Kiva - Microfinance Access to Banking $25 at A Time
The skills I demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
Kiva is a non-profit crowdsourcing microfinance platform, providing banking and financial services to underserved communities around the world. Loans are also offered to borrowers who aim to create social impact in their communities. Lenders may crowdfund the loan in increments of $25 or more.
Kiva works with Field Partners that disburse the loan to borrows and then post the loan on the Kiva website for crowdfunding. Field Partners are typically local micronfinance enterprises, schools, or other approved organizations.
This Shiny app is built based on a dataset provided by Kiva for a Kaggle competition. Information included is limited to loans posted during early 2013 to mid 2017.
The Global Multidimensional Poverty Index (MPI) published by the United Nations Human Development Program is an additional metric incorporated to the Kiva dataset. MPI uses healthy, education, and standard of living indicators to determine the degree of poverty experienced by a population.
The purpose of this Shiny app is for users who want to learn about the landscape of microfinance through Kiva and users those who would like to consider lending for the first time. Kiva can also use this analysis to assess performance by their Field Parters or whether they should feature a certain country or sector. This app explores various metrics such as general use of loans, loan balance, funded and unfunded amount, lender counts, etc. on a country by country basis, as well as on a global basis.
This app is built in R studio and please find the code here.
Datasets used for this app are as the following: Kiva Dataset, United Nations MPI Data.
Country by country data can be quickly visualized with a map. The metrics that can be displayed by hovering over the countries are total funded loan, total loan amount, total unfunded amount, average funded amount, and average loan amount. The max and min country for each metric as well as the metric mean are also shown. The intensity of the the coloring can be adjusted based on MPI, lender count, and funded ratio. This is an easy way to display 2 metrics at the same time.
The observations are as the following:
- The number and effectiveness of Field Partners may drive the total loans and funded amount.
- The metrics may be affected when Kiva became present in a country
It is noted immediately noted that MPI (i.e. a country's poverty level) does not have any correlation to any of the loan related metrics. Countries with low MPI may not have a sufficient number of or effective field partners help educate potential borrows and push through the loans. It may also due to the fact these countries are newly covered by Kiva. As mentioned previously, the dataset is a snapshot in time. If updated on a realtime basis, lenders who are MPI conscious may use this information to inform their decision on their lending decision. Kiva can use this information to evaluate their field partners and whether they should increase the number of field partners.
The Philippines has $55M of loan balance of which $54M is funded, while Guam only has $4.3K of total loan and the Virgin Islands has a zero funded balance. This could be that the Field Partners in the Philippines has a long standing partnership with Kiva and are relatively effective as writing loans while Guam and the Virgin Islands are relatively new to Kiva, thus their loan balance and funded balance are low respectively.
From an average loan balance and average funded loan perspective, there is a large range among countries, this could be based on sectors and the size of loans advocated by the Field Partners.
The funded ratio among most countries are quite consistent. The United States is actually on the lower end. Rather than helping underserved communities, Kiva aims to lend to individuals and groups who aim to make a social impact and fight for social justice in more developed countries. It is possible that lenders are more focused on underserved communities rather than communities in more developed regions.
Another feature of the app is the country by country comparison. This allows users compare two countries for the metrics included. In addition, this allows users to the see the top sectors for the loan balance, funded amount and unfunded amount, which would help users narrow down the sectors they may be interested in.
The app also explores metrics based on genders.
It is apparent that majority of the borrowers on Kiva are females or all female groups. This finding is consistent with Kiva's philosophy of empowering women via financial independence and ownership. Users can also see relative frequency of genders by unfunded loans. This could help users discern the loan they may want to fund based on borrowers' gender.
While the app is able to illustrate the observations stated above , the greatest limitation of the dataset that it only covered a static period of time. Further observations can be made with a dynamic realtime dataset. In addition, the below are examples of other features to be included in this dataset:
- Loan repayment information from Kiva.
- Regional analysis that is more granular than analysis at the country level.
- Measurements of economy on a country by country basis from other sources.
The above features may enable the analysis the relative measure of poverty, which could help Kiva set its investment priorities, better target communities, and help inform lenders.