Ask the Accounting Community

 View Only
  • 1.  Is anyone using AI for Financial Analysis, and how?

    Posted 09-18-2019 05:37 PM
    As finance professionals, we are bombarded by buzzwords like "AI," "Blockchain," "Machine Learning," etc.  Can anyone give me an example of how they might be using AI in their current roles?  Are finance folks developing analysis tools using AI (by programming neural networks??), or is this more of something that is inherent in software applications that finance teams are investing in?  Are these tools more useful to automate reconciliation or validation type work, or has anyone incorporated them into financial analysis tools?  Can they move historical profitability trending to predictive modeling?

    I'm just trying to get an practical understanding of where finance technology is headed, so that my team can get the correct education.  Do we need to become programmers of sorts?

    Thanks!

    ------------------------------
    Eliza Montgomery CMA, CSCA
    Analyst
    Peoria IL
    United States
    ------------------------------


  • 2.  RE: Is anyone using AI for Financial Analysis, and how?

    Posted 09-18-2019 07:05 PM
    Eliza, I consult in the Banking, Credit Union, and Finance Company space.  The AI see, create, and use tends to fall in two categories: 1) Financial Analysis, 2) Credit Risk Analysis.  My personal opinion, which is open to criticism is that AI is the next progression from the early concepts of Big Data.  In other words, if you do not have the data, and you do not use all of it than the use of AI will appear meaningless.

    If I can offer three examples.  On the Financial Analysis front, most institutions have a core system for accounting.  Most of the higher end product accommodates, budget numbers, actual numbers, and calculations of ratios.  However, most financial institutions do not think of them as departmentalized.  Is the Bank or Credit Union a depositor, a lender, a investor, etc? Departmentalization (Department Codes), labor costs, and outside service expenses are not often loaded at the Department level.  So the use of benchmarking, peer analysis, and forced place grading, which are the use of AI can not be used where the data doesn't exist.  Fortunately, programing is not often needed, just separating the data into available data codes and running the reports.

    Credit analysis is something of a different type.  The regulators are now using "Probability of Default" and "Loss Given Default" as assessment tools for Loan Loss Reserves.  Banks have never thought in terms of "What is the potential for loss in each loan?"  AI often need to be "programmed into the outputs such as cashflow and cashflow coverage plus collateral and collateral coverage to really assess risks.

    Finally, auditing has long ago moved away from random or systematic testing and migrated to managing the outliers.  Therefore, AI can be used as a predictive tool, it the data is sorted by, "What is not normal?"  This covers a vast array of uses, but the first criteria needed usually comes from the Finance Department of the institution. "What would normal look like if you saw it?" This usually breaks down to a statistical profile of common outcomes.  If you don't have that than AI will not do you much good.

    I hope my observations help.
    Sincerely,
    Jim
    James M. Wilber CMA, MBA
    IMA Chicago, President
    president.chicago.ima@...
    https://linkup.imanet.org/chicagochapter/home







  • 3.  RE: Is anyone using AI for Financial Analysis, and how?

    Posted 09-19-2019 08:07 AM
    This is great!  Thanks Jim!

    Risk management seems like the most obvious application.  In future, do you think that larger companies with big data repositories will be able to also incorporate the predictive capabilities of AI into financial forecasting models, or turn static KPI reporting into tools that can show predicted future KPI results as well?  Kind of like financial meteorology, lol.  Or is that overkill?

     Thanks again!

    ------------------------------
    Eliza Montgomery CMA, CSCA
    Analyst
    Belvidere IL
    United States
    ------------------------------



  • 4.  RE: Is anyone using AI for Financial Analysis, and how?

    Posted 09-19-2019 10:36 AM
    I really like the meteorology analogy.  Please note my perception of the predictive tools for weather is that forecasters don't use just one model.  They use a number of predictive formulas with several different outcomes, (remember how many landing sites there were for the hurricane Dorian?) and then the weathermen evaluate the data.  Predictive software is potentially difficult to build, but cheap to run.  It took me two weeks to build out a forecast model once.  When I was finished, I predicted the closure of a bank six months into the future and missed the date by one week.

    I believe the predictive capability of AI will become more frequently used.  An AI opportunity; where I don't see it used now, is the lack of stock price modeling within public corporations. Most equity and bond traders use predictive models, which causes extreme volatility in the financial markets.  The traders can't turn the program trading software off fast enough to stop trading based on misdirected data.  Unfortunately, public traded companies don't spend enough time building their own model to see what impact decisions may make on their stock price above and beyond earnings and cashflow.  Weeks are spent considering the impact of one and two cent differences between actual and forecasted earning per share guidance.  However, little impact is done to adjust corporate valuations based on human costs, environmental factors, press releases, corporate governance, etc.

    Personally, I like the use of dynamic KPI's.  What I mean is a number of companies understand the use of KPI's as a benchmark tool.  However, rolling a forecast out over multiple periods should call for adjusting decision making.  Growth, consolidation, inventory cost (changes in tariff rates) , labor rates (legislated minimum wage changes) etc. should cause static KPI's to be poor predictors over time.  Now is when the management accountant, (the weathermen) can apply their skill set.
    Sincerely,
    Jim
    James M. Wilber CMA, MBA
    IMA Chicago, President
    president.chicago.ima@...
    https://linkup.imanet.org/chicagochapter/home







  • 5.  RE: Is anyone using AI for Financial Analysis, and how?

    Posted 09-20-2019 09:47 AM
    Great food for thought!  thanks so much for your quick response.

    ------------------------------
    Eliza Montgomery CMA, CSCA
    Analyst
    Belvidere IL
    United States
    ------------------------------