New Beats Old Nearly Every Day: The Countervailing Effects of Renovations and Obsolescence on Hotel Prices Cornell Hospitality Report by John B. Corgel Vol. 8, No. 13, July 2008 www.chr.cornell.edu Advisory Board James C. Allen, Executive Vice President, Wines, Southern Wine and Spirits of New York Scott Berman, U.S. Advisory Leader, Hospitality and Leisure Consulting Group of PricewaterhouseCoopers Raymond Bickson, Managing Director and Chief Executive Officer, Taj Group of Hotels, Resorts, and Palaces Scott Brodows, Chief Operating Officer, SynXis Corporation Paul Brown, President, Expedia, Inc., Partner Services Group, and President, Expedia North America Raj Chandnani, Director of Strategy, WATG Benjamin J. “Patrick” Denihan, CEO, Denihan Hospitality Group Michael S. Egan, Chairman and Founder, job.travel Joel M. Eisemann, Executive Vice President, Owner and Franchise Services, Marriott International, Inc. The Robert A. and Jan M. Beck Center at Cornell University Kurt Ekert, Chief Operating Officer, GTA by Travelport Back cover photo by permission of The Cornellian and Jeff Wang. Kevin Fitzpatrick, President, AIG Global Real Estate Investment Corp. Gregg Gilman, Partner, Co-Chair, Employment Practices, Davis & Gilbert LLP Jeffrey A. Horwitz, Partner, Corporate Department, Co-Head, Lodgiing and Gaming, Proskauer Rose LLP Kenneth Kahn, President/Owner, LRP Publications Kirk Kinsell, President of Europe, Middle East, and Africa, Cornell Hospitality Report, InterContinental Hotels Group Volume 8, No. 13 (July 2008) Nancy Knipp, President and Managing Director, Single copy price US$50 American Airlines Admirals Club © 2008 Cornell University Gerald Lawless, Executive Chairman, Jumeirah Group Mark V. Lomanno, President, Smith Travel Research Cornell Hospitality Report is produced for Suzanne R. Mellen, Managing Director, HVS the benefit of the hospitality industry by Shane O’Flaherty, Vice President and General Manager, Mobil The Center for Hospitality esearch at Cornell Travel Guide University Carolyn D. Richmond, Partner and Co-Chair, Hospitality Practice, Fox Rothschild LLP David Sherwyn, Academic Director Richard Rizzo, Director, Consumer Intelligence, General Jennifer Macera, Associate Director Growth Properties, Inc. Glenn Withiam, Director of Publications Saverio Scheri III, Managing Director, WhiteSand Consulting Janice L. Schnabel, Managing Director and Gaming Practice Center for Hospitality Research Leader, Marsh’s Hospitality and Gaming Practice Cornell University Trip Schneck, President and Co-Founder, School of Hotel Administration TIG Global LLC 537 Statler Hall Barbara Talbott, Ph.D., EVP Marketing, Ithaca, NY 14853 Four Seasons Hotels and Resorts Elaine R. Wedral, Ph.D., President, Nestlé R&D Center and Phone: 607-255-9780 Nestlé PTC New Milford Fax: 607-254-2292 Adam Weissenberg, Vice Chairman, and U.S. Tourism, www.chr.cornell.edu Hospitality & Leisure Leader, Deloitte & Touche USA LLP Senior Partners Thank you to our American Airlines Admirals Club generous General Growth Properties, Inc. Corporate Members job.travel Southern Wine and Spirits of New York Taj Hotels Resorts Palaces TIG Global LLC Partners AIG Global Real Estate Investment Davis & Gilbert LLP Deloitte & Touche USA LLP Denihan Hospitality Group Expedia, Inc. Four Seasons Hotels and Resorts Fox Rothschild LLP HVS International InterContinental Hotels Group JohnsonDiversey Jumeirah Group LRP Publications Marriott International, Inc. Marsh’s Hospitality Practice Mobil Travel Guide Nestlé PricewaterhouseCoopers Proskauer Rose LLP Smith Travel Research SynXis, a Sabre Holdings Company Thayer Lodging Group Travelport WATG WhiteSand Consulting Friends American Tescor, LLP • Argyle Executive Forum • Caribbean Hotel Restaurant Buyer’s Guide • Cody Kramer Imports • Cruise Industry News • DK Shifflet & Associates • ehotelier.com • EyeforTravel • Fireman’s Fund • 4Hoteliers.com • Gerencia de Hoteles & Restaurantes • Global Hospitality Resources • Hospitality Financial and Technological Professionals • hospitalityInside.com • hospitalitynet.org • Hospitality Technology • Hotel Asia Pacific • Hotel China • HotelExecutive.com • Hotel Interactive • Hotel Resource • International CHRIE • International Hotel and Restaurant Association • International Hotel Conference • International Society of Hospitality Consultants • iPerceptions • KPMG Japan/Global Management Directions • Lodging Hospitality • Lodging Magazine • Milestone Internet Marketing • MindFolio • Parasol • PKF Hospitality Research • RealShare Hotel Investment & Finance Summit • Resort+Recreation Magazine • The Resort Trades • RestaurantEdge.com • Shibata Publishing Co. • Synovate • The Lodging Conference • TravelCLICK • UniFocus • WageWatch, Inc. • WIWIH.COM New Beats Old Nearly Every Day: The Countervailing Effects of Renovations and Obsolescence on Hotel Prices By John B. Corgel, Ph.D. About the Author The founding director of the Cornell Center for Hospitality Research, John B. Corgel, Ph.D., is Robert C. Baker Professor of Real Estate at the Cornell University School of Hotel Administration (jc81@cornell.edu). Formerly a visiting scholar at the Federal Home Loan Bank Board in Washington, D.C., he is a fellow of the Homer Hoyt Institute. He also maintains a consulting relationship with PKF Hospitality Research, where he is helping to develop new products for the hotel industry based on property-level financial performance information. He is the author of over 65 articles in academic and professional journals, including Real Estate Economics, Journal of Urban Economics, Journal of Risk and Insurance, Journal of the American Business Law Association, and Cornell Hospitality Quarterly. His textbook, Real Estate Perspectives (with Smith and Ling), was used throughout the nation for introductory real estate courses. An academic version of this paper is published as: John B. Corgel,“Technological Change as Reflected in Hotel Property Prices,” Journal of Real Estate Finance and Economics Vol. 34, No. 2 (2007), pp. 257–279. 4 The Center for Hospitality Research • Cornell University executive SummAry s is the case with other commercial real estate types, hotels begin to depreciate from the Atime they open, in a process largely driven by functional obsolescence. Unlike other asset types, however, hotel values hit an inflection point at which they begin to rise again. Average annual depreciation for the 3,810 chain-affiliated hotels in this sample was within the range found in other commercial types of real estate. Depreciation rates start off relatively brisk in the first few years, because hotel owners typically do not begin renovations until around year ten. When owners do begin renovation, those expenditures slow but do not stop the decline in the typical hotel’s value. Then, around year twenty-eight, the depreciation reverses for hotels that are still in business. Not only have renovations stabilized the loss in value, but other, unknown factors promote the hotel’s value— a phenomenon that could be called a vintage effect. Such fully depreciated properties may be located in particularly favorable sites, or they may have architectural or other features that make them attractive to investors. Cornell Hospitality Report • July 2008 • www.chr.cornell.edu 5 cornell hoSpitAlity report New Beats Old Nearly Every Day: The Countervailing Effects of Renovations and Obsolescence on Hotel Price By John B. Corgel, Ph.D. The large volume of hotel real estate transactions completed during the past several years signifies the culmination of an important change occurring in the lodging sector, in which hotel operations have become separate from property ownership. Major hotel companies that held and managed properties have been net sellers of those properties as they seek growth from domestic and international management and franchise contract expansion, while leaving others to realize the value of the real estate. This so-called “asset lite” strategy also was motivated by unprecedented escalation of real estate prices, coupled with an intense interest in real estate as an investment.1 Property buyers included traditional and new hotel investors, such as private equity funds and non-hotel REITs. These investors recognized opportunities for excess returns despite historically high prices. 1 See: “Asset Lite or Asset Right?,” Hotels Investment Outlook, June 2006, pp. 22–24. 6 The Center for Hospitality Research • Cornell University Substantial property transaction volume represents a to be 2 percent. With those values, the future price Pt,n movement toward greater efficiency in the industry as new would be $10.5 million, calculated as follows: owners find creative ways to enhance profits. Ironically, $10 million (1 + .07) - $10 million (1+ .02). [2] record transaction volume also represents considerable I discuss here the several determinants of the apprecia- disagreement about future asset price paths, since every tion rate, which has a number of alternative treatments in hopeful buyer is matched by a willing seller. As they decide research and practice. In contrast to the multiple determi- whether to dispose of or acquire properties, both sellers and nants to appreciation, depreciation is mainly driven by the buyers naturally concentrate on the assets’ present values age of the property. In this report I analyze the relationship from discounted cash flow (DCF) models and the recent sale between hotel prices and the age of property with the idea prices of comparable properties as benchmarks. Future sale of sharing insights into future price patterns of properties prices receive some attention in DCF models, but relatively of various ages to allow buyers and sellers to make better in- insignificant components of periodic cash flows often are vestment decisions. Also, the results presented in this report given far more attention. Despite the importance of future should assist owners who plan to hold their properties in prices to decisions regarding acquisition and disposition of making informed renovation decisions. hotel assets, the worlds of institutional investment and real estate investment research have been curiously ambivalent Conventional Wisdom about Property Prices and Age about understanding the determinants of future prices The conventional wisdom regarding a hotel’s age and its val- beyond cash flows. ue is that the prices of hotels naturally fall as the properties One way to think about the prospective prices of hotel age, holding others factors constant. As a related issue, then, properties is to separate the factors that will cause properties if a hotel’s value does decline with age, I wanted to deter- to appreciate from those that would be expected to cause mine whether that depreciation is a straight line or whether property prices to decline. We can express that concept as hotel price changes show some kind of curve. Asset values follows. Say that Pt represents the current price in period t, could, for instance, rapidly decline early in their operating Pt,n the future price, a the average appreciation rate from t to lives, similar to the depreciation of new automobiles, or, per- n for hotels reaching stabilized occupancy, and d the average haps a hotel asset’s value plummets at the end of its life, after rate of economic depreciation from t to n. The relationship a long, steady decline. Determining the nature of hotel price between current prices and future prices can be specified as depreciation requires a set of arguments for understanding follows: the conceptual relationships between time, hotel prices, and Pt, n = Pt (1 + a) - Pt (1+ d) [1] major renovations that are aimed at retarding age-related Now say that Pt equals $10 million, the holding period price erosion. This report presents empirical research to de- (n) is one year, a is estimated at 7 percent, and d is expected velop numerical estimates of the price and age relationship. Cornell Hospitality Report • July 2008 • www.chr.cornell.edu 7 Exhibit 1 possible curves to describe depreciation (d), or changes in hotel prices as a property ages A construction cost = price = value b c D e value of land t i m e Based on the study’s findings, I offer implications regarding Physical deterioration and the readily apparent result of the relationship between the prices of hotel properties for aging increase operating expenses, thus lowering cash flow. the benefit of asset owners and potential buyers. The effect on cash flows from obsolescence, which “results The Basics of Real Estate Depreciation when older things function as when they were new but otherwise lose their appeal or usefulness,” is less apparent.3 The standard economic theory of real estate depreciation, Note that an obsolescent hotel is still operable, even if it is which is defined as “the reduced ability of an asset to gener- no longer fashionable. ate future cash flows,” holds that financial benefits erode as The Appraisal of Real Estate identifies the following two properties naturally age and as market conditions change.2 types of obsolescence: functional and external. Functional 2 obsolescence usually occurs because of changes in consumer G.W. Blazenko and A.D. Pavlov, “The Economics of Maintenance for Real Estate Investments,” Real Estate Economics, Vol. 32, No. 1 (2004), preferences for physical features inside the property. These p. 57: “The many studies of capital asset depreciation performed during preferences are satisfied by technological changes introduced the past four decades greatly advanced the state of knowledge about into competing properties. Economists refer to a variation of how real estate loses its ability to produce future service or cash flows functional obsolescence as technical obsolescence. In either over time and about how to empirically measure the rate of economic depreciation.” For a review of the early housing literature, see: S. Malpezzi, case, technological change instigates obsolescence. L. Ozanne, and T.G. Thibodeau, “Microeconomics Estimates of Housing Businesses such as hotels, which operate at fixed loca- Depreciation,” Land Economics, Vol. 63, No. 4 (1987), pp. 372–385. An tions within an integrated real estate market, may experi- updated review appears in: B.C. Smith, “Economic Depreciation of ence external or locational obsolescence.4 This form of Residential Real Estate: Microlevel Space and Time Analysis,” Real Estate Economics, Vol. 32, No. 1 (2004), pp. 161-180. A recent discussion of 3 S.E. Margolis, “Depreciation and Maintenance of Houses,” Land Eco- research on the economic depreciation of commercial property is found nomics, Vol. 57, No. 1 (1981), p. 91. in: T.J. Dixon, N. Crosby, and V.K. Law, “A Critical Review of Methodolo- gies Measuring Rental Depreciation Applied to Commercial Real Estate,” 4 Appraisal Institute, The Appraisal of Real Estate, 13th Edition (Chicago: Journal of Property Research, Vol. 16, No. 2 (1999), pp. 153-180. Appraisal Institute, 2008). 8 The Center for Hospitality Research • Cornell University h o t e l p r o p e r t y p r i c e Exhibit 2 Depreciate curves for retail and hotel properties based on empirical studies v-shaped–hotels (per this study) Kinked–retail (per colwell and ramsland) 0 16 28 y e a r s See: P.F. Colwell and M.O. Ramsland, “Coping with Technological Change: The Case of Retail,” Journal of Real Estate Finance and Economics, Vol. 26, No. 1 (2003), pp. 47–63. obsolescence derives from factors external to the property, Colwell and Ramsland find empirical support for the for example, an oversupplied market, or adjacent-parcel main hypothesis derived from theory.6 They calculated that environmental contamination. Price effects due to external functional obsolescence during the initial sixteen years of a obsolescence generally do not result from technological retail property’s operation is 1.7 percent per year. By contrast, change. the long-run rate of commercial property obsolescence is Research Indicates Complex Relations between 0.9 percent per year. These estimates support the concept Real Estate Prices and Asset Age that obsolescence, assuming a constant rate of technological change, can be directly observed early in the life of proper- Colwell and Ramsland demonstrate how technological ties in the absence of renovation expenditures. Moreover, change, as the underlying cause of property obsolescence, their numbers imply that renovations to retail properties are affects retail real estate prices.5 They argue that even during effective in eliminating approximately one-half of all func- the years immediately following a new property’s opening, tional obsolescence in a commercial property. changing technologies begin to push that property toward Exhibit 1 shows five possible paths that Pt,n from obsolescence. Because renovations are seldom undertaken Equation [1] may follow over time, depending on the as- early in a hotel’s life, obsolescence may be observed during sumptions one applies. Note particularly the behavior of d this interval without the offsetting effects of capital expendi- (the depreciation rate) in paths C, D, and E, in which price tures for major renovations targeted to defeat obsolescence. always declines with the passage of time. Current Federal tax 5 law uses Path D as its assumption for depreciation calcula-P.F. Colwell and M.O. Ramsland, “Coping with Technological Change: The Case of Retail,” Journal of Real Estate Finance and Economics, Vol. 26, No. 1 (2003), pp. 47-63. 6 Ibid. Cornell Hospitality Report • July 2008 • www.chr.cornell.edu 9 h o t e l p r o p e r t y p r i c e Exhibit 3 capital expenditure patterns by age of hotel property panel A ratio of capital spending to total revenues by age of property and market segment Study Full Service 1995 2000 2007 Built before 1981 9.6% Built before 1983 6.4% Built before 1990 5.4% Built between 1981 & 1986 4.3% Built between 1983 & 1993 6.5% Built between 1990 & 2000 3.5% Built after 1986 2.6% Built after 1993 0.9% Built after 2000 2.3% Select Service 1995 2000 2007 Built before 1981 5.2% Built before 1983 6.5% Built before 1990 8.7% Built between 1981 & 1986 4.0% Built between 1983 & 1993 4.8% Built between 1990 & 2000 2.9% Built after 1986 2.3% Built after 1993 3.0% Built after 2000 0.7% panel b range of average annual capital expenditures for all hotels, by age (2000) Age in years minimum maximum 1 1.65% 4.51% 2 1.72% 3.29% 3 1.43% 3.15% 4 1.31% 3.64% 5 3.21% 6.23% 6 4.80% 6.77% 7 4.15% 5.85% 8 3.60% 5.23% 9 4.83% 7.01% 10 8.43% 11.94% 15 3.35% 5.72% 20 2.37% 8.68% 25 5.05% 10.24% tions for tax purposes. Properties placed into service now depreciation generally follows the V-shaped depreciation must be depreciated using the straight-line method, although curve shown in Exhibit 2, in contrast to the one that the depreciation periods vary for different classes of com- Colwell and Ramsland identified for retail properties. mercial property. For example, the period is 27.5 years for Summary of Statistical Results from this Study multi-family properties and 39 years for other commercial properties, including hotels. In contrast, economic research Having said that hotel depreciation follows a V-shaped and feasibility studies assume a concave path, such as Path C, curve, I note that the data in this study yield distinctive for depreciation. results regarding the relationship between a hotel property’s You’ll note that the graph in Exhibit 1 shows only simple price and its age. To begin with, the estimated rate of hotel and relatively smooth patterns. In reality, depreciation property obsolescence following construction is 1.9 percent patterns show combinations of all five of those paths. The per year, which aligns with the retail property estimate of obsolescence function estimated with retail property data by 1.7 percent. However, the breakpoint for when this gradual Colwell and Ramsland, for instance, has the kinked shape decline ends does not occur until year 28, compared to shown in Exhibit 2, on the previous page. They find that, year 16 for retail properties, as identified by Colwell and holding other factors constant, existing retail properties Ramsland. The late date for the inflection point occurs prices always decline as they mature relative to newer despite substantial follow-on investment in hotels, which properties. Indeed, as I explain below, I found that hotel begins around year 10 and continues to increase thereafter. 10 The Center for Hospitality Research • Cornell University The hotel data indicate that renovation investments do not self-service (or vanished entirely in the mid-price segment) significantly restrain price declines that stem from techno- represents an example of process change in hotels. logical change fairly until late in a hotel’s economic life. After Concerns by owners and managers about how much the breakpoint, the sale price appreciates by a surprising 0.7 money should be spent or put into reserve to keep hotels percent per year. Combining the two numbers yields a rough competitive prompted three surveys of hotel capital expen- estimate for the long-run obsolescence rate of 1.23 percent. ditures since 1995, conducted by the International Society An evaluation of the equation with statistical results from of Hospitality Consultants. The ISHC asks questions about this study (presented later in this report) at the average age actual expenditures for the following categories:9 of eighteen years indicates an obsolescence rate of 1.35 per- • Updating design and décor, cent. Both estimates modestly exceed the 0.9 percent found • Curing functional and economic obsolescence, thereby for retail real estate. extending both the physical and economic life of the I explain the upward sloping portion of the V-shaped asset, depreciation function presented in Exhibit 2 as a “vintage ef- fect” driven by a demand for surviving hotels. Goodman and • Complying with franchisors’ brand requirements, Thibodeau find a similar effect in housing markets.7 They • Technological improvements, attribute this result to some distinctive characteristic of the • Product changes to meet market demands, house or neighborhood that is correlated with age. Buyers • Adhering to government regulations, and may be willing to pay a premium for large porches on older homes, for instance, or houses near employment centers • Replacing all short- and long-lived building compo- may be in demand because they offer economies in trans- nents due to wear and tear. portation costs. Similar design or location factors may be Unfortunately, these reports do not identify expenditures by 10 responsible for the positive price-age relationship for hotels specific purpose. after the inflection point, but further investigation of this As shown in Panel A of Exhibit 3, the 2007 report phenomenon is not a part of the study reported here. summarizes capital expenditure ratios, age of property, and market segment from the three ISHC surveys. Expenditures Technological Change and Hotel Properties at full-service hotels either have exceeded or equaled those Even though the use pattern for hotels differs from that of at select-service hotels, except in the most recent survey for other commercial real estate, the large volumes of customers the oldest age category. Expenditure patterns by age of all who regularly pass through both hotels and retail establish- hotels, as shown in Panels A and B, confirm that relatively ments represent a common trait that makes both retail and small amounts of capital are spent during the initial years hotel real estate particularly vulnerable to technological following construction. Expenditures and expenditure vari- change. Colwell and Ramsland identify the following cat- ances increase steadily thereafter. In addition, expenditures egories of technological change in retail real estate: physical tend to be concentrated at points in the property life cycle (e.g., building materials and security cameras), contractual when renovations occur (e.g., year 10). The 2007 ISHC re- (e.g., percentage leases and commercial mortgage backed port shows expenditure spikes for full-service hotels in years securities debt, CMBS), and process innovations (e.g., live 6 and 14, with a large spike at year 23. demonstrations).8 Finally, property obsolescence is filtered by a hotel’s The hotel industry has experienced numerous innova- affiliation with a recognized hotel company’s brand. Hotel tions during the past few decades. From a design perspective, chains incur substantial monitoring costs to prevent proper- suite-style rooms increased in number relative to traditional ties from becoming obsolete. Consequently, responses to rooms, exterior-corridor hotels almost disappeared in favor changing technology occur fairly rapidly, incrementally, of interior corridors, atrium entrances gained (and lost) pop- and uniformly across brands within the same company and ularity, and the movement toward more wired and wireless across competing companies. These conditions create an environments has been a design focal point. Contractually, environment different from other property types with regard numerous advancements have occurred to strengthen man- to technological change and obsolescence. agement and franchise relationships. The manner in which 9 CapEx: A Study of Capital Expenditures in the U.S. Hotel Industry (Mem- food and beverage service delivery has evolved toward phis, TN: International Society of Hospitality Consultants, 1995); CapEx: A Study of Capital Expenditures in the U.S. Hotel Industry (Memphis, TN: International Society of Hospitality Consultants, 2000), p. 2; and CapEx: 7 A.C. Goodman and T.G. Thibodeau, “Age-Related Heteroskedasticity in A Study of Capital Expenditures in the U.S. Hotel Industry (Alexandria, VA: Hedonic House Price Equations,” Journal of Housing Research, Vol. 6, No. International Society of Hospitality Consultants, 2007. 1 (1995), pp. 25–42. 10 The reports do present detail on expenditures at various locations with- 8 Colwell and Ramsland, op.cit. in the hotel (e.g., the lobby) and for specific items (e.g., wall coverings). Cornell Hospitality Report • July 2008 • www.chr.cornell.edu 11 Exhibit 4 models of the effects of technological obsolescence and vintage attributes on hotel prices panel A panel b price effects of technological change, or obsolescence price effects of vintage attributes $ c(x) $ c(x1) c(x) p(x) p(x1) p1* p2* p(x) p* p* x* x1* x x* x2* x Conceptual Model of Optimal Property Configurations tage price effects (positive) appear as movements in oppos-ing directions from the optimal configuration. New optimal I begin the conceptual story that underlies the statistical configurations arise, therefore, either because of shifts in the analysis of hotel properties’ price changes with age by as- cost curve due to changes in input prices and technology suming that the cost of building hotels increases linearly as or because of shifts in the price curve from demand-related more rooms and amenities are added, but that the value or repricing of attributes and from changes in expenses associ- purchase price of hotels levels off and ultimately diminishes ated with owning the attributes. as more rooms and amenities populate the market (i.e., di- Exhibit 4, Panel A shows the optimal configuration of minishing marginal utility). Both the construction cost and a new property, x*, at the intersection of C(x) and P(x). An purchase price originate from a collection of the property increase in costs resulting from advancements in technology, attributes (notably, location, type of rooms), signified as x. for example, shifts the cost curve to C(x1), thereby produc- Thus, C(x) represents the cost of placing a new property in ing optimal configuration x1*. The higher revenue earned service with a modern collection of x attributes, from properties with x1* relative to x* translates into price By assuming that P(x) represents the current price differential p1* > p*. Most seasoned properties continue to of property in service with quantities of x attributes, the operate with obsolete configuration x* prior to renovation. equilibrium solution involves determining the property at- Unanticipated market changes appear as different tribute configuration, x*, that maximizes net present value, configurations for seasoned properties relative to new or PV(x). This is the optimal property configuration. Further, properties. Typically, older property configurations produce a competitive market is assumed so that P(x) = PV(x), and, lower prices than do newer property configurations, and therefore: the increment of depreciated price reflects the extent of a P(x*) = C(x*) [3] seasoned property’s obsolescence. A portion of this price The property’s purchase price and construction cost differential comes from technological change, while the functions reach a point of tangency at the optimal property balance comes from such other depreciation drivers as configuration. Obsolescence price effects (negative) and vin- physical deterioration and external obsolescence. Isolating 12 The Center for Hospitality Research • Cornell University Exhibit 5 Descriptive statistics for hotel property transaction sample panel A Statistics for Selected variables variable Symbol n mean σ minimum maximum Sale Price P 3,810 $12.4 M $24.6M $.5M $365M Number of Rooms RM 3,810 167 155 20 2940 Age A 3,810 18 15 1 202 Published Room Rate R 3,810 $94.18 $61.23 $19 $950 Per Capita Income PI 3,810 $26,572 $13,799 $6,428 $148,052 panel b Statistics by category Sale price category Symbol n mean σ minimum maximum Market Segment Deluxe DEL 44 $105M $73.7M $8.5M $355M Luxury LUX 409 $46.3M $45.9M $1.8M $365M Upscale UPS 400 $19.7M $15.1M $1.2M $96M Upper-Tier Extended Stay UES 92 $12.4M $7.6M $1.2M $74.5M Midscale W/ F&B MW 753 $7.1M $7.6M $.6M $80M Lower-Tier Extended Stay LES 237 $5.1M $3.9M $.6M $26.7M Midscale W/Out F&B MWO 800 $4.8M $3.8M $.5M $52M Economy ECO 586 $3.2M $3.5M $.5M $5.3M Budget BUD 489 $2.5M $1.6M $.5M $10.5M Age CAtegory Zero to Ten Years N/A 1274 $12.4M $21.6M $.5M $275M 11 to 20 Years N/A 1237 $12.1M $23M $.5M $355M 21 to 30 Years N/A 694 $11.9M $25.5M $.6M $321M 31 to 40 Years N/A 404 $8.5M $23.3M $.5M $365M Over 40 Years N/A 201 $23.3M $42M $.5M $243M yeAr of SAle 1996 T96 462 $14.2M $21.4M $.5M $198M 1997 T97 499 $15.1M $23.1M $.5M $190M 1998 T98 404 $17.5M $29.9M $.7M $197M 1999 T99 372 $12.1M $25.8M $.5M $275M 2000 T00 502 $10.2M $26.4M $.5M $365M 2001 T01 407 $9.9M $23M $.6M $250M 2002 T02 390 $8.2M $17.9M $.6M $214M 2003 T03 463 10.9M $22.9M $.5M $321M 2004 T04 311 13.1M $30.9M $.5M $355M the contribution that technological change makes to Empirical Study property obsolescence requires empirical models that The transaction data consist of 3,810 hotel real estate sales include variables to control for these other determinants of that occurred throughout the U.S. from January 1996 overall economic depreciation. through early 2004. Information about property sale prices The process of property obsolescence is complicated by and characteristics come from a database managed by PKF any shifts in the price curve, for example, due to demand- Hospitality Research. This firm obtains hotel transaction related repricing of x attributes. Panel B of Exhibit 4 shows information through subscriptions with CoStar and Hotel an upward shift of the price function from P(x) to P(x1) Brokers International. Transactions data also come from in- without technological change. An increase in the demand dustry publications, news reports, and the firm’s consultants. for seasoned properties with x2 attributes means that these This firm researches sales to verify and to fill in missing properties command p2, where p2 > p*. These properties information. Demographic data, such as ZIP code per capita thus benefit from a vintage effect. income, come from CACI. Cornell Hospitality Report • July 2008 • www.chr.cornell.edu 13 Starting with these data I removed full-service proper- Exhiibit 6 ties with fewer than 75 rooms, limited-service properties with under 20 rooms, and any hotel with a sale price of less variables used in hotel price regression equation than $500,000. Rather than treat conference center and DEpEnDEnt VariablE resort hotels as separate categories, I merged them into the Natural log of sale price divided by number of rooms (price per room) appropriate full-service segments. Finally, I retained only inDEpEnDEnt VariablEs hotels with a nationally recognized brand affiliation and Focus variables screened out properties with either no affiliation or a re- 1. Age of Property at Time of Sale (A) 2 gional brand. This step ensures reasonable consistency in the 2. A sample with respect to maintenance, repairs, and, to a lesser control variables extent, renovation.11 3. Series of Dummy variables (1, 0) for Hotel Market Segments (MS) Definitions of the variables and their summary statistics 4. Series of Dummy Variables for U.S. States (S) 5. Personal Income of ZIP Code in which Property is Located (PI) appear in Exhibit 5, on the previous page. Transactions are 6. Adjusted Published Room Rate (R) evenly distributed by the year of sale and by market segment. 7. Year Property Sold (T) Market segment assignments for each property are based on PKF Consulting’s assessment of its brand homogeneity, and, thus, like collections of property characteristics. As shown in Exhibit 5, the deviations of sale prices from the mean of $12.4 million are considerable. Using a price-per-room appear in the estimating equation to measure physical de- calculation narrows this variation for estimation purposes. terioration (i.e., condition) and external obsolescence, such that the price equation now becomes Summary of Age Statistics Pi = ƒ (Xi, Age, Condition, Location) [5] The average age of the properties in the study is 18 years, The data base available for this study generally lacks with a standard deviation of 15 years. As shown in Exhibit details on the properties’ attributes beyond the number of 5, two-thirds of the transactions involved hotels up to age rooms and age. However, each segment has its own set of 20, although every age cohort up to 40 years has at least 400 attributes that must be controlled through a fixed-effects transactions. The hotels under 20 years old tended to be treatment of each of the nine market segments. Thus, the smaller than the sample average and, not surprisingly given equation requires a separate dummy variable for each trends in hotel development during the past 25 years, those market segment, as well as a dummy variable for properties relatively new hotels were heavily represented in the luxury, with all-suite rooms. As mentioned earlier, market segment mid-price without food and beverage, and upscale seg- designations are assigned according to consistency of prop- ments. By contrast older hotels were heavily represented in erty attributes. Note that the size of the hotel as measured the mid-price with food and beverage segment. In sum, the by the number of rooms (RMi), which often accounts for 30 sample of property transactions used for this study appears percent or more of the variation in hotel sale prices, enters to resemble the population of U.S. hotels in operation during on the left side of the equation through the price-per-room 12 the study period. variable. Thus, (P / RM ) = P = ƒ (X , Age, Condition, Location) [6] Transaction Price Equation i i i iAdjustments for condition occur in two ways. First, The price model represents hotel property sale price as a limiting the sample to nationally affiliated hotels provides function of various property characteristics, X , and overall consistency for physical condition. This does not mean i property depreciation. That is, that every affiliated property has exactly the same level of Pi = ƒ (Xi, Overall Depreciation) [4] deferred maintenance, only that the level of deferred mainte- Real estate appraisers divide depreciation into the fol- nance does not substantially exceed that of other properties lowing three categories: physical deterioration (i.e., normal in the same brand and segment. In short, the obsolescent wear and tear), external obsolescence (i.e., location or eco- and often decrepit properties that have lost their flag are not nomic), and functional obsolescence. Property age accounts included here. Differences among non-homogeneous brands for price variation due to functional obsolescence, if controls are picked up by the market segment variable. 11 Monitoring of these standards occurs through inspections and an institutional process known as the Property Improvement Program (PIP). If a hotel has been ‘PIPed’, then the property meets all of the cur- 12 See: J.B. Corgel and J.A. deRoos, “The ADR Rule-of-Thumb as Predic- rent standards of the sponsoring company. This event ordinarily involves tor of Lodging Property Values,” International Journal of Hospitality technology and other physical upgrades, all except extensive renovations. Management, Vol. 12, No. 4 (1994), pp. 353–365. 14 The Center for Hospitality Research • Cornell University Second, the instrumental variable, Ri^, which is derived Exhibit 7 from the published room rate, contains information related to the condition of the property.13 The transaction database regression results for all hotels contains the published room rate for double occupancy, Dependent variables ln(p/rm) which overstates the actual average daily rate, but is posi- variable label cofficient Std. error tively and highly correlated with realized ADR.14 If directly A Age at Date of Sale -.0171* .0012 2 introduced into the price equation, the published room rate A Age Squared .0001* .0001 provides an effective control for quantity, quality, and condi- R^ Published Rate Instrument .1534* .0083 PCI ZIP Per Capita Income .0001* .0001 tion differences among the rooms and properties in this DEL Deluxe 2.0296* .0840 sample. Published room rate, however, contains a serious LUX Luxury 1.1722* .0358 statistical problem, because it correlates closely with other UPS Upscale .8131* .0364 explanatory variables, such as the age of the property, and UES Lower-Tier Extended Stay .1904* .0583 MW Midscale W/ F&B .2537* .0318 with the equation error term. Consequently, I instituted LES Upper-Tier Extended Stay 1.0261* .0412 econometric procedures to adjust published room rate and MWO Midscale W/Out F&B .2983* .0309 create an instrumental variable so that this variable can be ECO Economy .0300 .0367 used in the price equation without violating statistical rules. AS 1= All Suites .2931* .0367 T97 1 = Sold in 1997 .1467* .0330 Location adjustments are accomplished in a general T98 1 = Sold in 1998 .2145* .0352 way with fixed-effects treatment of the states in which the T99 1 = Sold in 1999 .1808* .0364 property sale occurred. More specifically, the per capita in- T00 1 = Sold in 2000 .1643* .0339 come level for each property’s ZIP code serves as a measure T01 1 = Sold in 2001 .1709* .0357 T02 1 = Sold in 2002 .1231* .0364 of locational obsolescence. Many of the ZIP codes with the T03 1 = Sold in 2003 .1193* .0365 lowest per capita income in this data base are in and around T04 1 = Sold in 2004 .1387* .0387 downtown areas, for example. C Intercept 10.2624* .1968 Both A and A2 enter the equation assuming a concave (S1,…,SJ) States (Not Shown) R2 Adjusted .5347 relation between asset price per room and age such that the RMSE Root Mean Square Error .5113 expected sign of A is negative and the sign of A2 is posi- Note: * Significant at p < .01. Ln is the natural log. tive. The coefficients on the age variables would normally indicate the rate of economic depreciation in hotels. Due to the controls in this model, however, age coefficients instead indicate the extent to which hotel properties lose value as a transactions and standard regression procedures. The inde- consequence of functional obsolescence. Exhibit 6 presents pendent variables in the model explain 53.47 percent of the the variables used in the regression model. variation in the natural log (ln) of price per room. Nearly Given the assumption that technological change occurs all of the independent variables in the price equation are at a constant rate over time, the pure effect of that change on significant and have the expected signs. All market-segment obsolescence can be estimated by comparing data on the sale variables except one (namely, economy) are statistically of recently constructed properties to those of older proper- significant at the 1-percent level. Note that the coefficients ties. In this study I replicated the advanced econometric are largest for the highest price market segments (i.e., deluxe procedure applied by Colwell and Ramsland to find relative and luxury). Also note that the coefficients for all of the T obsolescence rates. variables have a positive sign, indicating that hotel property Statistical Results prices increased every year relative to 1996 prices, control- ling for all other factors in this equation. Exhibit 7 presents the results from estimating the equations The estimated negative sign (which is significant) on the presented above using the entire sample of hotel property age coefficient and the positive sign (also significant) for age squared in the price-per-room equation confirm a concave 13 An instrumental variable closely substitutes for another variable either relationship between hotel property prices and age. This because the original variable cannot be collected or its inclusion creates general pattern appears as Path E in Exhibit 1 and is similar econometric problems. to economic depreciation rate patterns found for other prop- 14 Due to seasonal variation in room rates, industry analysts generally erty types.15 With controls in place for physical condition make cross-sectional comparisons using an annualized rate. Thus, when a hotel sale occurs and the room rate is identified that rate will be an annual and external obsolescence, hotels on average decline in price average. The published rates in these data are annual averages of seasonal rates cited in travel guides. Annualized published rate and ADR are highly 15 Blazenko and Pavlov, op.cit.; Smith, op.cit.;.and Dixon, Crosby, and Law, correlated (about 0.9), differing mostly by a scale factor. op.cit. Cornell Hospitality Report • July 2008 • www.chr.cornell.edu 15 A “vintage effect” counteracts depreciation for the average hotel, usually after about 28 years. through functional obsolescence at a decreasing rate. Nev- because renovation has already occurred to the extent profit- ertheless, the size of the coefficient on A2 is quite small. The able. Second, a higher rate of functional obsolescence occurs rate of functional obsolescence in the first year, derived from in the early years because renovations do not counteract the coefficients on A and A2, equals 1.69 percent (-.0171 + obsolescence. (2 (1)* .0001). By year 18, the average of the hotel transac- Finally, a vintage effect of approximately 0.7 percent is tion sample, the obsolescence rate has faded to 1.35 percent detected following the critical year. By contrast, Colwell and (-.0171 + (2 (18) * .0001). These estimates lie between the Ramsland found a continuation of price erosion beyond the long-run rate of obsolescence estimated by Colwell and critical year for other types of commercial property. These Ramsland for shopping centers of 0.9 percent and the rate of two paths were presented in Exhibit 2. economic depreciation of 2.7 percent estimated by Fisher et. Summary of Findings and Recommendations al. for apartments. (Note that an economic depreciation rate should exceed the obsolescence rate.)16 Technological change affects the prices of many seasoned Technological Change and Obsolescence products and assets by accelerating their obsolescence. One purpose of property renovations is to counter obsolescence. This study also applies the procedure used by Colwell and The main findings from this research indicate that hotel Ramsland to detect a breakpoint in the price and age func- renovations do, indeed, counter obsolescence. To begin with, tion. They introduced a variable in the form (A–Ā), where A the long-run or average annual rate of obsolescence for the is the age of the property at time of sale and Ā is an un- typical hotel lies between 1.23 and 1.35 percent. Stated dif- known critical age where a breakpoint occurs. The critical ferently, the typical hotel tracking along a straight-line path age comes from repeatedly running regressions each time would be totally obsolete in 77 years (1/77 = 1.3 percent). with a successively greater age until R2 reaches a maximum. Having said that, I note that typical hotel may be demolished Applying this procedure to the hotel property data returned well before the age of 77 due to factors unrelated to techno- a single critical value of 28 years. The significant coefficient logical change and functional obsolescence, such as external on the age variable decreases from -.0171 to -.0193. Inter- obsolescence. The hotel obsolescence rate lies between the pretation of these findings is taken as confirmation of two obsolescence rate generated from a recent study of retail hypotheses derived from the theory. First, the functional properties (0.9 percent) and the economic depreciation rate obsolescence observable in asset prices stops at some criti- found in another recent study of apartments (2.7 percent). cal age, and thereafter no additional obsolescence appears We see the effects of renovations in the shape of the obsolescence function in this study, which is not a straight 16 J.D. Fisher, B.C. Smith, J.J. Stern, and B. Webb, “Analysis of Economic line, but is instead concave. Thus, the obsolescence rate ex- Depreciation for Multi-Family Property,” American Real Estate and ceeds the range given above during the early years of service, Urban Economics Association, Atlanta, GA, January 4–6, 2002. because no renovations are typically made on brand-new 16 The Center for Hospitality Research • Cornell University properties. The estimates developed in this study indicate Investor Recommendations a range in the obsolescence rate of 1.7 and 1.9 percent per year for the early years of a hotel’s operation. Again using The investment implications of the findings from this study a straight-line-relationship assumption for simplicity, the are as follows: typical hotel would become obsolete in 56 years (1/56 = 1.8 • During periods of rapidly changing technology, the percent) or sooner if no major renovations were undertaken. newest and oldest properties are less susceptible The most telling finding is that the shape of the hotel to price declines due to aging than are hotels in property obsolescence curve does not follow that of other mid-life. commercial real estate, because of the inflection point in • Renovation expenditure decisions are the most the curve. Hotel prices fall at a rate of 1.9 percent for the difficult for properties over 10 years old and less first 28 years of operation then enter a period during which than 30 years old. Money will be well spent on they increase at 0.7 percent per year (until demolition). properties in this age group that are positioned in We could say that during the first 28 years, obsolescence good markets at proven site locations because they dominates renovations. Hotel industry data indicate that will be most likely move to maturity (i.e., 28 years aggressive renovations do not begin until around year ten. old) and hold their price after that point without The results from this study confirm the merits of such a extensive investment. Other properties should be strategy because obsolescence is not severe during the first allowed to “filter down” the ADR and chain scales ten years of operation. The renovation activity that occurs with minimal reinvestment. from years 11 through 28 has the appearance of “swimming • Projections of value increases in DCF modeling upstream.” The data indicate that renovations beginning in should directly reflect the age of properties. Two year eleven keep obsolescence from doing more damage to properties in the same market, one 15 years old property prices, but does not prevent aging from inflicting and the other 30 years old, will be affected differ- losses in property valuation. On average, renovations offset ently by age during a subsequent holding period of approximately 0.5 percent per year of price erosion due to five to ten years. Counter to conventional thinking, obsolescence. Renovations to properties that survive until smaller differences between going-in and terminal their twenty-eighth year will more than offset the effects of capitalization rates can be justified for a 30-year- obsolescence. old property than for a comparable 15-year old property. n Cornell Hospitality Report • July 2008 • www.chr.cornell.edu 17 ExecEdPathAd_chr-2c_ƒ.qxd 12/7/06 9:15 AM Page 1 Cornell Short Courses and Certifications for Hotel Industry Professionals: Anheuser-Busch The General Managers Program The Professional Development Program Tackle strategic hotel management issues and find Study and share experiences with peers from around the world relevant, specific solutions. Work with a global network in these intensive hospitality management seminars led by of managers and top Cornell faculty in an intensive Cornell faculty and industry experts. learning experience. Intensive three-day courses are held on the Cornell University Ten-day programs are held on the Cornell University campus in Ithaca, New York in June-July; in Brussels, Belgium campus in Ithaca, New York in January and June and at in June and at the Cornell Nanyang Institute in Singapore in the Cornell Nanyang Institute in Singapore in July-August. January and July-August. The Online Path The Contract Programs Available year-round, choose individual courses or Programs delivered by Cornell faculty for your company. Many combine courses to earn one of six Cornell Certificates. hotel and foodservice management topics available, both “off Interact with an expert instructor and a cohort of your the shelf” and custom developed to your needs and delivered peers to develop knowledge, and to effectively apply to your management team on the Cornell campus or anywhere that knowledge in your organization. in the world. Complete program information and applications online: www.hotelschool.cornell.edu/execed/chr PHONE: +1 607 255 4919 EMAIL: exec_ed_hotel@cornell.edu 18 The Center for Hospitality Research • Cornell University ExecEdPathAd_chr-2c_ƒ.qxd 12/7/06 9:15 AM Page 1 Cornell Hospitality Reports Index www.chr.cornell.edu 2008 Reports Vol. 8, No. 5 Optimizing a Personal Wine Vol. 7, No. 14 Why Trust Matters in Top Cellar, by Gary M. Thompson, Ph.D., and Management Teams: Keeping Conflict Vol. 8, No. 12 Frequency Strategies and Steven A. Mutkoski, Ph.D. Constructive, by Tony Simons, Ph.D., and Double Jeopardy in Marketing: The Randall Peterson, Ph.D. Pitfall of Relying on Loyalty Programs, by Vol. 8, No. 4 Setting Room Rates on Cornell Short Courses and Certifications for Hotel Industry Professionals: Michael Lynn, Ph.D. Priceline: How to Optimize Expected Vol. 7, No. 13 Segmenting Hotel Hotel Revenue, by Chris Anderson, Ph.D. Customers Based on the Technology Vol. 8, No. 11 An Analysis of Bordeaux Readiness Index, by Rohit Verma, Ph.D., Wine Ratings, 1970–2005: Implications for Anheuser-Busch Vol. 8, No. 3 Pricing for Revenue Liana Victorino, Kate Karniouchina, and The General Managers Program The Professional Development Program the Existing Classification of the Médoc Enhancement in Asian and Pacific Julie Feickertand Graves, by Gary M. Thompson, Ph.D., Region Hotels:A Study of Relative Pricing Stephen A. Mutkoski, Ph.D., Youngran Strategies, by Linda Canina, Ph.D., and Vol. 7, No. 12 Examining the Effects of Tackle strategic hotel management issues and find Study and share experiences with peers from around the world Bae, Liliana Lelacqua, and Se Bum Oh Cathy A. Enz, Ph.D. Full-Spectrum Lighting in a Restaurant, relevant, specific solutions. Work with a global network in these intensive hospitality management seminars led by by Stephani K.A. Robson and Sheryl E. of managers and top Cornell faculty in an intensive Cornell faculty and industry experts. Vol. 8, No. 10 Private Equity Investment Vol. 8, No. 2 Restoring Workplace Kimes, Ph.D. in Public Hotel Companies: Recent Past, learning experience. Communication Networks after Long-term Future, by John B. Corgel, Intensive three-day courses are held on the Cornell University Downsizing: The Effects of Time Vol. 7, No. 11 Short-term Liquidity Ph.D. on Information Flow and Turnover Measures for Restaurant Firms: Static Ten-day programs are held on the Cornell University campus in Ithaca, New York in June-July; in Brussels, Belgium Intentions, by Alex Susskind, Ph.D. Measures Don’t Tell the Full Story, by campus in Ithaca, New York in January and June and at in June and at the Cornell Nanyang Institute in Singapore in Vol. 8, No. 9 Accurately Estimating Linda Canina, Ph.D., and Steven Carvell, the Cornell Nanyang Institute in Singapore in July-August. January and July-August. Time-based Restaurant Revenues Using Vol. 8, No. 1 A Consumer’s View of Ph.D. Revenue per Available Seat-Hour, by Gary Restaurant Reservation Policies, M. Thompson, Ph.D., and Heeju (Louise) by Sheryl E. Kimes, Ph.D. Vol. 7, No. 10 Data-driven Ethics: The Online Path The Contract Programs Sohn Exploring Customer Privacy in the 2007 Reports Information Era, by Erica L Wagner, Vol. 8, No. 8 Exploring Consumer Ph.D., and Olga Kupriyanova Available year-round, choose individual courses or Programs delivered by Cornell faculty for your company. Many Reactions to Tipping Guidelines: Vol. 7, No. 17 Travel Packaging: An Internet Frontier, by William J. Carroll, combine courses to earn one of six Cornell Certificates. hotel and foodservice management topics available, both “off Implications for Service Quality, by Vol. 7, No. 9 Compendium 2007 Ekaterina Karniouchina, Himanshu Ph.D., Robert J. Kwortnik, Ph.D., and Interact with an expert instructor and a cohort of your the shelf” and custom developed to your needs and delivered Mishra, and Rohit Verma, Ph.D. Norman L. Rose peers to develop knowledge, and to effectively apply to your management team on the Cornell campus or anywhere Vol. 7, No. 8 The Effects of Organizational Vol. 7, No. 16 Customer Satisfaction Standards and Support Functions on that knowledge in your organization. in the world. Vol. 8, No. 7 Complaint Communication: with Seating Policies in Casual-dining Guest Service and Guest Satisfaction in How Complaint Severity and Service Restaurants, by Alex M. Susskind, Ph.D., Recovery Influence Guests’ Preferences Restaurants, by Sheryl Kimes, Ph.D., and Jochen Wirtz K. Michele Kacmar, Ph.D., and Carl P. and Attitudes, by Alex M. Susskind, Ph.D. Borchgrevink, Ph.D. Complete program information and applications online: Vol. 8, No. 6 Questioning Conventional Vol. 7, No. 15 The Truth about Integrity www.hotelschool.cornell.edu/execed/chr Wisdom: Is a Happy Employee a Good Tests: The Validity and Utility of Integrity Vol. 7, No. 7 Restaurant Capacity Effectiveness: Leaving Money on the PHONE: +1 607 255 4919 EMAIL: exec_ed_hotel@cornell.edu Employee, or Do Other Attitudes Matter Testing for the Hospitality Industry, Tables, by Gary M. Thompson, Ph.D. More?, by Michael Sturman, Ph.D., and by Michael Sturman, Ph.D., and David Sean A. Way, Ph.D. Sherwyn, J.D. Cornell Hospitality Report • July 2008 • www.chr.cornell.edu 19 www.chr.cornell .edu