You now have the background to learn about simulation as the essential tool for dealing with uncertainty in making investment and development decisions in real estate. What we call “Monte Carlo” simulation provides the means to analyze and evaluate alternative flexible management strategies.

In general, a simulation is a way to explore what might happen in the real world. It uses some model of reality—physical or mathematical—to run a “what‐if” analysis. That is, it provides an answer to the question: What would the outcome of a certain model be if certain input conditions apply?

As we have stressed, we cannot adequately consider uncertainty and estimate the value of flexibility by focusing only on the mean of the outcome distribution. We have to move beyond the traditional, single‐stream DCF model. We need to consider the entire distribution of future possibilities. Monte Carlo simulation allows us to do this.

6.1 Generating Scenarios

Think of Monte Carlo simulation as scenario analysis on steroids. It considers many, many scenarios, not just two or three or even a handful. Its major feature is that it draws on probability distributions to generate independent, random scenarios of the future—for example, future trajectories of prices, revenues, or resale yields. These probability distributions are our representations of the uncertainty that exists in the real world. From the perspective of the present, the future can contain many possible scenarios, any one—but only one—of which could actually occur.

The probability distributions that we specify in the inputs to the model govern the likelihood of occurrence of the outcomes. These distributions are similar to what we introduced in Chapter 5. But we make them more realistic for representing the nature of uncertainty and dynamics in real estate variables, as best we can understand them.

The objective of the Monte Carlo simulation is to obtain representative results, ones that collectively mimic what could happen in reality. In particular, we want to cover the entire range of possible outcomes (or close to it for practical purposes). In Monte Carlo simulation, we refer to each individual, independently generated random future scenario as a “trial.” The idea is that each trial has an equal chance of actually happening in the real‐world future, as we model it in the simulation. To achieve this, the process for generating scenarios must reflect the appropriate input probability distributions.

Luckily for practitioners, the mechanics for generating trial scenarios are not difficult. The ability to draw samples in exact accordance with a governing input probability distribution (representing the relevant uncertainties and dynamics) is a routine part of modern simulation and standard spreadsheet software (see Box 6.1).

These mechanics use a simple technical process called “random number” generation. This name provides the rationale for linking probabilistic simulation to the Monte Carlo gambling establishments. As shorthand, we routinely drop the “Monte Carlo” label when we refer to probabilistic simulation, which is the only kind of simulation we use in this text.

6.2 Real Estate Simulation in a Nutshell

The Monte Carlo simulation of real estate investment or development projects is a simple repetitive process. The simulation:

• Generates a trial scenario consisting of a future sequence of what might happen based on input probability distributions and a model of the functioning of the project in each period (for example, a DCF valuation over 10 years);
• Calculates the project performance metrics of interest for the project outcome resulting from that scenario (for example, the DCF present value reflecting the future cash flows in that scenario, or the internal rate of return that the project would yield at a given upfront investment price);
• Repeats this process many times (we usually run at least 2000 trials), thereby generating a “sample” (or a simulation “run”) of many outcomes (one for each trial); and finally
• Displays results as graphical and statistical summaries of the entire distribution of the outcomes for the sample (for example, the “target curves” we describe in Chapter 8).

6.3 Simulation Is an Efficient Process

It is important to appreciate that simulation is an efficient process. It is really a remarkable extension of human decision‐relevant analytical abilities. Our grandparents could only have dreamed of its capabilities. At first, the task of considering how to deal with all the uncertainties can be scary; there are so many possibilities. So many things could happen in Year 1, and then so many more in Years 2, and 3, and 4 and so on. As Box 6.2 indicates, millions of combinations are possible.

Simulation is able to inform management decisions quickly and efficiently in the context of the huge number of possible combinations by deploying three main features: big‐picture focus, speed, and sampling.

• First, we use simulation to build intuition and gain insight into the general nature, or the big picture, of tactical, strategic, and design and planning decisions. We do not specify highly detailed tactics or solutions. For example, the result of a simulation may indicate that management should be flexible about timing the resale of a property. It will suggest guidance on how to make the resale decision, and build intuition about the value implications. The result is not going to specify an exact rule or precise date. Simulation does not tell us exactly what to do in any given circumstance at any specific time; it provides general insight into how to manage real estate uncertainties. This means that simulation models should have a degree of abstraction and simplification. There is genuine practical value in elegance and parsimony.
• Second, simulation exploits the computer’s ability to execute DCF calculations nearly instantaneously. Standard laptop computers take seconds to produce sufficiently accurate simulations (with thousands of trials) of alternative strategies to manage real estate opportunities under uncertainty. We can therefore generate large samples very quickly—that is, thousands of equally likely, independent scenarios (trials) within a few seconds or less. (And this capability is getting better all the time.)
• Finally, the speed of the computer enables multiple sampling to explore and analyze alternatives and build decision‐relevant understanding of the project. We can generate different samples by varying probability and pricing dynamics inputs assumptions or model parameters, such as decision rules or cost assumptions. Doing this sort of analysis in real time is a great way to build intuition and gain insight about a project or investment.

6.4 Number of Trials

In practice, we use 2000 trials as our standard sample size for the example analyses of real estate projects in this book. This is generally quite enough to provide a sufficiently accurate analysis, even if the actual total number of possible combinations is many billions. Larger sample sizes would reflect the sample output distribution more precisely, but take longer to run any one simulation (sample). If the metrics of interest seem unstable in repeated simulation runs, we might increase our sample size to 10 000 trials. Small differences, say 1%, between the estimates of the value of alternatives are probably not economically or managerially significant. We know that the probability data we input into the spreadsheets is not so precise. In any case, the “big picture” perspective does not call for minute detail. We should properly pay attention to major differences in prospective performance and ignore tiny differences (see Box 6.3).

6.5 Conclusion

This chapter introduced simulation modeling for real estate investment and valuation analysis. We saw that simulation:

• Can efficiently explore a much fuller and richer range of possible consequences than the traditional, single‐stream DCF. It thus realistically reflects uncertainty and management decisions in real estate.
• Uses automated processes to cycle through representative scenarios.
• Delivers simulated future outcome distributions that can give insight into the value of flexibility and how to manage real estate assets under uncertainty.