Citi Community Capital (CCC), the community development and investment group of Citi, tasked the Capstone team to improve its financial risk model. CCC provided the team with more than a decade’s worth of data from its affordable housing loans which had never been analyzed comprehensively. The team organized, cleaned, and merged over 40 CCC datasets to produce: (1) a cross-sectional dataset for 2021; and (2) a panel dataset for 2012-2021. For both the cross-sectional and panel datasets, the team generated summary and descriptive statistics, and ran simple and multiple regressions to evaluate different loan characteristics as predictors of loan performance. 

The results of this study found that deals for more walkable projects with better occupancy rates and higher gross potential rents were more likely to perform well, while those in higher crime risk areas with higher principal balances were more likely to perform poorly. This was as expected. The team's regional analysis found that Chicago and Washington D.C. had the highest and second highest proportion of underperforming loans as of 2021. Unexpectedly, some project attributes, such as whether a project was senior-only, did not explain loan performance. Further, the measures of market conditions used by CCC had negative associations with loan performance, leading the team to recommend that CCC examine whether these are appropriate indicators for their risk model. Finally, based on the team’s experience with data wrangling, they have provided CCC with extensive recommendations to improve its data management.