Jay Gho
Jay Gho
Sep 11 · 5 min read

Most of us in Finance think of ourselves as good decision makers. We didn’t acquire our decision-making abilities overnight, of course, but through years of training, experience and too many failures to remember. Having put in years of work, many of us are now comfortable making daily decisions through a series of mental steps: observing data (including qualitative data), creating mental or actual models, evaluating options and identifying our risk tolerances (determining our “limit order”, essentially).

Now imagine a different you — an artificial you — that can analyze millions of years of life experiences in minutes — without fatigue, emotions or cognitive biases.

Artificial Intelligence, Machine Learning and “Deep Learning”

AI is frequently used alongside another term: machine learning. While most experts consider machine learning as a subset of broader AI, this article uses both terms interchangeably to describe computer-based software and algorithms used to observe and interpret data sets to arrive at conclusions.

A more complex form of machine learning — sometimes coined “deep learning” — structures algorithms in layers to create an artificial neural network in the same way humans make complex decisions. The most famous example of a deep learning model is Google’s AlphaGo, which successfully learned the complex techniques and strategies of the game to defeat several (human) world champions.

AI in Finance

AI has been researched and developed in academia for decades, but in recent years this interdisciplinary science has attracted billions in investments, hyperbolic media coverage and a cult-like following — most notably in autonomous vehicles. Within financial institutions, AI and machine learning methods are used in a myriad of processes:

  • Credit scoring of prospective borrowers
  • Pricing of insurance contracts
  • Automated sales calls
  • Automated customer service responses
  • Fraud detection
  • Trading and portfolio management
  • Capital optimization

AI and Municipal Bond Investing

The municipal finance industry is not known to lead on technology and new investment practices. That said, given the idiosyncratic, fragmented nature of the industry, AI and machine learning may yield more benefit to municipal bond investors and issuers than to equity investors and corporations. I’ll write about AI applications for municipal issuers in an upcoming article, but here are some AI applications for investors of municipal bonds:

Improving The Credit Analysis/Research Process

Credit analysis for municipal bonds hasn’t evolved very much. As I’ve written a year ago in Municipal Equity Value — Challenges and Applications, credit analysis of municipalities remain focused on the Municipal 3Cs: Cashflow, Coverage and Covenants (broadly, the security package including pledge, indenture and bond documents). Of the 3Cs, cash is the driver. When there is sufficient cash, coverage and covenants are usually non-issues.

Many research analysts model issuer cash flows on a few key “top down” metrics, like economic growth, personal income, employment, property values, enterprise revenues etc.

Properly-designed and with the use of reliable data sets, AI models can glean interconnections between credit-worthiness and less-understood metrics — like population migration, economic development investments or crime. Tools like sentiment analysis and text parsing (a machine learning model can read 100 Official Statements during a work day) may be valuable in certain cases, especially for issuers that issue frequent disclosure or make public statements.

Algorithms cannot supplant a good credit analyst’s judgement and experience. But experience can also be a crutch, blinding analysts to changed conditions or correlations. Properly used, AI models can point analysts to new directions in search of alpha.

Providing Monitoring and Surveillance to Fragmented Muni Market

One of the challenges specific to municipal finance is the fragmented nature of the industry.

Based on MSRB data, despite being half the value of the corporate bond market, the municipal finance industry has over 50,000 municipal issuers vs 10,000 corporations and over 1,000,000 municipal CUSIPs outstanding vs just 30,000 for corporate CUSIPs.

As asset managers continue to cut costs by hiring fewer young investment professionals, tracking the performance of small and infrequent issuers becomes a challenge.

Obtaining data sets with issuers’ financial data has never been an easy task. As the taxonomy of CAFRs gets standardized (there is an ongoing push to achieve this), the potential to utilize AI models to screen small issuers, monitor holdings and flag surveillance events will allow asset managers to increase diversification within their portfolios, while also reducing borrowing costs for small but high-grade issuers.

For fund managers with Environmental, Social and Governance (ESG) mandates, AI models may prove to be even more valuable, facilitating the surveillance of the hundreds or thousands of operational Key Performance Indicators (KPIs) monitored as part of external Green Bond certification or a complex’ in-house proprietary ESG scorecard.

Predicting Demand/Supply Technicals + Fund Marketing

The performance of municipal securities can often be driven by technicals, with fund flows ruling rather than credit fundamentals.

As AI expertise expands and quality data sets proliferate, asset managers can formulate algorithms to predict near and long-term demand for municipal securities, size portfolios accordingly and make appropriate investment decisions — perhaps to pre-empt a decline in demand through investor education?

For example, to answer the question “will Millennials and Gen Zs buy munis”, brokers will first have to answer: “why did Gen X and Baby Boomers buy munis”?

The answer is likely to be differentiated by state, issuer, demography and local economic conditions. Once again, robust AI models — with the help of an experienced data scientist and an experienced muni analyst — can help address such questions without being daunted by complexity and large data set challenges that can easily overwhelm smaller teams.

AI As Subset of The Broader FinTech Revolution

While the use of AI is still nascent in Finance, there is a broader “FinTech” force underway that is massively disrupting commercial banking, consumer lending and asset management industries. The behemoths of Finance — JP Morgan, Goldman Sachs and Bank of America among others — have been forced to respond by creating nimbler versions of their legacy business lines, acquiring startups or partnering with pure-play technology companies.

In view of the overwhelming revolution sweeping through Finance, it seems unlikely that municipals will be immune to change. For those in the industry eager to learn new skillsets and utilize technology to innovate, finance/munis could get really interesting again.


Jay Gho is a New Yorker who loves dogs, cooking and Orange is the New Black. Contact him at jay@jaygho.com.

T.E.C.H

Essays on Technology, Media and the Internet of Things

Jay Gho

Written by

Jay Gho

Family Man. Humanist. Immigrant. Finance/Tech/Policy Nerd. Former Banker/Private Equity. Open to gigs.

T.E.C.H

T.E.C.H

Essays on Technology, Media and the Internet of Things

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade