Valuation Engineering

The use of options is an extension to traditional NPV for calculating what is termed “Expanded” or “Strategic NPV.”   In both approaches the point of reference is always the current point in time or the effective date of appraisal.  Both approaches are therefore limited by current point in time perceptions of future risk.    The difference is that with Strategic NPV, there is an assumption that should a possible negative situation occur in the future, it will be avoided if possible.  This is called the “option,”  - the option not to develop or not to proceed down a negative path.  The “Option Value” is simply the difference between the Strategic NPV and the Passive NPV, that is:

                             Option Value = Strategic NPV - Passive NPV

In simple concrete terms, consider the purchase of land for possible development of a subdivision in one year.  There is reason to believe that there is a 30% chance that housing prices will fall to a point in one year that will result in a negative NPV, should development proceed.  There is a calculated 70% chance that prices will be such that a positive NPV of $800,000 would result.  The Passive or Traditional NPV approach will weight the two outcomes and discount to a present NPV.  That is, the Passive NPV approach will assume that should negative conditions prevail, development will nonetheless proceed.   Strategic NPV will assume, on the contrary, that the developer has the option in one year to cancel the project and sell the land should house prices be unfavorable.  What happens is that the Strategic NPV will likely be higher than the Passive NPV.  That is, that the land residual based on Strategic NPV will be positive and possibly higher than the land residual calculated on the basis of Passive NPV.   This is as it should be.  Passive NPV fails to take into consideration that the developer can change direction at some point in the future - that he can avert negative outcomes.  The consequence of employing Traditional or Passive NPV in these kinds of situations is that incorrect decisions are made regarding Highest and Best Use or, in more general terms, significant errors in valuation result.  

One characteristics of Strategic NPV formla is that values are often expressed using terms based on  max(V1, V2) functions.  For example,

                                                    NPV = max(I,0)/R

or NPV is the maximum of the expected Income I or 0 divided by the required rate of return.  In other words, if the expected income I is negative, the developer uses $0 as the income, that is, the developer exercises the option NOT to proceed with the associated activity.   One of the consequences of this is that by avoiding risk, the developer can use a smaller rate of return, a “riskless” rate of return.  Clearly this has a double impact on NPV:  (1) Negative income is removed and (2) this results in a lower required rate of return that reflects lower risk.  This typically results in a higher NPV or land residual, that does a better job of supporting accurate Highest and Best Use analysis.

An option may also involve choices with different costs, in which case occurances of the function min(cost1, cost2) function can be found.  

Calculations for simple options can be fairly straightforward.  However, options can fall into more complex patterns that include any of the following:

1. Option to defer investment.
2. Option to expand.
3. Option to contract.
4. Option to temporarily shut down.
5. Option to abandon for salvage value.
6. Option to switch use.  
7. Option to default on planned staged costs during construction.

(More to follow)

It has been my experience that the best adjustments for appraisal with a limited number of sales transactions in a complex real estate environment such as the San Francisco Bay Area can be obtained using multiple adaptive regression splines.   My experience has been primarily based on using MARS (r) from Salford Systems.  However, with a starting price of around $2,500 and yearly license fees of $1,200, it is a relatively expensive package for most appraisers.  Recently, I have tried using R language modules such as Earth and MDA that provide similar, although weaker functionality.

While Earth doesn’t provide all of the features found in the Salford-Systems implementations, it does provide plotting and printing methods.  The biggest downside is the lack of cross-validation, although this could be implemented with some additional programming.  The following information is taken from the documentation:

Limitations