Data Analytics in Accounting

hello everyone and thank you for joining us today on this webinar my name is matt kelly and i am your host for the next hour as we talk about data analytics and its applications in auditing and accounting we have several great speakers joining us today and i think they'll both talk about how they're using analytics in their daily work right now the skills and expertise that you would need to put analytics to its best use give us some good interesting examples of how they are using analytics in their own practice and daily work and we have i believe more than 350 people registered for our webinar today so this should be a great and lively discussion uh first let me introduce our speakers for the hour first we are going to have pranav guy he is co-founder and ceo of calc bench pranav does plenty of analytics already today in his work uh at calc bench and then prior to calc bench he had also worked on analytics and financial research at morgan stanley tiaa and itg he is also heavily involved these days in data quality and structure issues working with xbrlus the cfa institute and other bodies next we also have with us brian wollahan who is partner in charge of the audit innovation team at grant thornton ryan serves on the data analytics task force at the public company accounting oversight board as well and he is a board member for the data analytics research center at rutgers university uh brian are you out there hello i'm here hello hello glad to have you and uh we'll also have with us vern richardson he is the distinguished professor of accounting at the walton school of business at the university of arkansas byrne has been working in data analytics for many years and literally wrote the book on data analytics which is aptly named data analytics for accounting uh sovereign hello and welcome to you too hey thank you thanks for uh posting this session we're excited sure and so a quick order of our events here for the hour uh pranav is going to begin with a few welcoming remarks uh he promises to be brief so we can get right to our guest speakers then we're going to shift over to brian he is going to talk for maybe about 15 or 20 minutes on the power and potential of analytics in auditing and he's going to give us a few whiz-bang examples of what he's doing doing in his work at grant thornton uh then we'll switch to vern for another 15 or 20 minutes on how you can develop the right skills and expertise for analytics and also vern says he's going to show us some examples of the whiz-bang projects that he is working on in his teaching work at the university of arkansas after that we'll go and work our way through q a as often as possible you can submit questions any time in the hour and please do so we'd love to have them actually in fact let me just go through a few housekeeping details here before we really formally get started the webinar is being recorded uh everybody who is listening today you are muted by default uh as i said we love questions so submit them anytime using the q a function on your screen uh and then we will get to those either at the end of the hour or if there is a particularly apt question we might drop it in and try and answer it live uh and for anybody who is curious unfortunately we are not offering continuing education credit for this particular webinar but we still plan to have a lot of discussion going on here so with that uh pranav if you are out there and you want to give us our welcoming remarks we're happy to hear what you have to say and get rolling thank you matt uh welcome to everybody uh thank you for joining us this afternoon um last fall we hosted our first webinar calc bench on goodwill and impairments and that was with our friends uh pj patel at the valuation research corporation and dan goday who's a chair tenured professor at the stern school in nyu we're unsure how the webinar will be received given you know zoom fatigue in the environment but our clients loved it and asked for more so here we are today with our second webinar and we hope that you find it as informative as the first a bit about calc bench before we get started my co-founder alex rapp and i founded uh the company believing that there were other people just like us that had used sort of the standard products in their work lives things like factset and capital iq you know who you are and uh always found something missing from those platforms so we set out to sort of fill the gaps and uh many of the things that we do and the uh questions that we ask and the answers to those questions are hidden sort of in the dark corners of the financial statements and the disclosures uh commonly you know in the vernacular we refer to those as footnotes so we we built calcbinch you know to make it easier to access all of that data in the footnotes textual numeric whatever it is and so now almost a decade later uh we're thrilled that our financial platform has uh it's helped so many people go deeper into their own financial analysis so we've got fp a users of fortune 500 companies using this portfolio managers and research analysts at big major asset management firms and that's both on the hedge fund side and the traditional space we've got auditors at top accounting firms here using this regulators and of course uh you know something that's very near and dear to our hearts we've got professors in major universities using calc ranch and you'll hear from vern about things like this later uh in in the hour um all of them you know the universities and professors incorporate their platforms our platform into their curricula their teaching you can go in and do cross-sectional analysis you can do time series analysis looking for patterns and variances on as filed information lots lots more a couple of things i'd like to point out um just for those users and for if you're new to calc bench if you're just hearing us for the first time uh you know you can go to calcbench.com and check us out we've got a two-week free trial all of those things are in there um we also have a blog that is very useful um people uh comment on it all the time it's on the lower left hand side of the of the website or you can go to calcbench.com blog to see the history of their blog posts we do things there uh you know all the way from a micro level at the firm all the way to a macro level um looking at data specifically and so recently some of the things we've done you know in a micro level we looked uh yesterday actually at the 10k that came out from solarwinds to see what what was in there about the the data breach we also published yesterday a list of companies and uh their sales to china so as a portion of their their overall revenue so these are things that over the years have gotten more and more attention from people and so we thought we'd write about them and highlight those things in addition uh calc bench has been featured in a couple of recent wall street journal articles that you might find interesting uh one on february 23rd about covet adjustments to earnings and yes we've got that data another one on february 4th which was very interesting was uh it was uh predicated by all of the the market activity uh you know around uh the gamestop stuff and it was about the risks of accelerated share repurchase programs so those are two uh very recent articles as well and um so with all of that we're really really excited to have brian wallahan and vern richardson share their insights with us today we're thrilled to have matt kelly of radical compliance moderating it and uh and and i'll just turn it over to matt now thank you very much for joining us today uh yeah thank you very much as well let me call up the slides again i had a small technical failure here but let's see ryan you will be up first as soon as the slides come up all right perfect so um i will lead off then with um just kind of a quick quick introduction to um not just what we're doing in an audit today by way of innovation but more importantly the why i think particularly for uh for our audience today having that that framework to understand why we do what we do uh in audit in particular will be very very helpful and um i'll point to the top right corner of this slide just for starters and thinking back over the last seven years as we've developed there we go um as we've developed our our latest suite of innovations since i rejoined the firm we've really been focused on audit quality sort of as our north star that's been uh our principal purpose and focus of what we've developed we set out really thinking how can we use the new tools in tech and the new data we have available today to do the highest quality audits and we began looking at full population analytics uh coin to phrase whole ledger analytics which is now sort of a pervasive uh term uh in the uh profession um and uh in doing that founder found that the the samples that we were pulling the items we were testing were more meaningful we were finding notable transactions instead of you know testing items that didn't seem to have a whole lot of purpose under a random approach and in the process of doing that and executing these high quality audits we began to find things that we never would have found before insights coming out of the data interesting here hearing micro and macro insights there i use the same phrase internally here and just published an article on that thinking of micro insights as being the insights that come out of transactional level data that we can get on an audit and then macro being some of the big picture analytics uh you know gross margin kpis industry level benchmarking using tools like calc bench and and data we get from other vendors uh some of the insights that we find at this level can relate to manual entries that maybe we can automate or a client can automate um opportunities to streamline the close process accelerate that finding you know errors in the financial reporting process even if they're not material just working out the kinks there and helping clients think through how they can streamline the financial reporting process and then efficiency as we've evolved here in the process we've found that the analytics themselves can lead to efficiencies by disaggregating populations we can we can focus our efforts and sometimes get an improvement in the sample sizes and then automations obviously are geared almost principally at efficiency where we can kind of take some of the drudgery out and automate some of the manual steps that our auditors are going through and our clients appreciate that because you know we're saving them time with the streamline process and and better sample sizes and the samples that we're coming to them with uh they're they're asking us how we found them uh because they're they're the interesting activity from the prior year so moving to the left here just a quick overview of the what when we talk about innovation we find it helpful to bucket that into automation analytics and ai and i've been on a bit of a mission to demystify some of the buzzwords in the space and artificial intelligence in my mind is really just sophisticated automated analytics so it's not self-aware computers at this point but but rather just really a really good analytic in the middle here our final point on this slide is uh the how so we think of people process and tech is driving all of this and the process and the tech i found to be a lot easier um the the difficult part for the profession has been doing audits in a new way embracing tools in tech that require a level of effort to to get to understand to deploy to interpret uh to understand the risk the risks that come with various innovations and the opportunities that come with them and um and having those be in a position where where there's a level of comfort with the regulators with those new new tools so that's kind of a quick overview of i wanted to hang this whole conversation uh on these benefits of quality insights and efficiency we can go to the next slide let's see there we go this i thought would be helpful too just to show that the analytics really aren't just occurring in the audit itself there are analytics occurring in the profiling on the front end before we even do an audit we're running predictive analytics that can identify outsized risk in a portfolio of companies and risks are referring to there are risks of restatement or mw's tear weakness bankruptcy financial reporting fraud things like that using publicly available and historical data that everybody has access to and this in this bucket here is where we would use something like you know calc bench can give us you know industry level data uh that we can use in that in that area um i know vern's gonna talk about uh almond z and bankruptcy scores we do use that and merton distance to default and operational solvency liquidity metrics a battery of different analytics on the predictive front they all have strengths and weaknesses so we like to use them together and what we call an ensemble approach so see what they say as a group and i'll highlight two there just for the this may be of interest uh we're talking here on the profiling side about inductive uh reasoning in other words we're making references or inferences about a whole group of companies and cohorts of companies that may have risk characteristics we're not deducing that any one company has a problem if that makes sense second program area there is forensic support so um after we built the risk profiling program we stood up this program to respond to some of the the flags or the risks that were being potential risks that were being surfaced by the risk profiling so this was sitting down with teams and talking through the models talking through the risk the models were seeing talking about whether or not they made sense or where they were off they are a blunt instrument in many cases and really asking the engagement team what are they seeing in the field what do they see in the transactional level data that they have that the models don't have at the profiling level and then when we land on a handful of risks there what can we do about that you know advanced journal entry testing uh creating new analytics to respond to those risks in many cases those analytics then we brought to bear and released to the entire audit practice in this third bucket which is audit data analytics this area here is predominantly using client data gl data in the analytics there are transactional level analytics in many cases i'll talk about whole ledger a bit but that's kind of our flagship analytic and that includes a transactional scoring component an account combinations analysis which is like account pairings but looking at every unique combination of debits and credits numerical and digital analytics i know vern's got a reference to benford's we have something like that that's a really an extension of the benford's concept and then text and letter analytics looking at the words that are used in a gl if you can believe that we can see trending there and changes that might merit attention on an audit so i'll pause there for any questions so far and then we'll go to the next slide brian i did have one question just looking at all of the potential capabilities here um that data analytics can provide all sorts of answers assuming people know the right questions to ask and i just wonder when you are working with various teams trying to see how you can assist them like what are the challenges in them understanding we want to answer this we want to answer this like how how well are they um in a good position to be able to ask some good questions to take advantage of analytics yeah that's a really good question and i'll put that under sort of the change management bucket the the support of analytics bucket we rolled out um you know faqs and user guides and templates and we have trainings and there's a you every time you roll out a new tool you've got to make sure that it's well articulated what the tool where it fits best best what it does how to interpret the results so that's a great question and i'll provide an 80 20 view there on some of that which is that a lot of the the the tools we roll out can be very very effective right out of the gates using them exactly as we design them but there are let's call it 20 percent of the engagement teams might want to tweak that tool and we've got flexibility in the tools allowing them to promote demote or eliminate routines that are being used in the analytics so they can customize that for their audits all right in that case i think we have one other question here i'm not sure if you have to discuss this right now but somebody's asking can brian please share his paper about micro and macro insights or at least refer us to it after the presentation so somebody there's your past work i don't know if you want to absolutely so if you go to my linkedin page i actually just put that article up there under publications it's the first one all right hey brian i was going to ask you is is this something the line auditor is going to use i mean to what extent is analytics just going to be centralized and at what point will be decentralized i assume it's centralized now but is steadily being rolled out to the line auditor it's a bit of a blend there and and i can tell you've looked ahead at the slides vern we we do cover that a little bit um i'll say something like whole ledger analytics is used on almost all of our public company audits so that's pervasive it's in the hands of our teams um some of the uh the automations that we're developing as they're coming online they're being developed in a national central role piloted on a handful of audits rolled out to uh 50 audits and then released so many of the innovations are moving through that that life cycle from a centralized development or even an engagement team development and to to a national rollout [Music] so this is the last slide on analytics i just wanted to give kind of one image here on transactional scoring what we have on the screen here is is one of the outputs of whole ledger and this is effectively risk ranking every single journal entry against several dozen proprietary routines that are designed to detect the risk of management override we have each dot is color coded by user so that the colorful dots are manual entries and the gray dots are automated entries and we plot on the x-axis a composite score based on that customization of the routines that our engagement teams apply and on the y-axis we show the impact on the income statement we can we can scale this by balance sheet impact in other ways but this is kind of the default view in looking at the activity this way we can focus our attention in the upper right corner or quadrant or a couple of percent even of a population and draw entries from there to test for management override and i've got one example here which is redacted but it's a real story an entry on the last day of the year boosting that income by 216 217 000 and the description was to balance the sub ledger so this turned out to be okay it was a cleanup entry but it highlights the kind of the kind of activity we're looking for with these tools making sure that that wasn't a you know judgmental override that was a legitimate entry and sometimes insights coming out of an entry like that why did we need this level of cleanup uh what can we do to make sure that those entries are right on the front end next slide so moving on from analytics into automation i wanted to just kind of highlight that we are really automating the entire process from acquiring data from a client to work paper generation so it starts with the acquisition or the at least access to client data and i make that distinction because some of our clients just provide access to the erp and we can hop in there and see and analyze and and do whatever we want with the data but in most cases there is a uh data provided in a cloud or transferred into our software for for analysis um so that data acquisition step is a is the the most critical and uh oftentimes time-consuming part of an analytics engine is getting the right data and then transforming it which is effectively normalizing the data if we're getting data out of several thousand erps depending on the erp and the instance they may refer to things like revenue different ways they won't all say revenue so we effectively we have scripts that will normalize all the data from all of our audits into a common language a common data model that we can then feed that normalized data into our analytics engines at that point you know once we've transformed and validated the data it's as easy as uh letting the analytics run the outputs come out our teams interact with it they sort through it they pick items to test we also here on the top and the what is that the fifth node reporting automated work paper generation i use the word report there loosely because that can be work papers it can be spreadsheets it can be visualization it concludes sort of the audit step then when we get into augmentation what we're doing is providing a feedback loop where auditors can look at the outputs there and say hey this item was flagged or this item wasn't flagged and it should have been or shouldn't have been and we can take that feedback back into the analytics engine and approve that for the next year and then benchmarking is capturing you know you've got this um you've acquired transformed and analyzed and into standardized reporting uh work from thousands of audits in an anonymized redacted way that can provide a powerful set of data for understanding risks across a portfolio of similar companies brian can i ask a question when we're looking at this slide here the extracting and the transforming and the validating what are the i guess the protocols are the the safeguards to make sure that we do have the completeness and accuracy of the data is maintained as we're validating and transforming it into something suitable for analytics but how do we make sure that we're still looking at the original apples for an apples to apples analysis yeah that's a great question so there's all of the sort of traditional you know reconciliations and cross checks and uh you know validating the number of entries is right according to what was in the erp uh making sure it ties to the trial balance all of that stuff that we've been doing for 100 years we're doing the same thing but we're doing it with scripts so we've taught the bots to do that for us instead of having to do that manually and and that's that's a real that's a real coup because that can be that can be a tough has been a tough uh part um of auditing in years past is just getting getting the data yeah anything else on that one no next slide please okay so uh the topic du jour what do our what do our auditors need to know how to do how can we prepare this next generation of auditors and and in analytics and automation and things like that um and verna i'll give a hat tip to uh to your framework here i've developed this before you and i talked and i'm seeing tremendous similarity in in this list and your framework which is much simpler and easier to remember but when i think about what our auditors need to be able to do today it's it's asking the right questions as matt said and then thinking about what data might exist to help answer those questions what analytic techniques can be brought to bear to answer questions and i put a note in here to just listen to what the data has to say i wanted to highlight that we don't have to go to the data with a question the way we used to have to do we can slice and dice the data so many different ways and see what surfaces we might see see insights coming out of that that approach that we would have never thought to ask so it is a combination now of asking questions of the data but also just listening to the data sharing the results telling stories about what the data means helping folks that maybe aren't as familiar with analytic techniques to understand what the analysis is saying coding you know writing or at least being able to edit scripts in multiple languages or one language and learning new languages i will highlight we are language agnostic coding agnostic in fact there's a lot of tools out now we think of them as coding for non-coders coding has been made so easy by uh some of the new new software packages out there now like like alteryx um communicating questions when you have an analytic looking at the results and asking the people that may have insights into that output kind of what might be going on in there and then maybe often iterating and going back to the data with new questions collaborating with people of different perspectives this is more important than ever because the world has gotten more complicated the data's gotten more complicated and uh having multiple skill sets coming to bear accounting and auditing skills and finance skills data skills automation skills analytic skills together to create innovations is really a very powerful combination that's harder to find in one person today than it was in the past managing projects with multiple dependencies i mean audit is really a project management exercise and developing new innovations is too knowledge of us gap and audit standards certainly important and i know vern's going to touch on this but i feel like there's room in the profession now for for more folks that aren't cpas and that's where i've got these skills at the bottom that i think are incredibly helpful today all under the banner of data manipulation skills exhibited via proficiency with these kinds of uh tools and i don't want the i don't want to limit the list to this what's on the page but effectively the first row is dealing with large data sets understanding how to to load and populate normalized data a second row is on automation alteryx automation anywhere etc third row here is on analytics so acl python power query etc we were talking about r this morning actually in the context of regression of vern which i think you'll appreciate and then lastly the the the visualization engines tableau power bi click there's multiple packages out there we use power bi a lot because we can publish to our whole firm in one click where we can't do that in tableau as easily but um but really any visualization engine that helps you tell stories is a plus so i'll pause there for questions on that so brian probably following up with my earlier question and maybe help me in a little bit later is this required of 10 of auditors a hundred percent fifty percent and and when do we get to steady state i mean does everyone need to know this is this a special office centralized or yeah that's a that's a great question and uh my final slide will allude to that a little bit um but i think it's a continuum um so so we can certainly work with folks that are that are you know just know their cpas and no auditing and accounting right but i think that auditing and accounting is evolving into an analytics world so picking up some proficiency across those data wrangling skills is going to be a huge plus to any auditor and conversely we can hire folks that aren't cpas but are really really good at uh at wrangling the data and building tools so what i'm showing on this screen here you reverend you asked about a national office if you look at the gray column in the middle you can have a in my case we've got a centralized innovation team um that develops tools and rolls them out to the practice if you move down one row you may have engagement teams which are are spinning up innovations most commonly for a lot of firms automations using alteryx which is like i said very intuitive and easy to use so engagement teams that aren't in the national office might be creating innovations and in many cases we can we can take those and scale them and roll them out to the whole practice and then at the bottom here in the in the center column in gray is what every engagement team is going to see they're going to see whole edge of analytics they're going to have access to automations work paper automations for revenue inventory fixed assets all the common work papers so it's really a continuum and depending on where you start and how you move through your career you can end up in any one of these three buckets for audit or tax or advisory so that's it on my slides i'll pause there for questions and kick it back to vern you know uh we have a stream of questions coming in and i think i'll try and save most of those for the q a portion at the end of this year um so anybody listening if you do have questions that are on your mind by all means type them submit them i've got a stack of five or six or seven already that are piling up vern if you want to pivot into uh your turn at the in the cockpit go right ahead great so we've gotten a good example of of the skills that are needed the analytics that are performed at various levels and i guess academ we kind of take a more basic view and just say how did we get here and so just looking at the global data sphere we see the data is increasing and we see the machines are becoming more powerful so between data availability and the changing machines we're seeing a changed world so we see the number of zettabytes increasing and continue to increase at a very quick level and the statement that accounting could be argued is no longer the de facto provider of information regarding the firm or firm performance we're not the only providers anymore there's a lot of information out there that's outside of just the accounting or the financial statements then on the right really talks about computers and this is now a pretty old site but two authors fray and osborne essentially said hey we know computerization or machines are coming what's the chance there will be job losses and and you know i'm one that doesn't believe accounting go extinct but i do believe accounting will change and so if you're an athletic trainer machines aren't going to change your job but if you're a telemarketer or accounting or an auditor 94 chance of job losses within the next two decades so that kind of sets us up these two uh changes are changing our jobs and and looking at uh this is look closely this is the aicpa's 2019 accounting graduate supply and demand report and this is trends in new bachelor's and master of accounting hired into the accounting finance function and so you can kind of see the purple line there on top is the bachelors the blue line there is the ones those with master's degrees in accounting and then it's the total non-accounting and uh and so we just see this this surge and we would argue many of us that this is data science and and on econ and other areas that are being brought in and perhaps the accounting finance function is changing or perhaps there's an emphasis on machine learning or machines computerization and data and and harnessing that data and so as we look at those we see the world is changing so some of you have been in kind of educational psychology are familiar with bloom's taxonomy and this is critical thinking skills starting the bottom from remember and if you remember kind of what we learned in high school a lot of those memorizations even undergrad is often memorization and then it goes to understand and apply and analyze but the first three levels the remember understand and apply are where machines excel that's machines can remember they can follow wrote instructions we can program it hey they're really they're better remembering than i am so we need to move up our game to the more critical thinking skills so as academics we think hey where do we need to play and where do we need to train our students or our graduates to analyze okay analytics right analyze evaluate and create that's where we need to play it doesn't mean we don't have to take intro accounting and don't have to know the debits and credits but it does mean we've got to move up the learning curve and and get quicker to those analysis skills and so those are the skills we need to develop so we also know that the role of the accountant is changing before we were the the de facto information provider if you needed information you'd go to your accountant and now frequently there with so much data out there our roles are changing uh between us and and the decision maker you know essentially we're going to be the interpreter we're going to be the go between we know what the management questions are we know what data is available we know what the data can provide we can go and speak with the analysts for the data scientists and extract and get that data but we've got to have all those skills now we're going to be writing code every day probably not but are we actually going to be you know discussing and interpreting and trying to answer questions absolutely so we see our role changing there and so uh with all these new skills needed how are we changing accounting curricula okay so i'm a county professor been working at this for 23 years or so and essentially this is the first net new course in accounting in 30 years so the classes i took in 1987 at byu are are pretty much the same courses i'd take now and uh and so how are we teaching these new skills well as of a year or so back our textbooks hadn't changed for 30 years our courses hadn't changed in 30 years the cpa exam hasn't changed much in 30 years professors don't use data in the course even excel we don't use regression we don't do classification we don't do bankruptcy prediction we don't do profiling we don't do forecasting we're really good at measuring the past but not so good at predicting the future so better forecasting we may or may not cover benford's law we may talk about xbrl go calcbench you know essentially talking about you know the data that's available out there it wasn't really covered or even the etl which stands for extract transform and load is getting access to the data do we teach sql do we know how to access big data do we use you know power query or power pivot do we know how to to access big data sets how do we deal with missing data can we even use vlookups you know those types of things really not but there's been an incredible movement in the last two years and and essentially um we are moving that direction i would say 75 percent of accounting programs now cover analytics at some level and the asacsb that's the association the accreditation body for business schools and for accounting programs essentially now requires uh covering technology and analytics the cpa exam is changing you know more details are coming out all the time covering analytics so we're on our way uh to getting there and so i'm the one that believes that every accounting course will be infiltrated with analytics every single one and um that it's coming uh and so we're starting to get there you know when you come and try to uh develop a new course first net new course in 30 years you need to provide a foundation a basis a theoretic frame of some sort and this is the easiest one i could come up with called the amps model amps model has asked the question master the data perform the analysis share the story that brian referred to earlier and essentially this is a way of thinking we have a question can we carefully specify that question can we go get the data what data is appropriate will it be financial statements will be cost accounting data what would be standards you know will it be um you know social media data will it be macro economic data stock market data what is what are we going to use then we're going to perform the analysis so we're going to do descriptive analytics diagnostic analytics predictive analytics prescriptive analytics some sort of machine learning and then how are we going to share the story how we're going to communicate the results perhaps in visualizations or otherwise and so this is kind of the foundation that we've made uh in our books that we were publishing and starting to get out there in the field in this area i really wanted to highlight maybe i'll stop there i'm not as good of brian as stopping but any questions to this point matt or or brian that would help let me take a breath i guess i well we do have a couple of people here asking about the implications for less need of human auditors and i was going to save that up for the end but i guess maybe i'll throw that out here right now and i know you would touched on that before that analytics and automation are going to touch the accounting profession but uh i mean clearly when we've got multiple people asking about less need for human auditors it's something high on the mind i don't know if you've got other additional thoughts about but all you know is is when i audited now some years back you know there was a whole you know 15 people in the room just inputting invoices and uh you know doing basic bookkeeping those are basically gone right and but it's the higher level i mean my dad would always say there's always on room on top for the best i mean there's always room for those who have the appropriate skills but we have to continue to skill up if we don't have a net new course in the last 30 years we're not really you know if this is the first one and there's going to be many more courses you know we've just got to skill up and and be able to uh to complement the computers and complement you know the data and use the data to make us better and i think there will be job changes absolutely i'm not sure about job losses except to the extent that we don't keep up our skill set so that that's probably what i would argue i'll second that that was that was well said we haven't seen a diminution in the need for for auditors we've just seen a shift in the skill sets as vern said audits are looking differently now and increasingly so in the coming years and uh we need a new a new generation of auditors that are comfortable in that environment and not just doing it the way that we did it 20 years ago you know i'd i'm reminded of a story i once heard from a cfo and an investor conference talking about a new tech system he was implementing and the vendors said well look this is going to be a great system you'll be able to do whatever you want to do with financial reporting and cfo told the audience like i realized oh crap now i have to know what i want to do and people listening today like there is a business to be able to help cfos figure out what it is they want to do when they can do anything so i'm still very bullish on there's going to be a lot of demand for these higher function skills that you both are talking about and that's more interesting matt that's a more satisfying i think activity than than you know taking and tying right thinking through the possibilities of an analytic engine and what that could mean for financial reporting risk and insights coming out of that that's fun yeah so there are a number of things we could cover uh here i just wanted to kind of point out what is in a data analytics class what what would be included and so uh benford's law which many are familiar with the distribution of the first digit you know if you took the population of countries of the world guess what conforms with benford's law as does you know many financial transactions we actually use dillard's department store data we have access to that at the university of arkansas so we've got a ton of real world data that we use and you can apply benford's lot of that i'm going to show an example here of altman z but it gets students thinking about predicting bankruptcy what has been done what does the research show and it's a great way to apply some skills i'll show that momentarily but typically in a cost accounting class you say here are the cost drivers that the company has determined hey how about we actually run aggression and we find and we discover those cost drivers so that's what we do estimating cost behavior you know those textbooks still show the high low method hey how about we try regression and look at that and estimate cost behavior looking at outliers kind of diagnostic analytics variances controls testing even the use of conditional formatting you know doing gap detection forecasting is another area that i don't think the traditional accounting program has done very well in we do really good job we have three classes four classes five classes measuring in the past and zero classes forecasting the future how about we do that excel has a forecast sheet really easy to use talk about seasonality and and rolling forward look at the persistence of sales and persistence of earnings and make that part of the common dialogue you know the goal seek function you know determining break-even levels or even today with a student they said what grade do i need to get on the final because i bombed the midterm let's look at the goal seek function and figure out what you need to get uh scenario analysis fuzzy matching you know an exercise a good good chance to talk about type 1 and type 2 errors how many you know false positives fault negatives you want to look at um hypothesis testing you know just some basic statistics some t-tests you know higher returns we we look at predicting sales returns and and looking at uh kind of with dillard's data essentially look at predicting uh sales return on holidays or weekends or or so forth and then some just some basic sensitivity analysis you know a lot of these topics are ones we've covered for a long time but now it's with different framing and so uh it's it's really a way to think about uh analytics and and how we might cover it in the classroom i know we're running out of time so i'm gonna talk very quickly but we use uh aldman z uh bankruptcy prediction now very old but still very applicable and aldman nyu there uh in 1969 said hey we found five factors that predict bankruptcy and so what would i do is i give the students you know 100 line items out of calc bench or something or and essentially say hey let's compute fight the five factors let's apply weight uh to those factors based on the regressions that altman ran let's put them into buckets or histograms and then predict bankruptcy status and then based on the score we could put them in one of the zones in the distressed zone gray zone or safe zone so that would be an example of a lab and that lab we do 50 or so of those each semester of those labs predicting all sorts of things or prescriptive analytics or so forth so that's an example of what we might do an example in in in one of my textbooks so uh i think that uh is where we're at uh and so i'll open up for questions we have a bunch of questions man if you want to go back two slides i had a comment there i i couldn't resist making here go back one more so if you if you look here at benford's law i wanted to highlight that benford's law applies really really well for naturally occurring number sequences in nature but may not apply really well for a company uh if the company sells everything for 9.99 for example you're going to get a lot of nines and that you won't see in a naturally occurring number sequence so when i mentioned derivations of benford's what we're doing there is we're using the company's own data from prior years to create a benford's context to normalize the current year benford's against the company's own data that's cool yeah and that's great expectation that's diagnostic analytics setting an expectation and then seeing departure so excellent the second point i'll make vernon is i told you just a few minutes ago that we were looking at a regression this morning you'll be pleased to hear that that was we were predicting operating costs and instead of deciding in advance what we thought would drive that we ran a random forest model that highlighted the top five drivers of op x so a great real life example for your story there yeah excellent thank you so much really quickly very just one second um all of this the the great thing about all of these techniques is it requires the input of world-class you know data to to do to do these things and with with the advent of xbrl and you know tools like calcbench the input is just there now all you need to do is to consider you know is the stuff coming in clean enough to run the analysis and all you know years ago it was always about oh do i have enough data to run it it's not about that anymore it's it's it's kind of uh you know you're just like we are it's sort of like you know being in a candy store as a kid it's good stuff yeah we call that the democratization of data you get it clean and organized and give people tools they can explore yep so here's the first question we have from a listener is the innovation we're talking about essentially a u.s story right now because xbrl and data tagging are only starting to be mandatory in the eu and other jurisdictions and it has been here for the better part of a decade um brian i'll let you feel that first and vern if you have any thoughts i'm eager to hear them but brian what do you think to that sure when we talk about sort of the macro data and the risk profiling and things like that we do certainly have uh globally a leadership position in in uh and well organized homogenous data which is a a real privilege in the u.s i wouldn't say that the profiling opportunities are limited to the u.s but it's certainly uh there's certainly more data here that's accessible and clean um then when you look at the micro data and you know everybody's got a gl so you can use some of our you know whole ledger analytics and other engagement level analytics anywhere in the world you do run into nuances with language if you're using text analytics understanding characters versus letters and things like that but um but yeah it's a continuum uh certainly not limited to the us but it's this is definitely uh it's easier here with the with the with the macro data at least uh vern here's another question that i think might be more up your alley somebody is asking do you see a difference in the skill sets needed for assurance and audit students versus valuation projects and evaluation track for students we we don't really differentiate it that way obviously a different set i mean i would guess auditors should be more diagnostic analytics and i would guess valuation would be more predictive and prescriptive analytics would be their expectation uh you know i would guess more forecasting be needed in valuation than than in audit we don't really treat them that way we say we think you need to know it all at least at this point we know we're early on i mean we're writing first and second editions of textbooks that we know will change dramatically in the future and i think we'll be able to specialize but what's happening now in the textbook world is the analytics are starting to be spread to the the individual topic areas the the cost accountings the financial statement analysis the audit class instead of just an overall course you know funny enough we just have a comment from a listener here saying i think the solution is including data analytics across the curriculum creating new courses changing the way these courses are taught and presented which seems like it's right at the end and i just have a very quick question uh comment is the the the roadblock is professors it's professors that have taught the same course for 30 years same using the same powerpoint deck because you go teach analytics and you tell someone they're going to teach tableau or power bi and they're not familiar with that it's not going to happen and so it it it's it's just slowly getting there and and we have to have all the curriculum materials available and easy for professors to adopt yeah uh brian here's another question that i think might be well suited for you uh how do you see tax people using data analytics sure yeah it's a good question um it's it's a similar blend i think of of analytics and automation um i would say automation is probably a little bit heavier than analytics relative to what we see in audit um taxes is even more rules-based than than audit certainly judgment areas there that the tools can direct our professionals to to make judgments there but i think it's a lot of automation okay uh here's another question that might be suitable for you both is the goal of data analytic data analytics eventually to replace traditional audit sampling techniques how do you see traditional sampling and data analytics working together i don't know who wants to take that first vern maybe you what do you think sure you know uh obviously there's opportunity to sample the pool not i guess not even sampled but evaluate the full population but you still want to do it risk-based you're not going to be able to look in detail at every every observation and so a lot of times you want to do some data reduction you want to do kind of what grant thornton was just showing you know basically where you're highlighting the riskiest largest material you know uh transactions that have occurred and go back and out of those with a finer brush but certainly you know you're going to have your continuous monitoring continuous auditing scenarios which will kick out errors and then you you figure out you know which ones to evaluate further so this is a great a great question and certainly a a topic that i think regulators across the world are thinking through um when the question they're referred to testing 100 of populations in many cases we're analyzing 100 percent we may not be doing what would actually qualify as a test of details um so then uh if it's not a test of details where does it fit uh in in the audit framework and it may be risk assessment as vern said it may be a substantive analytic it may be that even after a really good sap we have to test a couple as a sample right so it right now i think it's probably a blend but i could see a future state where the analytics receive a full evidentiary credit in the literature well along those lines let me also ask um you had mentioned a solid grounding in uh financial reporting standards and an audit standards to pursue this uh career i'm just curious where is the pcaob in developing reliable standards for data analytics because this does seem like a point where the technology is ahead of the regulatory framework to get auditing done um i don't know if you have any thoughts on that uh brian or vern if you do but like how mature are the standards right now to accommodate all of this sure um i'll i'll speak to that because the the pcob's data technology task force was convened by the pcaob to help think through these issues yeah so they've been really good about about uh asking questions here and seeing what the state of play is and where we could be headed as i alluded to a minute ago everything we're doing today can fit within the standards i just think that the standards could evolve so that the some of these advanced techniques would receive more credit than they currently do yeah so and the pcob are really clear and they're correct that they're not uh you know prohibiting any of this new tech uh it's just a just a question of where does it fit and how much credit do you get yeah vern do you have any thoughts no i i uh it's just interesting to figure out how they're going to stand and you know it it doesn't appear to be ignorance it appears to be you know trying to figure out exactly what it can do and how to set a standard on analytics uh is an open question [Music] somebody else asking how far away are we from a future that has live audits since we're moving rapidly into live data streaming and analytics like why couldn't we just do continuous auditing like that or live auditing um i don't know which one of you wants to take a stab at that first i'll i'll take a crack at it and then turn over to vern um i would say we aren't far away from that capability i i often say that we can run the analytics that i talked about the automations as often as we can get access to the data if we wanted to do that quarterly monthly weekly daily hourly real time that can be done it becomes a question in my mind of what do investors really want what would be the value in in having access to audited data at that level at that time interval right so i think that's really more of a uh investor demand question than it is a technological question yeah yeah it seems like the technology there is there it's you know for me it's the the possibility of litigation you know an error is made and and it goes out and uh you know is there going to be a safe harbor or you know so is the demand heavier than the possible error in the litigation that might follow all right um i think then in the interest of time we are probably up against it here i don't know if maybe i'll give each of you uh like one more minute or so just in closing thoughts about what you would recommend to either students about what they should be trying to do for a career in analytics or brian if you want to talk about uh the the potential here and how how much more there is that we could mine but uh vern if you want to have a minute for closing thoughts and then i'll give it to brian too and then i think we'll wrap up sure data and machines are changing accounting and uh i i think we're up for a change we haven't had even bigger than sarvang's oxley and bigger than enron and worldcom you know i think it's just a dramatic change for us all and i think we're going to get there i think universities are up for the challenge we're just typically slow and i'm not sure kovat has sped us up or slowed us down but uh the curricular materials are going to be there and some of it is just the ability of professors to deliver because i can tell you the students want it and we will continue to go forward and if we don't figure it out in accounting the best students will go to another field so let's figure it out all right brian what's uh the last word there yeah i'll echo that uh things are changing and and i would encourage folks to be a part of the change help to to shape that change rather than let the change shape you uh get involved and be a part of that evolution it's a really i think making auditing more exciting than it's ever been so it's a really neat time to to get into the profession and you can peel off out of or back into audit in so many different ways now you mentioned valuation forensic work uh different kinds of advisory applications of the same analytics um i see a world of possibilities for new professionals to do a lot of different things starting with you know a career in audit or coming into a career in audit i think it's easier now than ever to uh to really to have a tapestry of skills in your toolbox than it was 20 years ago where maybe you learned how to audit and that's all you ever did well in that case um thank you everybody who's been listening we do still have a bunch of questions we didn't get to we'll try and ship those off to vern and brian after the fact and maybe they can follow up but vern richardson of the university of arkansas brian wallahan of grant thornton and pranav guy of calpinch thank you all for helping us out this has been a great hour i really appreciated everybody's time thanks for having me thank you we appreciate it bye everybody thank you thank you

Channel: Calcbench Inc.

Views: 5,055

Keywords:

Id: Y_MCocMOfFo

Length: 57min 32sec (3452 seconds)

Published: Thu Mar 04 2021