Smith Faculty Awarded Grants for Innovative Teaching Practices

The Smith Teaching/Learning Innovation Grants (STIGs) support the strategic plan of the School and the University. Grants awarded during the 2024-25 academic year are centered around the creation of a smart learning environment, aimed at developing innovative teaching practices and learning methodologies.

Cliff Rossi, professor of the practice, executive-in-residence and director of the Smith Enterprise Risk Consortium, was awarded a grant for the 2024-25 academic year.

Q&A:

Please tell us about your teaching innovation.

Students taking this ELP had a unique opportunity to conduct advanced statistical and machine learning analysis for a major financial services company. Skills learned in this project set students apart from other job candidates for analytic positions in the financial services industry as they learned how to produce and manipulate very large loan level databases for analysis, develop and interpret results from complex statistical and machine learning models, and present their findings in a formal presentation to senior industry professionals. Python software was used to develop databases, estimate models, and conduct comparative analysis of models. As a former C-level executive at major financial services companies, I can state that the experience students would take away from such a program is unique to anything else they are likely to take at Smith and will put them in a position to be very competitive in applying for industry positions. Over the course of 7 weeks the team of MFin and MQF students downloaded 15 million loans from the CFPB 2021 Home Mortgage Disclosure Act (HMDA) data. This data was combined with the FEMA NRI data that provides risk assessments for 18 different natural hazards at the census tract and county level.

This data was used to answer the following question: Do natural hazards impact mortgage loss severity? And if so, what is the nature of that impact? This is one of the hotly debated topics within the housing finance market relating to climate risk today. To empirically answer these questions, a variety of machine learning (ML) (including various types of supervised and unsupervised models) and parametric models (e.g., logistic regression) were developed and tested.

ML models have become popular in many industries including banking and so students will get a chance to develop and compare results from both statistical and ML-based methodologies. One of the benefits from ML models is their ability to identify important nonlinear relationships in data that may be overlooked by conventional statistical methodologies. Alternatively, one of the limitations from such models is that they lack transparency which is crucial to understanding how specific attributes affect risk outcomes such as default. As a result, beyond the development of various ML models such as neural networks or decision trees, students used the results from ML- based models to inform how more transparent results can be made by leveraging the results from ML models for statistical modeling. This project represents one of the most realistic and ambitious analytical projects graduate students can work on in an academic environment. It thus required substantial dedication of time and effort; however, it provided an excellent practical experience for students to showcase their skills. Weeks 1-3 were used to develop data and statistical and machine learning models as required. Weeks 4-6 were dedicated to working on enhancing or re-estimating the models with Week 7 allocated for final results review and presentation review and Week 8 (during finals week) was the final presentation to the corporate sponsor (in this instance the head of model development for Arch Mortgage Insurance). This ELP is not like a standard MFin/MQF course where each week a lecture is given, but rather the role of the faculty advisor was to provide guidance to the team and to solve problems that may arise throughout the project. The students “owned” the project including project management, making team assignments, and adhering to scheduled deliverables. Students were encouraged and expected to speak and interact during all our sessions. The project objectives were met for all the students, however, in a team environment it is always difficult to know exactly how much effort each student is making to the project. I used the class time each week to engage with all the students to find out what they were each working on, and also asked each person to complete a Team 360 survey where they had to rate each team member on the basis of their contributions to the project. As all but one of the students were “first-year,” there were some aspects of the project that prevented us from addressing additional technical models that they had not yet encountered in their coursework. However, all in all, the project was highly successful. In fact, the corporate sponsor was profusive in his praise of the technical abilities and domain expertise the students built up on mortgage credit risk in such a short amount of time.

What inspired you to create it?

Before coming to Smith in 2009 I had been in the financial services industry for nearly 25 years and had hired many analysts in the time and so I knew what I would want to see in a job candidate and wanted the ELP to showcase the critical skills students need to be successful in securing employment. The ELP was therefore structured to not just highlight the technical skills the students learn in their coursework, but also the intangible skills required to succeed in a dynamic business environment. Those include the ability to communicate technically complex analysis to nontechnical senior audiences, ability to apply business domain expertise to developing and interpreting model results, working in a team environment, and managing a complex project to a successful conclusion under demanding timelines.

Why is it important?

As we’ve seen with the recent nearly back-to-back hurricanes impacting Florida, North Carolina and the Southeast, extreme weather events are taking its tool on homeowners and renters in many cases displaced from their homes. Understanding how climate risk affects homeowners is of paramount importance in the housing industry and so our work showcases what effects such events have on the value of the largest asset most people will ever have, namely their home. Using well established publicly available datasets provides an opportunity to learn more about the risk to properties affected by such events so that policymakers, lenders and mortgage servicers can develop products and services to assist borrowers, particularly those that are most financially vulnerable to such events in guarding against the potential damages natural hazards can bring to property owners.

How did it impact the student experience?

The project, as have all of my prior ELPs help put our students in a great position to compete for top jobs at well known companies. We have regularly placed students coming out of their programs at companies such as Goldman Sachs, Freddie Mac, Fannie Mae, Wells Fargo, among many others and I hear quite often from these students that the ELP experience was what recruiters spent the most time during the interview discussing. One of the primary objectives of the ELP was to give students as close to a real-life work experience as possible and from what they tell me, we exceeded their expectations.