Spatial Finance: How Satellite Data Is Enhancing Financial Models, with Jorge Rojas
In this episode of Financial Modeler’s Corner, Paul Barnhurst sits down with Jorge Rojas, finance professor, consultant, and Advanced Financial Modeler, to explore how emerging technologies are reshaping financial modeling. Jorge shares insights from his latest research on spatial finance and Earth observation data, explaining how satellite imagery, AI, and alternative data sources are transforming forecasting, investment analysis, ESG reporting, and risk management.
Jorge Rojas is a finance professor, consultant, CFA Charterholder, and Advanced Financial Modeler (AFM) with a PhD in Finance from Tulane University, where he also serves as an Adjunct Professor. He has taught finance across the United States, Europe, Asia, and Latin America, and is the author of several books on corporate finance, risk management, and financial modeling.
Expect to Learn:
How satellite data can improve financial forecasting
Why Monte Carlo simulation is useful for decision-making
Practical applications of spatial finance
Tips for building reliable financial models
Advice for developing your financial modeling skills
Here are a few quotes from the episode:
"Most useful financial models should be forward-looking and based on current information." - Jorge Rojas
"Build your model as automated as possible. It takes more time upfront, but it will save you time later." - Jorge Rojas
Jorge Rojas shares practical insights on building stronger financial models by combining sound modeling principles with modern data sources and risk analysis techniques. Whether you're new to financial modeling or an experienced professional, this episode offers valuable lessons to help you make better-informed financial decisions.
Follow Jorge:
Linkedin: https://www.linkedin.com/in/jorgerojasphdcfa/
Company: https://orcid.org/0009-0002-4136-7995
Follow Financial Modeler's Corner:
LinkedIn: https://www.linkedin.com/company/financial-modeler-s-corner/
Newsletter - Subscribe on LinkedIn: https://www.linkedin.com/build-relation/newsletter-follow?entityUrn=7079020077076905984 Sign up for the Advanced Financial Modeler Accreditation today and get 15% off using the code “Podcast” at https://bit.ly/497oAqW
In Today’s Episode:
[00:00] – Trailer
[03:08] – Financial Modeling Horror Story
[04:35] – Why Jorge Enjoys Teaching
[07:38] – CFA & AFM Certifications
[12:18] – Introduction to Spatial Finance
[17:01] – Satellite Data in Finance
[25:41] – Monte Carlo Simulation
[32:20] – Modeling Agricultural Risk
[37:57] – Rapid Fire Questions
[45:29] – Advice for Financial Modelers
Full Show Transcript:
Guest: Jorge Rojas (00:00):
Most useful models are or should be forward-looking, real time. Current satellite data is unbiased. You cannot cheat a satellite around Earth. You cannot greenwash it. You cannot. I don't know if you have heard about the Hawthorn effect that mentions that people react differently when they know they're being observed. Additionally, if you have satellite data, it's a much better input to models as compared to historical data that could be biased manipulated or with too much of a lie, it could be delayed.
Host: Paul Barnhurst (00:33):
Financial Modeler's Corner is the world's premierModeling podcast. It is brought to you by the FinancialModeling Institute, the world's leading financialModeling accreditation organisation. Welcome to another episode of FinancialModelers Corner. I'm your host, Paul Barnhurst, AKA, the FP&A guy. This podcast is where we talk all about the art and science of financialModeling with distinguished financialModelers from around the globe. The FinancialModelers Corner podcast is brought to you by the FinancialModeling Institute. FMI offers the most respected accreditations in financialModeling and that's why I completed the Advanced financialModeler this week. I'm thrilled to welcome our guest to the show. Jorge Jorge Rojas, welcome to the show.
Guest: Jorge Rojas (01:28):
Thank you for having me, Paul. So
Host: Paul Barnhurst (01:30):
Let me give a little bit of Jorge's background and then we'll jump into our questions. So Jorge has a PhD in finance. He's earned his CFA and his AFM. His PhD is from Tulane University where he currently is an adjunct professor of finance and international programmes. In fact, we'll talk about one of the articles he recently published. He also has his CFA and he is a member of the Society in Mexico. He earned his advanced financialModeler and he's also a member of the Financial andModeling Institute's Global Leadership Council. He's been a lecturer at several different universities all over the world, including the US, Spain, China, Chile, Peru, Mexico, Panama, and Guatemala. He's authored several books in corporate finance, capital budgeting, risk management, and derivatives. He's published them in English, Spanish and Mandarin. He has been working with Excel and financial models for many years developing custom models for financial valuation in many industries such as agribusiness, real estate and land development, food and energy production, and others. He commonly uses Monte Carlo simulation in most of his models and his latest article, which we'll talk more about earth observation data for finance is a Biblio Metric review and Directions for Future Research on Earth Observation and was recently published in the Journal of Economic Surveys. Love the background, Jorge. Great background.
Guest: Jorge Rojas (03:07):
Thank you.
Host: Paul Barnhurst (03:08):
First question I'd like to ask you, everybody has that horror story. What's your worstModeling story ever dealt with?
Guest: Jorge Rojas (03:15):
Well, I think I have, it'll be fair to say that I have a recurring financialModeling in nightmare where students or customers send me their Excel model and what I do is I have an analogy of how you sometimes need to untangle that Christmas tree light series that looks like a big ball of wires and looks like a tumbleweed. That's exactly what I think when I get some of these Excels. Probably one of the worst I have to deal with is this customer who sends me an urgent model for a loan application and it's the perfect example of the antiModeler. It has all of the worst practices, no inputs, all sales hardcodes, the values everywhere, no formulas. Each scenario was a different sheet and it was a mess and no units, no texts, and those horrible data islands all around links to external war books and as usual, it was urgent. So sure enough, I was better off throwing it away and starting from scratch, which I did and eventually their loan was approved, but unfortunately it's very common for me to get models like that.
Host: Paul Barnhurst (04:23):
I hear you. It is not uncommon. It's amazing how messy it can get when somebody just thinks of excel as a playground and brings no structure.
Guest: Jorge Rojas (04:34):
Yes, exactly.
Host: Paul Barnhurst (04:35):
We've all seen that. So I'm curious, you've spent a lot of your career teaching about financialModeling. What is it you love about teaching? Why teaching?
Guest: Jorge Rojas (04:44):
I see teaching as helping people cross the bridge. A bridge that takes you from something that you do not know to something that you want to know and eventually learn. When you help someone transform in a positive way, they're better prepared to achieve their goals and that gives you satisfaction. And also what at the end of the day, what I want to do is provide them with a friendly and enjoyable journey in crossing that bridge in their learning journey. I like sharing my knowledge, that's what I like doing and I try to explain in a manner that I wish it had been explained to me when I learn it. Easy, digestible, non-complex way of explaining. Also teaching a subject forces you to really master that subject and that's why today I'm also a CFA and an FMI prep provider, which is really satisfying when the topic is really a tool that you will use every day. So overall it's an experience that I enjoy as we all have heard you need to do. What makes you happy, what you enjoy and what you are good at.
Host: Paul Barnhurst (05:48):
Love that. I heard a couple things there first, obviously enjoy it and that's one of the most important things. Nobody wants to go through life doing something that they hate. I've seen people like that and no thanks, I'll pass. And then the others. You mentioned you wish you had a teacher like you being able to teach students the way you wish someone had taught you. Incredibly valuable to have teachers that want to do that. We've all had the teachers where I used to joke in one of my classes, me and another person would turn to each other and it's like I feel dumber for having come to class today. It was always so complex and never explained in a way I could understand it. I always walked out going, what's wrong with me?
Guest: Jorge Rojas (06:29):
Yeah, just for example, it's like a convenience deal type of benefit where you can look at their faces and know if they're enjoying, if they're learning and if they come up to you and say, George, I was really inspired today. I'm not really well versed in finance, but you really made it easy and enjoyable and I'm really enthusiastic about this topic. So that has a lot of value for me.
Host: Paul Barnhurst (07:02):
I hear you. I love when a student, when I do training, they come to you and they say, oh, this will save me hours, or Thank you for teaching me this. Those are the moments that make it worth it versus the times when you wonder if anyone's listening, which we all have.
Guest: Jorge Rojas (07:18):
Yeah, exactly.
Host: Paul Barnhurst (07:19):
So I'm curious, you are both your CFA designation. Yeah, a hundred percent agree. You can tell no question. The worst is when I'm teaching remotely, nobody's on camera. You could tell they're paying no attention. You're just like, okay, when is this done?
Guest: Jorge Rojas (07:36):
Exactly, exactly. Yeah.
Host: Paul Barnhurst (07:38):
So I'm curious, you've earned your CFA designation and your advanced financialModeler, so you did both. What was your motivation for earning them?
Guest: Jorge Rojas (07:46):
Well initially my motivation for the CFA programme actually was mainly academic. I was then the director of the master of finance programme and I wanted to elevate it by making it CFA recognised as it was called then. But not too long after that, my encouragement was extended well beyond academics as I saw its potential in investments and consulting areas. And I would say that my main incentive for a FM was very specific. Through the years I had observed mainly through consulting projects that many people were good and they were bragging about how they would project p and ls and budgets, but very few people could project a balance sheet that actually balances five years into the future. That's not easy and that's one of the things that caught my attention about A FM overall. I also have an engineering background, so I'm very comfortable with numbers with Excel C-F-A-F-M. So I found it to be a perfect compliment for me both in the classroom and as a financial consultant since I use Excel in most of my day-to-day activities. So that's my motivation for both of them.
Host: Paul Barnhurst (09:06):
It makes a lot of sense and I really like what you said on the A FM of the balance sheet balancing that was the part I worried about the most. Most of my models have been p and l driven and when I first finished the test, they didn't balance. Fortunately, I found out why before time ran out.
Guest: Jorge Rojas (09:21):
And if I may ask Paul, what's your main motivation for a FM?
Host: Paul Barnhurst (09:25):
So originally when I first decided to take it, I wanted to work for some smaller companies and I knew I needed to do three statementModeling. Most of my career has been with large companies, so American Express where you're not building the balance sheet and the cashflow, you'reModeling the p and l and maybe you're providing a few key items from the balance sheet or other places. So that was my main motivation. I had originally signed up for it before COVID hit COVID hit it got cancelled and then I ended up staying at the job I was at and I delayed it and delayed it and then I started my own business, so I delayed it some more. I started the podcast and FMI sponsored the podcast and one of the agreements was, if you're going to run the podcast, you need to have your A FM. So I finally finished it, I had started, I signed up for it in 2020 I think, and I finally took it in 2024.
Guest: Jorge Rojas (10:14):
Yeah, I recall. I like your post about that. It's very satisfying. Yeah,
Host: Paul Barnhurst (10:19):
It is. And then I signed up to do the CFM and I ended up with some health challenges last year, so it got put on hold, but it's still on my list to do because I think it's just incredibly valuable to have a really good base forModeling.
Guest: Jorge Rojas (10:34):
Yes, for sure.
Host: Paul Barnhurst (10:35):
So yeah, I love that question. So that was my motivation. I'm curious, any advice you would offer for someone studying for either of these? For CFA or FMI?
Guest: Jorge Rojas (10:44):
Yeah, both the CFA and the A FM designations are outstanding and they do open professional doors that likely are more difficult to open. Otherwise. I use a very simple analogy. I haven't gone recently, but at Disney they used to have this fast pass where you could get faster into places and in a good sense, this will open doors for you and make processes and overall interactions smoother. If you're in either programme, just keep in mind that the rewards are well worth the effort. It's going to be tough, especially the CFA. It's more extensive and you need to finish the three levels, but if you can get them both, wow, I use the analogy of epoxy glue, epoxy glue. When you mix two components and the result, it's a super strong cement. That's how I see if you can get both designations. See the result will be just about as strong as powerful and as you can imagine, yeah,
Host: Paul Barnhurst (11:51):
I love the example. Great advice there. I think they're both great programmes. I actually started the CFA, I didn't finish it as I decided I'm really spending my time in corporate finance, and so I decided to focus elsewhere, but I did level one and I agree with you. You definitely have to put aside a lot of time. There's a lot of study that goes in. Yes,
Guest: Jorge Rojas (12:09):
But the satisfaction is just, and the results and the professional leverage that it gives you, it's well worth the effort.
Host: Paul Barnhurst (12:18):
Yeah, a lot of great benefits that can come into it. So I want to switch gears here and dig into spatial finance and earth observation. So you recently wrote an article on this topic, which you shared with me. I found it really fascinating. Can you start by telling our audience what spatial finance is? Earth observation. Let's start there.
Guest: Jorge Rojas (12:42):
First and foremost, I would like to mention and credit my co-author, professor Christian Pinto. He's from the University of Talca in Chile, who was actually the lead author of this article. I just want to mention that. So satellite data has been around for decades, both mostly for weather forecasting and climate models. We will know about that, but in the last few years, earth observation or special finance has been growing significantly and now further driven by AI making it one of the fastest growing alternative data sources for finance in general and also for financialModeling. What special finance, special finance refers to the incorporation of geospatial data and analysis into financial theory and practise basically is the convergence of earth observation ai, cloud computer with mainstream finance.
(13:38):
Its importance in financialModeling I believe is that most useful models are or should be forward looking, real time current satellite data is unbiased. You cannot cheat a satellite around earth. You cannot greenwash it, you cannot, I don't know if you have heard about the Hawthorn effect that mentions that people react differently, behave differently when they know they're being observed. And additionally, if you have satellite data, it's a much better input to models as compared to historical data that could be biassed, manipulated, or with too much of a lie, it could be delayed. I believe in a few years as analysts have today a Bloomberg terminal, there will be some sort of geospatial data available in different forms and sources. So basically geospatial data means data that you get from satellites and you can somehow process it to help you in financial decision analysis. That's in a nutshell what it's about.
Host: Paul Barnhurst (14:50):
It's funny, funny when you first mentioned it, I'd never thought of it because I'd never thought of having geospatial data for a model. But the more I think about it, I could see how it could be incredibly valuable if you take over a month's period seeing every day how many cars or what the traffic footprint is for a mall or a location or so many different things like that. Okay, what's the climate and how's that going to impact us? Building something here at a level you're not going to get without that data. So really I was fascinated by the article and just the whole idea when you brought it up. I
Guest: Jorge Rojas (15:26):
Think it will change and evolve exponentially in the next few years. And if you have AI now, it's going to be even stronger. I think it's going to be one of the main alternative data sources for a financial analyst in the near future.
Host: Paul Barnhurst (15:42):
You shared the example of having a terminal or having access to the right cloud where you can go out and just search for those type of satellite images. I'm sure the companies will be more than happy to offer subscription service to their images. I'm sure they're out there now and it'll just continue to grow.
Guest: Jorge Rojas (15:58):
And that's already happening, by the way. Yeah, yeah.
Host: Paul Barnhurst (16:01):
Well, my background is in fp and a. I am also passionate about financialModeling. Like many financialModelers, I was self-taught. Then I discovered the FinancialModeling Institute, the organisation that offers the advanced financialModeler programme. I am a proud holder of the A FM. Preparing for the A FM exam made me a betterModeler. If you want to improve yourModeling skills, I recommend the A FM programme podcast listeners save 15% on the A FM programme. Just use a code podcast. It totally makes sense. In the article, I know part of what you did is you reviewed a lot of the research that's been done, what's been done so far, and there were three dominant streams that you talk about in the article or that you recognise people are talking about. Can you share what those are different kind three key themes were
Guest: Jorge Rojas (17:01):
Certainly, yeah. We found three main clusters of research development. The first one is information asymmetry and corporate governance. The core idea of this first cluster is that satellites derived from the articles that we studied. Car counts, car counts from retail parking lots can tell investors what companies still don't report at that time. So imagine knowing how Walmart will do before Walmart tells you. So of course this works for certain retailers on certain types of locations that have open parking spaces, but still it gives you, it levels up the information that investors can retail, retail investors can get on real time. So that's one of the areas that we saw that it's being explored. The car counts on parking lots, that's one. Then the second one is, and it makes a lot of sense, alternative to government statistics, we'll know that governments manipulate, delay or outright lie about national statistics.
(18:11):
One of the most provocative findings in the field is that private satellite companies now produce economic estimates so good that they are reducing the market reactions to official government announcements. The articles I mentioned, I mentioned the country because it's in the articles. For example, in China there's a lot of announcements and these articles used as controlled variables, the cloud cover, so cloud cover was used to test when there are a lot of clouds, when you have clear skies, how it affects the government's preciseness on their information. So the articles are very robust statistically and it makes sense that governments usually their information is not quite reliable, it's not quite current or it may be even manipulated. And the third one, which is probably the one that is growing and getting the most attention, it's ESG, environmental risks, sustainable finance, and it's arguably the most policy relevant cluster since satellite information. It's independent and can verify if companies and governments are doing what they say they're doing about environmental performance. And that's a game changer for ESG investment and green finance, as I said before, you cannot greenwash a satellite. So satellite information overall is providing very interesting data that has been used and developed in different articles in these three areas. And that's basically the clusters that we,
Host: Paul Barnhurst (19:59):
And those were the three I remember reading about. I thought the example of the cloud cover in China and the government is fascinating. I think you make some great points. We've all seen numbers that we are skeptical of coming from a government or doubtful or we've seen, hey, we're going to revise the way we calculate this when the new person's in office or all those types of things. And given they have an incentive for the economy to be doing well, I think we all take some of those numbers with a grain of salt, especially given they're always revised six months later, they're almost always revised downward. Why are they always worse after the fact?
Guest: Jorge Rojas (20:36):
I
Host: Paul Barnhurst (20:36):
Think that's a fabulous use case is satellite data to help validate, augment, improve what you can get from government resources
Guest: Jorge Rojas (20:46):
And of course it has errors, it needs to be worked around and there's a lot of area for improvement. But in general, if you ask me, I will trust more satellite data on bias than a report from a government. It's pretty obvious, right? Yeah,
Host: Paul Barnhurst (21:03):
No, I'm with you. Right. It's not a perfect, no perfect system, just like there's a reason why we're making assumptions inModeling versus providing answers. We don't know the exact answer. I can give you an estimate and a range and we'll talk about that with Monte Carlo, but as I always like to say, if I can give a perfect answer, I'd be in Vegas right now, getting rich, not here on a podcast.
Guest: Jorge Rojas (21:28):
Yeah, probably me too.
Host: Paul Barnhurst (21:32):
So I think we're out of the same space there. What do you see as the biggest maybe real world example for the averageModeler to use this type of data? Where do you see it kind of most being used byModeler?
Guest: Jorge Rojas (21:47):
There are many areas of real estate insurance, and I'm going to give you one very specific example. If you incorporate earth observation data, your model may shift from relying only again on backboard looking information. Actually use real time objective physical truths. For example, let's say you're doing a model on commodity trading and supply chain traders usually require accurate global supply and demand forecasts so you can price future contracts or fund agricultural supply chains. Now satellites allow institutions to bypass opaque self-reported government data from foreign markets by directly measuring physical assets. For example, you can check the fullness of crude oil tanks, you can monitor shipping container congestion at major ports. You may observe crop health in key agricultural belts. For example, AI allows, again, tank measuring crude oil storage. And in agriculture for example, you can do macro crop gel forecasting that uses deep learning models to track agricultural indices like chlorophyll absorption.
(23:09):
You can tell the health of the agricultural crops from space how they are absorbing chlorophyll. So imagine having all this data available for your financial model, it will make it supremely more informative and current than we have today. And there are other areas like real estate or for example now there, I don't know how recently they coined this term, but in insurance you also have the ability, for example, earthquakes, floods. You can very quickly assess the damage that otherwise will take you probably months to assess. So there's different areas where you can get very quick and very current information like crops and agriculture for example.
Host: Paul Barnhurst (24:01):
Bad financial models can lead to bad decisions or worse. So how do you minimize the risk of a bad model? You make sure the models you build are great financialModeling Institute develop the advanced financialModeler accreditation programme to helpModelers like you. The A FM programme offers a step-by-step approach to building world-class financial models. The programme ensures that you know the best practises in model design and structure and will help you brush up on your excel and accounting skills too. Be the one on your team to build great models. If you want to impress your boss and your clients, get a FM accredited podcast listeners, save 15% on the A FM programme. Just use code podcast at fm institute.com/podcast. Thank you for sharing that. I hadn't thought of all of those, but crops make a lot of sense. Earthquake, right? An insurance company could go in and with the data, do an assessment of what they think the cost may be better than they'd be able to otherwise with a bunch of images to start looking at things and getting an idea of what the true damage is. Lots of great examples there. So I want to shift gears a little bit and talk about Monte Carlo simulation. I know you're a big fan of it, so I'm curious, let's start with this. How do you decide when to use Monte Carlo simulations in models you're building? Why do you use it? When do you use it? Maybe let's start there.
Guest: Jorge Rojas (25:41):
Yes, for sure. Indeed. I'm a Monte Carlo fan. Let's put it this way. I'd rather be without my cell phone than without my Monte Carlo plugin. That's how extreme I use Monte Carlo and actually for me it's the other way around. I will use Monte Carlo in most of my financial models and think about which ones I will not use. I do it the other way around. A good example is for example, projecting financial statements. It does not make a lot of sense to use Monte Carlo as the projected results are accounting static measures, estimates. So it doesn't make much sense to simulate them, at least for a projection of the typical financial statements. Now my simple answer is I use Monte Carlo when I have stochastic variables. Stochastic by the way, comes from the Greek stochastics, which means skillful in aiming stocks is target, aim or guess.
(26:38):
So that's exactly what Monte Carlo does. So the way I do it is since most of my models have anywhere from five to 20 inputs with very random probabilistic distributions, uniform triangular, normal log normal, these variables may be interest rates, exchange rates, commodity prices, inflation growth rates, cost of capital, and so many more. So since there is no way, there is no human being that can forecast all these variables and they all play into your model. If you rely on Monte cargo simulation the same way you rely on a pilot training on a flight simulator or a doctor, a medical doctor training on a dummy before they actually cut a human being or do surgery the same way the simulation allows you to test, to stress test, to do scenario sensitivity analysis and see what could go wrong and these variables will usually produce. The key insight of the Montecarlo simulation is that you need to understand that it relies on the law of large numbers and the central limit theory. If you know that and you can grasp that idea, the output, be it net percent value, the intrinsic value of a common share, you'll have a very accurate number with probabilistic ranges and you will understand much better what's going on. You will decide better what's in front of you. I dunno, by the way, do you use Monte Carlo simulation often?
Host: Paul Barnhurst (28:27):
I didn't use it in my career. I've seen it a little bit. I learned it in school. If I was to go back now, I'd use it more. I think it is valuable. I've definitely seen people who used it, but it was just something I saw nobody using in my work, and so it's not something I really use. I did do some statisticalModeling but not Monte Carlo, but I do think it's valuable. That's why I love to hear about it is if I went back and did, I don't do muchModeling nowadays with my business, but if I did more, I would definitely want to do more Monte Carlo.
Guest: Jorge Rojas (28:56):
It's very powerful.
Host: Paul Barnhurst (28:58):
Why do you think it's not used more? Do you think it's people who fear it or they just kind of go with an estimate and a best and a worst case? It feels like to me it's not used near as much as it could be.
Guest: Jorge Rojas (29:10):
I totally agree. I think one of the reasons may be that the Monte car plugins are expensive, they're very expensive, they're not cheap. They usually run only on windows. I think that's by design for Mac systems. I think it's more limited, but I don't think that that will be the main reason. One of the other reasons that I have thought about it is because I've learned from CFA behavioural studies and that area is that people usually are not very good at statistics. People usually don't like statistics. They think it's like a black box and some mysterious formulas, but it's actually, it's so powerful that I share your question. I don't know why people use it more every time I teach it. Every time people are like, their eyes are like this and say, wow, I didn't know this was available and it's really powerful. For example, one of the typical cases is if I tell you that a project has a positive net personal value and I ask you should we go ahead and do it?
(30:28):
You would probably say, yes, of course financial theory tells you to go ahead with a positive net person value. But if I tell you the net percent value is positive, but the probability that the net percent value is zero or negative is 45%, then you will probably see, even though the theory tells me to go ahead, most investors will say, oh no, no, I'm going to take a step back because this looks too risky. Or if I tell you the intrinsic value of a common share is $25 versus I tell you it's $25 and there is a 90% probability that it's in the range of 22 to 28. So that's the power of the simulation technique and I agree that I wonder why more people don't use it because it's so powerful. Maybe it's an expensive software, but at the company level it shouldn't be that expensive. I mean for personal level maybe yes, but at company level it should be paid. It should pay itself very quickly when you use it to make decisions. Yeah,
Host: Paul Barnhurst (31:36):
Yeah. I was talking to the guy who created one of the main ones at risk, which you may have used the Monte Carlo simulation. He was the founder of the company originally. I was chatting with him the other day and they're coming out with some new Excel sensitivity type stuff. And I agree with you. I think statistics, I saw that, not even Carlo, but just I'd use some statistics regression and stuff and it was woefully under underutilised. And I think you're right. A lot of it is people don't understand it or they just don't want to spend time on it. It takes some time to understand it and to be used to it, and it's different than just doing the basic model or whatever. But when done right, it can be incredibly valuable.
Guest: Jorge Rojas (32:19):
Yes, for sure. Yeah.
Host: Paul Barnhurst (32:20):
So I'm with you. Alright, so I want to ask a question. When you and I chatted, you talked about some of the challenges you faceModeling agriculture projects, particularly in Guatemala, and the risks to crop growth is something I haven't thought about because I've never modeled anything like that. So it was fascinating hearing you talk about it. So I'd love to hear you to share a little bit of what are some of the challenges, the risks you face when you'reModeling agriculture?
Guest: Jorge Rojas (32:47):
I have developed several agribusiness, financial projections, sugar cane, palm oil, coffee, and they're very particular businesses and projects. First of all, the families and companies that usually have this have more than just agricultural businesses. They have, for example, hydroelectric power generation. So sometimes they mix, they want to know what's the free cash flow of their combined businesses. So it makes it even more challenging. But going back to specifically agricultural projects, they have several characteristics. One is that their price, usually it's set in international commodity markets, palm oil, you use the rotor, dam, sea price, coffee, sugar, all that is in international prices. So price is a big metric that you cannot just say, well, I'm going to assume it's going to be that. No, that's not good enough. Then agricultural projects are exposed to significant risks such as tropical storms. We have one coming our way right now.
(33:51):
Seismic events, earthquakes, Guatemala is split in half by continental tectonic plates. So we're very seismic. Guatemala is about give or take, about the size of Louisiana or Tennessee and we have 37 volcanoes out of which three or four active all the time. Insect lakes, the poor plants take a lot of bidding around here. So what I did with this product is, well, there's no way I can predict all of this. So what we did is we went back with the owners, the experts and say, okay, how often do you have one of these exogenous events, one significant event every four to five years. So in this case, for example, I model a oli variable on Monte Carro, which may take a zero one value, but you can assign a probability. So every year you have a oli exogenous event variable that may be zero or one with a 20% probability, one every five years.
(34:57):
So the owner knows and tells me if this happens, the agricultural result is that the crop is 25% less efficient or whatever he tells me, he or she tells me what's the effect on the plants or how much more he has to spend on whatever needs to be fixed. So in this case, I use the commodity prices model as typically a log normal or exponential distribution. The good thing is that the Monte Carlo software can give you, if you give the data, it can suggest distribution probability. So you can use that as an input for the commodity item. Then you have the external events and then you have your usual cashflow, sales costs, expenses, taxes, and all of the free cashflow to the firm. But basically Monte car and for example in the hydroelectrical business you have data of the quantity of rain. So that affects the level of the dam and at the same time you can model it together with the agriculture business and have a joint free cash flow. And you model that in Monte Carlo and of course the model needs to be garbage in, garbage out. You have to have a great model before you apply simulation. But that's basically the typical process I use with this kind of projects here. When
Host: Paul Barnhurst (36:29):
You were mentioning the Benli variable, I could see you, you're going to go with 20% once every five years and that may lead to 50% less crop or 30% more expense plus less crop. Like people say, well I can't model because I never know when a black swan event's going to happen. But you can at least say, how would I behave if there were some kind of catastrophic event? What's the probability of that? You could still do some work, you can't predict it, but you can at least try to plan for it. And that sounds like what you're really trying to do with these probabilities is plan for these events that we know are going to happen at some point. We just don't know when or exactly how bad they'll be.
Guest: Jorge Rojas (37:09):
You have to work around expected values. I always start any course saying, listen, the goal here is to be as little wrong as possible. You will never know if your net person value is correct, how there's no way you can know it. So we need to be rigorous so we can be the least possible. We have the least possible error. Other than that, there's no way you can be sure that your number is perfect. There's no way.
Host: Paul Barnhurst (37:37):
I appreciate all that. And that gets back to what we talk about Monte Carlo, spatial finance, all these things are tools, ways to improve your estimate and to better understand the range of possibilities.
Guest: Jorge Rojas (37:54):
Yes, totally agreed. Yeah.
Host: Paul Barnhurst (37:57):
What I want to do now is move into some standard questions we ask every guest. So first, before we get to rapid fire, there's a couple I like to ask. Do you have a favourite Excel shortcut?
Guest: Jorge Rojas (38:08):
Yes, I have built-in shortcuts. I use a lot of control, right? Control D to copy sales. I use it all the time. I would say it's my favourite built-in shortcut. And then I have personalised shortcuts that I used to format my local currency Zales. I don't want to go three layers with my mouse. So I developed a macro shortcut for my currency and also for centre across selection, which is a very common function that I use, but I have developed my own personalised circle for that.
Host: Paul Barnhurst (38:42):
Yeah, currency totally makes sense. If you're dealing with a lot of currencies. Yeah, you have to drill down. I'm not surprised to hear you say centre across selection since they bury it in Excel. Maybe one day they'll put it on the front page.
Guest: Jorge Rojas (38:55):
Should yeah,
Host: Paul Barnhurst (38:58):
That's a common one. Alright, what's the most unique or kind of fun thing you've done with the spreadsheet? It could be for personal life, hard work, but what would you say is the most unique thing you've created with the spreadsheet?
Guest: Jorge Rojas (39:10):
One very unique and fun thing was that we recently went with my family to, we took a road trip actually from Albuquerque all the way to Denver and my kids design an Excel with ai, so it suggested the national parks where we should go, the cost, the budget and everything was fed into an Excel. So I told them, you plan this trip, go ahead, we will just get into the car and you will drive. And that was a very fun and useful way that we use an Excel and it was just a couple of months ago.
Host: Paul Barnhurst (39:42):
That's awesome. I love the Excel and I love that ai. It's amazing the different ways we can use it. I don't think anyone realised three, four years ago, 2022 when it first came out of all these use cases.
Guest: Jorge Rojas (39:55):
Yes.
Host: Paul Barnhurst (39:56):
Alright, so this next section is called rapid fire. I do this with every guest. So I'm going to lay out the ground rules of how this works and then we'll go through it. So the idea is you have to pick a side. I realised there's nuance in all these questions, so you could give the answer. It depends, but that doesn't make for fun listening, so you have to choose in most situations, yes or no. So if I was to ask circular references, you would say which one? Then when we get through all the questions, you can elaborate on a couple of them because I recognise there's nuance to all of them. So ready?
Guest: Jorge Rojas (40:29):
Yes.
Host: Paul Barnhurst (40:30):
Circular references in models, yes or no?
Guest: Jorge Rojas (40:33):
No.
Host: Paul Barnhurst (40:35):
VBA, yes or no?
Guest: Jorge Rojas (40:38):
Yes.
Host: Paul Barnhurst (40:41):
Lambdas in financial models, yes or no?
Guest: Jorge Rojas (40:43):
Yes.
Host: Paul Barnhurst (40:46):
External workbook links? Yes or no?
Guest: Jorge Rojas (40:49):
No.
Host: Paul Barnhurst (40:50):
Yeah, I get a lot of that of no, please don't do it. ShouldModelers primarily use keyboard shortcuts? Yes or no?
Guest: Jorge Rojas (40:57):
Yes.
Host: Paul Barnhurst (40:59):
Okay. Should models always be print ready, yes or no?
Guest: Jorge Rojas (41:03):
Yes.
Host: Paul Barnhurst (41:04):
Are merge cells, is there ever a situation where you should use merge cells? Yes or no?
Guest: Jorge Rojas (41:10):
No. There should be jail time for using merge cells.
Host: Paul Barnhurst (41:14):
I had one guess that his saying was use merge cells, you go to hell. So kind of along with your time, your jail time. So yeah, that's one of the strongest ones yet. Do you think financialModelers should learn Python in Excel?
Guest: Jorge Rojas (41:29):
Yes.
Host: Paul Barnhurst (41:30):
What about power query?
Guest: Jorge Rojas (41:33):
Yes.
Host: Paul Barnhurst (41:35):
How would bi I?
Guest: Jorge Rojas (41:37):
Yes.
Host: Paul Barnhurst (41:38):
Do you believe every financialModeler should be able to build a fully integrated three statement model?
Guest: Jorge Rojas (41:45):
That's a painful no, but I will take no,
Host: Paul Barnhurst (41:48):
I actually say no, but I can understand the pain. So I imagine we'll talk more about that one in a minute. Will excel ever die? Yes or no?
Guest: Jorge Rojas (41:57):
No.
Host: Paul Barnhurst (41:57):
Have you used AI to help you build a model in Excel?
Guest: Jorge Rojas (42:01):
Not so far, no.
Host: Paul Barnhurst (42:03):
Not
Guest: Jorge Rojas (42:03):
To build it, no.
Host: Paul Barnhurst (42:04):
If you had to pick just one, what financial statement is most important forModelers? Is it the p and l, the balance sheet or the cashflow statement?
Guest: Jorge Rojas (42:14):
In my case, the cashflow statement.
Host: Paul Barnhurst (42:16):
Do you have a favourite? LLM? Do you like Claude's copilot chat? GPT. Do you have one you turn to the most?
Guest: Jorge Rojas (42:22):
Claude by far. Yeah,
Host: Paul Barnhurst (42:25):
That seems to be becoming a favourite for a lot of people. Alright, if you could pick only one for all your models, would you choose sensitivity analysis or scenario analysis?
Guest: Jorge Rojas (42:36):
Sensitivity analysis With Monte Carlo.
Host: Paul Barnhurst (42:40):
I figured that's where you're going to go. See you're the first. Everybody else has picked scenario, so I love that. I thought you might go there and say something about Monte Carlo. So I love the different answers. Do you believe financial models are the number one corporate decision-making tool?
Guest: Jorge Rojas (42:55):
Yes.
Host: Paul Barnhurst (42:55):
That's always a tough one. People think about that one for a minute. And then what's your favourite lookup function?
Guest: Jorge Rojas (43:00):
X lookup.
Host: Paul Barnhurst (43:02):
Alrighty. I like it. X lookup. It's funny you're interviewing aModeling nerd. When I interviewed the financialModeling World Cup champion and I gave him four options, he goes, you do realise there's more than four lookups. I'm like, yes. Immediately he wanted to go, well you could do this one and that one and see how many he could list. It was kind of funny. Alright, I think you wanted to elaborate. It sounded like on the three statement model you said you gave a painful no. So what was your thinking there?
Guest: Jorge Rojas (43:30):
The reason is that since I did a FM, and I'm not an accountant, but doing the projection of the three statement model, it's so powerful, so productive that I have used that knowledge in other applications that are none related to financial statement projections. So I would say no because it would be a little unrealistic to expect anyone to project free statement it very useful and very powerful. I was between a yes and a no, but I would say probably would be inferred that anyone needs to do that because you can do a lot of good models without doing that project statement project. So that's one of the, I
Host: Paul Barnhurst (44:08):
Think you and are
Guest: Jorge Rojas (44:08):
Ones that
Host: Paul Barnhurst (44:09):
I think it's incredibly valuable. I think it would be good for anyone to be able to do it, but I spent my whole career building p and ls. I didn't do my first three statement models until I started my own business and I worked with a small company. And so I recognise that there's people that can build great models, just don't do it. I'm with you, but it's incredibly valuable. Any others you want to elaborate on?
Guest: Jorge Rojas (44:28):
Yeah. The other one is that it's up my alley. Usually when I tell people you have this what if analysis you need to do, they're not mutually exclusive. You need to do scenario sensitivity and a Monte Carlo simulation if you happen to have deploying. For me the three are critical and important and I actually do the three. Of course this exercise is different, but in the real world applications I do the three of them, the three of them are so powerful that you get a lot of information from doing the three. But other than that, I'm clear on that.
Host: Paul Barnhurst (45:01):
Yeah, it makes a lot of sense and I realise this is all just kind of situational. It's not reality. So as we wrap up here, first, thank you so much for joining me. I loved hearing about spatial finance. Like I said, I hadn't even thought about satellite data and how it would impact finance. So it was great to hear what you shared and I'd like to give you an opportunity before we wrap up, any final advice you'd like to offer people to help 'em withModeling?
Guest: Jorge Rojas (45:29):
Well, yeah, first of all, and I'm doing that, it's incorporate AI in your learning. I mean everyone is learning AI now on how to incorporate it in your model. Second, you learn by doing and practising and mainly listening to guys like you value content from everyone. Many different expertModelers. I see Excel as a blue ocean. You need to go out and shop around knowledge to become an efficientModeler. There's always something you can learn from people that shared their knowledge on Excel and build your model as automated as possible. Probably will take you more time upfront, but at the end I promise you it will save you time down the road. I have what I call the Friday 5:00 PM syndrome. It's when you do a model, your boss will call you a Friday 5:00 PM and tell you I need this in an hour. So you better build your model so you can respond and whatever needs to be done very quickly in less than a couple of hours. If your model is built automated and well thought you will have a model that will last for a while. So that's what I would like to share.
Host: Paul Barnhurst (46:35):
Great advice and I can appreciate the five o'clock Friday syndrome. We've all been there. Thank you for that. If our audience wants to learn more about you or maybe get in touch with you, what's the best way for them to do that?
Guest: Jorge Rojas (46:47):
LinkedIn I think is the way to go. Yeah.
Host: Paul Barnhurst (46:48):
Alright, well we'll make sure to put your profile in the show notes so everybody can find that on there. And thank you so much for joining me. I really enjoyed the conversation, appreciate you taking an hour and spending it with me and my audience. So thank you so much Jorge.
Guest: Jorge Rojas (47:04):
No, thank you. It was an honour and thank you for having me. Thanks.
Host: Paul Barnhurst (47:08):
Financial Modeler's Corner was brought to you by the Financial Modeling Institute. This year, I completed the Advanced Financial Modeler certification, and it made me a better financial modeler. What are you waiting for? Visit FMI at https://fminstitute.com/podcast/ and use Code Podcast to save 15% when you enroll in one of the accreditations today.