CW 1041 - Insider Sign for Housing Market, Labor Shortages & Science Behind Automated Valuation Models with Dr Andy Krause, Part 1

I do very much credit you with getting me involved in real estate investing. I had tried to do a flip originally that went sideways and I ended up having a condo that I had rented out after a bad flip. And I realized that being a landlord wasn't all that tough. And eventually I found you on the web and that really got me into real estate investing. So thank you for that.

Welcome to the Creating wealth show with Jason Hartman. You're about to learn a new slant on investing, exciting techniques and fresh new approaches to the world's most historically proven asset class that will enable you to create more wealth and freedom than you ever thought possible. Jason is a genuine, self made multimillionaire who's actually been there and done it. He's a successful investor, lender, developer and entrepreneur who's owned properties in 11 states, had hundreds of tenants, and been involved in thousands of real estate transactions. This program will help you follow in Jason's footsteps on the road to your financial Independence Day. You really can do it. And now, here's your host, Jason Hartman with the complete solution for real estate investors.

Welcome Creating Wealth Nation. This is your host, Jason Hartman, and this is episode 1041. 1041. Thank you so much for joining me today as I am being chauffeured. I've been driving most of the day, but right now I am being chauffeured by our client and venture alliance member who you heard on the podcast before. That is Carmen. Carmen, say hello.

Yes, hello.

Okay, so we are going. We've been driving from San Diego up to Long beach to see a little bit of my childhood roots there in Long Beach. We are headed up to Manhattan beach and then, oh, I don't know, maybe Santa Barbara, Ojai. I've had a pretty long vacation this summer. But you know, look, I never really take vacation and I never really want to because I love, I love to work. I have a huge passion for it. Do you know, after I spent a month in Europe and talked to you many times from Europe, whether we were on the yacht in Croatia or we were on our traveling around Italy, Positano, Capri.

Where else did we go? Naples, etc. Oh, and what's that place called? You know, where the ruins are?

Pompeii.

Pompeii, that's right. Yes. For some reason I couldn't think of it. Thank you. Yes. And I thought, well, I did talk about Pompeii on the podcast.

Yes, you have talked about it.

You said I forgot to talk about Pompeii. Anyway, in traveling all Around. Do you know that while I was traveling around Europe, I hired about a dozen people remotely on that trip and. And got a whole bunch of things done. I just refinanced one of my apartment complexes, the one I own with our client Steve, and we just did a $2.3 million refinance on that unit. And I did that from Kansas City, signed the docs there. And, you know, it's amazing how you can just be geographically independent nowadays. And it's pretty neat. It's not without its hassles and its cost. I'm certainly not saying that it is as productive as sitting at a desk all day. You know, that can be a lot more productive. But, hey, you got to get out and enjoy life and see the world a little bit. But Carmen had an interesting remark. It was about vacations. Now, you stayed in Europe a little longer than me. How long were you there?

I was there for a month.

You were there for a month? Okay, so I was there like four days short of a month, I think. Okay, and so you stayed a little longer. But while you were still there, I remember we were communicating back and forth, and you said something to me that I thought was. It was an interesting statement. And it was about how you were kind of sick of vacation. And now we're still on vacation. You know, we're driving up the California coast. We might even make it up to the Hearst Castle. And hey, Hearst and Pulitzer, those are the two icons who really, in large part save the world. Why? Because they were the first, at least the first sort of known people that were into investigative journalism, investigative reporting. And I tell you, the media should not be a lap dog. The media needs to be a watchdog. And that's what we intend to do on this show. And our other media outlets, you know, tell you about scams and things that are hurting people. You know, that's what we do. And it's not without great cost. I mean, hey, my troll is one of those great costs. And I know he's out there listening because he keeps commenting on my show. So thank you for subscribing. Appreciate it. Glad you're listening. So the interesting thing about vacation. Carmen, tell the listeners what you said. It was pretty cool thing.

Well, let me give you a little background. So I spent a month in Europe. First we were in Croatia, which was beautiful. We visited, visited all the islands. Then after that we went to the Amalfi coast.

Those Croatian islands are pretty awesome. And Croatia is gorgeous. And having our own boat that we could go wherever we wanted. Was a pretty neat thing, too.

Yeah. So after that, we went to the Amalfi coast in Italy. I mean, that's just beautiful. It's probably one of the most beautiful places in the world. After that, I headed. I was headed to Spain. I spent a week and a half there. And then after that I went to Portugal.

So this is the part where we split up. So you did half the trip with me, half the trip without me?

Yeah. So when I was in Portugal, I have no idea what I was going to find. And Porto was just another beautiful place. Great food, great wine, and all of a sudden I found myself sitting in a bus stop. And that's when I left you that message.

And I said, now wait, what were you doing at a bus stop when you were staying in, like $400 a night hotels? I mean, you sent me a picture of one of your hotels. It was gorgeous.

I was just sitting there. I wasn't waiting for the bus.

Okay. Because, you know, I know what you're like. You probably have a limo escorting you around Europe.

No, I was.

I just high maintenance.

No, that's not correct. Anyway, I just stopped there and that's when I left you a voicemail. And I said, listen, I think I have a little bit of a vacation saturation.

Vacation saturation. That is the phrase of the month. Vacation saturation. And what else did you say after that? It was pretty funny.

Well, I mean, you know, I had seen so many beautiful cities and so many beautiful churches and eaten so much good food. I mean, there's a point where you.

Just had enough, you know, And I agree with that, by the way. Look, at the point of this show is to help everybody become financially independent. And financial independence is something I achieved way back in 2005. I certainly don't have to work, but I love to work. I mean, it's, you know, look, it. We are meant to do things as people. We are meant to contribute. We are meant to have a purpose in life. You know, who wants to sit around and lay on a beach or vacation all the time? It's really purposeless. I mean, that's why even when I'm on, quote, unquote, on vacation, you know, I'm always working because I just. I can't get enough of it. It's just. It's just very fulfilling. I. I just. I don't know, it's. Maybe it's not always balanced or healthy, but I don't know. Whatever. That's where I am now. That's the stage I'm in. But hey, let me. Before we get to our guest today, we are going to. This is going to be a fascinating subject, by the way, we're going to talk today with a guest who is a specialist in avm. Avm. What does that mean, you ask? Automated Valuation modeling. Automated Valuation modeling. So when you go to Zillow or Rent O Meter or Rent Range or any of these websites that give you valuation estimates, there's a thing in the real estate industry called a BPO that stands for broker price opinion. And they've largely automated these things. You know, they used to. When I was in traditional real estate, I remember banks and various organizations used to send me BPO requests. And, you know, they'd send you 50 bucks and you're 100 bucks, and they'd say, do a BPO on this property. And you know, you really, you didn't do it for the 50 or 100 bucks. You did it. So maybe they would list the property with you or list some of their other properties with you for sale. And this stuff is becoming largely automated. Everything is becoming largely automated. And so this is a fascinating discussion today with our guest. Make sure you notice that we are many times producing more than three episodes per week. So last week we had four episodes. The week before that we had four episodes. And this week we probably will as well because sometimes we have to split these into two segments because the show runs too long. So always keep an eye out for that. We definitely produce episodes for you every Monday, Wednesday, Friday, but occasionally we'll throw in an extra one or even two per week. Okay, so automated valuation modeling. We'll talk about that. We'll talk about what's behind it, how accurate it is, and I think you'll really enjoy that. I also wanted to give you a reminder, be sure to subscribe to our propertycast. I'm getting astoundingly good feedback on that one. Everybody seems to love it. So you just, on whatever podcast platform you're using, look up Jason Hartman, PropertyCast, like podcast, but instead of a podcast, it's a property cast. And you can get the latest and greatest properties right in your RSS feed on your podcast app. So be sure you take advantage of that. And also if you are an Alexa user, if you have the Alexa in your home, be sure you download our Alexa skill for my daily real estate update. And that can be part of your Flash briefing or your news briefing every day on Alexa. So check that out as well. Okay, so I want to share with you a sign. Now, these are. This is an insider Sign on the direction of the market. And I'm not sure how to interpret it, but I'm just going to share it with you. So personally, in terms of my own money, I have approximately $2 million in loans out there on pieces of real estate. And these are all hard money loans. In other words, they are short term loans with fairly high interest rates. And I'm mostly loaning money. And this is just my own money. It's not, you know, I don't get the money anywhere. It's just literally out of my bank account. Okay. I loan my own money to a lot of times our local market specialists so that they can acquire more properties and they can have more inventory for, for you to buy and invest in. This is a sign, okay? And it's an insider sign that you will not hear anywhere else unless you're in the business. But I thought this was meaningful and I thought I should share it with you. Here's the sign. And then let's talk a little bit about how we might interpret it. And I don't have a conclusion on this. I just want to tell you, before I tell you the sign, what's going on and how it might be interpreted one of two ways. It really could go either way. The sign is that these loans that I make are typically six month loans. They're six month in duration and I will be happy to extend them. And many times if they want to extend and have another six months or any portion thereof, they pay a little extra extension fee. The borrower does. And the borrower again is buying a property, doing a rehab and then selling it ultimately to a retail buyer or an investor. And they might put it in the MLS and sell it to a regular home buyer or they might sell it to an investor. Either way, and what is happening, I've noticed with my own portfolio, this is just my own money that I've loaned out there about approximately 2 million bucks. Currently.

The loans are paying back more slowly than they should. I'm getting extension requests. A lot of them are going over six months. That is a little bit of a concern. It's not too terribly much of a concern. And I'll tell you why. I actually don't want them to pay me back very quickly because the interest rates are good and I want to keep the money in play. I want to keep it loaned out because hey, if they pay me back, then I'm not earning interest from the loan. So that's one thing. I mean, I ultimately of course want to be paid back. I don't want them to default on the loan, but I'd like them to keep the money. Okay. And just pay interest because that's obviously the investment. But they are getting slow. The paybacks are late. And again, I don't much care about this because I want to keep the money in play. But it is a sign. Now the question is, what is the sign? How do we interpret the sign? I think one way you could interpret is you could say, oh, God, the market is slowing down. It's the sign of a recession. They're not able to unload these properties, and that's why they haven't paid me back on the loan. And that's why they're going overtime and then why they need extensions. Okay, that could be one way to interpret it. It could be the sign of a legitimate real estate market slowdown. That's one sign. Pregnant pause. Intentional. Okay, what is the other possible interpretation? And there are many, but I'm just sharing two of them with you. The other interpretation is that the market is booming at such a crazy pace that it is so difficult to source labor and materials that their rehabs are taking longer and they are having to carry the properties longer. And that is costing them. The rehabbers, the ultimately local market specialists that you might be buying properties from. It is causing them to hold the properties longer. And if they hold the properties longer, they have not only an opportunity cost, they have higher management costs because when they own the property, they got to manage the construction and manage the. The rehabilitation of that property and manage the sales process of that property and the rent up and, you know, the lease up of the property and so forth. And also the labor is very scarce. There is a huge labor shortage right now in the construction field. These trades are very, very busy. And they have raised their prices a lot. Because, look, the oldest thing in free market economics is raise the price until the demand falls off, okay? And that's what you do. You keep raising it. You raise it. You raise it until the demand sloughs off, and then you don't want to slough off too much because you want the work, obviously, and you stop raising the price. This is true with rentals. Also, as you're an investor, and some investors, kind of stupidly not thinking about it, will brag about things like, oh, you know, I have a 2% vacancy rate or a 0% vacancy rate on my rental properties. Well, okay, okay. You sure that's good, right? It's not good. Okay, you want some vacancy rate? We perform a 8%, about one month per year. So a 92% occupancy rate, 8% vacancy rate. Why do you want some vacancy rate? You might think, oh, that's bad. Well, no, you want some vacancy rate because if you don't have any vacancy rate, it means your rents are too low. Raise the rent, you know? Remember that guy? The rent is too damn high. Remember that funny guy? He was funny. Carmen, did you think that was funny?

Yes.

That's all we get here. Yes. Okay, comment. Now she's giggling. So that's the thing. So the same happens with the construction trades, right? All these contractors that are rehabbing the properties that many of you are buying. And so that labor is scarce and it's making the project take longer. And materials are also scarce. There's shortages of materials. So I just wanted to share that with you from my very own portfolio of loans that I have out there. Of course, I prefer owning the actual real estate to doing the loans, but I do both because the funding the loans with my own money helps increase our inventory. So we have more properties for you to buy. So it's kind of a little ecosystem and that's one of the major reasons I do it. Okay. So just wanted to share that with you. Be sure to join us in Hawaii. And by the way, we are selling tickets at a nice pace for Hawaii. I'm feeling that event might sell out. At first I was a little worried, but we've also got. I am now looking at various bands and we've got some good proposals from bands. We're going to have some fun there. We're going to do probably a luau. We're working on planning that. So we'll have some good community building and networking where you can network with, with your fellow investors and so forth. And it should just be a great time for you. So you can register for that@jason hartman.com. and what else do I want to tell you? Property Cast, Alexa app, Hawaii event. I think that should do it. Hey, let's go to our guest and talk about automated valuation models. And let's understand what's behind a zestimate. Okay, here we go.

It's my pleasure to welcome Dr. Andy Kraus. He is principal data scientist at Greenfield Advisors and AVM Analytics. Andy, welcome. How are you?

I am well, thank you, Jason.

Good. Good to have you on the show. And we were talking a little bit offline here before we started about automated valuation models. That's what AVM stands for. This is the bots and the algorithms are taking over the world, aren't they?

They certainly are. It's invading all aspects of our lives. So I think that it's happening in real estate is no surprise.

Yeah, Well, I think the biggest AVM that everybody listening will be familiar with is Zillow, you know, the zestimate. Right. Whether it be a rent zestimate or a sale price zestimate. And everybody's got their opinions on how accurate they are, how inaccurate they are, and, you know, I'd say they're both. What I like to tell our listeners is, look, you know, take it with a grain of salt. It's better than what we used to have, and we used to have Nothing. So these AVMs, they're a guide, but they're not the Holy Grail yet, are they? Each house is so different. I remember I looked at on Zillow recently at one of my old homes that I owned in Orange County, California, and I sold it for, I think, gosh, what I sell for, like, $911,000. The property was pretty unique. It had a pretty unique lot, unique view, not similar to the rest of the homes in the neighborhood. And I think that's where models really struggles when you get a really unique property. Kind of take us through the science of this and the accuracy of it.

Sure, sure. I think the example you just gave is really illustrative in terms of how and when AVMs are good and when they're not. So in terms of the science of it, it is, after all, a model. And for any model avm, or any sort of mathematical model that's out there, it's only as good as the data that goes into it. So when you do run against a unique property like that, the model is going to struggle. And in terms of the science of it, I think people often get the impression that there are these massive kind of black boxes that are doing some sort of strange magic underneath it. And to some extent, that's true. The formulas can be quite complex, and the terms and the models are things that a lot of people have never heard of. But at the end of the day, it's really trying to mimic what what an appraiser does. It's really trying to look at what have sales in the local neighborhood done recently, what are the trends there, how do various factors influence value, and how do we convert all that into a prediction? So at the heart of it, it's not much different than what your local appraiser's doing. It's just doing it at scale.

And certainly in fairness to AVMs, we should say that appraisers are. They make mistakes all the time.

They have their own accuracy problems, don't they?

Right. And I think what often gets left out of the conversation is that we don't know how wrong appraisers are. We don't ever actually do tests to see. We don't do, hey, let's have seven appraisers appraise this home and then see what it sells for two weeks later. We don't do that.

Well, you know, that is a real. I'm so glad you said that, because I want to sort of couch that a slightly different way. Wouldn't it be nice if there was a database? And I don't think there is one. I may be wrong, but I really doubt there is one of appraised, final appraised values on properties versus actual sales prices within, say, a month or something like that. So you knew it was the same time. Because remember, folks, the ultimate appraisal on any property is a meeting of the minds of between a ready, willing and able buyer and a ready, willing and able seller. When those two minds meet and the contract is signed, that is the best appraisal in the world, bar none. Everything less than that is an educated guess, isn't it?

Absolutely. And I think that's very well put. Everything short of the two people meeting really is just one person's estimate. Or in this case, a model's estimate.

Yeah, absolutely. So to go back to what you just said, which is really fascinating, and the fact that there is, I guess, no database of appraised value versus sales prices, I always like to talk about how you can't hear the dogs that don't bark. And like you said, there is no test to test the accuracy of appraisers because there's no database. I'm guessing there needs to be a database like this, right?

Yeah. Yeah. And I think there are two things. Stopping it. And then I'll maybe talk about how it might be a reality. I think that the first thing stopping is obviously price. I think a lot of people would love to do these sort of tests. But in order to have seven, let's say we're having seven appraisers do a home, okay? So that's a couple thousand dollars for one home. And if you actually want to come up with some statistically meaningful estimate of how, if there's appraisal error and how much you need to do that thousands and thousands of times. So there's just kind of a cost and time prohibition to it. And not only that, you need a house that's probably hasn't been listed yet. That truly is sort of off the market to do this to and then have it turn around and sell quite quickly. You just have market dynamics that are kind of preventing that as well. Two things that make that really difficult. Now, that's not to say it'll never happen. I mean, what I think we see in the industry of AVMs and real estate data more in general is like a lot of places there's consolidation and many of the mortgage big appraisal management corporations, AMCs are slowly being bought up. CoreLogic has purchased some of those companies. Some of the other big players are looking at that as well. I don't know the rationale, but there's a lot of data that comes with those appraisal management companies.

Right, right. And that could lead to what I just hope for that database if you do see a roll up. Now, I hate the big corporatocracy and the monopolistic or duopoly type of thing, but it does have some benefits too. And that is you can get really clean good data when it's not so fragmented. Right. And so something like all these buyouts with CoreLogic and so forth, that could lead to some better data to make better APMs. Right?

Yeah, absolutely. That is the other side of that coin. As consumers, sometimes we can suffer from monopolistic movements, but with it, sometimes some.

Good stuff comes out of it too. It's a mixed bag.

Yeah, absolutely, absolutely.

Okay, go ahead with what you're saying though.

Yeah, so I think just with the database, I think as a researcher and a developer in this space, I would love that database. There's many of us who would absolutely love to have that sort of data, but it just really doesn't exist at the moment. So when it hopefully does, it'll be interesting to see how well appraisers do. I suspect that they do quite well in many places.

Yeah, absolutely.

I don't think appraisers should fear that. I think that they should relish that because they have the local knowledge that the models often don't.

Right, right. Use it as a tool rather than fearing it. Okay, so how do these AVMs work? I mean, take us through kind of some of the details. I mean, probably everybody sort of has a simplistic view. I do. At least I'll speak for myself. You know, hey, look, there's a computer database. You've got all these prices in there. You've got maybe certain remarks in the remarks section of the MLS listing. Got bedroom and bath count square Footage. What do they really do behind the scenes to come up with these valuations?

I think you hit on one of the most important things, and I think I mentioned this earlier, is really the data that goes into it. A lot of the bigger firms that are doing AVMs, you've got to have a really solid database behind it. It's a combination of your local tax assessor who has information on the home size, all that biased as they may be, combining that with MLS data, that was sort of, if you look at AVMs a decade ago, that was kind of where the data stopped. What we see moving forward is this movement into, as you hinted at, text analysis, kind of the natural language processing of the remarks. You also see people starting to use image data. So whether it's an aerial image or Google street view, trying to just slowly gather more information. And the two pieces of information that are always the hardest for AVMs to measure are condition and quality. It's really hard to know what's inside a house, and it's really hard to understand the quality of that home without having an inspection done or, you know, all the things that an appraiser does. So AVM companies have been really working hard to gather information on condition and quality, and that's through the photos and through the listing information. So all the improvements, you see that, you know, Zillow is listing other errors going down, and House Canary says, you know, they've got an error around 3% and these people are publishing their error rates. A lot of those improvements are made by image processing and looking at the natural language processing of the comments.

Okay, so they process the natural language, but what else happens really? What is the software doing to come up with a value?

Sure. So in sort of a nutshell, it's trying to find comps. I mean, that's maybe the easiest way to do it. It's not in the traditional way that an appraiser says, oh, here's the three most common, most common mathematical models within an AVM are going to look at somewhere, let's call it 10 to 100, maybe even more. But, you know, they're looking at a bit of a larger set of comps because statistical models need more than three observations to work properly. So in some ways they're just trying to find what homes are similar to this home, what did they sell for, and how do we adjust for differences. It's really that simple. If you really look at an adjustment grid in an appraisal, you kind of follow that same logic through. The ABMs are just using A number of different statistical techniques to get to that same underlying logic.

Okay, good. Do you want to tell us anything about the possible flaws with these models? And, and then maybe some use cases. We talked about that a little bit off air and we've seen the residential single family home real estate business become much more institutionalized since the Great Recession. You know, this is the AVM thing is I think a critical component of making that possible. So maybe you want to speak to that a little bit.

So your first question, what are some possible flaws? The first thing that we all say in the modeling industry is garbage in, garbage out. So if the data in your area is not particularly accurate or has a bias in one direction or another, the model's going to suffer for that. So it really is a question of having good data. It raises a lot of problems actually in the non disclosure states because actually getting the true sales price can be a bit difficult. So you know, if you ever wonder.

Why, by the, by the way, speak to that if you would. How many non disclosure states are there? I think this has to end. This is one of the problems with real estate around the world. You know, we've had many clients ask us, why don't we talk about international investing. And those markets are so opaque, it's ridiculous. You just, it's very hard to understand the value because the data is just not there. The US is a very special market in that way. But there are still some places where you have these non disclosure areas, which areas, if you know, are non disclosure. And gosh, there's got to be a movement to stop that. We've got to make this stuff public so everybody can really know value. But go ahead.

Yeah, I think the number's not exact because some states are sort of on the border. I think there's some somewhere around 15 to 17 non disclosure states, mostly in the mountain west and southwest, south central U.S. so Texas, Oklahoma, Utah, places like that. I don't know the full list offhand, but it's generally focused there. Your data is pretty good on the coasts and in the upper Midwest. Most of that is all publicly released. But in those nondisclosure states, as you hinted at, the sale prices is not a matter of public record. And to get that, you need to do a number of ways to kind of derive what you think the best estimate of that sale price was. So you can see how not having that exact value can lead to higher errors in your automated valuation models.

Okay, so how does the algorithm work to try and get data in non disclosure markets? Now see, even though the public record may be non disclosure, there are still other records, there's still multiple listing services, there are still various websites that do have data. So it's not completely opaque, is it?

No, no, it is exactly what you said. You're looking at the MLS and depending on where you are, sometimes upwards of 90% of single family residences go through an MLS. So if you have access to that MLS as a data aggregator or an AVM builder, you've got certainly some good information there. There are other public records, such as when the mortgages are recorded, you can assume some loan to value ratios based on the way the mortgage has been recorded.

I'm glad you mentioned that. I wanted to speak to that a little bit, but don't forget your train of thought. Okay, so you know, look at if the typical mortgage is 90% loan to value or 80% loan to value, the software can estimate what the purchase price was because it knows the recording amount of the mortgage. So if the mortgage is $80,000, you know, hey, look, that home probably sold for 90 or 100,000, right?

Correct, correct. That's something that you then triangulate. You'd say, okay, I had that property address. What did I find in the MLS? Oh, do I have an MLS sale in there? $400,000. Yeah, I've got a match here. This looks pretty good, you know, or oh, this was a for sale by owner property. Maybe there's a little less weight I need to put on this one. Or maybe we won't consider this sale. So there's some sort of internal algorithms or scoring that goes into the data to judge the quality of that imputation or estimate of price. But just kind of circling back around to something you said. I do really think there should be a movement to make that public information and I think it needs to start with real estate professionals in those states who say, look, our industry in this state is suffering. We're potentially losing out on investors. The market is a lot more opaque than it is in one state over. You know, why can't we have this information?

This will be continued on the next episode. Thank you for listening and happy investing.

Thank you so much for listening. Please be sure to subscribe so that you don't miss any episodes. Be sure to check out the show's specific website and our general website, hartmanmedia.com for appropriate disclaimers and terms of service. Remember that guest opinions are their own. And if you require specific legal or tax advice or advice in any other specialized area, please consult an appropriate professional and we also very much appreciate you reviewing the show. Please go to itunes or Stitcher Radio or whatever platform you're using and write a review for the show. We would very much appreciate that. And be sure to make it official and subscribe so you do not miss any episodes. We we look forward to seeing you on the next episode.