Shivam Arora, CPA, is an expert data scientist with dual Master’s degrees in Accounting and Business Analytics, and this combination has come to define the most innovative aspects of his career.
Arora is an applied Artificial Intelligence consultant who specializes in applied AI for accounting and finance. He’s currently working with a global consulting firm, managing the reporting analytics workstream and overseeing the development of ancillary data layers.
Most importantly, Arora has earned a great deal of notoriety by being one of very few people who apply data science expertise and methods to the typically business-minded fields of accounting and finance.
This represents an enormous opportunity, especially given the many recent advances in data science related to AI and Machine Learning.
Today we’d like to walk you through some of Arora’s most important practitioner papers and discuss why they’re poised to have a major impact on the worlds of data science and business.
Table of Contents
Before we cover some of Arora’s more recent practitioner papers, we would like to talk about an earlier paper that has received a great deal of attention.
The paper in question is called “Persistent, Historic Factors Correlated with Airport Bond Returns,” and Arora published this paper as a recipient of the ACRP Graduate Research Award.
For the unfamiliar, the ACRP is the Airport Cooperative Research Program, and this award is meant to inspire research and discussion within the community and also serve as a way of sourcing individuals who may become involved with airport operations, coordination, and relevant policymaking in the future.
Following his receipt of this award, Arora was selected to serve on the expert panel of the ACRP Graduate Research Award, alongside fellow academics and industry leaders.
As a member of this panel, Arora is currently overseeing the research of two doctoral students, providing feedback and guidance to these research efforts, both of which focus on the application of analytics to airports.
“Traditionally, only renowned professionals who are either veterans in their fields or possess critical expertise vital to the aviation industry are invited to serve on ACRP expert panels. Other distinguished individuals in this year’s panel include Elliott Black, Planning/Programming Branch Manager at the Federal Aviation Administration, and Danielle Rinsler, Public Policy and Planning leader at Amazon.”
As for why Arora’s award-winning research paper was groundbreaking for the industry, that will take a bit of explaining.
The paper explains the returns on bonds issued by US airports, as a function of six different factors that have been historically correlated with airport bonds, and which can then be used to understand the types of returns investors expect on these kinds of bonds.
With this advanced understanding in hand, airports can then make more informed, intelligent decisions about which types of bonds to issue at specific times.
Airport-issued bonds hadn’t been formally explored, at least not until Shivam’s work. He used regression analysis to create multiple models based on data available to the public.
Shivam’s paper is the first of its kind in this field, and many believe that it will continue to be influential, especially for academics and practitioners, and it seems likely that the work will also influence future developments in airport bond pricing.
Once again, the core of Arora’s groundbreaking work on this front is the application of data science methods to accounting and finance. Arora has made it clear that, by using advanced tools, there are many opportunities to better understand financial developments, which can in turn inform future financial and accounting decisions.
With this idea well-established, let’s move on to some of Arora’s more recent work.
Two of Arora’s most recent practitioner papers are centered on Natural Language Processing (NLP) in Accounting and the estimation of current expected credit losses (CECL), respectively.
While we can’t cover every detail of these papers, we’ll do our best to provide an overview of some of the key concepts presented in these papers, with Arora’s assistance and commentary.
The first paper we’d like to mention here is titled, ‘Natural Language Processing in Accounting,’ and it was published in Strategic Finance magazine in March of this year.
The paper takes into account that Machine Learning has started to make its way into the accounting profession, becoming much more common in recent years.
Further, there’s a specific branch of Machine Learning known as Natural Language Processing (NLP).
As the name implies, NLP involves the processing of spoken languages, and this technology has a huge amount of potential within the accounting industry, specifically as a tool for developing models to process text-related tasks.
“My practitioner paper on the subject was published in Strategic Finance (SF), the award-winning magazine of the Institute of Management Accountants (IMA). The paper details an end-to-end pipeline for developing an NLP model to predict customer sentiments such that the model can ingest a customer review and predict its sentiment as ‘Positive,’ ‘Negative,’ or ‘Neutral.'”
Automating the analysis of customer feedback could have a large number of benefits for airports, making it much easier to compile that feedback into useful data sets, which can then inform future decisions and the likely impact of those decisions.
The paper provides details on each step of the process, from the initial data collection and preprocessing work all the way to training and fine-tuning the NLP model.
This paper clearly made a splash in the industry, with Strategic Finance selecting the paper for the inner cover of its print version. The Institute of Management Accountants also quoted Arora on multiple social media channels, including Twitter and LinkedIn.
But this isn’t the only practitioner paper that Arora turned out recently.
The second paper we want to highlight is titled, ‘A Regression Approach to Estimate Credit Loss.’ This paper was published in Analytics, which is the analytical magazine of the Institute for Operations Research and the Management Sciences (INFORMS).
First, a brief explanation of Current Expected Credit Loss (CECL). Credit loss, on its own, refers to the amount of debt that a given creditor doesn’t expect to collect. This is also known as bad debt. This calculation is labeled as CECL.
Previously, accounting standards only required a creditor to estimate and accrue credit loss only in cases where it was likely that a debtor would default on their payments toward debt.
However, there is a new Accounting Standards Update, specifically ASU 2016-13, that altered these requirements. Now, an entity is required to estimate CECL right when they issue credit, regardless of how likely the debtor is to default.
For accounting professionals, that can be a difficult task, especially when dealing with debt that has a long maturity period. It’s difficult to consider circumstances that far into the future.
This is where Arora’s paper comes into play. Arora outlines a new approach for estimating CECL that can make the task far easier.
“In this paper, I propose a novel approach to estimate CECL using Vector Autoregression (VAR). VAR is a regression analysis that uses the past values of a variable, which here would be CECL, to predict its future values. Because prediction is based on past values, an entity can be confident that all factors that could impact the future loss amount have been taken into consideration.”
For more detailed information, interested parties can read through the full paper by following the link we listed above.
As we’ve already discussed, the application of advanced data science within business contexts, and even more specifically within the context of airport accounting, is hugely beneficial.
Data science technologies allow for the collection and analysis of mass amounts of data, and emergent technologies will continue to present even more opportunities.
We’ve also talked about how, thus far, Arora is one of the only people exploring these applications in detail, and while his groundbreaking work is clearly the result of his accounting and data science expertise, we were still left wondering why these applications haven’t been explored previously.
Arora feels that it’s because aviation finance, in particular, is extremely complex.
“It is traditionally difficult to apply advanced analytics to business domains like accounting and finance. Aviation finance is an incredibly complex topic in itself. Applying analytics to it requires a very high level of skill and sophistication. One has to understand the nuances of financial concepts as well as statistical techniques. This, in my opinion, is the reason why many airports do not apply data science to their finances.”
Thankfully, Arora has proven that he’s more than willing to tackle these complexities through his data science knowledge, and he’s also fostering the continuation of this work through his work as a member of the ACRP Graduate Research Award panel.
As a way of closing out, we’d like to thank Arora for assisting us with this article and also mention some of his upcoming work.
Arora has another practitioner paper set to be published soon, and it will apply fuzzy matching to an accounting use case.
Even more exciting is the upcoming release of Arora’s book, titled ‘Python for Accounting and Business,’ the first of a two-volume series on the application of programming language Python to accounting, finance, and business domains. It’s set for release later this month, so keep an eye out for it!