Monte Carlo Method in CRE Models

https://www.adventuresincre.com/category/re-modeling/stochastic-modelin…

Go bananas.

Youtube Spencer Burton, he has some way to incorporate Monte Carlo models into excel without add ons.

I honestly don't see how you will ever use it. Not saying people shouldn't be more complex and really analyze the CRE market, there are just so many people who would not understand it.

Also from my understanding there are like 4 CRE firms who get this deep into their underwriting.

only place in CRE that I've seen it used if is you are a bank and are running risk ratings to satisfy the fed.

I have to run these every quarter for development projects and mezz positions and it's asinine IMO. A simple sensitivity analysis is all you really need for RE. The plus is that it makes you look smart (it's basically mental masturbation). The minus is checking the work. Not as simple as checking a DCF, and time consuming when you think about the number of people who look at it and actually know what it's saying. My best advice to you is read into probability theory and create your own formulas.

If an analyst brought me a model with Monte Carlo simulation I would think great this kid did zero market research has no idea about the supply pipeline / forecasted rent growth / recent lease and sale comps and literally thinks random number generation is better than anything he can produce

Spencer Burton & AdventuresInCRE is an amazing resource. I'm glad other people are reading his stuff.

I can only see a Monte Carlo simulation being useful in modeling resi projects due to the remaining asset types drawing such a large % of their income from contractual rent.

And to add onto IRRelevant's post, the localized nature of CRE should lead to an analyst making a list of well-thought out assumptions versus using random generation. Perhaps it could be useful projecting future expenses, but that might be getting too intricate in an underwriting.

Remember people - real estate is simple.

I've done it. We used the crystal ball plug-in to excel. Basically, we needed to know exactly how much headroom we had on a certain covenant and which operational metrics put us at the most risk of busting that covenant. In the end, it didn't tell us anything we didn't already know intuitively, but it quantified exactly how much certain metrics could put you at risk and provided charts and distributions on those metrics. Everyone thought it was pretty fucking cool, but we haven't had the time or need to get into since.

Someone came on here a few weeks ago with a scenario that I thought begged for monte carlo. They wanted to take a rent roll on an apartment they'd bought and forecast exactly how much cash they needed to reserve throughout the year for rehab capex based on the expiration schedule and the likelihood of turnover. You could of course back of the envelope a number, but I think monte carlo would help you really optimize it. Of course the top reply was that turnover didn't matter in multifamily, which is the dumbest damn thing I've read in a long time.

From a developer's standpoint, I can't see a marginal benefit of using MC models.

Just budget for what you think you'll need plus a margin of safety the project can support and other investors will buy off on. Assuming the odds of the project being a failure or success fall in a normal distribution, the margin of safety should help you achieve better than average success barring force majeure.

This method isn't as complicated as MC modeling but it works.

I've built out an office acquisition model that utilizes the concepts from Spencer regarding monte carlo simulations. So far the response that I've received is that it's awesome, we need to implement this internally, how do we best explain this to banks and investors for them to buy off on the project running so many variables.

It was a pain in the ass to build and it still isn't being used full time so I'm not sure whether the older management group will fully buy into it. Guess we'll see.

False Precision my man.

You can build all the MC scenarios you want, but you are still going to IC with simple numbers they understand and put your company in the promote

What are your guys thoughts on time series forecasting/ stochastic modelling to support growth assumptions/ exit yields and try to find autocorrelation and predict where the market is going. I'm not referring to a simple monte Carlo simulation, but looking at big data sets including population growth in the area, regional/national consumer spending, GDP, Household debt levels, Bank lending volume towards targeted market, interest rates, inflation, Governmental Changes and how these have effected RE asst prices in past cycles in an attempt to derive future consensus. As a statistics student currently in uni and working part time for a REPE fund I'm astonished at the lack of empirical support the underwriters use in their assumptions. My boss would underwrite a deal and just say "ah yeah 2% annual comp growth in ERV and a 1% yield compression over 3 years seems about right", It's complete speculation. People say you need to have a "strong" idea of your market, but look how many brokers and sophisticated investors got it so wrong in the last cycle. One of my aims when I finish college is to talk the head of what ever fund I'm at to hire a statistician to support big data analysis as I quite frankly think CBRE/JLL etc are incapable of seeing beyond their heard mentality. Correct me if I'm wrong but I think Davidson and Kempner are big into this.