Matlab/R & statistics skills in Real Estate Finance
Hi all,
I am currently working as an analyst at proprietary trading firm. At work I mainly develop trading systems and analytical tools in R. Is there a place in real estate industry where I could do some proper statistical analysis in R/Matlab using prices of real estate deals like I do now with stock prices (I know that the data is much more noisy and it's different from analyzing stocks)? A perfect job for me would be, for example, predicting changes in real prices in certain districts or real estate types, predicting crashes in real estate market, building valuation models, estimating correlations with real estate in other districts / with other asset classes for portfolio construction, using cointegration, possibly some technical analysis, etc. to make predictions of price movements. Are there many statisticians working in real estate finance? Where should I look for such jobs: REIT, REPE, IB or something else? How such positions are usually named and at what career level are they located? Would be great to get some company names as well.
Thanks!
There are definitely areas of real estate finance that utilize statistics, but in my experience those type of roles are on the credit side rather than the investing side. If you are really into real estate and want to build predictive statistical models, the only areas that do that i know of are credit risk for a commercial lender or doing credit risk for a third party analytics group.
If you were going to work in risk analytics at a commercial lender, you could be building some models but a lot of the day to day is more ad hoc deal related work. A lot of times these groups even outsource the model building to third party consulting firms.
As such, I think your best be t would be to work in credit risk at a consulting firm. In this role you would be responsible for building models that predict CRE probabilities of default and loss given default based on the real estate market, the building's occupancy rate, cash flow, etc. Usually these models are simulation base, so if you have experience building simulation models that would be a huge plus.
If you are located in nyc, PM me and I'd love to give you more detail on the industry and recommend a few groups you should consider.
Structured finance may be of some interest to you also.
Credit side structured finance would probably offer the most opportunities. I worked in MBS trading and my firm was also involved in the origination of such loans and eventual issuance of structured notes better known as mortgage backed securities. We employed statistical analysis in three ways: one was in analyzing which loans to buy from the market based on geographical/macro/micro economic analysis, borrower credit profiles, property analysis, etc. The loans would become the “raw material” for the next step.
Once we would buy these large blocks of loans from various banks, we would then structure them into a complex investment vehicle called an MBS security. In this step, we had mathematicians working on the team who would package loans together in a manner which would produce a specific return/risk profile which attracted sophisticated wall street investors such as major banks, hedge funds, etc. The structure would produce a certain return for its investors, provide default protection through the use of tranches, would have to sell at a market competitive rate, and would need to pass scrutiny by all levels of regulators. Additionally some of these structures were going to be bought and guaranteed by the government so they had to meet additional criteria to be GNMA eligible (Ginnie Mae). The investors buying this security would also have significant influence on the eventual structure we created. For example for a certain period of time our investors would say “no Florida loans” and our team would create a structure eliminating any Florida residential loans.
The third way, but a less formal way, stats/math was applied was by the traders of the proprietary trading desk. They would analyze which securities to buy and hold depending on various market and pricing variables and their job was to earn a return on the money by analyzing the short term interest and price/duration changes that occur in the daily trading of structured products. Since our business was very complex, we even had risk analysts overseeing our entire operation and they would analyze how much of the firms capital were we putting at risk in the various courses of operation. The guy overseeing my team used Matlab and other risk analysis programs to measure things such as Var for our trading operations.
Guess thats 4 ways we used. Sorry for the long post but I want to stress that there is space for statisticians/mathematicians in RE finance..but some of the brightest opportunities are in the structured credit products. If you do some research on MBS, CMBS, HMBS, or simply structured finance and start digging into university level PhD papers, you will see the level of math this shit can require. Another great source to get a general feel for things is the actual Ginnie Mae website.
You could go work for a consulting firm. Green Street, PPR those guys love that type of stuff. I'm also a fan.
I work for a large life company lender and we have a few guys in a group we call applied research that build models, crunch numbers and do this sort of thing for our current equity and debt portfolios as well as the overall market. They issue internal research on what they think various property types and msa's are going to perform like and other such tasks.
Right, I know Pru used to have an internal research division where these skills would come in handy. Also, besides who I mentioned above, CBRE-EA, RCA, REIS, CoStar, basically all the data providers could use someone with your expertise. There are also small consulting firms like Chandon Economics.
Exactly, my company is structured much like Pru and Principal. CBRE or Costar would be a great fit for you if you want to "crunch numbers" like you described.
Real Estate and Statistics (Originally Posted: 03/08/2016)
Recently took a modeling class that covered risk quantification (basically running different scenarios of a development project to figure out an expected return and stdev, then using those to figure out the probability of a loss of investment). I haven't used this stuff since college. Do any of you focus on stats in your current RE jobs and could you talk more about it if you do? Also curious to hear what types of RE roles would focus on this. Seems like risk quant would be helpful if you're in some kind of research position and you're analyzing a large set of historical data...but not sure if it's really "needed" at the property level.
Most of the people I know who work in RE are sales, sadly. The few ones that do analysis on quantitative stuff are the owners/partners of the firm.
Hiten
You kind of answered your own question. I could see how with a large data set this would be useful for market research, but at the property level for an individual asset you'll do a sensitivity analysis to determine your risk depending on X Y & Z scenarios.
You will see those kinds of risk measures used more frequently in the debt world. Most shops have risk groups that set standards for what the various metrics mean and how they are used, but some examples:
Banks have to rate all of their debt (it is a Fed requirement). Most choose to use some sort of matrix that takes into account things like probability of default, loss given default, etc. CMBS rating agencies do the same thing when the rating issues and the various tranche ratings will effect the pricing. Some private lenders price loans in a similar manner. So if your deal rates an AA in the model, then your pricing might be narrower than a deal that rates a B in the model.
Expected loss calculations are used by banks to set loss reserve figures (their capital reserve minimums).
A lot of the big institional shops use Moody's CMM product to calculate these measures and then have a customized grid setup by their risk departments as to what those measures mean to them.
Garbage in, garbage out--that's the problem with modeling a development project's risk profile. Also, an expected "statistical" return will never actually happen in reality, so it wouldn't be super useful. Running a development IRR with conservative inputs is the closest we would have to risk quantification.
I've done a lot of RE development deal with my fund and I'm not sure I understand what you're trying to do. Are you running monte carlo simulations and stress testing the development timeline to ascertain the risk of completion? Most dev modelling is based on the S-curve, and that's how you forecast cash flows which drive your IRR. Also, bear in mind that there is contingency baked into any dev budget so you have some runway if there are cost overruns or delays to the project. Also, bear in mind there are non-quantifiable factors, such as indemnification and completion gaurantees that reduce risk of the principal owners, along with GC relationships with the trades, which, if threatened, can increase the risk to the development tanking. I'm not sure how you can quantify those without signficant assumptions most of which are likely arbitrary given the lack of historical data available.
You said it better than me. There are aspects of "risk" that are qualitative, not quantitative, which makes risk "modeling" not particularly useful. Risk PROFILING is useful, but it can be largely subjective, and even arbitrary, which is OK because, for example, maybe you "feel" more comfortable developing in one neighborhood where you have deeper personal expertise; therefore, a deal with objectively the same or less risk might be labeled more risky if it's in a neighborhood in which you lack expertise.
There is a book 'Real Estate Mathematics: Applied Analytics and Quantitative Methods for Private Real Estate Investing' that discusses exactly your concerns - I would look into it.
$699 on Amazon. Ouch.
In qui dolores porro assumenda delectus perspiciatis alias. Eveniet qui libero recusandae non pariatur. Asperiores voluptas ipsam temporibus quam qui mollitia aperiam.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...