Veterinary Vertex is a weekly podcast that takes you behind the scenes of the clinical and research discoveries published in the Journal of the American Veterinary Medical Association (JAVMA) and the American Journal of Veterinary Research (AJVR). Tune in to learn about cutting-edge veterinary research and gain in-depth insights you won’t find anywhere else. Come away with knowledge you can put to use in your own practice – along with a healthy dose of inspiration to remind you what you love about veterinary medicine.
Learn about the veterinary economic landscape with insights from industry experts Matthew MacLachlan and John Volk. Learn how historical data and economic models shape the future of veterinary services, as our guests dissect the supply and demand dynamics impacting veterinarians today. With John highlighting the influence of pet ownership and disposable income, and Matt delving into the effectiveness of the ARIMA framework, this episode promises a comprehensive guide to navigating economic instability and forecasting with confidence.
Explore how data-driven strategies can revolutionize veterinary practice management. Gain practical advice on building robust data systems and retaining employees and clients. This engaging discussion, peppered with personal anecdotes, offers invaluable perspectives on staying ahead in the ever-evolving veterinary field. Join us as we chart the course for future success in veterinary economics.
You are listening to Veterinary Vertex, a podcast of the AVMA Journals. In this episode we chat about incorporating model selection and uncertainty into forecasts of economic conditions and companion animal clinical veterinarian labor markets with our guests Matthew McLaughlin and John Volk.
Lisa Fortier:
Welcome to Veterinary Vertex. I'm Editor-in-Chief Lisa Fortier, and I'm joined by Associate Editor Sarah Wright. Today we have Matt and John joining us. Thank you guys both so much for taking time out of your busy schedules to be with us here today.
John Volk:
Thank you. Thank you for having us.
Sarah Wright:
All right, let's dive right in, John. Your JAVMA article discusses incorporating model selection and uncertainty into forecasts of economic conditions in companion animal, clinical, veterinary and labor markets. Can you talk our listeners through what data fed into this model?
John Volk:
Yes, thank you. So, we looked at the primary drivers for supply of veterinarians and demand for veterinary services for this article. On the supply side, the number of veterinarians in the US is well established. AVMA has records of that going back decades. And then we also have information about the rate of growth of the number of veterinarians, both from existing US veterinary schools as well as from other sources. And then, as you know, there are four new schools that are coming online in the near future and several others that have been proposed. So, based on the pace at which those schools may be entering and adding new graduates, we factored those into the supply projections as well, so that we really have projections from existing sources as well as how the population might change based on new schools coming on board.
John Volk:
On the demand side for veterinary services, we looked at four primary factors. First, was the number of pet-owning households, and again we have annual data going back more than 25 years on the number of households owning pets, and the number of pets per household has been remarkably stable over that entire time. So, it gives us a really good estimate of the population of pets and how that's changed. Then we also looked at household disposable income, because all pet care comes out of disposable income, and so we have a history of how disposable income has changed over time and how it may fluctuate or change in the future. And then there are two other factors. One is the spending on pets per household, as well as the spending on veterinary services. So, all of those factors gave us data on the likelihood of demand for veterinary services.
Lisa Fortier:
Oh, that's really fascinating, John. John talked a lot about some. This data is pretty stable, but we sure learned from COVID that we can't really predict a whole lot. Can you tell us more about in that time of uncertainty, what models, tools did you select, how did you choose them and how did you consider uncertainty?
Matthew McLaughlin:
Yeah, this was an important part of this analysis. We had a couple of periods of relatively unstable economic forces going on behind the veterinary economy, and so it was important to have a modeling system that was straightforward, understandable, that we could build on in the future as we learn more about how this system moves through time, but also that explains both stable and unstable periods. What we moved towards was a framework that is commonly used within the time series econometric literature. It's called an ARIMA framework. That stands for Autoregressive Integrated Moving Average. This is simpler than some of the machine learning counterparts that folks might have encountered while browsing the internet, and it is much more interpretable. It's much easier to implement, which has some big advantages, and oftentimes it just performs better than some of the more complicated counterparts. Also, what it allows us to do is use the same system to model how each of these very distinct series may move through time. As John pointed to, the number of graduates, the supply of economists is very, very, very stable over time. In contrast, things that we're looking at like disposable income, changed a lot during COVID and we need to be able to model both of these sorts of variables over time, as they change, and so how uncertainty gets built into these.
Matthew McLaughlin:
All models are wrong, but some are useful. This is a great phrase. So they will fit the data so well. But we will always have unexplained variation, so we take that into consideration directly. Looking backwards, in our modeling system, we're going to have residuals or distances between our data and what we would predict using our modeling system. This helps us explain how much of the variation we can reasonably pick up.
Matthew McLaughlin:
Looking forward, we use something called Monte Carlo simulations. All this does is take our forecasting model results and the uncertainty we've seen in the past. Combine them so that we might have a better prediction of a range of values that we could see going forward. If you make a prediction I've worked in food price forecasting, commodity price forecasting you're going to be wrong at the point estimate 100% of the time. But if you have a range, you can more confidently say that it's very likely that our value in the future is going to fall within this range. So, both looking backwards, we have to acknowledge how much of this variation we can really explain and looking forward. It helps us to credibly, and in a way that's helpful for consumers of the forecast, understand uncertainty around forecasts. It gets really important with long forecasts like we produced in this paper.
Lisa Fortier:
Yeah, that's a great explanation, Thank you. And using all those tools and looking forward and backwards, how does modeling help you differentiate a short shift in something compared to a long-term trend?
Matthew McLaughlin:
Yeah, this is a great question. Sometimes systems will move back to equilibrium very quickly and so you will usually only have you know. Whatever the frequency of the data is a month or a year, we have a high value and then it typically comes right back towards trend. Usually wouldn't model that explicitly. It just wouldn't make any sense to have a model that accounts for a single observation. Other systems much more gradually revert back to those long-term trends. It takes a long time. So, where appropriate, we would apply a moving average which would allow us to more slowly and gradually come back to equilibrium going forward so it accommodates these more persistent changes. So again, our supply of veterinarians looks very much like a straight line. We probably wouldn't need something like that, but if we're thinking about consumer sentiment, disposable income, some of these things that move a lot more slowly, we do need to be able to account for these more gradual coming back to equilibrium patterns.
Sarah Wright:
Very interesting. Thank you for sharing and for those of you just joining us. We're discussing incorporating model selection and certainty into forecasts of economic conditions in companion animal, clinical, veterinary and labor markets with our guests Matt McLaughlin and John Volk.
Sarah Wright:
John, what are the important take-home messages from this JAVMA article?
John Volk:
Thank you. There were a couple of major takeaways for me. One is and Matt mentioned equilibrium, and that is you have to keep in mind that over the long term, supply and demand are constantly seeking equilibrium. So, there's another factor involved, which is price right. So, if demand grows faster than supply, then prices go up, which will in turn suppress demand and bring it back in line. On the other hand, if supply grows faster than demand, then it suppresses price, which then stimulates demand. So, in the broad scheme of things, supply and demand always seek equilibrium. The other thing is that you know, if we model out likely growth in supply and likely growth in demand, if in fact you know right now they've been growing at a pretty constant rate. But if you increase supply rather dramatically, you can put economic hardship on the economics of the profession.
Sarah Wright:
So, John and Matt, did any of the findings from this JAVMA article surprise you? Something that you didn't think you'd find, but you did.
John Volk:
The one thing that was really surprising, in effect, was that over the last 25 years, supply and demand have actually grown at very, very similar rates. So, we did not see much variation. And again, there were short-term variation. For example, during the Great Recession and following, demand was certainly suppressed, and during the COVID years, when the government pumped $5 trillion into the economy, demand was stimulated. But overall, over those 25 years, they grew at very similar rates.
Lisa Fortier:
Matt, do you have anything to add to that? Were you surprised by anything in the manuscript and your findings?
Matthew McLaughlin:
You know I've been in food for a long time and I'm transitioning over to veterinary medicine and evaluating these markets, so you know I'm learning as I'm coming into this field. One of the things that struck me is similar to what John talked about is I'm used to looking at systems that are highly volatile and in some ways, the veterinary, clinical veterinary profession is extraordinarily stable. In some ways, we have gone through some recent periods of rapid changes which have prompted re-evaluation of how we're assessing these markets, but for a lot of the history it has been remarkably consistent and I think some of these periods of change that we've gone through more recently are probably more unsettling than they would be in areas where, you know, 40% change in price or employment is the norm. These are big changes for what has historically been an extremely stable system.
Lisa Fortier:
That's fascinating. Thank you for sharing that, especially with your background in food. Nice to compare it to something else, to know just how stable the veterinary market is. And, John, you and Matt have all talked about these modeling and forecasts of the labor market and how this might improve decision making by veterinarians, relevant business owners, professional associations like the AVMA and educational institutions. What would you say is the one piece of information all of these decision makers should know about incorporating this model selection and uncertainty into forecasts of economic conditions in the companion animal labor market.
John Volk:
Yeah, I think the thing I would say is that, as Matt said, in reality no one can predict the future. All forecasts are forecasts and things do change over time, but the fact that supply and demand ultimately seek equilibrium, I think it's just important that you know some forecasts are going to be more accurate than others, but I certainly think it would be inappropriate to forecast any major deviations in supply and demand well into the future. It's just that data do not support that kind of thing.
Matthew McLaughlin:
I would tend to agree. I think having a nice general system like the one John, Chris, and I were able to build up to allows for assessment of other variables. As data resources improve, we can assess more granular data and if there are more specific questions that come up, we can also start looking at those as well. So, it's always better to, I think, have a good guess of what the near and long-term look like rather than a rough guess, and I think we have developed a nice system that can be iterated on and improved going forward, but a great starting point through this paper for thinking about what the near and medium term look like.
Sarah Wright:
So, John and Matt, if you could recommend one action for these decision makers to take, what would it be?
John Volk:
You know, I think the thing that strikes me is that all this data reflects veterinary medicine in the aggregate right, and I think it's important to remember that all management decisions are local, so you can have shortages of veterinary services in local areas, you can have surpluses in local areas, but over time these will even out. But I think that if I'm talking to a practice manager, I'd say what you need to do is to really work hard at employee retention and customer retention, client retention, and make these fluctuations in the aggregate market as meaningless as possible for you and your practice.
Matthew McLaughlin:
And I would just piggyback on that. I think John makes a very good point there. I think, building up data resources. I'm going to show my background in economics. I think we all have the same answer in these situations, but the more data you can provide, the better insights, the better inference, the better forecasts that you can produce that are better adjusted to local conditions or specific types of veterinary practices. The more granular the data, the more specific questions you can start answering.
Lisa Fortier:
Yeah, really fascinating. Thank you, guys. I learned a lot reading your article, but certainly even learned more today having this conversation. So thank you again for your time. It's really appreciate it.
John Volk:
You bet.
Matthew McLaughlin:
Thanks for having us.
Lisa Fortier:
And as we wind down, we like to ask kind of a fun sort of question, Matt. So, we'll start with you. When you put together a puzzle and this is the holiday season, so my family is like fighting over the puzzle table. Do you start with the exterior border or do you do the interior pieces? Or sometimes we hear people do a mix.
Matthew McLaughlin:
Yeah, I think I'm a middle of the road guy and always I think a mix is best. I think, as you're putting, I think a lot of people start with the boundary, but as I do that, I start sorting pieces into different piles that I think will fill out the interior the best, so I use a blended approach.
Matthew McLaughlin:
Let's say, to my puzzle solving
Lisa Fortier:
Well, Sarah and I so far, we're going to gather this data. I'm a very data-driven person as well. It's all about evidence, not eminence. We tend to think that surgeons do the outside and internists do the inside. And then there's some people who do this hybrid approach, but we're not sure about economists, so we'll start to ask that question more. But data coming forward.
Matthew McLaughlin:
You got one data point now.
Lisa Fortier:
John, could you share with our listeners what is your favorite animal fact?
John Volk:
You bet. So my favorite animal fact is that a polar bear is really a black bear, so the only thing white on a polar bear is its hair and its teeth. If you shaved a polar bear, you would find a black bear underneath, and I think a lot of people really don't realize that, and I've had the opportunity twice to go up to the Arctic and photograph polar bears in their native environment. They are extremely impressive animals, but in doing so I've learned a few things about them. The other thing you might want to know is that polar bear hair is actually hollow there, so they are remarkably adapted to their environment because that long hollow hair provides tremendous insulation for them. But they're very impressive animals and I've had the opportunity to see them up close. So yeah, that was my fun animal fact.
Sarah Wright:
That's super cool. One of my husband and I's holiday traditions actually is to go visit Hudson the polar bear at Brookfield Zoo, chicago, for a year in the winter. So, yeah, super fun. Just thank you both, Matt and John, for being here today, for contributing your article to JAVMA. Really appreciate it.
John Volk:
You bet Thanks. Thank you for having us. Thanks for having us.
Sarah Wright:
And to our listeners. You can read Matt and John and Christopher's article in JAVMA. I'm Sarah Wright with Lisa Fortier. Be on the lookout for next week's episode and don't forget to leave us a rating and review on Apple Podcasts or whatever platform you listen to.
