Research constitutes a core activity for APAR. Below you can read more about our approach to research and guiding principles, current and upcoming research projects, and Sponsored Projects opportunities.
Our Research Principles
One major area of focus for APAR is its novel approach to empowering leading-edge, cross-disciplinary, and collaborative research not only for the intellectual pursuit of complex questions, but also to find applicable solution to address industry and societal needs.
Our aim is to question the inherent assumptions about current best practices, explore wide and deep the possibilities for revolutionary discoveries ("0 to 10 Solutions") as well as evolutionary improvements of existing frameworks ("1 to 10 Solutions"); in short, we strive to fundamentally challenge the status quo and change it for the better. The collective expertise of our faculty coupled with our intellectually-curious and highly-talented students allows APAR to be at the forefront of research, solutions output, and impactful projects that could positively affect the industry and the society as a whole.
In doing so, we are breaking some precedents to foster and support such research endeavors. For instance, we provide no-cost, open-access to our growing library of unique proprietary datasets (one cannot even buy such datasets!). We undertake the heavy-lifting efforts in dataset management (cleaning, structuring, hosting, etc.) so to liberate researchers from as many time-consuming administrative and set-up tasks, including technical, as possible so they could focus on pure research and run at the speed of thought. Additional support is provided to our researchers and collaborators on a case-by-case basis in the forms of unrestricted funding, technical research assistants, hardware and software, cloud/high-powered computing (CPU and GPU), and more.
If you are faculty, researcher (including Post-Docs and PhDs), or an graduate or undergraduate student interested in our work, we would love to hear from you. Write us at [email protected] to inquire about our datasets, research assistant opportunities, and technical research internships.
Select Research Projects
A selection of APAR's recent or ongoing research projects we can post publicly are listed below - collaborators and/or co-authors are listed alphabetically.
Some of APAR’s current projects, originated and spearheaded by our Executive Director, include:
"How Does Household Spending Respond to an Epidemic? Consumption During the 2020 COVID-19 Pandemic."
Co-authored by Scott R. Baker (Northwestern), R.A. Farrokhnia (Columbia), Steffen Meyer (U. of Southern Denmark), Michaela Pagel (Columbia), Constantine Yannelis (Chicago). Posted on April 2020 - NBER.
"Income, Liquidity, and the Consumption Response to the 2020 Economic Stimulus Payments."
Co-authored by Scott R. Baker (Northwestern), R.A. Farrokhnia (Columbia), Steffen Meyer (U. of Southern Denmark), Michaela Pagel (Columbia), Constantine Yannelis (Chicago). Posted on May 2020 - NBER.
Integrating Machine learning and Psychometrics to devise more effective nudges for desirable personal financial outcomes (completed).
In this project, we are aiming to create scalable, deployable, and impactful methods to help individuals save more money, pay down debt faster, and be financially savvy in managing their savings/retirement. Working with a large dataset of consumer financial transactions, this project uses a variety of analytics and predictive ML models to test and validate our hypotheses as well as explain the results. The project's Principal Investigators are R.A. Farrokhnia and Sandra Matz, both from Columbia Business School, with Suwen Ge (CS) as Technical Research Assistant.
Consumer financial behaviors using bank vs. non-bank lending products (in progress).
Our objective is to understand consumer financial behavior in regard to using bank-provided credit/loans products vs. non-bank providers (e.g. merchant financing). We would then aim to devise methods to improve consumer financial welfare and provide better access to less onerous financial products, improve financial inclusion, and nudge consumers toward more financially-sound outcomes in managing their personal finances. The project's Principal Investigators are R.A. Farrokhnia, Michaela Pagel, and Sharada Sridhar (PhD), all from Columbia Business School, with Suwen Ge (CS) as Technical Research Assistant.
Reinforcement learning and deep neural network tree search for an asymmetric multi-agent card game (in progress).
In light of the latest research in reinforcement learning and the ability to bypass supervised learning in areas without much (or any!) training data, this project’s goal is to leverage the potential of reinforcement learning to build a smart, adaptive game engine for a multi-agent card game (think of Bridge, but with no bidding and more fun!). In this game, there are both collaborative and adversarial player dynamics at play simultaneously, with luck and strategy playing a role in a game setting where all the states of the game cannot be fully known (esp. at the early stages of the game). This is a very promising and yet quite challenging task, but equally rewarding given the complexity and creativity needed to solve it. The Principal Investigator of this project is R.A. Farrokhnia (Columbia), with Suwen Ge (CS) as Technical Research Assistant.
Application Specific Tokens and Protocol Layer Development (upcoming):
This fast-developing and innovative field has come to prominence only recently, and is immensely interesting from both academic and practitioner perspectives. It offers opportunities for original and imaginative products and business models as well as solutions to difficult problems that until recently were deemed too costly and complex to address. In general, we are actively participating in the general education to widen the audience, building practical libraries of tools for finance professionals and technical developers, and contributing to the open-source community. In particular, we pursue ideation, validation, and development of original protocols and application-specific tokens (including ICOs) within and without the financial services industry. The Principal Investigator of this project is R.A. Farrokhnia (Columbia).
Examining legal, regulatory, governance and security issues that enabling broad application of blockchain technologies as well as factors needed to ensure security and fidelity of blockchain and smart contracts and the data therein (part 1 completed - part 2 in progress).
It will also delve deep into matters relate to legal, policy and regulatory issues affecting catalyzation of key components of the crypto-economy such as the underlying blockchain technologies and smart contract applications, and providing a roadmap for ensuring legal certainty, application security, fairness and transparency. Principal Investigators are R.A. Farrokhnia and Dr. Leon Perlman, both from Columbia Business School. This project is funded through Columbia-IBM Center for Blockchain and Data Transparency.
Dynamic use of D3.js for interactive, self-paced learning modules (in progress).
Using inspiration from R2D3 and scrubbing calculators (Cruncher,io, Numi, Soulver), we aim to explore how to push the boundaries of forms, functions, and applications in expressing statistical and technical thinking with interactive design (incl. for educational purposes). This project primarily will focus in offering an accessible primer on Demystifying Blockchain without requiring any technical background or knowledge by the part of the reader. The Principal Investigator is R.A. Farrokhnia (Columbia), with Chen Chen (CS) and Mitali Juneja (CS) as Technical Research Assistant.
Alternative Trading Systems in Fixed Income (upcoming):
With the advent of Regulation ATS in 1998, equity trading has witnessed vast changes and shifts in ways it operates, whether for retail or institutional clients. While the rapid technological development continues in equity-trading realm, the traditional models in fixed-income trading have lagged behind. In addressing this gap, this research project focuses on the experimentation, development and in-field deployment of novel trading protocols, specifically decentralized systems and algo-driven models.
Edge computing and inline analytics on elemental data (upcoming):
Within large enterprises, the use of elemental data that are required for other downstream functionalities is typically hampered by lack of interoperability and limitation of encryption frameworks between various databases and information silo (sold and supported by different vendors). Furthermore, the use and ownership of such elemental data as identity could be streamlined greatly while offering higher security and control compared to existing systems. We aim to explore how this challenge could be addressed by both rethinking the next-gen architecture (bottom up, "blue sky" approach) as well as developing potential solutions that could be implemented within the existing operational and infrastructure constrains of financial institutions. Success in addressing this issue will also enable the deployment of new edge-computing frameworks that could potentially provide faster, more efficient inline analytics in a number of settings, from equity trading to security and fraud detection to consumer identity protection.
Micro- and modular insurance, including smart contracts (upcoming):
A rich realm of research in which we are conducting a deep-dive into: 1) modern approaches to need-specific insurance products, with AI and algorithmic-driven claims submission, processing and handling, and 2) market, regulatory, and technological implementation analysis in both developed and developing countries, with an eye toward protocols that are executable with minimally-sophisticated hardware and software requirements, ensuring wider and easier mass adoption where costs and infrastructures may be bottlenecks.
Sponsored Projects: "Academic + Industry" Partnerships
On a select basis, APAR works with financial services firms on sponsored projects, on a varying levels of complexity and originality. Sample projects include exploring dynamic pricing and revenue optimization for credit cards, new distribution and agency management systems in insurance, and deploying Machine Learning and AI-based systems for more efficient and cost-effective customer acquisiton and service in banking. If you and your organization are interested in learning more or would like to pitch a collaborative sponsored project, please contact us at [email protected].