to achieve these goals is detailed below:

Stage 0: Prototype the innovation and household intervention, estimate costs and prices that firms can offer these services at. (Davao City)

Inputs are equipment for data collection, engineering simulations, and low-cost retrofitting materials. Outputs are the personalized retrofitting recommendations. The community level intervention is the retrofit 10 homes with constructional professionals and supplement this knowledge with focus grouping and testing within the community. Intermediate outcomes are whether the retrofitting recommendations passed the back-checks, and whether the retrofitting of 10 homes was implementable and feedback from the families. Final outcomes are the estimates of costs to firms to procure materials and offer climate-resilient retrofitting.

Stage 1: Pilot and demand side estimates. Measure willingness to adopt the Resilience 1.0 algorithmic recommendation and implement. Estimate the efficacy to generate realistic cost-benefit analysis to the government. (Davao City).

The inputs to the pilot are climate-resilient housing materials and personalized retrofitting recommendations generated by Resilience 1.0. The outputs are the retrofitted homes. The interventions to measure willingness to adopt are messaging and information delivery interventions. Willingness to pay may be boosted by messaging, information interventions, and by providing loans. The intermediate outcomes are household willingness to accept and willingness to pay for the retrofitting services generated by Resilience 1.0. The final outcomes are the actual number of houses implemented, and 1-year post-intervention performance of the Resilience 1.0 retrofitting recommendation technique, as measured from satellite data. From this, we can generate actual estimates of cost-benefit of the retrofitting innovation. We can also conduct a 1-year second endline survey to gauge whether dynamic investments in climate-resilience to housing infrastructure increased after the pilot. 

Stage 2: Determine which market solution is feasible, and then attempt firm-side research on how to develop capacity to offer Resilience 1.0. If market prices are less than or equal to willingness to pay, then our solution can be entirely supported within the market. If prices minus willingness to pay is less than costs of reconstruction to local governments, then our solution can be supported via the free market and government subsidies. 

Our team is working with Innovations for Poverty Action Philippines to collect robust data on our impact goals. We use a combination of qualitative and quantitative surveys from the families, and satellite data to track how many houses were retrofit, how the retrofit houses survived over time, and what the impact of boosting climate resilience to houses was on families. The exact measurements are discussed in the impact plan section.

The long-term goal of our solution is to boost the proliferation of climate-resilient housing in informal communities within the Philippines. This needs to come from correcting the market failure: demand side willingness to pay for climate-resilient retrofitting services is lower than the market price, so firms do not offer these services to poor SES households. Resilience 1.0 is an innovation that substantially lowers the costs to firms for offering climate-resilient retrofitting services for informal communities. We will first determine how to design delivery of the services to maximize household willingness to pay in order to best address the market failure in our prototype and pilot. Then, using information on costs to firms, we will determine whether the market can entirely support climate-resilient retrofitting or if the prices need to be supported via government or donor intervention. 

Boosting the ability to adopt climate resilient retrofitting should have a ripple effect, working to further boost climate-resilient household investments in housing infrastructure. The increase in investments in housing infrastructure from families will continue to boost the probability of the house surviving a natural disaster, which may increase willingness to invest in insurance.

Resilience 1.0 uses the methodology to boost climate-resilience of buildings which was proposed in Mata et al 2022. However, it is specifically adapted to slum housing. Instead of relying on an individual civil engineer to walk through each housing unit to recommend climate-resilient retrofitting, the proposed method relies on use of cameras and drone footage to capture the baseline structure of the buildings. It then utilizes genetic algorithm to optimize over the best roof angle and other structural components (as highlighted by family use) conditional on properties of the underlying structure to maximize the probability of the home surviving a magnitude 6 earthquake or typhoon winds of 110mph. We will test different ways of delivering the recommendations to families to make Resilience 1.0 via methods like SMS or door-to-door visits.

https://www.mdpi.com/2075-5309...

We are a team of researchers committed to finding a scalable solution that will help boost climate-resilience in informal communities. Our end goal is to have Resilience 1.0 completely supported within the market without need for support from us as an organization. If this isn’t possible, we can also see this becoming a for-profit business that we contract out to a different set of practitioners. We can send different teams of freelancers to collect data from the communities on building structures, materials, and familial demands around housing. We can then generate climate-resilient retrofitting recommendations for each community at a time and charge per-house for the service. Implementation can then be contracted out to a local construction firm and done in partnership with the families.

Our solution is best positioned to sustain itself within the free market, or via government subsidies. We will determine which solution is most viable through our prototyping and piloting. Discussion on exactly how this will be done is in the impact section.

So far, our work has received funding from grants supporting research in scalable solutions for development economics research. Financial viability of our solution goes hand in hand with both scalability, and hence we have gotten a lot of interest from both practitioners and academics. We have raised enough funds to complete our prototype, and intend on using our results to generate more interest and funds for the pilot.