The financial crisis has had many consequences for the hotel industry, one of them being the re-evaluation of current appraisal practices. While some have tried to lay blame for the troubles in commercial real estate on appraisers (for one of the most forceful assaults, see: "Appraisals are inherently flawed" by Joel Ross published on 4 August 2009 by HotelNewsNow.com), others, more constructively, have discussed potential ways to improve the appraisal process and methodology. For instance, some argue for developing new estimates of reserves that more realistically reflect the condition of the assets, and the timing and amount of Property Improvement Plans (PIP). Others suggest estimating capitalization rates and discount rates from the bottom-up using a cost of capital approach. Further, some advocate cash flow projections of less than 5 years in lieu of the traditional 10-year window, to better reflect the buyers' hold periods.  
 
In spite of the importance of trying to improve upon the current practices, the major problem with any recommendation involving projections - whatever projection it is - is the very nature of the future: it is uncertain. Hence, instead of apprehending "wrong" forecasts and trying to get them "right" by developing excessively complex models dealing with every detail of hotel operation, asset management and investment, appraisals would benefit more from an increased rigor in evaluating uncertainty and risk as they pertain to the key assumptions driving hotel value. I am not saying here that an improved assessment of capital reserve for replacement or PIP is not an important task, or that a more thorough analysis of rates of return would not be beneficial. My point is that no matter what is estimated today and however diligently the work has been performed, any estimate made today will likely not be the true value in the future (I can actually bet my savings on it).
 
Take for instance discount rates estimated based on the cost of capital. Both components - equity and debt - are somewhat related to base interest rates set by federal banks. As a matter of fact, these rates change as a function of what central bankers decide to pay attention to, inflation or economic growth (or deficit). In addition, opportunity costs of equity investors and spreads used by lenders also vary as a function of the perceived level of risk. The forecast of these factors - base rates, opportunity costs and yield spreads - is certainly easier said than done and carries a fair amount of uncertainty. Hence, independently from the quality of the estimate, the reliance on a single number (note that the number may vary over time, but remains a single number per time period) is likely to yield an appraised value that is not less prone to criticism than one that is based on a market-derived discount rate or one that is based on experience.
 
In reality, the major problem with appraisals and valuation techniques is not to be found in the way the assumptions are estimated (even though it sometimes is a problem), but rather in the deterministic approach used and asked for by most investors. Deterministic systems have a single result, or a set of results, given a set of input parameters (i.e. best case, worst case, expected case). On the contrary, probabilistic systems incorporate the uncertainty in the state of each assumption and yield results that vary as a function of their likelihood. Such method permits scenarios that might appear fairly unlikely at first. For instance, a scenario in which base rates decreased rapidly but where yields continued to increase despite a bear stock market, would have been envisioned, yet with a low probability of occurrence (remember, this took place a year ago).  
 
Probabilistic valuation models - such as Value-at-Risk (VaR) used in the financial service sector - would have to include a measure of uncertainty in the key assumptions used, for instance a standard deviation or some measure of likelihood and intensity of variation. Adding this layer of complexity in the valuation process would make more sense than trying to detail each and every cost and revenue without paying attention to their probable variability. It would also add to the difficulty of performing the valuation which means add cost to the task. The question will then be: are investors willing to pay for that additional cost?
 
Unless investors are willing to suffer the additional cost of performing appraisals that deal with the uncertainty inherent to any forecast, then they will still get what they pay for, which is a number, or set of numbers, that will always be "wrong" (but not necessarily useless though).
 
Will investors be willing to pay the price? As long as risk-aversion dominates markets, then they might be willing to do so. But once risk-taking and the need for more liquidity and short-term hold of assets comes back, don't be surprised if they revert to boilerplate and deterministic appraisal. Should ones be worried? No, they'll get what they pay for. Risk-taking has a price, whether you pay it first to reduce some of the uncertainties, or whether you pay it later: someone's got to pay the premium.
 
Nicolas Graf
 
About the author
Nicolas Graf, Ph.D., joined the ESSEC faculty in January 2010 after having spent two years at the Hilton College of the University of Houston, Texas, and two years as Professor of finance and strategy at the Ecole hoteliere de Lausanne, Switzerland. He specializes in strategic management and finance in the hospitality industry, and his research and consulting activities concentrate on competitive strategies, supply and demand analysis, and risk modelling. He can be reached by email at: graf@essec.fr