e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

30 

 

 

Sameer Mathur 
Carnegie Mellon University, Qatar 

 

 

An Empirical Analysis of US Hotels Including Free Breakfast with Room Rent 

The likelihood of a US hotel including free breakfast with its rent is empirically measured as a 

function of hotel star rating, capacity, and rack rate. We used mixed-effects logistic regression 

to analyze the choices made by 1607 US hotels regarding whether to offer or not offer free 

breakfast with their room rent. Our analysis reveals that the probability of a hotel including 

free breakfast with its rent decreases with an increase in the star rating, decreases with an 

increase in hotel capacity, and decreases with an increase in the rack rate. These results have 

practical implications for hotel strategy.  

 

Keywords: hotels; rent; free breakfast; logistic regression; mixed effects; rating; capacity; 

rack rate; 

 

Sameer Mathur 

Associate Teaching Professor of Marketing and Business Technologies 

Business Administration 

Carnegie Mellon University, Qatar 

Education City, PO Box 24866 

Al Luqta St, Ar-Rayyan 

Qatar 

Phone: [974] 66845011 

Email: smathur@andrew.cmu.edu 

 

Sameer Mathur is an Associate Teaching Professor of Marketing and Business Technologies 

at Carnegie Mellon University, Qatar. His research uses quantitative techniques such as 

regression analysis and game theory to study marketing problems, particularly pricing and 

promotion strategies.   

 

Acknowledgment: The author is grateful to Mr. Aryansh Gupta for providing research 

assistance with data collection and analysis.  

 

 

 

 

 

 

 

http://ertr.tamu.edu/
mailto:smathur@andrew.cmu.edu


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

31 

 

1. Introduction 

Hotels are a critical part of the travel and tourism industry and the service economy, especially 

in the United States. According to market research portal, Statista and the 2019 US Travel and 

Hospitality Outlook by Deloitte, the revenue in the US hotels industry is expected to hit 

US$9.85 Billion in 2020, with an expected CAGR of 5-6% over 2020-2023. In fact, the average 

revenue per user (ARPU) is approximately US$629.78 (Statista.com; Deloitte.com). In the 

hotel industry, offering free breakfast with the room is a popular strategy adopted by an 

increasing number of hotels. For example, hotels such as Best Western, Canopy by Hilton, 

Comfort Inn, Embassy Suites, Fairfield Inn and Suites, Hampton Inn, Holiday inn, La Quinta 

Inn, and Wingate by Wyndham all offer their guests free breakfast (TripAdvisor.com; 

Creditcards.com). However, not all hotels offer guests free breakfast, as indicated in our 

survey. We surveyed a large variety of 1607 hotels situated in 34 cities in the United States of 

America and recorded whether or not they offered free breakfast with their room rent. We 

observed that 761 hotels (i.e., 47.4% hotels) included free breakfast, while the remaining 846 

hotels (i.e., 52.6% hotels) excluded breakfast from room rent. This dichotomous distribution 

of choices among hotels prompts a deeper investigation regarding the likelihood of hotels 

including free breakfast. On what factors does the likelihood of a hotel offering free breakfast 

depend? What is the profile of hotels more likely to include free breakfast, and how does their 

profile differ from those excluding free breakfast from room rent? This paper partially answers 

these questions. 

In order to formally analyze the likelihood of a hotel including free breakfast with rent, we 

analyze a mixed-logit regression model, estimated using data derived from hotels renting rooms 

on the popular www.hotels.com portal. We focus empirical attention on how three important 

hotel attributes factors – (i) star rating, (ii) capacity (i.e., number of rooms), and (iii) rack rate 

– influence this likelihood. Our regression analysis leads to the following three insights: 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

32 

 

(i) The probability of including free breakfast with rent decreases with an increase in hotel star 

rating. 

(ii) The probability of including free breakfast with rent decreases with an increase in hotel 

capacity. 

(iii) The probability of including free breakfast with rent decreases with an increase in hotel 

rack rate. 

 

 

 

 

2. Literature Review 

 

Many lodging operations offer a complimentary breakfast as a sales tool. Our research 

is closely related to Nicolau and Sellers (2012), who study the zero price effect in tourism, 

extending the research by Shampanier, Mazar, and Ariely (2007). Here, a free product can 

become so extraordinarily attractive that another much more preferred alternative is foregone 

by consumers. This insight holds critical managerial implications for multicomponent tourism 

products because fixing a zero price for a specific component (e.g., breakfast) included in a 

product (e.g., hotel room plus breakfast) can abnormally raise the demand for the product. 

In related research, Monty and Skidmore (2003) evaluate consumer willingness to pay for 

specific characteristics of free breakfast with the room, applying the hedonic price technique 

to the bed and breakfast market.  

Acknowledging that a hotel is a complex, experience-based product, Xiang and Usyal 

(2015) use text analytics to determine what leads to guest satisfaction. They analyze online 

customer reviews for hotels and report that the word ‘Breakfast’ appears with a frequency of 

2.9 times per hotel and is the 9th most frequently occurring word among the top 80 primary 

words. Lee et al. (2017) identify which bundle of breakfast items is most preferred by travelers. 

Their findings suggest that managers should keep fruit bowls, waffles, scrambled eggs, and 

coffee on their menus. 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

33 

 

Mixed-Logit Regression: We estimate the probability of offering free breakfast using 

the well-known mixed logistic regression model (McFadden and Train, 2000; Andrews et al., 

2002; Ben-Akiva et al., 2002; Hensher and Greene, 2003; Hess and Polak, 2005; Fiebig et al., 

2010). We model the probability of a hotel including free breakfast with rent as a function of 

the star rating, capacity, rack rate, spatial characteristics across different cities, and seasonality 

across ten dates. We use the mixed logit model since it is known to overcome some crucial 

limitations of the classical logit model. In particular, it allows us to model random variation 

and possible correlation in unobserved factors over time and across spatial locations. It also 

allows us to relax the strong assumption of the Independence of Irrelevant Alternatives (IIA) 

present in the classical logit model. Moreover, this model framework is flexible enough to 

reasonably approximate many random utility models (McFadden & Train, 2000). One 

limitation of mixed-effect models is that they tend to be computer resource intensive. However, 

given recent advances in computing power, simulation algorithms, and computer packages, it 

has become feasible to run them (Alfnes, 2004; Espino et al., 2008). Although the mixed logit 

model can be derived under a variety of behavioral specifications (Train, 2003), the most 

commonly used version is the random coefficient version of the model, and the same version 

is used in this paper. 

Our paper contributes to the wide spectrum of past research on choice model in the 

hotel industry context (Kim & Perdue, 2013; Mei & Zhan, 2013; Masiero et al., 2015, 2016; 

Kim and Park, 2017). In particular, it is noteworthy that the mixed logit estimation approach 

has been previously successfully used by researchers in Tourism Management (e.g., 

Albaladego-Pina & Diaz-Delfa, 2009; Choi et al., 2009; Kim and Park, 2017). For example, 

Choi et al. (2009) study the economic valuation of cultural heritage sites using choice 

modeling; Albaladego-Pina and Diaz-Delfa (2009) analyze tourist preferences for rural house 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

34 

 

stays in Spain using discrete choice modeling; while Kim and Park (2017) study the moderating 

role of context in hotel choice in a mixed-logit framework.  

Discrete choice modeling has often been utilized to empirically analyze the relationship 

between product characteristics and choice in tourism and hospitality domains (Crouch & 

Louviere, 2000). For example, such models have been frequently used in analyzing the 

destination choice of tourists (Morley, 1994; Eymann & Ronning, 1997; Seddighi & 

Theocharous, 2002), while other applications include analyzing restaurant selection (Kim & 

Geistfeld, 2003; Kim & gu, 2006), recreational choice (Bhat & Gossen, 2004). Consistent with 

the methodology used in previously published papers, our research also used a mixed-logit 

model framework to analyze the factors driving the likelihood of US hotels offering free 

breakfast with room rent.  

3. Empirical Analysis of Hotels Including Free Breakfast with Room Rent 

The overall objective behind the empirical study presented in this paper is to investigate 

the likelihood of hotels offering free breakfast. The study evaluates the probability of a hotel 

including free breakfast in a mixed logit model framework. The goal is to analyze how the (i) 

star rating, (ii) capacity, (iii) rack rate of a hotel influence and impact its probability of 

including free breakfast with room rent. 

3.1 Data 

 

We collected hotel data from the popular travel portal www.hotels.com. Our dataset 

consists of a total of 𝐼 = 1603 unique hotels located in 𝐽 = 34 major cities across the United 

States, such as Chicago, New York, Los Angeles, Houston, Pheonix, and Boston. The dataset 

includes (i) the name of the hotel; (ii) the city in which it is located; (iii) the star rating on a 5-

point scale; (iv) the capacity, i.e., maximum number of available rooms; (v) whether or not the 

hotel includes free breakfast with its room rent; (vi) the rack rate, which is the "published rate," 

or "the maximum rent that a property charges for a room. During Nov 1-3, 2017, we recorded 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

35 

 

(vii) the rent for a standard double occupancy room for (viii) 𝑇 = 10 dates (Nov 29, 2017 - 

Dec 8, 2017), as posted on www.hotels.com. After cleaning the data with missing values, this 

yielded a dataset of 𝑁 = 15057 rows. The meaning of the data columns is summarized in 

Table 1. Table 2 summarizes the room rent, rack rate, and hotel capacity with respect to the 

star rating and whether or not free breakfast is included with the rent. Table 2 also includes the 

correlation matrix between star rating, capacity, and rack rate, and the correlations raise the 

possibility of multicollinearity in the data.  It is important to acknowledge that the data used in 

this paper is from pre-Covid times (Nov 2017). 

 

Table 1: Data variables and description. 

𝑖 Index to track a hotel 

𝑗 Index to track the city in which hotel 𝑖 is located, (1 ≤ 𝑗 ≤ 𝐽, where  𝐽 =

34) 

𝑡 Index to track the date (1 ≤ 𝑡 ≤ 𝑇, 𝑤ℎ𝑒𝑟𝑒 𝑇 = 10) 

𝐶𝑖𝑡𝑦𝑗 The name of the city in which hotel 𝑖 is located. 

𝐷𝑎𝑡𝑒𝑡 The date for which hotel rent was recorded, where 𝐷𝑎𝑡𝑒𝑡 ∈ (𝑁𝑜𝑣 29 −

𝐷𝑒𝑐 8, 2017)  

𝑆𝑡𝑎𝑟𝑅𝑎𝑡𝑖𝑛𝑔𝑖𝑗 The star rating of hotel i, in City j, where 𝑅𝑎𝑡𝑖𝑛𝑔𝑖𝑗 ∈

{2.0,2.5, 3.0,3.5,4.0,4.5,5.0} 

𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦𝑖𝑗 The capacity (i.e., maximum number of rooms) of hotel 𝑖, in 𝐶𝑖𝑡𝑦𝑗. 

𝑅𝑎𝑐𝑘𝑅𝑎𝑡𝑒𝑖𝑗𝑡 The rack rate (i.e., maximum rent) for a standard, double occupancy 

room at hotel 𝑖, in 𝐶𝑖𝑡𝑦𝑗, for 𝐷𝑎𝑡𝑒𝑡 (measured in USD) 

𝐹𝐵𝑖𝑗𝑡 Indicator variable for whether hotel 𝑖, in 𝐶𝑖𝑡𝑦𝑗  on 𝐷𝑎𝑡𝑒𝑡  bundled free 

breakfast with rent.  

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

36 

 

𝑅𝑒𝑛𝑡𝑖𝑗𝑡 The rent for a standard, double occupancy room at hotel 𝑖, in 𝐶𝑖𝑡𝑦𝑗, for 

𝐷𝑎𝑡𝑒𝑡 (measured in USD) 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

37 

 

Table 2: Summary Statistics of room rent, rack rate, and hotel capacity, with respect to the 

star rating and whether or not free breakfast is bundled with rent.  

 
StarRating 

(S) 

Free 

Breakfast 

(F) 

No. of 

Hotels 

% Hotels 

including Free 

Breakfast 

Mean 

Capacity 

Mean Rack 

Rate, in 

USD 

Mean Rent 

in USD 

2.0 No 114 47.2 79.8 $88.82 $71.33 

2.0 Yes 123 52.8 76.9 $93.26 $73.69 

2.5 No 88 21.4 99.9 $125.25 $98.72 

2.5 Yes 314 78.6 97.1 $126.62 $100.52 

3.0 No 190 48.0 167.0 $171.58 $127.68 

3.0 Yes 204 52.0 119.5 $168.27 $129.68 

3.5 No 167 71.6 254.0 $210.91 $156.01 

3.5 Yes 68 28.4 166.4 $181.10 $147.61 

4.0 No 195 84.1 314.3 $270.76 $204.71 

4.0 Yes 37 15.9 250.0 $230.60 $184.12 

4.5 No 58 85.1 295.8 $381.08 $278.47 

4.5 Yes 10 14.9 323.1 $314.78 $259.46 

5.0 No 36 81.3 199.6 $875.15 $623.94 

5.0 Yes 8 18.7 195.6 $592.07 $452.79 

All No 846 52.6 209.19 $229.72 $171.32 

All Yes 761 47.4 116.54 $148.70 $117.29 

All All 1607 100.0 164.74 $190.85 $145.40 

 

 
StarRating Capacity RackRate 

StarRating 1 0.4871*** 0.5593*** 

Capacity 0.4871 1 0.1936*** 

RackRate 0.5593*** 0.1936*** 1 

 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

38 

 

3.2 Model 

We propose a Generalized Linear Mixed Model (GLMM) to empirically measure the 

probability of a hotel bundling free breakfast along with the rent.  The GLMM is an extension 

of logistic regression to include fixed and random effects (Gilmore et al. 1985; McCulloch, 

1997). They are an extension of the class of generalized linear models in which random effects 

are added to the linear predictor. This allows the modeling of correlated, possibly nonnormally 

distributed data with flexible accommodation of covariates (McCulloch and Neuhaus 2005). 

The general form of the GLMM model, in matrix notation, is  

𝒚 = 𝑿𝜷 + 𝒁𝒖 + 𝝐      (1) 

Here, 𝒚 is an 𝑁 × 1 column vector of the outcome variable. Our model's outcome variable is 

binary, indicating whether a hotel includes or excludes free breakfast with its room rent. 

Specifically, 𝐹𝑟𝑒𝑒𝐵𝑟𝑒𝑎𝑘𝑓𝑎𝑠𝑡𝑖𝑗𝑡 = 1, if hotel 𝑖, located in city 𝑗 includes free breakfast with 

its room rent on date t, with 𝐹𝑟𝑒𝑒𝐵𝑟𝑒𝑎𝑘𝑓𝑎𝑠𝑡𝑖𝑗𝑡 = 0, otherwise.  

Also, 𝑿  is an 𝑁 × 𝑓 matrix of f predictor variables, while 𝜷 is an 𝑓 × 1 column vector 

of the fixed-effects regression coefficients. In our model, the fixed-effects include the hotel 

rating, capacity, rack rate, and the interaction between the rating and capacity. Specifically, 

𝑅𝑎𝑡𝑖𝑛𝑔𝑖𝑗 and 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦𝑖𝑗 represent the star rating and capacity (i.e., no of rooms) for hotel 𝑖, 

located in city 𝑗. 𝑅𝑎𝑐𝑘𝑅𝑎𝑡𝑒𝑖𝑗𝑡 represents the rack rate (i.e. maximum rent) for hotel 𝑖, located 

in city 𝑗 on date 𝑡.  

It should be noted that 𝒁 is the 𝑁 × 𝑞 design matrix for the 𝑞 random effects, while 𝒖 

is a 𝑞 × 1 vector of the random effects. In our model, the random effects include the 𝑞 = 34 

cities. We model the cities as random effects because we expect that the decision to include 

free breakfast with rent may be correlated across cities.  

Finally, 𝜺 is an 𝑁 × 1 column vector of the residuals. It is the part of 𝒚 that is not 

explained by the model, 𝑿𝜷 + 𝒁𝒖.  Note that 𝒁 is a large, 15057 × 34 sparse matrix. Since 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

39 

 

we are only modeling random intercepts, the matrix 𝒁 codes which city a given hotel is located 

in. Also, the random effects are modeled as deviations from the fixed effects 𝜷 and therefore 

have a mean zero. We assume that 𝒖 ∼ 𝑵 (𝟎, 𝑮), where 𝑮 is the variance-covariance matrix of 

the random effects. Since we only have a random intercept, 𝑮 is just a 1×1 matrix, the variance 

of the random intercept. 

Let 𝑝𝑖𝑗𝑡 represent the probability of hotel i, in city j, date t, bundling free breakfast along 

with its rent. Then, the log-odds ratio of bundling free breakfast with rent is 𝑦𝑖𝑗𝑡 = 𝑙𝑜𝑔 (
𝑝𝑖𝑗𝑡

1−𝑝𝑖𝑗𝑡
). 

We are interested in the influence of hotel rating, hotel capacity, and the rack rate on a 

hotel's decision to bundle free breakfast with rent. We adopt the logistic regression framework 

to model this relationship as follows. 

𝑦𝑖𝑗𝑡 = 𝑙𝑜𝑔 (
𝑝𝑖𝑗𝑡

1 − 𝑝𝑖𝑗𝑡
)

=  𝛽0 + 𝛽𝑠𝑅𝑎𝑡𝑖𝑛𝑔𝑖𝑗 + 𝛽𝑐𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦𝑖𝑗 +  𝛽𝑠𝑐𝑅𝑎𝑡𝑖𝑛𝑔𝑖𝑗 ∗ 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦𝑖𝑗

+  𝛽𝑟𝑅𝑎𝑐𝑘𝑅𝑎𝑡𝑒𝑖𝑗𝑡 + 𝜖 

 (2) 

Table 3 gives the output of the mixed logit regression model described above.  

Table 3: Mixed Logit Regression output to estimate the probability of a hotel including free 

breakfast, as a function of hotel star rating, capacity, and rack rate 

 

 

 

 

 

 

 

 

 

 

 

Signif. codes:  ‘***’ p < 0.001;  ‘**’ p <  0.01; ‘*’ p < 0.05; ‘.’ p <  0.1 

 

Predictors Estimate Std. Error  z value Pr(>|z|) 

(Intercept)         2.8039 0.23238 12.06 < 2e-16 *** 

StarRating (S) -0.70548 0.05472 -12.892 < 2e-16 *** 

Capacity (C) -0.00257 0.00123 -2.088 0.0368 *   

RackRate (R) -0.000335 0.00019 -1.768 0.0771.  

StarRating (S) * Capacity 

(C) -0.00072 0.00035 -2.052 0.0401 *   

 Random Effects 

Groups Name Variance Std.Dev. 

City 0.9578 0.9787 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

40 

 

 

 

 

 

 

Correlation of Fixed Effects 

 

 

 

 

3.3 Results  

The mixed logit regression explains how the hotel star rating influences the likelihood of 

bundling free breakfast. 

Result 1: (Effect of Star Rating on the Probability of Bundling Free Breakfast) 

The probability of including free breakfast with rent decreases with an increase in hotel 

star rating. 

Illustration of Result 1 for Chicago, and New York City, USA: Figure 1 plots Result 1 for 

the city of Chicago. It depicts the decrease in the probability of Chicago hotels of average 

capacity (165 rooms), including free breakfast with their rent, as the star rating increases. As a 

robustness check, Figure 1 also depicts a similar trend for smaller capacity Chicago hotels (15 

rooms) and larger capacity Chicago hotels (316 rooms). In fact, as summarized in Table 4, the 

estimated probabilities of 2, 3, 4, and 5-star hotels in Chicago, including free breakfast with 

their rent, are 43.2%, 25.0%, 12.8%, and 6%, respectively.   

The trend in New York City hotels is qualitatively similar to Chicago. Also summarized 

in Table 4, the estimated probabilities of 2, 3, 4, and 5-star hotels in New York City, including 

Date 0 0 

Observations (N) 15,057  
Deviance 16134.4  
Log Likelihood (LL) -8067.18  
AIC 16148.36  
BIC 16201.7  

 Intercept StarRating Capacity RackRate 

Rating -0.653    

Capacity -0.528 0.627   

RackRate 0.063 -0.313 0.076  

StarRating*Capacity 0.545 -0.691 -0.981 -0.065 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

41 

 

free breakfast with their rent, are 68.5%, 48.8%, 29.5%, and 15.5%, respectively. These 

probabilities have been derived from the mixed logit regression output shown in Table 3.  

Table 4: Probability of Chicago and New York City hotels bundling free breakfast with rent 

(illustrating Result 1) 

 

City 
Hotel 

Capacity 

Star 

Rating 

Probability of 

bundling Free 

Breakfast 

Chicago 165 

2 43.2% 

3 25.0% 

4 12.8% 

5 6.0% 

New York 165 

2 68.5% 

3 48.8% 

4 29.5% 

5 15.5% 

 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

42 

 

Figure 1: Probability of hotels in Chicago, USA bundling free breakfast as a function of their 

star rating for varying levels of hotel capacity. 

 

 

15 rooms 

316 rooms 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

43 

 

The mixed logit regression also describes how the hotel capacity, i.e., the number of available 

rooms in a hotel, influences the likelihood of including free breakfast, as follows. 

Result 2 (Effect of Hotel Capacity on the Probability of Bundling Free Breakfast) 

The probability of including free breakfast with rent decreases with an increase in hotel 

capacity. 

The regression output shown in Table 3 also indicates a statistically significant negative 

interaction between hotel star rating and hotel capacity. This suggests that the probability of 

bundling free breakfast with hotel rent is the highest for low-capacity, low-star-rated hotels.  

Illustration of Result 2 for Chicago and New York City, USA: Figure 2 depicts the 

probability of bundling free breakfast with rent in Chicago declines as the hotel capacity 

increases. As a robustness check, we verify that this trend is true for 3-star, 4-star and 5-star 

hotels. In fact, as summarized in Table 5, the estimated probabilities of 3-star hotels located in 

Chicago, having a capacity of 100, 200, and 300 rooms offering free breakfast, are 31%, 22%, 

and 15%, respectively. Similarly, the estimated probabilities of 5-star hotels located in 

Chicago, having a capacity of 100, 200, and 300 rooms, respectively, offering free breakfast 

are 8.7%, 4.9%, and 2.7%, respectively.   

Table 5 further indicates that the trend in New York City hotels is qualitatively similar to 

Chicago, consistent with Result 2.  

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

44 

 

Table 5: Probability of Chicago and New York City hotels bundling free breakfast with rent 

(illustrating Result 2) 

 

City 
Star 

Rating 

Hotel 

Capacity 

Probability of 

bundling Free 

Breakfast 

Chicago 

3 

100 31.0% 

200 22.0% 

300 15.0% 

4 

100 17.3% 

200 10.8% 

300 6.5% 

5 

100 8.7% 

200 4.9% 

300 2.7% 

New York 

City 

3 

100 56.4% 

200 44.7% 

300 33.5% 

4 

100 37.4% 

200 25.7% 

300 16.7% 

5 

100 21.5% 

200 12.9% 

300 7.4% 

 

 

 

 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

45 

 

Figure 2: Probability of hotels in Chicago bundling free breakfast as a function of hotel 

capacity for 3-star, 4-star, and 5-star rated hotels.  
 

 
 

 

3-Star 

4-Star 

5-Star 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

46 

 

It is noteworthy that similar trends can be verified for the remaining 32 US cities in our dataset, 

lending credibility and robustness to Results 1 and 2.  

Lastly, the mixed logit regression also describes how the rack rate of hotels, i.e., the 

maximum posted rent for hotels, influences the likelihood of including free breakfast. 

Result 3 (Effect of Rack Rate on the Probability of Bundling Free Breakfast) 

The probability of including free breakfast with rent decreases with an increase in hotel 

rack rate. 

Here, it is important to acknowledge that we were able to find limited evidence to support 

Result 3. Specifically, we observed a weakly significant relationship between the probability 

of including free breakfast with rent and the corresponding rack rate (p < .1). We acknowledge 

that the p-value is 0.0771. This suggests that the influence of rack rate on the probability of 

offering free breakfast is relatively weaker than the corresponding influence of star rating and 

capacity, characterized in Result 1 and Result 2.   

Illustration of Result 3 for Chicago and New York City, USA: Figure 3 illustrates Result 3 

for Chicago and New York City. Figure 3 shows that the probability of bundling free breakfast 

with rent in Chicago declines slightly as the rack rate increases. This trend is true for 3-star, 4-

star and 5-star hotels. The trend in New York City hotels is qualitatively similar to Chicago, as 

summarized in Table 6.  

Table 6: Probability of Chicago hotels bundling free breakfast with respect to rack rate 

(illustrating Result 3) 

 

Hotel 

Capacity 

Star 

Rating 

Rack Rate Probability of bundling Free 

Breakfast 

165 3 

50 25.97% 

100 25.64% 

150 25.32% 

200 25.01% 

165 5 250 5.95% 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

47 

 

300 5.85% 

350 5.76% 

400 5.67% 

 

 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

48 

 

Figure 3: Probability of hotels located in Chicago, USA bundling free breakfast as a function 

of rack rate 

 

 
 

 

  

3 Star 

4 Star 

5 Star 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

49 

 

4. Discussion 

We discuss the marketing implications for practitioners and the research implications arising 

from this paper.  

Marketing implications for practitioners: We begin by discussing how practitioners can use 

the findings in the paper from a marketing perspective. These marketing implications for 

practitioners are justified based on the empirical findings in the paper. Our first insight for 

practitioners is that relatively higher-rated hotels (e.g., five-star hotels) are significantly less 

likely to include free breakfast with their rent than relatively lower-rated hotels (e.g., three-star 

hotels). This makes sense. It is well known that five-star rated hotels offer high quality of 

service and luxury for correspondingly very high rents. Consequently, the consumers staying 

at such hotels are relatively less price sensitive and are more willing to pay for hotel rooms. 

Therefore, compared to lower-rated three-star hotels, five-star hotels do not benefit much from 

offering their consumers free breakfast. In contrast, lower-rated three-star hotels offer lower 

service quality and are far more of a commodity. They face heavy price competition. The 

consumers who prefer to stay at such hotels are more price sensitive and are less willing to pay. 

Therefore, it makes more sense for three-star hotels to tempt their consumers to stay at their 

hotel by including a freebie such as breakfast as part of the deal.  

Our second insight for practitioners is that larger capacity hotels are less likely to 

include free breakfast with rent compared to relatively lower capacity hotels with fewer rooms. 

One explanation for this is that when a large number of rooms are involved, the corresponding 

profit from selling breakfast is a relatively large amount since the profit will be the average 

margin made on each breakfast multiplied by the number of rooms occupied. This suggests 

that businesses managing hotels having a large number of rooms should not bundle free 

breakfast along with room rent.  

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

50 

 

Our third insight for practitioners is that hotels with larger rack rates are less likely to 

offer free breakfast with rent than hotels with relatively low rack rates. An explanation is that 

the rack rates of highly rated hotels, such as five-star hotels, tend to be significantly larger than 

the rack rates of comparatively lower-rated hotels. Given this positive correlation between star 

rating and rack rate, and our previous explanation for why higher-rated hotels are less likely to 

offer free breakfast compared to lower-rated hotels, we expect to see a similar relationship 

emerge for the rack rate. However, given the magnitude of the findings, it appears that the 

influence of rack rate is relatively less severe than the influence of star rating and hotel capacity 

in driving the hotel's decision to include or exclude free breakfast with rent. 

Research implications: This paper began with the anecdotal observation that approximately 

half the hotels in the US market include free breakfast with their rent, while the remaining half 

exclude free breakfast from room rent. This observation is prima facie quite puzzling with no 

obvious answer and therefore worthy of deeper analysis and discussion. After all, if one 

strategy strictly dominated the other strategy, we would observe nearly all hotels following one 

strategy. Instead, we have a near lack of unanimity on the best strategy – including or excluding 

free breakfast with rent. Our research contribution lies in using the mixed logit regression 

analysis framework in exploring this issue and systematically characterizing how hotel star 

rating, capacity, and hotel room rack rate jointly influence this decision.   

 

5. Research Limitations and Conclusion 

This paper has investigated the popular hotel strategy of including free breakfast with 

room rent, which is an important strategic decision. Our research suffers from several 

limitations, a few of which are as follows. We confine our attention to the US market. It will 

be interesting to see how robust our results are in European and Asian markets. Our focus was 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

51 

 

on the impact of three factors -- hotel star rating, capacity, and rack rate set by hotels, in 

determining the likelihood of their including free breakfast with rent. We acknowledge that a 

variety of factors besides these three factors may also influence this decision, and we leave this 

exploration for future research. For instance, future research could address the impact of other 

Socioeconomic factors in determining this choice.  

It is also important to acknowledge that the data used in this paper is from pre-Covid 

times (Nov 2017), and it is possible that Covid may have influenced the results and implications 

presented in this paper. 

To summarise, our research has shown that, relatively speaking, three-star hotels having 

relatively few rooms and relatively low rack rates and relatively low rack rates are the most 

likely to include free breakfast with rent. On the other extreme, five-star hotels with large 

capacity and relatively large rack rates are the least likely to include free breakfast with room 

rents. 

 

References 

 

Albaladejo-Pina, I. P., & Díaz-Delfa, M. T. (2009). Tourist preferences for rural house stays: 

Evidence from discrete choice modelling in Spain. Tourism Management, 30(6), 805-

811. 

Alfnes, F. (2004). Stated preferences for imported and hormone-treated beef: application of a 

mixed logit model. European Review of Agricultural Economics, 31(1), 19-37. 

Andrews, R. L., Ainslie, A., & Currim, I. S. (2002). An empirical comparison of logit choice 

models with discrete versus continuous representations of heterogeneity. Journal of 

Marketing Research, 39(4), 479-487. 

Ben-Akiva, M., McFadden, D., Train, K., Walker, J., Bhat, C., Bierlaire, M., ... & Daly, A. 

(2002). Hybrid choice models: Progress and challenges. Marketing Letters, 13(3), 163-

175. 

Bhat, C. R., & Gossen, R. (2004). A mixed multinomial logit model analysis of weekend 

recreational episode type choice. Transportation Research Part B: 

Methodological, 38(9), 767-787. 

Choi, A. S., Ritchie, B. W., Papandrea, F., & Bennett, J. (2010). Economic valuation of cultural 

heritage sites: A choice modeling approach. Tourism Management, 31(2), 213-220. 

Crouch, G. I., & Louviere, J. J. (2000). A review of choice modeling research in tourism, 

hospitality, and leisure. Tourism Analysis, 5(2-3), 97-104. 

Espino, R., Martín, J. C., & Román, C. (2008). Analyzing the effect of preference heterogeneity 

on willingness to pay for improving service quality in an airline choice 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

52 

 

context. Transportation Research Part E: Logistics and Transportation Review, 44(4), 

593-606. 

Eymann, A., & Ronning, G. (1997). Microeconometric models of tourists' destination 

choice. Regional Science and Urban Economics, 27(6), 735-761. 

Fiebig, D. G., Keane, M. P., Louviere, J., & Wasi, N. (2010). The generalized multinomial 

logit model: accounting for scale and coefficient heterogeneity. Marketing 

Science, 29(3), 393-421. 

Free breakfast, free dinner: Hotels that offer them, card tips to get them. (2019, July 18). 

Retrieved from https://www.creditcards.com/credit-card-news/hotels-free-breakfast-

free-dinner.php 

Gilmour, A. R., Anderson, R. D., & Rae, A. L. (1985). The analysis of binomial data by a 

generalized linear mixed model. Biometrika, 72(3), 593-599. 

Hensher, D. A., & Greene, W. H. (2003). The mixed logit model: the state of 

practice. Transportation, 30(2), 133-176. 

Hess, S., & Polak, J. W. (2005). Mixed logit modelling of airport choice in multi-airport 

regions. Journal of Air Transport Management, 11(2), 59-68. 

Hotels - United States: Statista Market Forecast. (n.d.). Retrieved from 

https://www.statista.com/outlook/267/109/hotels/united-states 

Kim, E. J., & Geistfeld, L. V. (2003). Consumers' restaurant choice behavior and the impact of 

socio-economic and demographic factors. Journal of Foodservice Business 

Research, 6(1), 3-24. 

Kim, H., & Gu, Z. (2006). Predicting restaurant bankruptcy: A logit model in comparison with 

a discriminant model. Journal of Hospitality & Tourism Research, 30(4), 474-493. 

Kim, D., & Park, B. J. R. (2017). The moderating role of context in the effects of choice 

attributes on hotel choice: A discrete choice experiment. Tourism Management, 63, 

439-451.  

Kim, D., & Perdue, R. R. (2013). The effects of cognitive, affective, and sensory attributes on 

hotel choice. International Journal of Hospitality Management, 35, 246-257. 

Lee, S. H., Lee, J., & Neilson, S. M. (2018). Exploring guest preferences of breakfast menu: 

conjoint analysis. Journal of culinary science & technology, 16(2), 149-164. 

McCulloch, C. E. (1997). Maximum likelihood algorithms for generalized linear mixed 

models. Journal of the American Statistical Association, 92(437), 162-170. 

McCulloch, C. E., & Neuhaus, J. M. (2005). Generalized linear mixed models. Encyclopedia 

of Biostatistics, 4. 

McFadden, D., & Train, K. (2000). Mixed MNL models for discrete response. Journal of 

Applied Econometrics, 15(5), 447-470. 

Masiero, L., Heo, C. Y., & Pan, B. (2015). Determining guests’ willingness to pay for hotel 

room attributes with a discrete choice model. International Journal of Hospitality 

Management, 49, 117-124. 

Masiero, L., Pan, B., & Heo, C. Y. (2016). Asymmetric preference in hotel room choice and 

implications on revenue management. International journal of hospitality 

management, 56, 18-27. 

Mei, H., & Zhan, Z. (2013). An analysis of customer room choice model and revenue 

management practices in the hotel industry. International Journal of Hospitality 

Management, 33, 178-183. 
Monty, B., & Skidmore, M. (2003). Hedonic pricing and willingness to pay for bed and 

breakfast amenities in Southeast Wisconsin. Journal of Travel Research, 42(2), 195-

199. 
Morley, C. L. (1994). Experimental destination choice analysis. Annals of tourism 

research, 21(4), 780-791. 

http://ertr.tamu.edu/


e-Review of Tourism Research (eRTR), Vol. 19, No. 01, 2022 

http://ertr.tamu.edu 

 

53 

 

Nicolau, J. L., & Sellers, R. (2012). The free breakfast effect: an experimental approach to the 

zero price model in tourism. Journal of Travel Research, 51(3), 243-249. 

Schwartz, Z., & Uysal, M. (2015). What types of hotels make their guests (un) happy? Text 

analytics of customer experiences in online reviews. In Information and 

communication technologies in tourism 2015 (pp. 33-45). Springer. 

Shampanier, K., Mazar, N., & Ariely, D. (2007). Zero as a special price: The true value of free 

products. Marketing Science, 26(6), 742-757. 

Seddighi, H. R., & Theocharous, A. L. (2002). A model of tourism destination choice: a 

theoretical and empirical analysis. Tourism Management, 23(5), 475-487.  

The 10 Best Hotels in New York City 2020 (with Prices). (n.d.). Retrieved from 

https://www.tripadvisor.in/Hotels-g60763-New_York_City_New_York-Hotels.html 

Train, K. E. (2009). Discrete choice methods with simulation. Cambridge university press. 

US Travel and Hospitality Outlook 2019. Retrieved from 

https://www2.deloitte.com/content/dam/Deloitte/us/Documents/consumer-

business/us-consumer-2019-us-travel-and-hospitality-outlook.pdf 

 

http://ertr.tamu.edu/