- Email: [email protected]
Hotel values
Hotel Vdues In the Aftermath
of
September II,2001
This model of hotel values once again demonstrates the importance of maintaining solid fundamentals, such as occupancy percentage and average daily rate.
BYJOHN W. O’NEILL
AND
ANNE Ft. LLOYD-JONES
M
uch has been written about the effects of the events of September 11 on various aspects of life in the United States. One part of that discussion has examined the plight of the U.S. hotel industry. In that vein, we have developed a model that gives specific estimates of the financial effects that those attacks will have on hotel values. Our estimates indicate that in the next two years certain property types will clearly be more vulnerable to losses in value than will others. We believe that the information presented here will be useful for hotel owners and investors who are contemplating buying or selling hotels and for appraisers and consultants who must determine hotel values. Although all facets of United States society have been affected by the September 11 attacks, it is hard to cite an industry that has felt those effects more immediately than the hospitality industry. The airlines, for instance reported a 320 2001, CORNELL
UNIVERSITY
10 Cornell Hotel and Restaurant Administration Quarterly
percent drop in passenger traf& in September.’ With the sudden drop in air traffic, many hotels saw their customer base evaporate. While the hotel business has gradually recovered, we believe that business will not return to the strong levels seen in 1999 and 2000-at least, not in the near term. Because hotel values are generally based on revenues, we believe that future hotel values will be depressed for some time to come. Without doubt, the industry’s performance was abysmal in the weeks following September 11. In the eight weeks immediately following September 11, weekly occupancies were down between 10.6 and 25.9 percent nationally, average daily rates (ADRs) were off between 5.3 and 15.4 percent, and RevPARs were down between 16.0 and 37.3 percent-all compared to
I S. McCartney, “At Airlines, Both Fares and Traffic Drop Sharply,” WaU Stwet Journal, October 24, 2001, p. A2.
DECEMBER 2001
HOTEL VALUATION
the previous year. 2 (Remember, too, that even willing travelers were prevented from flying for several days immediately after September 11 while all commercial airline flights within the United States were grounded, and inbound international flights were diverted elsewhere.) Buffering those disappointing statistics are such hopeful ones as the report by the American Automobile Association that, after an approximate 50-percent fall-off in bookings in the weeks immediately following September 11, its travel agents’ bookings have returned to within 7 percent of prior reservation levels.3 U.S. airlines have reported that the 32-percent year-to-year decline in passenger traffic in September tapered to a 26percent drop in October, 20 percent in November, and only 14 percent in December.4 Furthermore, in November, weekly hotel occupancies were down by as little as 4.1 percent, weekly ADRs by as little as 6.2 percent, and weekly RevPARs by as little as 10.2 percent, all compared to a year earliet5 Even before September 11, many hotels and markets were experiencing a softening of occupancy and ADR as the economy slid into recession.6 At this writing, most economists seem to agree that a recovery will be well underway by the end of 2002. Given those expectations, we anticipate an increase in hotels’ net income levels by mid 2002. Our analysis is that the anticipated increases in revenue will be enhanced by greater stability on the supply side. The pace of supply growth was already decelerating markedly by the middle months of 2001, in response to the overbuilding that had occurred in the preceding five years.’ 2 J.N. Ader, R.A. LaFIeur, and T. McCoy, “Outside the Box: Exploring Important Investor Issues,” Bear Stearns Equity Reseurch, November 200 1. 3Ibid. * S. McCartney, “Strength of Christmas in Passenger Traffic at U.S. Airlines,” January 7,2002, p. A2. 5 J.N. Ader, “Lodging Intelligence Equiy Research, December 200 1.
Travelers Slows Fall Wall Street Journuf,
Report,”
Bear Steam
6J.N. Ader, R.A. LaFleur, and T. McCoy, “Lodging Industry: Monthly Trends Chronicle,” Bcur Steam Equity Research, August 2001, p. 5. ’ l?H. Ford, “Hotel Real Estate White Paper on the State of the Industry,” Lodging Econometrics, Quarter 2, 2001.
DECEMBER 2001
I
INDUSTRY ANALYSIS
The uncertainty arising from the terrorist attacks led bankers to shut off credit, thereby stanching much new construction. A similar stalling in supply growth amplified revenues in the years immediately following the 1990-l 99 1 recession, and we believe that this pattern is likely to be repeated in the next several months.
Hotel Valuation Index HVS International established the Hotel Valuation Index (HVI) as a measure of trends in median hotel values for the nation. The HVI is
Even before September were experiencing
2001, many hotels
a softening of occupancy
and ADR as the economy slid into recession.
developed through an income approach, using market-area data for 46 major U.S. markets provided by Smith Travel Research (STR), together with operational, capitalization-rate, and capital-availability’ information derived from HVS International’s database. Consequently, the value trends reflected here incorporate the influences of both income and capital on valuation. Exhibit 1 (on the next page) sets forth the yearly HVI for the United States from 1987 to 2000. The exhibit also shows STR’s data regarding the changes in supply, demand, occupancy, ADR, and RevPAR for those years. A review of those national data indicates that hotel values fell by 9.8 percent in 1990. In 1991, with the recession in full force and the Gulf War underway, hotel values fell by an additional 27.7 percent. Moreover, the volume of hotel-property sales decreased significantly in 1990 and 1991. Values stabilized in 1992, rising by a minimal 1.3 percent. In 1993, the industry saw the beginning of a five-year surge, during which hotel values increased at an average annual rate of well over 20 percent, and the volume of hotel-property sales soared. It is worth noting that during the previous recession, the decreases in value coincided with declines in average hotel occupancy. Moreover,
Cornell Hotel and Restaurant Administration Quarterly
11
INDUSTRY ANALYSIS
I
HOTEL VALUATION
Trends in hotel values, occupancy, ADR, and RevPAR
Year
Change in Demand
Change in SUPPlY
1987
ADR
Change in ADR
$53.12
-
1988
4.8%
4.6%
$54.98
3.5%
1989
5.2%
3.7%
$56.84
3.4%
1990
2.3%
3.6%
$58.88
3.2%
1991
-0.9%
1.7%
$58.78
0.2%
1992
2.1%
0.9%
$59.61
1.4%
1993
1.9%
0.4%
$61.27
2.8%
1994
3.1%
1.2%
$63.54
3.7%
1995
2.1%
1.6%
$66.65
4.9%
1998
2.2%
2.5%
$70.93
6.4%
1997
2.8%
3.6%
$75.31
6.2%
1998
3.0%
4.2%
$78.62
4.4%
1999
3.3%
4.0%
$81.87
4.1%
2000
3.8%
3.1%
$85.24
4.1%
Sources: HVS International, Smith Travel Research
the only other decrease in the HVI during this period also coincided with an occupancy decrease. In 1998 occupancy fell by 1.2 percent, and the HVI decreased by 0.4 percent. These figures reflect the correlation between hotel values and hotel performance. That is, although changes in the aggregate supply-and-demand relationships in a market are an important macroeconomic issue, they drive hotel values only to the extent that they affect hotel performance. Further proof of this correlation can be found by examining the RevPAR data. Since 1987 nationwide RevPAR has decreased in only one year, 1991. That was also the year of the single greatest decrease in hotel values.
12 Cornell Hotel and Restaurant Administration Quarterly
In general, we expect the same pattern of dedine, stability, and growth to emerge in the U.S. hotel market in conjunction with the current recession. However, we do not expect the decrease in hotel values to be as dramatic as was seen in 199 1, for several reasons. For one thing, the industry in 2002 has a different capital structure than what existed in 1991. Coming out of the 1980s many hotel assets were highly leveraged, with logn-to-value ratios in the 75- to 80-percent range and impossibly low debt-coverage ratios.8 Equity investors had
* M. Billig, “Tougher Times May Mean More Hotels Will Change Hands,” Hotel Businm, October 7-20, 2001, pp. 1-13.
DECEMBER 2001
HOTEL VALUATION
little capital invested and many owners had little long-term commitment to the hospitality industry. Exacerbating those &umstances was the fact that many of the lending institutions that had underwritten the hotel loans were themselves undercapitalized. The failure of the savings and loans, together with the Resolution Trust Corporation’s subsequent wholesale disposition of hotel assets at fire-sale prices, put significant downward pressure on values and contributed materially to the dramatic decline reflected in the HVI data. By contrast, the industry’s 2002 capital situation is structurally much more sound. Lenders were careful during the 1990s. They required the investment of real equity, realistic debt-coverage ratios, and loan-to-value ratios in the 60- to 65 percent range.9 Moreover, a substantial amount of hotel real estate is now owned by hospitality companies, which have a long-term commitment to the industry. These companies are in the main financially sound and should be able to weather this recessionlo Furthermore, the financial institutions and vehicles that have underwritten the hospitality sector are much less vulnerable to nonperforming loans than were many of the savings and loans. Offsetting those positives is the uncertainty associated with the terrorist attacks and the ongoing terrorist war. One of the most obvious differences between the Gulf War and the current one is geography. The Gulf War was distant, but the September 11 attacks were all too close to home. Even so, geography may not make as much difference in travel patterns as one might expect. The 1991 downturn in travel was essentially the result of the fear of terrorist attacks on U.S. airplanes, and that is the same concern that we have today-nly this time those fears were sadly realized. Whether travel activity returns to the levels experienced in 2000 remains to be seen, We think that travel growth will likely be influenced more by the recession than by the attacks. What is clear is that travel patterns are changed at least in the short term, with drive-in destinations achieving renewed popularity among leisure travelers. Those destinations that are heavy with governg Ibid. lo Ader, LaFleur, and McCoy, op. cit.
DECEMBER 2001
I
INDUSTRY ANALYSIS
ment or military activity may also benefit in the short term, as a function of increased government defense spending. Destinations that depend on international travel may have to seek alternate sources of demand, at least in the short term.
Our Methodology To conduct our analyses we relied on the database compiled by one of us, namely, John O’Neill. Over the past ten years he has developed a database of verified sales of hotels in the United States that includes operating information for the
Offsetting the hotel industry’s many positives is one notable negative-uncertainty.
12 months preceding the sale. These properties represent all hotel types (classified as upper upscale, upscale, midscale with food and beverage, midscale without food and beverage, and economy) and all regions of the United States (i.e., New England, middle Atlantic, southeast, upper midwest, lower midwest, southwest, and west). This database contains a sample of 245 properties for which complete operating information is available, including occupancy percentage, ADR, net operating income, capitalization rate, and room-revenue multiplier, as well as sale price, sale date, opening date, and number of guest rooms. Using multiple-regression analyses, O’Neill developed automated valuation models (AvMs) that predict hotel-sale prices based on the above factors. The sample size of 245 properties is sufficient because AVMs that have been developed by other researchers for other types of real property (and have had statistical significance using multiple regression analyses) have had sample sizes ranging from 143 to 219 properties.” Although computerized valuation meth-
” J.E. Eckert, P.M. O’Connor, and C. Chamberlain, “Computer-assisted Real-estate Appraisal: A California Savings and Loan Case Study,”ApprakalJournal, October 1993, pp. 524-532.
Cornell Hotel and Restaurant Administration Quarterly
13
INDUSTRY ANALYSIS
I
HOTEL VALUATION
Examples of hotel brands by segment
Upscale
Adam’s Mark Courtyard Doubletree Hotels Homewood Suites Radisson Residence Inn Midscele with F&B
Best Western Clarion Holiday Inn Howard Johnson Quality Inns and Suites Ramada
Source: Smith Travel Research
odologies are increasingly available for commercial real-estate properties,12 the development of such methodologies for lodging properties has progressed much more s10wly.‘~ For this analysis, we found that occupancy percentage and ADR for the 12 months prior to sale are significant predictors of a hotel’s sale price. We also found that the number of guest rooms was a significant predictor of sale price. Therefore, we evaluated sale prices on a per-room basis, which follows a long-established industry tradition. The model, explained in the sidebar, at right, achieved a correlation coeffkient (R) of .888 and a regression coefficient (R*) of .789. Our model resulted in an R2 that is within the range of R2 results ofAVMs previously developed for other types of real property, where R2 results have ranged between .772 and .888.‘* Other variables.We examined other variables for inclusion in our model, but we found that those variables either did not explain hotel prices or were subsumed by the effects of ADR and occupancy. For example, we found that a variable capturing time was not significant. The database shows a clear trend of increasing hotel sale prices from its inception in 1990 through 200” a period that includes both the trough of an economic recession and the peak of an economic expansion. However, we found that those increasing sale prices correlated with the changes in occupancy and ADR. The increasing sale prices resulted from the improved operating performance of hotels during those years, rather than from any hind of time value. We found that when occupancy and ADR are included in the model, capitalization (cap) rate is also not a significant predictor of a hotel’s selling price. Thus, while a cap rate is a direct reflection of a property’s value given its income, our
‘* S. Kleege, “Will Computers Take Over the Appraisal Game?,” American Bunkcc June 13, 1997, p. 10. I3 J.B. Corgel and J.A. deRoos, “Pure Price Changes of Lodging Properties,” Cornell Hotel and Restaurant Adnainhratbn Quarter& Vol. 33, No. 2 (April 19921, pp. 70-77. ‘* See: J.H. Detweiler and R.E. Radigan, “Computerassisted Real-estate Appraisal: A Tool for the Practicing Appraiser,” Appraisal/ournal, January 1996, pp. 91-101; and J.H. Detweiler and R.E. Radigan, “Computer-assisted Real-estate Appraisal: A Tool for the Practicing Appraiser,” AppraisalJourna~ July 1999, pp. 280-286.
14 Cornell Hotel and Restaurant Administration Quarterly
DECEMBER 2001
HOTEL VALUATION
model appears to capture its effect through the inclusion of occupancy and ADR. Thus, we excluded cap rates from our model. Location. We also found that the region in which the hotel is located is not a significant fattor in predicting its sale price. We did, indeed, find differences in the summary statistics for sale price by region (e.g., properties in New England tended to sell for higher prices per room than properties in the southeast). Once again, however, when hotel occupancy and ADR are included in the model, region ceases to be a significant predictor of sale price, because the model captures variations in economics for each region through the variations in property occupancy and ADR. Size doesn’t count. We found similar results when we included a variable for large metropolitan areas (over one million permanent residents). While properties in the largest metro areas may sell for higher prices, our model once again captures those differences through the markets’ variations in occupancy and ADR. In short, occupancy, ADR, and number of rooms appear to capture important characteristics of hotels as assets and investments, including not only performance characteristics, but level and style of service, type of market, and other location characteristics. Notably, other previous research has also found that occupancy, ADR, and number of guest rooms are significant predictors of hotel values.15 Thus, our model used just occupancy and ADR as variables to predict sale price per room.
INDUSTRY
ANALYSIS
A Regression Model for Hotel Valuation The multiple regression model discussed in the accompanying article is summarized as follows: Y = (-$89,148 + $781(X) + $1,345(Z)) x R where Y = predicted hotel value (F=451 .148, p < .OOl for the overall model), X = annual occupancy percentage x 100 @ < .OOl), Z = annual average daily rate in dollars @ < .OOl), and R = number of available guest rooms. The Y-intercept is represented by the figure -$89,148. Thus, a hotel with an annual occupancy percentage of 70 percent, an average daily rate of $100, and 100 available guest rooms would be valued at approximately $9,882,400, or $98,824 per room, calculated as follows: (-$89,148 + $781(70) + $1,345($100)) x 100. Similarly, a hotel with an annual occupancy percentage of 50 percent, an average daily rate of $80, and 100 available guest rooms would be valued at approximately $2,980,400, or $29,804 per room. That calculation is: (-$89,148 + $781(50) + $1,345($80)) x 100.
Post-attack Changes To estimate the change in hotel values based on September 11, we analyzed changes in projections of occupancy and ADR before and after September 11. Numerous organizations develop prognostications of this kind for hotels in the United States, including PricewaterhouseCoopers, PKF Consulting, Smith Travel ReI5 See: J.B. Corgel and J.A. deRoos, “The ADR Rule of Thumb as a Predictor of Lodging Property VaIues,” International Journal of Hospitality Management, Vol. 12, No. 4 (19931, pp. 353-365; and S. Rushmore and J.A. deRoos, “Hotel Valuation Techniques,” in Hotel Invemnmts: Issues & Penpectives, second edition (Lansing, MI: Educational Institute of the American Hotel & Motel Association, 1999), pp. 151-180.
DECEMBER 2001
Cornell Hotel and Restaurant Administration Quarterly
15
INDUSTRY ANALYSIS
HOTELVALUATION
I
Hotel operating projections
Occupancy
I
before September
ADR
1
RevPAR
11
1
Annual% Chsnae
2000 2001 2002 2003
63.7% 62.9% 63.2% 63.1%
$85.24 $88.44 $91.90 $95.64
$54.30 $55.63 $58.08 $60.35
-2.5% 4.4% 3.9%
2000 2001 2002 2003
72.7% 71.2% 71.9% 71.7%
$155.56 $162.97 $170.47 $179.10
$113.09 $116.03 $122.57 $128.41
-2.6% 5.6% 4.8%
2000 2001 2002 2003
70.2% 69.7% 69.6% 69.0%
$99.24 $102.92 $106.51 $110.64
$69.67 $71.74 $74.13 $76.34
-3.0% 3.3% 3.0%
2000 2001 2002 2003
59.9% 59.3% 59.2% 51.1%
$73.41
$43.97 $45.09 $46.49 $41.57
2000 2001 2002 2003
64.7% 64.7% 64.7% 64.0%
$67.22 $70.09 $72.94 $76.20
$43.49 $45.35
2000 2001 2002 2003
58.1% 58.4% 58.6% 58.6%
$47.60 $48.96 $50.31 $51.85
$27.66 $28.59 $29.48 $30.38
2.5% -3.1% -10.6%
-4.3% 4.1% 3.3%
-3.4% 3.1% 3.1%
Note: Numbersfor2000 areactual figures according toSmithTravel Research; 2001figures are year-end estimates basedon two quarters ofactualdata. Sources: PricewaterhouseCoooersand SmithTravel Research
search, and Ernst & Y~ung.‘~ The Pricewaterhousecoopers (PwC) forecasts have the advantage of being updated throughout the year based on changing economic conditions (rather than annually at the beginning of each year). They are also detailed by segment (upper upscale, upscale, midscaie with food and beverage, midscale without food and beverage, and economy), and they are based on statistical models that have been refined over the course of their ten-year development.17 Therefore, we used PwC’s forecasts of occupancy and ADR that were developed immediately prior to September 11 for our base-case analysis, and the forecasts developed immediately after September 11 as the gauge for the anticipated change in occupancy and ADR due to the terrorist attacks. We should point out that the projections immediately prior to September 11 reflected downward adjustments to PwC’s earlier forecasts for 2001, due to the effects of the already softening economy. The PwC forecasts are based on a roomdemand equation that returns an R2 of .996, an ADR equation with an R2 of .930, and a roomstarts equation (new hotel-room construction) with an R2 of .880. Thus, all three of those equations capture a considerable level of the variation in their particular factor. The PwC forecasts of occupancy and ADR immediately prior to September 11 are presented in Exhibit 3, while Exhibit 4 shows its forecasts immediately after September 11. (Exhibit 2, on the previous page spread, gives examples of hotel brands that are included in each segment.) Exhibits 3 and 4 show how PwC lowered its overall projections of occupancy, ADR, and RevPAR for 2001 through 2003 in the wake of September 11. The analysts made a downward adjustment for each of the identified hotel segments, with the high-price segments (upper upscale and upscale) projected to experience the most dramatic reversal. The occupancy and ADR forecasts before and after September 11 were the inputs for our multiple-regression equation. Thus, we calculated value-per-room estimates based on PwC’s occuI6See, for example, “News Briefs,” Hotel Bwiners, October 21-November 6, 2001, p. 1. ‘7 Coopers 81 Lybrand, Hospidiry Directions: Forecasts & Analysesfir the Hospitality Industry, Vol. 1, No. 1 (199 1).
16 Cornell Hotel and Restaurant Administration Quarterly
DECEMBER 2001
HOTEL VALUATION
pancy and ADR forecasts just before and just after September 11. We also calculated the change in value per room based on the pre- and postSeptember 11 value estimates to gauge the change in value due to the September 11 attacks. Those estimates of value change were calculated for 2001 through 2003. The results indicate that values will continue to lose ground through 2003 relative to anticipated values prior to September 11 (see Exhibits 5 and 6, overleaf). What It Means Asshown in Exhibit
5, our model indicates that the average U.S. hotel was worth 3.6 percent less at the end of2001 than it was in 2000 ($71,313 per room at the end of 2001 versus $73,978 per room at the end of 2000). Based on forecasts developed prior to September 11, the model projects that in the absence of the attacks, the average hotel would have instead been worth 5.0 percent more at the end of 2001 than it was at the end of 2000 ($77,673 per room in 200 1 versus $73,978 per room in 2000). That calculation shows a net loss in mean value of $6,360, or 8.6 percent, due to the events of September 11 ($6,360+$73,978). Looking ahead to 2003, the model indicates that the average U.S. hotel will lose a net $11,050 in value per room by the end of 2003 as a result of September 11 (albeit most of that loss had already occurred in 200 1). Could have been. Let us emphasize that our calculation of the loss in hotel value is based on the pre-September 11 expectations for the industry’s performance. While hotels are expected to continue to lose ground compared to “what would have been” through 2003, at this moment it appears that 2001 is the only year in which U.S. hotel values would experience an overall year-to-year decrease. We expect valuation to increase by 1.3 percent in 2002, and our model predicts an overall increase of 5.8 percent in 2003. Among the hotel segments that we examined, the model indicates that the largest overall change in value due to September 11 occurred among upper-upscale hotels, which declined in value by 8.0 percent in 2001 compared to 2000. Also, these hotels are expected to experience the largest value change compared to “what would have been” through 2003, with an overall value loss of $35,854 per room as of that year.
DECEMBER
2001
Hotel operating projections
OccuDancv
1
INDUSTRY
after September
ADR
RevPAR
ANALYSIS
11
1
Annual% Change
2000 2001 2002 2003
63.7% 60.5% 60.1% 61.2%
$85.24 $85.07 $86.01 $88.50
$54.30 $51.47 $51.69 $54.16
--5.2% 0.4% 4.8%
2000 2001 2002 2003
72.7% 68.8% 69.3% 71.3%
$155.56 $147.32 $148.27 $152.65
$113.09 $101.42 $102.74 $108.90
--10.3% 1.3% 6.0%
2000 2001 2002 2003
70.2% 68.3% 67.5% 68.7%
$99.24 $96.56 $95.35 $97.12
$69.67 $65.96 $64.32 $66.72
--5.3% -2.5% 3.7%
2000 2001 2002 2003
59.9% 58.6% 59.2% 55.3%
$73.41 $71.83 $72.51 $72.51
$43.97 $42.10 $42.91 $40.11
--4.3% 1.9% -6.5%
2000 2001 2002 2003
64.7% 64.3% 65.5% 66.9%
$67.22 $66.85 $68.09 $69.47
:z $44162 $46.45
-1.1% 3.7% 4.1%
2000 2001 2002 2003
58.1% 57.4% 57.4% 58.5%
$47.60 $47.03 $47.05 $47.95
$27.66 $27.00 $27.02 $28.06
--2.4% 0.1% 3.9%
- _... _I_,.... l.l-. _I
Note: Numbersfor2000 areactual figures according toSmithTravel Research; 2001figures are year-end estimates basedon three quarters ofactualdata. Sources: PricewaterhouseCoopersand SmithTravel Research
Cornell Hotel and Restaurant
Administration
Quarterly
17
INDUSTRY ANALYSIS
HOTEL VALUATION
I
Estimated changes in value per room Value Before S/l1
2001 2000 2002 2003
2000 2001 2002 2003
2000 2001 2002 2003
$77,873 $73,978 $82,555 $87,509
Annual
% Change
5.0% -6.3% 6.0%
$206,307
5.9%
$97,754 $102,323
-4.7% 4.6% 4.8%
Value After 9lll
f?;,;;; $72:267* $78,459
Annual
K Change
-3.8% -1.3% 5.8%
-$8,3!: -$10,287 -$11,050
$175,407
--8.0% 1.0% 4.6%
$97,754 $92,708 $90,435 $93,757
--5.2% -2.5% 3.7%
-$9,6:! -$16,641 -$18,417
$50,472
-5.5%
-$6,1:: -$8,115 -$8,685
$50,502 $49,734 $52,303 $55,174
--1.5% 5.2% 5.5%
;;;;g
--6.8% 0.3% 11.5%
..-.-I____ _. 2000 2001 2002 2003
2000 2001 2002 2003
$50,502 $54,362 $58,195 $82,047
-7.6% 7.1% 6.6%
2000 2001 2002 2003
$19,090 $21,148 $23,116 $25,187
-10.8% 9.3% 9.0%
$17:845 $19,888
-$4,8!t
T&g p&aces. Upscale hotels represent the only segment where we expect values per room to continue declining in 2002. This likely decrease in value arises for essentially the same reason that upper-upscale hotels lost value in 2001-to wit, travelers are expected to trade down to the next lower tier of properties. As a consequence of the trade-down effect, midscale hotels with food and beverage are expected to experience relatively minor value changes in 2001 and 2002. In fact, these hotels are expected to experience occupancy, ADR, and value increases in 2002. However, as the economy improves by 2003, we anticipate that this trend will reverse, and guests will be lost back to upscale hotels. with the shift in travelers will come a shift in hotel valuation. Under normal economic conditions, upscale travelers select hotels primarily for reasons other than price-to the disadvantage of the midscale hotels.“’ Consequently, the m&ale-with-F&B segment is the only one for which we predict a value decrease in 2003 compared to 2002. Thus, owners of midmarket properties who are considering selling may be well advised to sell in 2002. Limited service, escalating value. For the first time ever, the model shows that midscale hotels without food and beverage are likely to be worth more on a per-room basis in 2003 than are midscale hotels with F&B. As a group, the properties without F&B are newer than their fullservice counterparts, in part because the fullservice hotels are suffering from functional obsolescence as their designs become increasingly outmoded.‘y The valuation problem for this segment is that investors do not attribute much value to food and beverage outlets, particularly if those amenities are not seen as generating a profit (as is often the case). *OWe find it ironic that hotels that are less expensive to build (that is, those withI8B.J. Knutson, “FrequentTravelers: MakingThem Happy and Bringing Them Back,” Cornell Hotel and Restaumnt Administration Quart&y, Vol. 29, No. 1 (May 1988), pp. 83-87. I9 Appraisal Institute, TheApprairalofRealEte, edition (Chicago: Appraisal Institute, 1996).
eleventh
2oThis is by no means a recent phenomenon. See: Bjorn Hanson, “Hotel Food Service: Where’s the Profit?,” Cornelf Hotel and Restaurant Administration Quarter&, Vol. 25, No. 2 (August 1984), pp. 93-96.
18 Cornell Hotel and Restaurant Administration Quarterly
DECEMBER 2001
HOTEL VALUATION
I
INDUSTRY ANALYSIS
Estimated values per room (year 2003)
Values before September 11
Values after September 11
Ufwr
Upscals
lIpScale HOTEL
Midecale with F&B
Midscale without F&B
Economy
SEGMEWT
Note: Values shown on the above graph correspond to the year-2003 figures shown in Exhibit 5.
out F&B facilities) are seen as being more valuable than those that are more expensive to build (i.e., those with full F&B facilities). Our model indicates that the midscale hotels without F&B are emerging as the “ideal” hotel improvement in many lodging markets. At $5,000 per room, the value decrease anticipated for economy properties is the least in terms of dollars. However, this equates to a substantial percentage-approximately 25 percentof the total average value of the hotels in this segment. In considering this outcome, however, note that our model is highly sensitive to fluctuations in occupancy and ADR Thus, the average economy hotel operating with a typical 200 1 annual occupancy of 57.4 percent and ADR of $47.03 would be worth only $17,79 1 per room in this model. However, an economy hotel with an annual occupancy of 75 percent and ADR of $55, for example, would be worth $41,904 per room based on our model. If such a superior performing economy hotel loses $5,000 per room
DECEMBER 2001
in value, that decrease represents just 12 percent of its value, rather than the 25-percent value loss expected for an economy property with average statistics. While we have used economy properties in this example, the model shows that superior performance within any class should mitigate a loss in value, just as one might expect. Like all multiple-regression models, ours is effective at predicting results in the aggregate and for overall segments, but it is not so effective at predicting individual property values. This situation is particularly true of outlying properties with extremely low or high values. In this case, our model would not accurately value lodging properties with values below $17,791 or above $175,407 per room.
Hoping That the Worst Is Over Assuming that no further attacks occur on American soil, we expect that most of the drop in hotel values occurred in 200 1. As the model indicates, the shaky performance of the industry in terms
Cornell Hotel and Restaurant Administration Quarterly
19
INDUSTRY ANALYSIS
HOTEL VALUATION
I
of occupancy and ADR throughout 2001 is the primary cause of this decline. In addition, the lack of liquidity that we have noted is also contributing to this value decline. In the weeks immediately following September 11, virtually all transaction activity came to a halt, with buyers, sellers, and lenders reevaluating their investment strategies. At present, most lenders have a waitand-see attitude concerning loans to this and
Values
and transaction
that are not overly
activity
stressed
should
pick up in concert
pected
economic
for properties
by the recession
with the widely
ex-
recovery. I
many other real-estate sectors.Z1 In this environment, the transactions that are likely to occur would most likely reflect capitalization and discount rates above historical norms, mirroring the high level of uncertainty currently affecting the sector as well as the cost of low or un-leveraged transactions.22 In these market conditions, we do not expect to see many market-value transactions. Achieving a market-value exchange requires both a willing buyer and a willing seller,23 and in the prevailing investment climate, we see few willing sellers. We consider it unlikely that any owner not under duress would contemplate a sale until the market stabilizes and financing becomes more widely available. Most of those transactions that do occur in the next few months will likely involve distressed properties, for which the current situation is the last straw. These assets are expected to trade at liquidation prices, and may represent a prospect for opportunistic buyers. On the other hand, because we do not expect there to be a massive volume of these types of transactions, 21Jones Lang LaSalle Hotels, Focus on the Zmpactoft/w September Ii, 2001, Terrorist Attacks on the U.S. Hotel Real Estate Market (internal document, October 2001). l2 Ibid. 23Appraisal
Institute,
op. cit.
20 Cornell Hotel and Restaurant Administration Quarterly
such sales are not expected to have a substantial effect on national valuation averages. On the other hand, properties that have strong track records and that demonstrated some durability in their cash flows as the economy slowed should continue to be attractive prospects for both lenders and equity investors. While value adjustments necessarily reflect the current economic conditions, the availability of capital should limit any downward pressure on the value of those assets. Two markets. We expect to see a dichotomy in terms of hotel values in 2002. Those properties that succumb to the financial pressures of the recession will likely trade at prices that represent a discount to both pre-2001 values and construction costs. However, we do not expect to see a large number of deeply discounted transactions, as occurred in the early 1990s. In contrast to the poor fundamentals of many of those deals, the more responsible underwriting criteria of the past five years should help limit the sector’s vulnerability (although properties undeiwritten on the basis of their financial performance in 2000 may face dif%culties). Moreover, competition for the assets of failed hotels should limit downward pressure on prices. Recognizing that history can repeat itself, a number of opportunistic funds are already on the lookout for properties that may succumb to the current financial pressures. Competition among such funds, which are more numerous than they were in the early 199% should limit the magnitude of the actual discounts. For those properties that are not overly stressed by the recession, including those that are benefiting from changes in travel patterns, we expect that values and transaction activity will pick up in concert with the widely expected economic recovery. While financing from traditional sources may be slow to return in force, some financing will be available. The U.S. lodging industry continues to be the highest-yielding realestate sector,24 and history has shown that there is always an appetite for high-yield loans. Spreads are likely to increase, but the Federal Reserve’s easing of credit should allow the hospitality sector to tolerate some increases. 24Real Estate Research Corporation, November 2001.
RERC SpcciaI Survey
DECEMBER 2001
HOTEL VALUATION
The scenario for 2003 should be a favorable one for the hotel industry, as demand growth and limited supply additions should support increases in occupancy and ADR Moreover, managers will maintain the operating efficiencies they’ve achieved over the past several years, as well as control expenses during the current downturn and continue cost-control measures as revenue levels improve. Financing will become more readily available, as lenders under pressure to generate loans are lured by high yields and attractive market performance. Competition for hospitality properties will gain momentum on both the debt and equity side. The net effect of those influences will be a continual increase in values, which we ultimately expect to equal or exceed the levels achieved in the past three years. The purpose of this analysis is to provide an overview of hotel-value trends on a national level.
INDUSTRY ANALYSIS
As is usually the case, any analysis of the national lodging industry represents an aggregate of numerous individual markets. National trends are not always reflected directly in individual markets, which are much more subject to local economicsz5 Consequently, so m e markets may be more adversely affected by the current situation, while others will be hurt to a much lesser degree. Since our model indicates that high-rate, upperupscale hotels will experience the greatest declines in value per room, we would expect that the highprofile urban and resort markets will experience the greatest level of pain, whereas markets with historically more moderate occupancies and average daily rates should be more insulated. n 25For example, see: CathyA. Em, Linda Canina, and Kate Walsh, “Hotel-industry Averages: An Inaccurate Tool for Measuring Performance,” in this issue of Cornell QuurrerJy (pp. 22-32).
John W. D’Neill, MAI, CHE, Ph.D., is an assistant professor in the School of Hotel, Restaurant, and Recreation Management at The Pennsylvania State University (jwo30psu.edu). Anne R. Lloyd-Jones, CRE, is senior vice president at HVS International (ALloyd-JonesOHVSlnternationaLcom). 0 2001, Cornell University;
refereed paper: submission receivedNovember 20, 2001; accepted with minor revisions on January 11, 2002,
DECEMBER 2001
Cornell Hotel and Restaurant Administration
Quarterly
21
of
September II,2001
This model of hotel values once again demonstrates the importance of maintaining solid fundamentals, such as occupancy percentage and average daily rate.
BYJOHN W. O’NEILL
AND
ANNE Ft. LLOYD-JONES
M
uch has been written about the effects of the events of September 11 on various aspects of life in the United States. One part of that discussion has examined the plight of the U.S. hotel industry. In that vein, we have developed a model that gives specific estimates of the financial effects that those attacks will have on hotel values. Our estimates indicate that in the next two years certain property types will clearly be more vulnerable to losses in value than will others. We believe that the information presented here will be useful for hotel owners and investors who are contemplating buying or selling hotels and for appraisers and consultants who must determine hotel values. Although all facets of United States society have been affected by the September 11 attacks, it is hard to cite an industry that has felt those effects more immediately than the hospitality industry. The airlines, for instance reported a 320 2001, CORNELL
UNIVERSITY
10 Cornell Hotel and Restaurant Administration Quarterly
percent drop in passenger traf& in September.’ With the sudden drop in air traffic, many hotels saw their customer base evaporate. While the hotel business has gradually recovered, we believe that business will not return to the strong levels seen in 1999 and 2000-at least, not in the near term. Because hotel values are generally based on revenues, we believe that future hotel values will be depressed for some time to come. Without doubt, the industry’s performance was abysmal in the weeks following September 11. In the eight weeks immediately following September 11, weekly occupancies were down between 10.6 and 25.9 percent nationally, average daily rates (ADRs) were off between 5.3 and 15.4 percent, and RevPARs were down between 16.0 and 37.3 percent-all compared to
I S. McCartney, “At Airlines, Both Fares and Traffic Drop Sharply,” WaU Stwet Journal, October 24, 2001, p. A2.
DECEMBER 2001
HOTEL VALUATION
the previous year. 2 (Remember, too, that even willing travelers were prevented from flying for several days immediately after September 11 while all commercial airline flights within the United States were grounded, and inbound international flights were diverted elsewhere.) Buffering those disappointing statistics are such hopeful ones as the report by the American Automobile Association that, after an approximate 50-percent fall-off in bookings in the weeks immediately following September 11, its travel agents’ bookings have returned to within 7 percent of prior reservation levels.3 U.S. airlines have reported that the 32-percent year-to-year decline in passenger traffic in September tapered to a 26percent drop in October, 20 percent in November, and only 14 percent in December.4 Furthermore, in November, weekly hotel occupancies were down by as little as 4.1 percent, weekly ADRs by as little as 6.2 percent, and weekly RevPARs by as little as 10.2 percent, all compared to a year earliet5 Even before September 11, many hotels and markets were experiencing a softening of occupancy and ADR as the economy slid into recession.6 At this writing, most economists seem to agree that a recovery will be well underway by the end of 2002. Given those expectations, we anticipate an increase in hotels’ net income levels by mid 2002. Our analysis is that the anticipated increases in revenue will be enhanced by greater stability on the supply side. The pace of supply growth was already decelerating markedly by the middle months of 2001, in response to the overbuilding that had occurred in the preceding five years.’ 2 J.N. Ader, R.A. LaFIeur, and T. McCoy, “Outside the Box: Exploring Important Investor Issues,” Bear Stearns Equity Reseurch, November 200 1. 3Ibid. * S. McCartney, “Strength of Christmas in Passenger Traffic at U.S. Airlines,” January 7,2002, p. A2. 5 J.N. Ader, “Lodging Intelligence Equiy Research, December 200 1.
Travelers Slows Fall Wall Street Journuf,
Report,”
Bear Steam
6J.N. Ader, R.A. LaFleur, and T. McCoy, “Lodging Industry: Monthly Trends Chronicle,” Bcur Steam Equity Research, August 2001, p. 5. ’ l?H. Ford, “Hotel Real Estate White Paper on the State of the Industry,” Lodging Econometrics, Quarter 2, 2001.
DECEMBER 2001
I
INDUSTRY ANALYSIS
The uncertainty arising from the terrorist attacks led bankers to shut off credit, thereby stanching much new construction. A similar stalling in supply growth amplified revenues in the years immediately following the 1990-l 99 1 recession, and we believe that this pattern is likely to be repeated in the next several months.
Hotel Valuation Index HVS International established the Hotel Valuation Index (HVI) as a measure of trends in median hotel values for the nation. The HVI is
Even before September were experiencing
2001, many hotels
a softening of occupancy
and ADR as the economy slid into recession.
developed through an income approach, using market-area data for 46 major U.S. markets provided by Smith Travel Research (STR), together with operational, capitalization-rate, and capital-availability’ information derived from HVS International’s database. Consequently, the value trends reflected here incorporate the influences of both income and capital on valuation. Exhibit 1 (on the next page) sets forth the yearly HVI for the United States from 1987 to 2000. The exhibit also shows STR’s data regarding the changes in supply, demand, occupancy, ADR, and RevPAR for those years. A review of those national data indicates that hotel values fell by 9.8 percent in 1990. In 1991, with the recession in full force and the Gulf War underway, hotel values fell by an additional 27.7 percent. Moreover, the volume of hotel-property sales decreased significantly in 1990 and 1991. Values stabilized in 1992, rising by a minimal 1.3 percent. In 1993, the industry saw the beginning of a five-year surge, during which hotel values increased at an average annual rate of well over 20 percent, and the volume of hotel-property sales soared. It is worth noting that during the previous recession, the decreases in value coincided with declines in average hotel occupancy. Moreover,
Cornell Hotel and Restaurant Administration Quarterly
11
INDUSTRY ANALYSIS
I
HOTEL VALUATION
Trends in hotel values, occupancy, ADR, and RevPAR
Year
Change in Demand
Change in SUPPlY
1987
ADR
Change in ADR
$53.12
-
1988
4.8%
4.6%
$54.98
3.5%
1989
5.2%
3.7%
$56.84
3.4%
1990
2.3%
3.6%
$58.88
3.2%
1991
-0.9%
1.7%
$58.78
0.2%
1992
2.1%
0.9%
$59.61
1.4%
1993
1.9%
0.4%
$61.27
2.8%
1994
3.1%
1.2%
$63.54
3.7%
1995
2.1%
1.6%
$66.65
4.9%
1998
2.2%
2.5%
$70.93
6.4%
1997
2.8%
3.6%
$75.31
6.2%
1998
3.0%
4.2%
$78.62
4.4%
1999
3.3%
4.0%
$81.87
4.1%
2000
3.8%
3.1%
$85.24
4.1%
Sources: HVS International, Smith Travel Research
the only other decrease in the HVI during this period also coincided with an occupancy decrease. In 1998 occupancy fell by 1.2 percent, and the HVI decreased by 0.4 percent. These figures reflect the correlation between hotel values and hotel performance. That is, although changes in the aggregate supply-and-demand relationships in a market are an important macroeconomic issue, they drive hotel values only to the extent that they affect hotel performance. Further proof of this correlation can be found by examining the RevPAR data. Since 1987 nationwide RevPAR has decreased in only one year, 1991. That was also the year of the single greatest decrease in hotel values.
12 Cornell Hotel and Restaurant Administration Quarterly
In general, we expect the same pattern of dedine, stability, and growth to emerge in the U.S. hotel market in conjunction with the current recession. However, we do not expect the decrease in hotel values to be as dramatic as was seen in 199 1, for several reasons. For one thing, the industry in 2002 has a different capital structure than what existed in 1991. Coming out of the 1980s many hotel assets were highly leveraged, with logn-to-value ratios in the 75- to 80-percent range and impossibly low debt-coverage ratios.8 Equity investors had
* M. Billig, “Tougher Times May Mean More Hotels Will Change Hands,” Hotel Businm, October 7-20, 2001, pp. 1-13.
DECEMBER 2001
HOTEL VALUATION
little capital invested and many owners had little long-term commitment to the hospitality industry. Exacerbating those &umstances was the fact that many of the lending institutions that had underwritten the hotel loans were themselves undercapitalized. The failure of the savings and loans, together with the Resolution Trust Corporation’s subsequent wholesale disposition of hotel assets at fire-sale prices, put significant downward pressure on values and contributed materially to the dramatic decline reflected in the HVI data. By contrast, the industry’s 2002 capital situation is structurally much more sound. Lenders were careful during the 1990s. They required the investment of real equity, realistic debt-coverage ratios, and loan-to-value ratios in the 60- to 65 percent range.9 Moreover, a substantial amount of hotel real estate is now owned by hospitality companies, which have a long-term commitment to the industry. These companies are in the main financially sound and should be able to weather this recessionlo Furthermore, the financial institutions and vehicles that have underwritten the hospitality sector are much less vulnerable to nonperforming loans than were many of the savings and loans. Offsetting those positives is the uncertainty associated with the terrorist attacks and the ongoing terrorist war. One of the most obvious differences between the Gulf War and the current one is geography. The Gulf War was distant, but the September 11 attacks were all too close to home. Even so, geography may not make as much difference in travel patterns as one might expect. The 1991 downturn in travel was essentially the result of the fear of terrorist attacks on U.S. airplanes, and that is the same concern that we have today-nly this time those fears were sadly realized. Whether travel activity returns to the levels experienced in 2000 remains to be seen, We think that travel growth will likely be influenced more by the recession than by the attacks. What is clear is that travel patterns are changed at least in the short term, with drive-in destinations achieving renewed popularity among leisure travelers. Those destinations that are heavy with governg Ibid. lo Ader, LaFleur, and McCoy, op. cit.
DECEMBER 2001
I
INDUSTRY ANALYSIS
ment or military activity may also benefit in the short term, as a function of increased government defense spending. Destinations that depend on international travel may have to seek alternate sources of demand, at least in the short term.
Our Methodology To conduct our analyses we relied on the database compiled by one of us, namely, John O’Neill. Over the past ten years he has developed a database of verified sales of hotels in the United States that includes operating information for the
Offsetting the hotel industry’s many positives is one notable negative-uncertainty.
12 months preceding the sale. These properties represent all hotel types (classified as upper upscale, upscale, midscale with food and beverage, midscale without food and beverage, and economy) and all regions of the United States (i.e., New England, middle Atlantic, southeast, upper midwest, lower midwest, southwest, and west). This database contains a sample of 245 properties for which complete operating information is available, including occupancy percentage, ADR, net operating income, capitalization rate, and room-revenue multiplier, as well as sale price, sale date, opening date, and number of guest rooms. Using multiple-regression analyses, O’Neill developed automated valuation models (AvMs) that predict hotel-sale prices based on the above factors. The sample size of 245 properties is sufficient because AVMs that have been developed by other researchers for other types of real property (and have had statistical significance using multiple regression analyses) have had sample sizes ranging from 143 to 219 properties.” Although computerized valuation meth-
” J.E. Eckert, P.M. O’Connor, and C. Chamberlain, “Computer-assisted Real-estate Appraisal: A California Savings and Loan Case Study,”ApprakalJournal, October 1993, pp. 524-532.
Cornell Hotel and Restaurant Administration Quarterly
13
INDUSTRY ANALYSIS
I
HOTEL VALUATION
Examples of hotel brands by segment
Upscale
Adam’s Mark Courtyard Doubletree Hotels Homewood Suites Radisson Residence Inn Midscele with F&B
Best Western Clarion Holiday Inn Howard Johnson Quality Inns and Suites Ramada
Source: Smith Travel Research
odologies are increasingly available for commercial real-estate properties,12 the development of such methodologies for lodging properties has progressed much more s10wly.‘~ For this analysis, we found that occupancy percentage and ADR for the 12 months prior to sale are significant predictors of a hotel’s sale price. We also found that the number of guest rooms was a significant predictor of sale price. Therefore, we evaluated sale prices on a per-room basis, which follows a long-established industry tradition. The model, explained in the sidebar, at right, achieved a correlation coeffkient (R) of .888 and a regression coefficient (R*) of .789. Our model resulted in an R2 that is within the range of R2 results ofAVMs previously developed for other types of real property, where R2 results have ranged between .772 and .888.‘* Other variables.We examined other variables for inclusion in our model, but we found that those variables either did not explain hotel prices or were subsumed by the effects of ADR and occupancy. For example, we found that a variable capturing time was not significant. The database shows a clear trend of increasing hotel sale prices from its inception in 1990 through 200” a period that includes both the trough of an economic recession and the peak of an economic expansion. However, we found that those increasing sale prices correlated with the changes in occupancy and ADR. The increasing sale prices resulted from the improved operating performance of hotels during those years, rather than from any hind of time value. We found that when occupancy and ADR are included in the model, capitalization (cap) rate is also not a significant predictor of a hotel’s selling price. Thus, while a cap rate is a direct reflection of a property’s value given its income, our
‘* S. Kleege, “Will Computers Take Over the Appraisal Game?,” American Bunkcc June 13, 1997, p. 10. I3 J.B. Corgel and J.A. deRoos, “Pure Price Changes of Lodging Properties,” Cornell Hotel and Restaurant Adnainhratbn Quarter& Vol. 33, No. 2 (April 19921, pp. 70-77. ‘* See: J.H. Detweiler and R.E. Radigan, “Computerassisted Real-estate Appraisal: A Tool for the Practicing Appraiser,” Appraisal/ournal, January 1996, pp. 91-101; and J.H. Detweiler and R.E. Radigan, “Computer-assisted Real-estate Appraisal: A Tool for the Practicing Appraiser,” AppraisalJourna~ July 1999, pp. 280-286.
14 Cornell Hotel and Restaurant Administration Quarterly
DECEMBER 2001
HOTEL VALUATION
model appears to capture its effect through the inclusion of occupancy and ADR. Thus, we excluded cap rates from our model. Location. We also found that the region in which the hotel is located is not a significant fattor in predicting its sale price. We did, indeed, find differences in the summary statistics for sale price by region (e.g., properties in New England tended to sell for higher prices per room than properties in the southeast). Once again, however, when hotel occupancy and ADR are included in the model, region ceases to be a significant predictor of sale price, because the model captures variations in economics for each region through the variations in property occupancy and ADR. Size doesn’t count. We found similar results when we included a variable for large metropolitan areas (over one million permanent residents). While properties in the largest metro areas may sell for higher prices, our model once again captures those differences through the markets’ variations in occupancy and ADR. In short, occupancy, ADR, and number of rooms appear to capture important characteristics of hotels as assets and investments, including not only performance characteristics, but level and style of service, type of market, and other location characteristics. Notably, other previous research has also found that occupancy, ADR, and number of guest rooms are significant predictors of hotel values.15 Thus, our model used just occupancy and ADR as variables to predict sale price per room.
INDUSTRY
ANALYSIS
A Regression Model for Hotel Valuation The multiple regression model discussed in the accompanying article is summarized as follows: Y = (-$89,148 + $781(X) + $1,345(Z)) x R where Y = predicted hotel value (F=451 .148, p < .OOl for the overall model), X = annual occupancy percentage x 100 @ < .OOl), Z = annual average daily rate in dollars @ < .OOl), and R = number of available guest rooms. The Y-intercept is represented by the figure -$89,148. Thus, a hotel with an annual occupancy percentage of 70 percent, an average daily rate of $100, and 100 available guest rooms would be valued at approximately $9,882,400, or $98,824 per room, calculated as follows: (-$89,148 + $781(70) + $1,345($100)) x 100. Similarly, a hotel with an annual occupancy percentage of 50 percent, an average daily rate of $80, and 100 available guest rooms would be valued at approximately $2,980,400, or $29,804 per room. That calculation is: (-$89,148 + $781(50) + $1,345($80)) x 100.
Post-attack Changes To estimate the change in hotel values based on September 11, we analyzed changes in projections of occupancy and ADR before and after September 11. Numerous organizations develop prognostications of this kind for hotels in the United States, including PricewaterhouseCoopers, PKF Consulting, Smith Travel ReI5 See: J.B. Corgel and J.A. deRoos, “The ADR Rule of Thumb as a Predictor of Lodging Property VaIues,” International Journal of Hospitality Management, Vol. 12, No. 4 (19931, pp. 353-365; and S. Rushmore and J.A. deRoos, “Hotel Valuation Techniques,” in Hotel Invemnmts: Issues & Penpectives, second edition (Lansing, MI: Educational Institute of the American Hotel & Motel Association, 1999), pp. 151-180.
DECEMBER 2001
Cornell Hotel and Restaurant Administration Quarterly
15
INDUSTRY ANALYSIS
HOTELVALUATION
I
Hotel operating projections
Occupancy
I
before September
ADR
1
RevPAR
11
1
Annual% Chsnae
2000 2001 2002 2003
63.7% 62.9% 63.2% 63.1%
$85.24 $88.44 $91.90 $95.64
$54.30 $55.63 $58.08 $60.35
-2.5% 4.4% 3.9%
2000 2001 2002 2003
72.7% 71.2% 71.9% 71.7%
$155.56 $162.97 $170.47 $179.10
$113.09 $116.03 $122.57 $128.41
-2.6% 5.6% 4.8%
2000 2001 2002 2003
70.2% 69.7% 69.6% 69.0%
$99.24 $102.92 $106.51 $110.64
$69.67 $71.74 $74.13 $76.34
-3.0% 3.3% 3.0%
2000 2001 2002 2003
59.9% 59.3% 59.2% 51.1%
$73.41
$43.97 $45.09 $46.49 $41.57
2000 2001 2002 2003
64.7% 64.7% 64.7% 64.0%
$67.22 $70.09 $72.94 $76.20
$43.49 $45.35
2000 2001 2002 2003
58.1% 58.4% 58.6% 58.6%
$47.60 $48.96 $50.31 $51.85
$27.66 $28.59 $29.48 $30.38
2.5% -3.1% -10.6%
-4.3% 4.1% 3.3%
-3.4% 3.1% 3.1%
Note: Numbersfor2000 areactual figures according toSmithTravel Research; 2001figures are year-end estimates basedon two quarters ofactualdata. Sources: PricewaterhouseCoooersand SmithTravel Research
search, and Ernst & Y~ung.‘~ The Pricewaterhousecoopers (PwC) forecasts have the advantage of being updated throughout the year based on changing economic conditions (rather than annually at the beginning of each year). They are also detailed by segment (upper upscale, upscale, midscaie with food and beverage, midscale without food and beverage, and economy), and they are based on statistical models that have been refined over the course of their ten-year development.17 Therefore, we used PwC’s forecasts of occupancy and ADR that were developed immediately prior to September 11 for our base-case analysis, and the forecasts developed immediately after September 11 as the gauge for the anticipated change in occupancy and ADR due to the terrorist attacks. We should point out that the projections immediately prior to September 11 reflected downward adjustments to PwC’s earlier forecasts for 2001, due to the effects of the already softening economy. The PwC forecasts are based on a roomdemand equation that returns an R2 of .996, an ADR equation with an R2 of .930, and a roomstarts equation (new hotel-room construction) with an R2 of .880. Thus, all three of those equations capture a considerable level of the variation in their particular factor. The PwC forecasts of occupancy and ADR immediately prior to September 11 are presented in Exhibit 3, while Exhibit 4 shows its forecasts immediately after September 11. (Exhibit 2, on the previous page spread, gives examples of hotel brands that are included in each segment.) Exhibits 3 and 4 show how PwC lowered its overall projections of occupancy, ADR, and RevPAR for 2001 through 2003 in the wake of September 11. The analysts made a downward adjustment for each of the identified hotel segments, with the high-price segments (upper upscale and upscale) projected to experience the most dramatic reversal. The occupancy and ADR forecasts before and after September 11 were the inputs for our multiple-regression equation. Thus, we calculated value-per-room estimates based on PwC’s occuI6See, for example, “News Briefs,” Hotel Bwiners, October 21-November 6, 2001, p. 1. ‘7 Coopers 81 Lybrand, Hospidiry Directions: Forecasts & Analysesfir the Hospitality Industry, Vol. 1, No. 1 (199 1).
16 Cornell Hotel and Restaurant Administration Quarterly
DECEMBER 2001
HOTEL VALUATION
pancy and ADR forecasts just before and just after September 11. We also calculated the change in value per room based on the pre- and postSeptember 11 value estimates to gauge the change in value due to the September 11 attacks. Those estimates of value change were calculated for 2001 through 2003. The results indicate that values will continue to lose ground through 2003 relative to anticipated values prior to September 11 (see Exhibits 5 and 6, overleaf). What It Means Asshown in Exhibit
5, our model indicates that the average U.S. hotel was worth 3.6 percent less at the end of2001 than it was in 2000 ($71,313 per room at the end of 2001 versus $73,978 per room at the end of 2000). Based on forecasts developed prior to September 11, the model projects that in the absence of the attacks, the average hotel would have instead been worth 5.0 percent more at the end of 2001 than it was at the end of 2000 ($77,673 per room in 200 1 versus $73,978 per room in 2000). That calculation shows a net loss in mean value of $6,360, or 8.6 percent, due to the events of September 11 ($6,360+$73,978). Looking ahead to 2003, the model indicates that the average U.S. hotel will lose a net $11,050 in value per room by the end of 2003 as a result of September 11 (albeit most of that loss had already occurred in 200 1). Could have been. Let us emphasize that our calculation of the loss in hotel value is based on the pre-September 11 expectations for the industry’s performance. While hotels are expected to continue to lose ground compared to “what would have been” through 2003, at this moment it appears that 2001 is the only year in which U.S. hotel values would experience an overall year-to-year decrease. We expect valuation to increase by 1.3 percent in 2002, and our model predicts an overall increase of 5.8 percent in 2003. Among the hotel segments that we examined, the model indicates that the largest overall change in value due to September 11 occurred among upper-upscale hotels, which declined in value by 8.0 percent in 2001 compared to 2000. Also, these hotels are expected to experience the largest value change compared to “what would have been” through 2003, with an overall value loss of $35,854 per room as of that year.
DECEMBER
2001
Hotel operating projections
OccuDancv
1
INDUSTRY
after September
ADR
RevPAR
ANALYSIS
11
1
Annual% Change
2000 2001 2002 2003
63.7% 60.5% 60.1% 61.2%
$85.24 $85.07 $86.01 $88.50
$54.30 $51.47 $51.69 $54.16
--5.2% 0.4% 4.8%
2000 2001 2002 2003
72.7% 68.8% 69.3% 71.3%
$155.56 $147.32 $148.27 $152.65
$113.09 $101.42 $102.74 $108.90
--10.3% 1.3% 6.0%
2000 2001 2002 2003
70.2% 68.3% 67.5% 68.7%
$99.24 $96.56 $95.35 $97.12
$69.67 $65.96 $64.32 $66.72
--5.3% -2.5% 3.7%
2000 2001 2002 2003
59.9% 58.6% 59.2% 55.3%
$73.41 $71.83 $72.51 $72.51
$43.97 $42.10 $42.91 $40.11
--4.3% 1.9% -6.5%
2000 2001 2002 2003
64.7% 64.3% 65.5% 66.9%
$67.22 $66.85 $68.09 $69.47
:z $44162 $46.45
-1.1% 3.7% 4.1%
2000 2001 2002 2003
58.1% 57.4% 57.4% 58.5%
$47.60 $47.03 $47.05 $47.95
$27.66 $27.00 $27.02 $28.06
--2.4% 0.1% 3.9%
- _... _I_,.... l.l-. _I
Note: Numbersfor2000 areactual figures according toSmithTravel Research; 2001figures are year-end estimates basedon three quarters ofactualdata. Sources: PricewaterhouseCoopersand SmithTravel Research
Cornell Hotel and Restaurant
Administration
Quarterly
17
INDUSTRY ANALYSIS
HOTEL VALUATION
I
Estimated changes in value per room Value Before S/l1
2001 2000 2002 2003
2000 2001 2002 2003
2000 2001 2002 2003
$77,873 $73,978 $82,555 $87,509
Annual
% Change
5.0% -6.3% 6.0%
$206,307
5.9%
$97,754 $102,323
-4.7% 4.6% 4.8%
Value After 9lll
f?;,;;; $72:267* $78,459
Annual
K Change
-3.8% -1.3% 5.8%
-$8,3!: -$10,287 -$11,050
$175,407
--8.0% 1.0% 4.6%
$97,754 $92,708 $90,435 $93,757
--5.2% -2.5% 3.7%
-$9,6:! -$16,641 -$18,417
$50,472
-5.5%
-$6,1:: -$8,115 -$8,685
$50,502 $49,734 $52,303 $55,174
--1.5% 5.2% 5.5%
;;;;g
--6.8% 0.3% 11.5%
..-.-I____ _. 2000 2001 2002 2003
2000 2001 2002 2003
$50,502 $54,362 $58,195 $82,047
-7.6% 7.1% 6.6%
2000 2001 2002 2003
$19,090 $21,148 $23,116 $25,187
-10.8% 9.3% 9.0%
$17:845 $19,888
-$4,8!t
T&g p&aces. Upscale hotels represent the only segment where we expect values per room to continue declining in 2002. This likely decrease in value arises for essentially the same reason that upper-upscale hotels lost value in 2001-to wit, travelers are expected to trade down to the next lower tier of properties. As a consequence of the trade-down effect, midscale hotels with food and beverage are expected to experience relatively minor value changes in 2001 and 2002. In fact, these hotels are expected to experience occupancy, ADR, and value increases in 2002. However, as the economy improves by 2003, we anticipate that this trend will reverse, and guests will be lost back to upscale hotels. with the shift in travelers will come a shift in hotel valuation. Under normal economic conditions, upscale travelers select hotels primarily for reasons other than price-to the disadvantage of the midscale hotels.“’ Consequently, the m&ale-with-F&B segment is the only one for which we predict a value decrease in 2003 compared to 2002. Thus, owners of midmarket properties who are considering selling may be well advised to sell in 2002. Limited service, escalating value. For the first time ever, the model shows that midscale hotels without food and beverage are likely to be worth more on a per-room basis in 2003 than are midscale hotels with F&B. As a group, the properties without F&B are newer than their fullservice counterparts, in part because the fullservice hotels are suffering from functional obsolescence as their designs become increasingly outmoded.‘y The valuation problem for this segment is that investors do not attribute much value to food and beverage outlets, particularly if those amenities are not seen as generating a profit (as is often the case). *OWe find it ironic that hotels that are less expensive to build (that is, those withI8B.J. Knutson, “FrequentTravelers: MakingThem Happy and Bringing Them Back,” Cornell Hotel and Restaumnt Administration Quart&y, Vol. 29, No. 1 (May 1988), pp. 83-87. I9 Appraisal Institute, TheApprairalofRealEte, edition (Chicago: Appraisal Institute, 1996).
eleventh
2oThis is by no means a recent phenomenon. See: Bjorn Hanson, “Hotel Food Service: Where’s the Profit?,” Cornelf Hotel and Restaurant Administration Quarter&, Vol. 25, No. 2 (August 1984), pp. 93-96.
18 Cornell Hotel and Restaurant Administration Quarterly
DECEMBER 2001
HOTEL VALUATION
I
INDUSTRY ANALYSIS
Estimated values per room (year 2003)
Values before September 11
Values after September 11
Ufwr
Upscals
lIpScale HOTEL
Midecale with F&B
Midscale without F&B
Economy
SEGMEWT
Note: Values shown on the above graph correspond to the year-2003 figures shown in Exhibit 5.
out F&B facilities) are seen as being more valuable than those that are more expensive to build (i.e., those with full F&B facilities). Our model indicates that the midscale hotels without F&B are emerging as the “ideal” hotel improvement in many lodging markets. At $5,000 per room, the value decrease anticipated for economy properties is the least in terms of dollars. However, this equates to a substantial percentage-approximately 25 percentof the total average value of the hotels in this segment. In considering this outcome, however, note that our model is highly sensitive to fluctuations in occupancy and ADR Thus, the average economy hotel operating with a typical 200 1 annual occupancy of 57.4 percent and ADR of $47.03 would be worth only $17,79 1 per room in this model. However, an economy hotel with an annual occupancy of 75 percent and ADR of $55, for example, would be worth $41,904 per room based on our model. If such a superior performing economy hotel loses $5,000 per room
DECEMBER 2001
in value, that decrease represents just 12 percent of its value, rather than the 25-percent value loss expected for an economy property with average statistics. While we have used economy properties in this example, the model shows that superior performance within any class should mitigate a loss in value, just as one might expect. Like all multiple-regression models, ours is effective at predicting results in the aggregate and for overall segments, but it is not so effective at predicting individual property values. This situation is particularly true of outlying properties with extremely low or high values. In this case, our model would not accurately value lodging properties with values below $17,791 or above $175,407 per room.
Hoping That the Worst Is Over Assuming that no further attacks occur on American soil, we expect that most of the drop in hotel values occurred in 200 1. As the model indicates, the shaky performance of the industry in terms
Cornell Hotel and Restaurant Administration Quarterly
19
INDUSTRY ANALYSIS
HOTEL VALUATION
I
of occupancy and ADR throughout 2001 is the primary cause of this decline. In addition, the lack of liquidity that we have noted is also contributing to this value decline. In the weeks immediately following September 11, virtually all transaction activity came to a halt, with buyers, sellers, and lenders reevaluating their investment strategies. At present, most lenders have a waitand-see attitude concerning loans to this and
Values
and transaction
that are not overly
activity
stressed
should
pick up in concert
pected
economic
for properties
by the recession
with the widely
ex-
recovery. I
many other real-estate sectors.Z1 In this environment, the transactions that are likely to occur would most likely reflect capitalization and discount rates above historical norms, mirroring the high level of uncertainty currently affecting the sector as well as the cost of low or un-leveraged transactions.22 In these market conditions, we do not expect to see many market-value transactions. Achieving a market-value exchange requires both a willing buyer and a willing seller,23 and in the prevailing investment climate, we see few willing sellers. We consider it unlikely that any owner not under duress would contemplate a sale until the market stabilizes and financing becomes more widely available. Most of those transactions that do occur in the next few months will likely involve distressed properties, for which the current situation is the last straw. These assets are expected to trade at liquidation prices, and may represent a prospect for opportunistic buyers. On the other hand, because we do not expect there to be a massive volume of these types of transactions, 21Jones Lang LaSalle Hotels, Focus on the Zmpactoft/w September Ii, 2001, Terrorist Attacks on the U.S. Hotel Real Estate Market (internal document, October 2001). l2 Ibid. 23Appraisal
Institute,
op. cit.
20 Cornell Hotel and Restaurant Administration Quarterly
such sales are not expected to have a substantial effect on national valuation averages. On the other hand, properties that have strong track records and that demonstrated some durability in their cash flows as the economy slowed should continue to be attractive prospects for both lenders and equity investors. While value adjustments necessarily reflect the current economic conditions, the availability of capital should limit any downward pressure on the value of those assets. Two markets. We expect to see a dichotomy in terms of hotel values in 2002. Those properties that succumb to the financial pressures of the recession will likely trade at prices that represent a discount to both pre-2001 values and construction costs. However, we do not expect to see a large number of deeply discounted transactions, as occurred in the early 1990s. In contrast to the poor fundamentals of many of those deals, the more responsible underwriting criteria of the past five years should help limit the sector’s vulnerability (although properties undeiwritten on the basis of their financial performance in 2000 may face dif%culties). Moreover, competition for the assets of failed hotels should limit downward pressure on prices. Recognizing that history can repeat itself, a number of opportunistic funds are already on the lookout for properties that may succumb to the current financial pressures. Competition among such funds, which are more numerous than they were in the early 199% should limit the magnitude of the actual discounts. For those properties that are not overly stressed by the recession, including those that are benefiting from changes in travel patterns, we expect that values and transaction activity will pick up in concert with the widely expected economic recovery. While financing from traditional sources may be slow to return in force, some financing will be available. The U.S. lodging industry continues to be the highest-yielding realestate sector,24 and history has shown that there is always an appetite for high-yield loans. Spreads are likely to increase, but the Federal Reserve’s easing of credit should allow the hospitality sector to tolerate some increases. 24Real Estate Research Corporation, November 2001.
RERC SpcciaI Survey
DECEMBER 2001
HOTEL VALUATION
The scenario for 2003 should be a favorable one for the hotel industry, as demand growth and limited supply additions should support increases in occupancy and ADR Moreover, managers will maintain the operating efficiencies they’ve achieved over the past several years, as well as control expenses during the current downturn and continue cost-control measures as revenue levels improve. Financing will become more readily available, as lenders under pressure to generate loans are lured by high yields and attractive market performance. Competition for hospitality properties will gain momentum on both the debt and equity side. The net effect of those influences will be a continual increase in values, which we ultimately expect to equal or exceed the levels achieved in the past three years. The purpose of this analysis is to provide an overview of hotel-value trends on a national level.
INDUSTRY ANALYSIS
As is usually the case, any analysis of the national lodging industry represents an aggregate of numerous individual markets. National trends are not always reflected directly in individual markets, which are much more subject to local economicsz5 Consequently, so m e markets may be more adversely affected by the current situation, while others will be hurt to a much lesser degree. Since our model indicates that high-rate, upperupscale hotels will experience the greatest declines in value per room, we would expect that the highprofile urban and resort markets will experience the greatest level of pain, whereas markets with historically more moderate occupancies and average daily rates should be more insulated. n 25For example, see: CathyA. Em, Linda Canina, and Kate Walsh, “Hotel-industry Averages: An Inaccurate Tool for Measuring Performance,” in this issue of Cornell QuurrerJy (pp. 22-32).
John W. D’Neill, MAI, CHE, Ph.D., is an assistant professor in the School of Hotel, Restaurant, and Recreation Management at The Pennsylvania State University (jwo30psu.edu). Anne R. Lloyd-Jones, CRE, is senior vice president at HVS International (ALloyd-JonesOHVSlnternationaLcom). 0 2001, Cornell University;
refereed paper: submission receivedNovember 20, 2001; accepted with minor revisions on January 11, 2002,
DECEMBER 2001
Cornell Hotel and Restaurant Administration
Quarterly
21