11-23-2020 Hot Topic: How To Shave 20.6 Days From Loan Cycle Time With Tom Showalter

Facebook Twitter LinkedIn

In this episode of Lykken on Lending, David interviews a special guest for the Hot Topic segment with the aim of having a discussion on how Tom Showalter, Founder & CEO of Candor FinTech has applied Aerospace Technology experience to shave 20.6 days from the mortgage manufacturing process.

Quick background of Tom Showalter, Founder & CEO of Candor FinTech

Tom has been in the industry for quite some time, having held key executive roles at industry notables, including Digital Risk, Core Logic, and First American.

Tom has enjoyed performing in a mixture of disciplines: sales, market development, product management, analytics systems design and development, database design and development and other Big Data and Big Analytics experiences.

His focus has been on the financial services space with a special focus on credit card and mortgage industries, both in the primary and the secondary markets.

Included in his background is a tour at NASA, where he developed a variety of aerospace technologies for use in civilian and military aircraft, as well as the former Space Shuttle program. So, we are especially interested in hearing how a Rocket Scientist ended up helping solve some of our industry’s tough challenges.

Talking Points:

Why breaking underwriting into tasks is a doomed strategy
Understanding the true bottleneck in mortgage manufacturing
How Aerospace Technology is being applied to mortgage lending

—

Hot Topic: How To Shave 20.6 Days From Loan Cycle Time With Tom Showalter

Our Hot Topic is Tom Showalter coming on. He’s the Founder and CEO of Candor FinTech. I’m so excited to have him. This is the fourth in a four-part series of episodes dealing with technology, what we’re doing, and helping underwriters underwrite more loans. We’re certainly paying them a lot more money and many would say it’s well deserving. We’re not going to argue with that, but now that you’re making the extra money, we did find a way to make them as effective and get the best return out of our investments.

We’ve been doing a series of episodes talking about underwriting technology. We’ve had some great episodes. We had Joe Tyrell lead off the first week. The second week was Brent Chandler of FormFree and he did a great job, then we had Frank Poiesz for Black Knight, and we’ve got Tom Showalter. I’m very excited to have him on. Tom was a rocket scientist in the Aerospace technology industry. We’re going to be having some fun with that a little bit here, especially with Frank’s comment, “This isn’t rocket science. It’s worse.” Anyway, more on that. Stay tuned to the second half.

We’re also proud to be a part of IndustrySyndicate.com. Check it out. As well as Mortgage Media. Two great places where you can read other great episodes. Again, we promote shows everywhere. We’re also pleased with our sponsors. Thank you, sponsors, especially the Mortgage Bankers Association of America. On September 14th, we have Marina Walsh talking about the profitability numbers and what’s going on there. That was a great episode.

Also, a special thank you to Finastra’s Fusion Mortgagebot Solution. It’s just so powerful. With Steve Hoke on August 24th, 2020 talking about the things that Finastra is doing, especially with data. Lenders One, we had Justin Demola back on June 1st, 2020. Also, we’ve got the Mortgage Collaborative. We’ve got a pre-recorded episode coming up here with Tom Gallucci. Mortgage Collaborative will be with us. Also, the Community Mortgage Lenders Association. We also had Michael Jones on September 21st, 2020. He did a great job. Go back to that episode.

Also, Indecomm. I’m so grateful for them as sponsors. We had Linda Bomar. I’m grateful to have all of our sponsors here with us. Thank you very much. Also, a special thank you goes out to Insellerate. Thank you to Josh Friend. He was on August 17th, 2020. Go back to that episode. Also, Ainsworth Advisors. Read the updated show you’ll find on the Ainsworth Advisors’ website. September’s 7th episode, we had a good discussion about that all.

Also, AI Assist. Read Roman Vinfield’s episode that we did on July 15th, 2020 about how you can use AI to artificially go out and connect with borrowers. It’s on the marketing side, so it’s very good stuff. Celebrity Home Loans, as well as Innovient. Check out what Ted Kramer and his team is doing for some of the top mortgage companies in getting the best pricing and being competitive in the marketplace. Also, Knowledge Coop, MobilityRE, Modex, Velma, Vendor Surf, and Vidyard are so grateful to have all of them. We’re also so grateful to have Alice, Allen, and Matt here on each and every episode of the show. We’re going to get right into the Hot Topic segment now.

—

It’s so good to have you joining us for the Hot Topic segment on November 23rd, 2020. It’s good to have as our special guest a special, interesting guy. He’s a rocket scientist. Before we go and introduce you to Tom Showalter, I want to give a shout-out to Brandy Young, who reached out to me when she heard we were doing these segments.

Again, this is the 4th in a 4-part series on underwriting and underwriting technology, what’s going on in the FinTech world to help underwriters. We’re paying them more money, as I said at the beginning of the show. We need to get more productivity out of them. Underwriters should be happy about this and get back to what they used to do and the volumes they used to do. We need technology to help them. Brandy was reading that and reached out to me. She said, “Dave, you got to have Tom on.” I’ve listened to Tom speak at conferences, so it’s a real honor to have joining us Tom Showalter, Founder and CEO of Candor FinTech, a true rocket scientist. Tom, good to have you here.

Thank you for having us.

It is a real pleasure. You also have a notable career. You are a digital risk core logic first American. You’ve got a notable career already in the mortgage industry and beyond that. I’m particularly interested in the comment that Frank said. He was working with a rocket scientist. For those of you who say, “This is not rocket science,” implying that it’s not as bad, he says, “It’s worse. It’s even more difficult than rocket scientists.” Tom, it’s so good to have on the show. We could go on about the background for people to check out and get to know you.

You are an aliquant speaker. You speak a lot of conferences, and we’re honored to have you here. In the past, some interesting discussions around what you think is a real bottleneck in underwriting and how to solve that. Can you bring some of the real improvements to lending? Talk a little bit about the bottleneck and what you’ve been seeing. You’re an expert. We want to get your perspective on it.

Thank you so much for having us, and this is an excellent question. When we were founding Candor, we wanted to target something that had true economic significance and what has process value for the average mortgage maker. We identify opportunities in the back-office space. This is where the bottlenecks were occurring. We took a hard look at the bottleneck in the back-office space. We found out, based on our analysis, it was the dependence upon the critical thinking of the underwriter. That was the bottleneck.

Loan Cycle Time: We really had to model the critical thinking of an underwriter. We must teach a computer to think critically and we must teach a machine to react as an underwriter would. This was not easy as you might guess.

You couldn’t get around. You couldn’t move along downstream. You couldn’t get it across the goal line unless you applied the underwriter’s critical thinking to the loan. It’s reliance upon critical thinking that is a true bottleneck to creating a fast and efficient underwriting process. It’s our assertion to revolutionize the economics of lending. We had to model the critical thinking of an underwriter. We must teach a computer to think critically, and we must teach a machine to react as an underwriter would.

This was not easy, as you might guess. When we got into it and took a hard look at it, we realized that some loans require a little critical thought to fund and others require a great deal, but all loans require some. Therefore, we had to have a path to successful underwriting and funding alone. We found that to be volatile and prone to change. This new technology had to have some dynamic nature to it.

As new data arrives, the complexion of the loan will change and as new investor guidelines are encountered, complexion also changes. It’s the job of the underwriter to assess and reassess the loan in this dynamic fast, changing environment and to adapt accordingly in order to successfully underwrite and fund a high-quality loan. This is no small task. No two loans are alike. Each loan is unique. The path to a successful underwriter is complex and dynamic. That is why underwriters are so well paid.

Until now, only a bright, adaptable, insightful, and experienced person could perform the underwriting successfully. We believe Candor has changed that. At Candor, we have a successfully modeled underwriter through life and institutionalized that in a computer with what’s called expert system technology. This was not easy, as I mentioned. Candor has absorbed over 2 years in its design, another 3 years in development, and at least 18 months in customer development and testing. Back to your previous guest comment about QA, we did a lot of it.

Now, Candor runs over 1 million lines of code. In the months of August, September, October, and November 2020, Candor has processed over 70,000 live loans. The economic benefit to our clients has been enormous, and we partnered with some great lenders. These folks have done their share of the heavy lifting. Cycle time from application to funding had been reduced by 20.6 days for every loan that Candor cuts. Pull-through rates have increased 18.1%. Borrow satisfaction has increased by 11% and over 50% of these loans, over 36,000 of them, absorbed no underwriter labor at all. Our clients were amazed and we were grateful for the opportunity. That’s the story of the bottleneck.

That’s so interesting. Also, what I understand about AI is it’s constantly learning. The fact that you’ve run 70,000 live loans and you’ve had 36,789 that does not require any underwriter labor, that is pretty amazing. At the core of this is this thing you reference called expert system. How does this replicate thinking? How can we have confidence in it? It sounds pretty futuristic.

It is futuristic in a way, but expert systems have been around for some time. What is an expert system’s job? It models the thinking of a subject matter expert. It does this using a methodology called knowledge engineering which depicts the expert’s thinking to best represent how he or she organizes and analyzes a problem.

An expert system models the thinking of a subject matter expert. It does this using a methodology called knowledge engineering. Click To Tweet

This thinking cannot be modeled by simple if-then statements. It’s more complex than that. I have an example of a medical diagnostic system. It gives you a feel for how expert system technology employs its critical thinking to diagnose a patient. A medical diagnostic expert system has modeled the thought life of a doctor who was a trained diagnostician. It contains two sources of information. This is part of modeling how the expert thinks about the problem.

One source of information is the library of diseases and their symptoms. It also contains a second source of information, which is data to represent the patient’s symptoms. As you might imagine, this expert system must adapt constantly to new information. Patient symptoms change all the time, and therefore, it must adapt. When the software achieves a state with the symptoms are consistent with the specific disease, the system is reached to diagnosis and this is not unlike what underwriting mortgages look like.

When Candor models the thought life of an underwriter, it considers resources of information, not just two. In Candor, the underwriter thought life is represented by the interaction of the 1003 data. That’s the loan application, and Candor requires this information by integrating directly with your LOS. The second source of information is third-party data about borrowing the property, EG pay stubs, appraisals from every information, and so forth.

Candor gets this information from many sources, OCR key documents, and XML data feeds from third-party data vendors for FormFree again. The third source is the investor guidelines, the representation of the type of loan, the investor robotic. These guidelines are resident in Candor’s database. They are supplies by Candor’s clients, and we assess their investor guidelines. These three sources of information interact.

The interaction of these three sources, when it’s completely consistent, meaning the data about the borrower, the data the borrower has represented as an official statement in the 1003, and the requirements in the investor guidelines. When those three are consistent, you have a loan that can be underwritten, and it’s ready to close. However, the path to that level of consistency is very hard. It’s complex, volatile and often fraught with inconsistencies. They are often unique to that loan. Candor had to invent a new technology to address this phenomenon. It’s called CogniTech.

Now that’s another new word we got. I want to go back to the guidelines thing because that is always a big challenge for so many people. When Ellie Mae bought AllRegs, how does your system access the underwriting guidelines?

If you were a client and you signed up with Candor, we’d ask you for the list of investors that you’d like, so your loans, too. We would get from that list of investors the contact points so that we can get a download of all the relevant guidelines. We also get put on the update methodology so that when investor number 3 issues the change to guideline number 37, we’re informed and then we can update the Candor database with the new guy.

You touched on CogniTech. I love that term. What does it do? How is it different than, say, artificial under AU?

CogniTech is the technology that enables Candor to adapt in real-time to a dynamic and complex problem. Contact enables Candor to respond quickly to new information with the most appropriate next step. As an example, when CogniTech determines that the income represented by a pay stub is not consistent with the income stated on the 1003, it launches a process to resolve this inconsistency.

It may, in this case, ask the letter to correct the income on the 1003. However, if the correction then leaves the borrower unable to meet the investor’s income requirements, it may ask for additional sources of income. If they’re not forthcoming, Candor will suspend the loan. However, if the borrower can supply another source of income, Candor can then clear the condition and move on.

What CogniTech is doing is taking a look at all the inbound information and reassessing the state of the loan. Is this something new and different? If so, what’s the next best step that Candor must take in order to fully underwrite this loan? That’s the dynamic part of CongiTech. Now, in terms of Candor, it fundamentally differs from an AUS. The AUS only uses a subset of the information process by Candor.

An AUS focuses upon 1003 information in the investor guidelines namely Fannie and Freddie. It ignores the data from third-party data providers. The AUS has no means to do that and validates the information from third-party sources. As a result, the AUS has no logical capability to issue clear conditions with any authority. By using only two legs of the stool, an AUS consistently presents a list of conditions that are inconsistent with the third-party data. Hence, the need for Candor will issue clear conditions in order to identify and resolve any inconsistencies across the three sources of information.

This is a form of intelligence that is unique to Candor. It was immensely difficult to build, very complex. Only an expert system that incorporates CongiTech can conduct a complete underwriting, given the nature of an underwriting problem, which is dynamic, fast-paced, changing new data all the time. That’s CogniTech. I hope it makes sense.

It does. It’s fascinating. How does aerospace technology play a role in something like this?

We were selected in the airspace technology, but we chose some stuff that’s very relevant. When we were challenging ourselves to come up with a way to automate the underwriting process and the critical thinking, we realized that we could automate the critical thinking unless we also found a way to make sure that we were making, in effect, the defect-free loan every time and that’s hard to do.

Loan Cycle Time: When we were challenging ourselves to come up with a way to automate the underwriting process, we realized that we could automate the critical thinking.

There are so many potential defects that are potential in this, and that’s the part that’s got to be unnerving.

One of the things that we found out was that airspace technology has a methodology. It is a technology designed to make the effects free stuff. If you think about what an aerospace system has to do, for example, in a commercial airliner, there are over 1 million interacting parts that perform perfectly every time. That’s not an accident. The airspace industry had to develop a science to make sure that the systems were all performing flawlessly all the time. Lives depend on it and if you’re flying commercially, your life depends on it.

What we did we took the science of system safety engineering from the airspace world and brought it over to Candor, where its job is to make defect-free loans. The same technology that would land aircraft safely will now land your loan safely, but that’s what it does. It sits there, and it enables you to sit there and determine what the quality of the loans going to be as you’re making it. You no longer have to wait. You no longer have to sit there and make the loan then subject to a manual QC and determine what the quality was. You will know what the quality is at the time you make it.

The same technology that would land aircraft safely will now land your loan safely. Click To Tweet

In that process, Candor has gone through a number of corroborations. For example, we check names and address across multiple sources. We check net pay on the pay stub and cross-reference it with a C on the bar bank statement. We detect and close self-employment. We also corroborate and asset balance listed on channel three, etc. We do this with every loan. In fact, probably the number of cross-checks and validations that Candor performed in every loan is an excess of 10,000. It’s very rigorous.

This is something only a system could do. You couldn’t ask an underwriter to do 10,000 cross-checks but you can ask a machine to do that, especially if all the data are there. Candor is so confident in this approach to defect estimation and resolution that it ensures its work with a loan defect policy from a major underwriter. There’s a lot of work that went into making sure that when we say the loan has an extremely low rate of defect that, we can prove.

You put in insurance policy behind it. That’s very impressive. You look at the bigger picture. A lender’s dream could be coming true with what you’re saying. It sounds very promising. Can you give me a bigger picture of the road map for Candor? What are some of the things you have coming up?

First of all, the road map, we’re rolling out the productivity tool, the thing we’ve been talking about that models the underwriter’s critical thought life and institutionalized that. The machine uses that to automate a significant portion of the underwriting problem. As we’re doing that, we also have Candor. Candor is designed to write all its information to a blockchain. Therefore, that provides this indelible record of the underwriting. What Candor has been able to do is catalog data about how the loan was made. How many income sources are needed? How long did it take to verify? How much of a challenge was it to get a particular loan funded and so forth? All of that is in the Candor’s database.

What we offer our clients going forward is not only a productivity tool but a data source about how they manufacture the loan so that they can use that to improve their manufacturing processes. They love the part where they can go back and see empirically how long something took, the different nature of the underwrite and how all of that has been compiled, analyzed, and verified. They have all of that in the Candor database. That’s one of the services we offer as they go home.

As we go further with the data part, we also get into what the airspace folks love, and that’s the analytics part. What we’ve done Is introduce aerospace precision to the mortgage space in terms of mortgage manufacturing. We leveraged in analytics to predict future outcomes in real-time. I mentioned about long quality. We do that now, but funding and pull-through will be able to start predicting what loans will be funded and how the likelihood of approved pull-through has affected and then month-end financials. We will be able to start talking about midway through the month, what the end of month going to look like, why, how many of these loans are going to fund, and what profit margins you’re looking at. We go from the productivity tool to the data to the analytics here as we evolve.

I’m going to toss the mic to Alice and get a couple of questions from you.

Thanks, Dave. My question was something that you had mentioned earlier. Is this borrower the same thing the way that you described it earlier?

You’re spot on. First of all, presently, it’s underwriter-facing or loan processor-facing. Perhaps, even loan officer-facing, but the structure can be the borrower and some of our clients want to move that direction. That’s all true.

Allen, you know the space so well. Tom, we both listened to him speak numerous times. Any questions you have for him?

If we had another hour, I could have some good questions, but the one that sticks in my mind now, Tom, is there are so many bottlenecks in the process. Even when people use technology in underwriting, there are still bottlenecks. Maybe just talk about the bottlenecks. How do you eliminate bottlenecks?

I’ll take you back to an interview that we did with one of our customers when we talked to them about what they wanted the underwriting system to do and what philosophy they wanted to pursue. There were two paths. Path number one was, we wanted to present the underwriter with all the analyses, all the data, all the conditioning, and everything that the engine was deciding and get the underwriter to review and approve. In a very detailed level, the underwriter had to understand what was going on and get involved with all the presentations going forward. Also, all the data and all the analysis as the system went about rendering a recommended approach. Let’s call that path A.

Path Bravo was the system does all of this in the background and the underwriter sits there as a surveyor of the progress on a loan and looks at the outcomes, namely, the conditions that are still outstanding to clear that the underwriter has to address. Those that had been cleared by the machine and therefore, crossing in the machine that has done the proper analysis and so forth so that they could move their loan forward faster. They wanted plan B.

The reason was very simple. She said, “If I have my underwriters sitting there, looking at all the data and analyses, and rethinking all the things the system is done, I’m not going to remove any labor. I’m not going to accelerate the thinking, nor have I changed the underwriter’s job to be a higher-level job. He still has the same regular job he had before.” This executive’s vision was to identify and rely upon the machine and then send the information to the appropriate person in the process. If any approval and review was necessary, they could sit there and exercise their authority fully informed about our specific items.

The bottlenecks were, A) Do you want to present the underwriter with huge amounts of information that they then have to review or B) Do you want to present them with an overview of a process and how it’s been processing concluded and get their endorsement going forward? Plan B was the one that we followed. Plan B is also the one that resulted in the tremendous economic benefit you heard about at the start of my presentation.

Allen, anything else?

I want to commend you on having those analytics and statistics up front. You’re a rocket scientist because he said Bravo instead of A and B. Those are real statistics that you’re getting from real loans. Do you have a target? Where would you like to be knowing we can only do so much because other things need to evolve? You can’t process all digital data unless the lenders using a digital one. Where are you hoping? Are you hoping to get to 50%? What are you looking at?

First of all, we have a very robust OCR, Optical Character Recognition component to handle PDFs and other kinds of documents that are scanned and faxed in. We have that dimension of our data gathering. As well as any form of XML where we get something like a form to prevent statements in over the transom and it goes directly with the day four. We have those two.

First, let me comment on something. The economic benefit that we enjoyed that I described at the outset, not that we weren’t expecting it but we were not expecting to have that big an impact so early in our life. We’re probably at year 3 of the economic impact we were open to creating, which speaks to 2 things. First of all, the viability of our solution and secondly, the quality of our initial customers. These folks have been great because they figured out how to use it. They figured out how to employ this capability called Candor and to use it in very innovative ways.

It’s been a two-pronged effort. The economic benefits that we’re enjoying now, we weren’t expecting to be able to offer people for another couple of years. The fact that it’s here now is as much a credit to our clients as it is, so we thank them. The next part is where we expect to go. We have the normal road map things like we now do conventional conforming. We also have that FHA program ready to go. After that, we’ll do VA jumbo and QM.

We also had people lined up clients to go test and verify all of that as we roll it out. You’ll get that Q1 and Q2 of 2021. Those are the rudimentary things. You got to do them. That’s the business. As we go forward, we’re also looking to improve the data arm of our business model and the analytics arm of our business model. We’re also seeking to give our customers more of a facility so that they can figure out better how to improve their manufacturing process.

For example, in Candor, there’s something called a smart code. That tells you for every condition what a source was, what it means, and why you need it. Some of our clients are starting to take the smart codes and build a matrix that says, “If we see this smart code, we got to send it to this person because this is the most efficient and appropriate person to process this problem for a loan.”

Getting to that smart code-driven manufacturing process which will again, decrease the cycle time, increase the productivity, and increase the throughput, all of those things will happen. That’s why our next major innovation is to go from the level of economic impact to a smart code-driven impact that will have even more impact, we believe.

Loan Cycle Time: Our next major innovation is to go from the current level of economic impact to a smart code driven impact.

Give us a little bit of idea of the customers you have. Don’t name anybody if you don’t want to, but you’ve been around for a while. It’s not new.

We have a very robust marketing effort. We are boarding approximately three clients a week. Planning to increase that to 4 to 5 a week, and then eventually eight. Our boarding process takes no more than 30 days. We’ve managed to figure out how to make the presentation of Candor and the joining of it with a lender’s process to be very efficient, easy, and straightforward. We’re able to do a board now in 30 days and eventually probably down to 10 or 15 days per client. We’re looking forward to that. That’s where we are with.

Is there anything else that we should be thinking about as we wrap this up?

The only thing I would offer is that this is not my first expert system. It’s more like my fourth. We didn’t have to do a dramatic amount of rewriting, re-engineering, and redesign work. We just didn’t. That’s not the norm with the expert system. That’s the unusual past. We’ve been very fortunate in being able to get a robust solution out quickly without having to rewrite it and redesign it. That’s been one of the things that we’ve benefited from.

Another thing we hit on something when we talked about underwriting, the underwriter’s critical thinking. Modeling that has been our target. That seems to have released the economic engine here and unleashed it, enabling it to perform. Candor’s a very unique piece of software. It has many components, and you have to have them all, the expert system. You need to have knowledge engineering, CogniTech, and the system safety engineering. All those things are needed to give a robust solution.

If you don’t, if you shortchange any of those with RS, you probably have a problem. Those are so few thoughts that I have, but again, we’re very fortunate that our initial vision was upheld. We didn’t have to redo it, which you can imagine not only the challenge of having to do that but then going to the investors are saying, “It didn’t work as well as we thought. We need more money. Please send the checks here.” That tends to be not a very fun conversation. We’ve avoided that one as well. We’ve been very fortunate and we have great clients. They’ve been a great contributor to me, making us successful so far.

You’ve done a great job. I love the forward-thinking that you have. I’m not sure but I’m hearing other people doing it. One of the things I did hear that you talked about is blockchain. Brent Chandler at FormFree was talking about how blockchain is going to play a bigger part in it. Kudos to you. I wish you continued success, Tom. You’re doing an outstanding job. We desperately need this. Let us know if there are any new developments as you continue along the way. We’d love to have you back on again. Kudos to Brandy for calling me and making sure we got you included in this series of episodes. I’m grateful for your time. Thank you.

Thank you very much.

You bet. We’ve had as our special guest Tom Showalter, Founder and CEO of Candor. You have to check out this system. If you have not taken a look and considered how it can help you and your underwriters become more productive, be sure to check out their website. In the next episode, we’re going to have Cyndi Danko on with Fannie Mae. I’m very excited. We had this interview with Cyndi, and it was good. You’re going to enjoy what Fannie Mae has done.

The thought that’s gone into the forbearance program, everything related to how to respond to the pandemic. It’s been interesting. Special thank you to our sponsors Finastra, Community Mortgage Lenders Association, Indecomm, Insellerate, Ainsworth Advisors, Mobility, and Modex, all these companies. Check out all of our sponsors on our sponsorship page. Thank you so much. I look forward to having you back here.

11-23-2020 Hot Topic: How To Shave 20.6 Days From Loan Cycle Time With Tom Showalter

Hot Topic: How To Shave 20.6 Days From Loan Cycle Time With Tom Showalter

Important Links