The Economic Feasibility of
a
New Luxury Hotel in
Downtown Waco
by
Tom Kelly, Ph.D.
Baylor Center for Economic
Analysis
June 1994
Introduction
Texas ranks fourth in total travel
expenditures among the 50 states, behind only California, Florida, and New
York. The 5.7 percent growth in Texas
during 1992 exceeded the 5.1 percent growth rate for the nation. Among domestic travelers, Texas also experienced
a 5.7 percent growth rate, compared with a 4.1 percent increase for the entire
nation.
Waco,
due to its geographic location on Interstate 35, is in a unique position to
benefit from a rapid growth in visitor spending in the future. However, overnight stays may be limited by
the lack of hotel space. This study
examines that possibility based upon the projected future demand for hotel
space in Waco, Texas.
The study
consists of following four phases:
Phase one determines a
baseline forecast of the demand for future hotel space in the Waco market based
upon past performance.
Phase two considers forces
that can alter this baseline forecast, including the competitive position of
Waco's lodging facilities compared with other metropolitan areas in the state
of Texas.
Phase three adjusts the
baseline forecast for projected changes in local visitor attractions, including
Big 8+ athletic events, an additional sports complex, added tourist attractions
(such as an outdoor theater), and a more competitive convention package.
Phase four compares
projected revenues with the costs involved in construction and operation of a
new hotel in order to determine if a new hotel is feasible within the
foreseeable future.
Baseline Forecast of the Demand for
Hotel Capacity in Waco
The
purpose of this phase is to determine future hotel/motel performance in the
Waco MSA based upon time series information.
This projection is a necessary step in formulating a plan for additional
hotel capacity to be located in downtown Waco.
Although past time series data can be viewed as representative of
changes in the hotel/motel market, past data will not provide information about
the possible loss of convention traffic that may have been due to the lack of
adequate hotel/motel room space.
Projections of past lodging industry performance must also be adjusted
for projected changes in local visitor attractions. For this reason, the projection of past data provides only a
baseline forecast that must be adjusted for changes in both visitor attractions
and past capacity constraints.
The Forecast Methodology
Time series data consists of four components
that may explain variation from one period to the next. These four factors can be described as trend forces that change in the long
run, business cycle forces that
change with the level of economic activity, seasonal
forces that change within the year but regularly repeat from one year to the
next, and irregular forces that
consist of outside influences that either are not expected to occur again or
can be viewed as random. In general,
trend and seasonal forces can be projected into the future more easily than
cyclical influences. Irregular
influences may be removed from past experiences to provide a better picture of
past changes, but they do not enter into future projections since they are
viewed as either random or unlikely to reoccur in the future. (An example of a irregular influence would
the be the Branch Dividian standoff, when government officials and media
crowded into Waco due to an unusual event that is unlikely to occur again in
the future.)
In
order to determine the relative importance of these four sources of change in
hotel/motel performance in the Waco market, quarterly data for lodging within
the Waco Metropolitan Statistical Area is examined for the period 1988 through
1993 using the decomposition method.
Both trend and seasonal factors are removed and the cyclical-irregular
residual is identified. Removal of past
irregular influences allows for the projection of trend values for hotel room
revenues that provides a baseline forecast of future annual hotel performance
in the Waco market. From these baseline
projection the impact of future changes in the Waco travel industry are
estimated.
Quarterly Hotel Performance
in Waco
Room revenue (the combination of occupancy
and average room rate) is the primary value used to determine changes in the
demand for hotel space in the Waco market.
Room revenue, occupancy rates, and average room prices for Waco's
lodging industry for each quarter beginning in 1988 through 1993 are shown in
Table 1. A visual inspection of the
data shows that during the first quarter of 1993 the Branch Dividian holdout
resulted in an irregular (exogenous) influence that contributed to
substantially higher hotel performance than predicted by trend, seasonal, or
cyclical influences.
Table 1: Quarterly Hotel Performance in Waco MSA
|
Yr/Qtr |
Nights |
Room |
% OCC* |
$ Rate** |
|
|
Sold |
Revenue |
(nts sold/ |
(avg price |
|
|
(thous.) |
($thous) |
nts avail) |
per nt sold) |
|
|
|
|
|
|
|
88/1 |
91.5 |
$3,313 |
43.5 |
$36.22 |
|
88/2 |
100.7 |
3,842 |
47.3 |
38.16 |
|
88/3 |
108.3 |
4,061 |
49.4 |
37.50 |
|
88/4 |
99.5 |
3,610 |
45.1 |
36.29 |
|
89/1 |
93.4 |
3,341 |
44.2 |
35.77 |
|
89/2 |
106.4 |
3,935 |
50.2 |
36.99 |
|
89/3 |
113.3 |
4,223 |
52.1 |
37.26 |
|
89/4 |
106.3 |
3,770 |
49.6 |
35.47 |
|
90/1 |
98.7 |
3,563 |
47.7 |
36.11 |
|
90/2 |
103.9 |
3,896 |
49.7 |
37.50 |
|
90/3 |
110.3 |
4,072 |
47.6 |
36.90 |
|
90/4 |
96.9 |
3,533 |
46.4 |
36.47 |
|
91/1 |
95.0 |
3,617 |
44.8 |
38.07 |
|
91/2 |
110.4 |
4,343 |
50.4 |
39.33 |
|
91/3 |
118.1 |
4,317 |
53.4 |
36.54 |
|
91/4 |
108.2 |
3,958 |
49.6 |
36.59 |
|
92/1 |
100.8 |
3,784 |
47.2 |
37.53 |
|
92/2 |
113.5 |
4,574 |
51.3 |
40.31 |
|
92/3 |
120.6 |
4,530 |
54.9 |
37.56 |
|
92/4 |
109.6 |
4,132 |
51.5 |
37.72 |
|
93/1 |
114.7 |
4,616 |
52.5 |
40.25 |
|
93/2 |
121.0 |
5,180 |
53.9 |
42.65 |
|
93/3 |
130.0 |
5,033 |
56.6 |
38.86 |
|
93/4 |
116.0 |
4,474 |
55.2 |
38.67 |
* Occupancy rate is nights sold divided by nights available for sale (x 100). ** Rate is the average price for each roomnight sold. Source: Texas Department of Commerce from Market Share/Source Strategies, Inc.
Table
1 also shows that the upward trend in room revenue in the Waco market is more a
function of higher occupancy rates than an increase in average room
prices. (One of the questions that will
be addressed is the question of the relationship between occupancy rates and
room prices and whether or not Waco can support a hotel with higher
"luxury" prices.) In general,
higher average room prices also accompany higher occupancy rates. Hence, room revenue is more representative
of market demand than either occupancy rates or room prices and, therefore,
will be the primary measure of hotel performance projected into the future.
Decomposition of Hotel
Performance
Forces
explaining changes in hotel performance consist of trend factors, such as
demographic forces that affect tourism; seasonal factors, such as the weather,
normal vacation periods, or Baylor University events that repeat within each
year; cyclical factors, such as business performance that affects convention
traffic; and irregular forces, such as the Branch Dividian holdout that are unusual
events that are unlikely to occur again in the future. A starting point in time series forecasting
is the application of the decomposition method to past time series data in
order to determine the relative importance of each of these four sources of
variation.
The
long run secular trend may be estimated with a least-squares regression line,
either in linear or non-linear form.
Seasonal changes are eliminated using seasonal indexes derived by the
ratio-to-moving-average method.
Seasonal factors are reported that coincide with the Census X-11 program
for time series decomposition. The
combined R-squared of a model using trend and seasonal factors measures the
percentage of past variation in hotel performance explained by these two
influences. The remainder of one
hundred percent is due to cyclical or irregular influences. This residual may be expressed as an index
by dividing the observed value by the forecast value based upon trend and
seasonal influences and multiplying by 100.
The resulting cyclical-irregular relative will exceed 100 when actual
performance exceeds the amount predicted by trend and seasonal factors and will
be less than 100 when actual performance falls below the predicted value. To the extent that trend and seasonal forces
are most important (high R-squared), future projections are more accurate than
if business cycle conditions or irregular influences dominate. Cyclical influences are difficult to project
very far into the future, and irregular influences cannot be predicted, since
they are viewed as unusual events that may or may not occur again in the
future.
Seasonal
indexes for quarterly hotel revenues, occupancy rates, and average price per
room were calculated for the period 1988 through 1992. (The year 1993 was not used because of the
unusual performance during the first quarter as a result of the Branch Dividian
holdout.) The results, shown in Table
2, indicate significantly greater seasonal variation in hotel revenues than in
occupancy rates. Average room prices
fluctuate even less from quarter to quarter due to seasonal influences. (Seasonal indexes average 100 for the year,
exceed 100 when seasonal forces are favorable, and fall below 100 when seasonal
forces are unfavorable.)
Table 2: Seasonal Indexes for Measures of Hotel
Performance
|
Quarter |
Room Revenue |
Occupancy Rate |
Average Room
Rate |
|
|
|
|
|
|
I |
91.5 |
94.2 |
99.5 |
|
II |
105.1 |
102.9 |
103.0 |
|
III |
108.0 |
104.1 |
100.0 |
|
IV |
95.4 |
98.7 |
97.5 |
Source: Computed from Quarterly Data from Texas Department of Commerce, Tourist Division
Trend - Seasonal Forecast of
Hotel Revenue
A least squares regression model that assumes
a nonlinear trend with a constant growth rate and a seasonal factor for each
quarter, based upon the ratio-to-moving-average, explained 83.3 percent of the
variation in quarterly hotel room revenue in the Waco Metro Area over the
period from 1988 through 1993. (A
nonlinear trend-seasonal model slightly outperformed a linear trend-seasonal
equation that explained 82.9 percent of the variation in room revenue.)
Table 3: Sources of Variation in Quarterly Hotel Room
Revenues
|
Year.Qtr |
Room |
Revenue |
Cyclical- |
Year.Qtr |
Room |
Revenue |
Cyclical- |
|
|
Actual Value |
Trend-seasonal |
Irregular Relative |
(cont.) |
Actual Value |
Trend-seasonal |
Irregular Relative |
|
|
($thous) |
Forecast |
(percent) |
|
($thous) |
Forecast |
(percent) |
|
|
|
|
|
|
|
|
|
|
88.1 |
$3,313 |
$3,319 |
99.8 |
91.1 |
3,617 |
3,752 |
96.4 |
|
88.2 |
3,842 |
3,753 |
102.4 |
91.2 |
4,343 |
4,242 |
102.4 |
|
88.3 |
4,061 |
3,847 |
105.6 |
91.3 |
4,317 |
4,392 |
98.3 |
|
88.4 |
3,610 |
3,534 |
102.2 |
91.4 |
3,958 |
3,994 |
99.1 |
|
89.1 |
3,341 |
3,458 |
96.6 |
92.1 |
3,784 |
3,908 |
96.8 |
|
89.2 |
3,935 |
3,909 |
100.7 |
92.2 |
4,574 |
4,419 |
103.5 |
|
89.3 |
4,223 |
4,047 |
104.3 |
92.3 |
4,530 |
4,575 |
99.0 |
|
89.4 |
3,770 |
3,681 |
102.4 |
92.4 |
4,132 |
4,161 |
99.3 |
|
90.1 |
3,563 |
3,602 |
98.9 |
93.1 |
4,616 |
4,071 |
113.4 |
|
90.2 |
3,896 |
4,072 |
95.7 |
93.2 |
4,574 |
4,603 |
99.4 |
|
90.3 |
4,072 |
4,216 |
96.6 |
93.3 |
4,530 |
4,766 |
95.0 |
|
90.4 |
3,533 |
3,834 |