CONTENTS
BACKGROUND ............................................................................................................................................ 1 OBJECTIVES ............................................................................................................................................... 1 METHODOLOGY.......................................................................................................................................... 2 DISCUSSION ................................................................................................................................................ 3 FINDINGS .................................................................................................................................................... 3 DENSITY ..................................................................................................................................................... 4 ESTIMATING TOTAL LODGING UNITS ............................................................................................................. 4 APPENDIX A: INDIVIDUALLY ADVERTISED UNITS AND HOUSING UNITS ......................................... 6 APPENDIX B: INDIVIDUALLY ADVERTISED UNITS AND HOUSING UNITS MAPS ........................... 10 APPENDIX C: STUDY METHODS ............................................................................................................ 15 SELECTING WEBSITES ............................................................................................................................... 15 DEFINITIONS.............................................................................................................................................. 15 DATA ........................................................................................................................................................ 16 Data from Internet Listings .................................................................................................................. 16 Housing Data ...................................................................................................................................... 16 DATA COLLECTION .................................................................................................................................... 16 DATA PROCESSING.................................................................................................................................... 16 DATA CLEANING ........................................................................................................................................ 17 DUPLICATE LISTINGS ................................................................................................................................. 17
LIST OF TABLES TABLE 1: TOTAL NUMBER OF INDIVIDUALLY ADVERTISED UNITS IN 2014............................................................ 3 TABLE 2: DENSITY, RATIO OF INDIVIDUALLY ADVERTISED UNITS TO HOUSING STOCK IN HAWAI‘I, 2014 ............... 4 TABLE 3: NUMBER OF LODGING UNITS IN THE STATE OF HAWAI‘I BY TYPE, 2014 ............................................... 5 TABLE A-1: HAWAI‘I ISLAND: INDIVIDUALLY ADVERTISED UNITS BY ZIP CODE ..................................................... 6 TABLE A-2: KAUA‘I INDIVIDUALLY ADVERTISED UNITS BY ZIP CODE ................................................................... 7 TABLE A-3: MAUI INDIVIDUALLY ADVERTISED UNITS BY ZIP CODE ..................................................................... 8 TABLE A-4: MOLOKA‘I AND LĀNA‘I INDIVIDUALLY ADVERTISED UNITS BY ZIP CODE ............................................. 8 TABLE A-5: O‘AHU INDIVIDUALLY ADVERTISED UNITS BY ZIP CODE ................................................................... 9
LIST OF FIGURES
FIGURE B-1: FIGURE B-2: FIGURE B-3: FIGURE B-4: FIGURE B-5: FIGURE B-6: FIGURE B-7: FIGURE B-8:
HAWAI‘I ISLAND NUMBER OF INDIVIDUALLY ADVERTISED UNITS BY ZIP CODE ............................... 10 HAWAI‘I ISLAND: INDIVIDUALLY ADVERTISED UNITS DENSITY BY ZIP CODE ................................... 11 KAUA‘I NUMBER OF INDIVIDUALLY ADVERTISED UNITS BY ZIP CODE ............................................ 12 KAUA‘I INDIVIDUALLY ADVERTISED UNITS DENSITY BY ZIP CODE ................................................. 12 MAUI COUNTY NUMBER OF INDIVIDUALLY ADVERTISED UNITS BY ZIP CODE ................................. 13 MAUI COUNTY INDIVIDUALLY ADVERTISED UNITS DENSITY BY ZIP CODE ...................................... 13 O‘AHU NUMBER OF INDIVIDUALLY ADVERTISED UNITS BY ZIP CODE ............................................. 14 O‘AHU INDIVIDUALLY ADVERTISED UNITS DENSITY BY ZIP CODE.................................................. 14
BACKGROUND The individual vacation rental market has rapidly evolved in recent years and demand for accommodations beyond the traditional hotel, condominium-hotel, or timeshare rooms, has increased globally. •
Renting out individual condominiums or homes (also known as vacation rentals, individual vacation units, or IVU) has long existed in Hawai‘i.
•
These vacation rental units were a very small part of Hawai‘i’s total lodging supply. Units were marketed using traditional methods such as yellow pages, print advertising, and directories.
•
Internet marketing (individual websites) and internet-based distribution channels which specifically market and book individual vacation units have allowed the demand to rapidly increase.
•
Non-traditional units now represent approximately 25 percent of Hawai‘i’s total visitor unit count.
The Hawai‘i Tourism Authority (HTA) desired to have more information about vacation rentals. The study was conducted by SMS Research in phases at year-end 2013 and between August and September of 2014 to determine the estimated number of vacation rental units statewide being advertised individually on the internet. Information reported here includes estimates of the number of individually advertised units, number of bedrooms, unit capacity, and unit density (individually advertised units per 100 residential housing units). These are reported for the State as a whole, each of the six major islands, and for smaller areas called “cities” which are agglomerations of zip code areas. Finally, results are used to estimate the total number of all types of lodging units in Hawai’i in 2014.
OBJECTIVES Objectives of this study include the following: •
Research the evolving use of visitor accommodations in Hawai‘i;
•
Understand the impact of the internet distribution channels and the shared economy on the availability and use of visitor accommodations;
•
To determine the number of lodging units in Hawai‘i; and
•
Quantify how these changes affect the structure of Hawai‘i’s lodging supply.
VPI Ancillary Study © SMS
Page 1 December, 2014
METHODOLOGY This study was designed to develop a reasonable estimate of independently marketed visitor units by examining selected internet marketing sites. A summary of the methods are described below. •
Sites were chosen for its large number of listing, uniqueness of listings between sites, availability of address or zip code data. Websites chosen for the project were: – vacation rentals by Owner (VRBO): http://www.vrbo.com/vacationrentals/usa/hawaii – Clearstay.com: http://www.clearstay.com/Vacation+Rentals/US-Hawaii – TripAdvisor: http://www.tripadvisor.com/hawaii-vacation-rentals.html – Airbnb: https://www.airbnb.com/s/Hawaii--United-States
•
The units identified by this search were referred to as Individually Advertised Units.
•
The number of bedrooms was taken directly from listings on the four websites. In rare cases the number of bedrooms was not given and was estimated from the unit capacity under an assumption of two persons per bedroom.
•
Most listings also included unit capacity, the maximum number of guests that can be accommodated by the unit. In some cases, the listing gave the capacity in terms of the number of persons who can sleep there. That figure could include the use of cots or rollaway beds. In cases where only the number of bedrooms was given, the assumption was made that each bedroom contained two beds and could sleep two people. Because of this latter assumption, the estimate given here may be a conservative one.
•
A preliminary study was conducted at the end of 2013 and a second wave of data collection occurred between mid-August and mid-September 2014. This report mainly focuses on the analysis of the 2014 data.
VPI Ancillary Study © SMS
Page 2 December, 2014
DISCUSSION Findings 22,238 Individually Advertised Units in Hawai‘i were identified between mid-August and midSeptember 2014 1. Table 1 shows the number of Individually Advertised Units located on each island. Table 1: Total Number of Individually Advertised Units in 2014
Island Hawai‘i Kaua‘i Lāna‘i Maui Moloka‘i O‘ahu State of Hawai‘i
•
1
11,155 7,466 57 15,113 605 9,103
22,238
43,499
117,607
Total Estimated Number of Bedrooms
Unlike traditional hotel, condominium hotel, and timeshare units, the units identified in this study may be offered and withdrawn for visitor rental at the owners’ discretion and are typically not available for visitor rental on a year-round basis. –
•
4,986 3,614 22 8,840 365 4,411
Estimated Number of Visitors that Could Be Accommodated 28,106 19,481 133 43,877 1,676 24,334
Number of Individually Advertised Units
For example, individual owners may make their units available in response to seasonal demand or withdraw their units from the market during periods of personal use.
22,238 units is the best estimate of the number of Individually Advertised Units in Hawai‘i for 2014. Given the relative precision of the methods used, the estimate should be treated as 20,000 Individually Advertised Units in high season, plus or minus 2,500 units. –
The preliminary study done at end of 2013 yielded 18,998 units
–
The increase in identified units could have resulted from: •
Improved data collection techniques
•
Number of listings might increase during peak summer season
•
Increasing demand has increased listings
This period represents the second phase of fieldwork. Methodology and techniques were refined between phases.
VPI Ancillary Study © SMS
Page 3 December, 2014
Density The density for Individually Advertised Units was based on existing residential housing stock rather than an area measure such as square miles. Table 2 presents the number of Individually Advertised Units identified, housing unit counts, and the ratio of identified units to total housing stock on each island. Table 2: Density, Ratio of Individually Advertised Units to Housing Stock in Hawai‘i, 2014
Island Hawai‘i Kaua‘i Lāna‘i Maui Moloka‘i O‘ahu State of Hawai‘i
Total Number of Individually Advertised Units 4,986 3,614 22 8,840 365 4,411 22,238
Total Number of Housing Units* 82,323 28,790 1,545 65,232 3,312 306,622 487,824
Ratio of Individually Advertised Units to Total Housing Units 2 6.1 12.6 1.4 13.6 11.0 1.4 4.6
•
Across the State, there were 4.6 Individually Advertised Units per 100 residential housing units.
•
Maui’s 13.6 units per 100 residential housing units was highest for the State. Kaua‘i’s ratio was 12.6 per 100. O‘ahu and Lāna‘i had the lowest ratio at 1.4 Individually Advertised Units per 100 residential housing units.
The ratio of Individually Advertised Units to housing units by zip code area is presented in Appendix A to this report.
Estimating Total Lodging Units Results of the data collection were compared to the data reported in the 2014 Visitor Plant Inventory Report. The objective was to estimate the total number of all types of lodging units in Hawai’i in 2014. •
Assessment of the data collected in this study indicated that many of the Individually Advertised Units were also included in the 2014 Visitor Plant Inventory as VR-Condo, VR-House, Bed & Breakfast, or as part of a traditional condominium hotel pool.
•
However, not all vacation rentals included in the Visitor Plant Inventory advertise on the selected Internet sites.
2
Ratio is expressed as the number of Individually Advertised Units per 100 residential housing units. Does not include housing units from geographic areas where there were no Individually Advertised Units found.
VPI Ancillary Study © SMS
Page 4 December, 2014
•
The methods and definitions used in this study and the 2014 Visitor Plant Inventory are different; however the 22,238 Individually Advertised Units identified in this study were used as an estimate for Vacation Rentals and B&Bs reported in the Visitor Plant Inventory.
•
Table 3 combines these Individually Advertised Units with traditional visitor units reported in the 2014 Visitor Plant Inventory. If all of the identified units were available for visitor use at the same time, these units would account for up to 25 percent of Hawai‘i’s total lodging inventory. Table 3: Number of Lodging Units in the State of Hawai‘i by Type, 2014 Lodging Type Hotel Condo Hotel Timeshare Hostel Apartment Hotel Other Individually Advertised Units (Vacation Rentals) 3 Total
3
Units
% Mix
43,575 10,560 10,647 303 325 393
49.5% 12.0% 12.1% 0.3% 0.4% 0.4%
22,238
25.3%
88,041
100.0%
Assumes all identified units are available for visitor use at the same time.
VPI Ancillary Study © SMS
Page 5 December, 2014
APPENDIX A: INDIVIDUALLY ADVERTISED UNITS AND HOUSING UNITS Table A-1: Hawai‘i Island: Individually Advertised Units by Zip Code
Hawai‘i Island City/Area Waikoloa Kailua-Kona
Zip Code
Individually Advertised Units
Housing Units
Individually Advertised Units per 100 Housing Units
96738
946
4421
21.4
96740 / 96739
2166
16843
12.9
Nīnole
96773
10
105
9.5
Volcano
96785
156
1776
8.8
Hōnaunau
96726
21
271
7.8
Kamuela
96743
417
5668
7.4
Hakalau
96710
20
275
7.3
Pāhala
96777
39
575
6.8
Hāwī
96719
43
655
6.6
Pāhoa
96778
397
6685
6.0
Captain Cook
96704
170
2938
5.8
Hōlualoa
96725
55
1469
3.7
Laupāhoehoe
96764
11
357
3.1
Nā‘ālehu
96772
29
1089
2.7
Papaaloa
96780
5
202
2.5
Honoka‘a
96727
43
1857
2.3
Pāpa‘ikou
96781
14
654
2.1
Honomu
96728
5
253
2.0
Kapa‘au
96755
25
1384
1.8
Kea‘au
96749
119
6645
1.8
Pepeekeo
96783
14
789
1.8
Kealakekua
96750
25
1466
1.7
Paauilo
96776
7
607
1.2
202
17770
1.1
Hilo
96720 / 96721
Ookala
96774
1
124
0.8
Ocean-View
96737
19
2450
0.8
Mountain View
96771
23
3660
0.6
Kurtistown
96760
4
1335
0.3
VPI Ancillary Study © SMS
Page 6 December, 2014
Table A-2: Kaua‘i Individually Advertised Units by Zip Code
Kaua‘i City/Area
Zip Code
Individually Advertised Units
Housing Units
Individually Advertised Units per 100 Housing Units
Kōloa
96756
1,286
3,247
39.6
Princeville
96722
947
2,464
38.4
Hanalei
96714
355
959
37.0
Anahola
96703
80
899
8.9
Kapa‘a
96746
635
8,134
7.8
Kīlauea
96754
103
1,706
6.0
Kekaha
96752
55
1,382
4.0
Waimea
96796
22
887
2.5
96766 / 96715
97
5,296
1.8
Makaweli
96769
3
185
1.6
Kealia
96751
1
69
1.4
Lāwa‘i
96765
3
210
1.4
Kalāheo
96741
26
2,370
1.1
Hanapēpē
96716
1
982
0.1
Līhu‘e
VPI Ancillary Study © SMS
Page 7 December, 2014
Table A-3: Maui Individually Advertised Units by Zip Code
Maui City/Area
Individually Advertised Units
Zip Code
Housing Units
Individually Advertised Units per 100 Housing Units
Lahaina / Kapalua
96761
3,845
11,928
32.2
Kīhei
96753
4,183
18,059
23.2
Pā‘ia
96779
151
1,292
11.7
Hāna
96713
67
964
7.0
Haiku
96708
190
4,394
4.3
Wailuku
96793
332
10,564
3.1
96768 / 96788
47
6,729
0.7
Kula
96790
23
3,664
0.6
Kahului
96732
2
7,638
0.03
Makawao / Pukalani
Table A-4: Moloka‘i and Lāna‘i Individually Advertised Units by Zip Code
Moloka‘i and Lāna‘i City/Area
Zip Code
Individually Advertised Units
Housing Units
Individually Advertised Units per 100 Housing Units
Moloka‘i: Maunaloa
96770
147
757
19.4
Moloka‘i: Kaunakakai
96748
217
2,159
10.1
Moloka‘i: Hoolehua
96729
1
396
0.3
Lānai
96763
22
1,545
1.4
387
4,857
8.0
Moloka‘i and Lāna‘i Combined
VPI Ancillary Study © SMS
Page 8 December, 2014
Table A-5: O‘ahu Individually Advertised Units by Zip Code
O‘ahu Zip Code
Individually Advertised Units
Housing Units
Individually Advertised Units per 100 Housing Units
Kahuku
96731
263
1,297
20.3
Hale‘iwa
96712
480
3,028
15.8
Hau‘ula
96717
195
1,826
10.7
Lā‘ie
96762
123
1,188
10.4
Honolulu: Waikīkī
96815
1,725
22,750
7.6
Ka‘a‘awa
96730
35
617
5.7
Waialua
96791
99
2,776
3.6
Kailua
96734
525
16,548
3.2
Kapolei
96707
372
12,461
3.0
Waimānalo
96795
73
2,494
2.9
Wai‘anae
96792
194
13,376
1.4
Honolulu: Hawaiʻi Kai
96825
67
11,592
0.6
Kāne‘ohe
96744
70
17,803
0.4
Honolulu: Kāhala & Kaimukī
96816
60
18,914
0.3
‘Ewa Beach
96706
48
18,319
0.3
Honolulu: Aina Haina & Niu Valley
96821
17
7,295
0.2
Honolulu: Ala Moana
96814
16
11,187
0.1
Honolulu: Downtown
96801 96812 96813
9
10,542
0.1
Honolulu: Mō‘ili‘ili
96826
8
15,948
0.05
Mililani
96789
8
18,650
0.04
Honolulu: Mānoa
96822
8
19,372
0.04
Aiea
96701
4
14,008
0.03
Honolulu: Nu‘uanu
96817
5
20,157
0.02
Honolulu: Moanalua
96819
3
12,399
0.02
Waipahu
96797
3
19,986
0.02
Pearl City
96782
1
12,089
0.01
City/Area
VPI Ancillary Study © SMS
Page 9 December, 2014
APPENDIX B: INDIVIDUALLY ADVERTISED UNITS AND HOUSING UNITS MAPS Figure B-1: Hawai‘i Island Number of Individually Advertised Units by Zip Code
VPI Ancillary Study © SMS
Page 10 December, 2014
Figure B-2: Hawai‘i Island: Individually Advertised Units Density by Zip Code
VPI Ancillary Study © SMS
Page 11 December, 2014
Figure B-3: Kaua‘i Number of Individually Advertised Units by Zip Code
Figure B-4: Kaua‘i Individually Advertised Units Density by Zip Code
VPI Ancillary Study © SMS
Page 12 December, 2014
Figure B-5: Maui County Number of Individually Advertised Units by Zip Code
Figure B-6: Maui County Individually Advertised Units Density by Zip Code
VPI Ancillary Study © SMS
Page 13 December, 2014
Figure B-7: O‘ahu Number of Individually Advertised Units by Zip Code
Figure B-8: O‘ahu Individually Advertised Units Density by Zip Code
VPI Ancillary Study © SMS
Page 14 December, 2014
APPENDIX C: STUDY METHODS Selecting Websites Internet websites that list vacation rentals were identified and evaluated for their effectiveness for this study. Websites chosen for the project were: • • • •
vacation rentals by Owner (VRBO): http://www.vrbo.com/vacationrentals/usa/hawaii Clearstay.com: http://www.clearstay.com/Vacation+Rent als/US-Hawaii TripAdvisor: http://www.tripadvisor.com/hawaiivacation-rentals.html Airbnb: https://www.airbnb.com/s/Hawaii-United-States
The VRBO site was the most comprehensive site in that it has a greater number of listings than any other site. VRBO is owned by the Homeaway and the two lists had a very large number of the same listings. We eliminated the Homeaway site from our study due to this duplication. TripAdvisor had about half the number of listings as VRBO. Clearstay.com and Airbnb each had about 1,000 listings for units in Hawai‘i. Clearstay.com and Airbnb were selected primarily because they did not have a high degree of overlap with VRBO, and because they included important data that was not available from other websites. Clearstay.com, for example, lists the address of most properties. Airbnb listings included zip codes for all listings in 2013. That information was not available in 2014, but we decided to use Airbnb to maintain comparable methods for the two phases of fieldwork.
Definitions Preparation for this report revealed that most of the studies of vacation rentals in
VPI Ancillary Study © SMS
Hawai‘i have used different definitions of terms. This stems from the fact that the authors had different objectives, used different methods, and applied their methods different data sources in each study 4. While this is understandable, it makes comparing results of different studies problematic. We note below the definitions used for this study with comments on how those definitions compare to the primary study available at this time, the VPI. •
Visitor Destination Area (VDA): A geographic area with significant visitor accommodations infrastructure, such as Waikīkī on O‘ahu. The term VDA is used on some islands as an official land use classification and those classifications were applied where available. Where they were not available, the district classifications for VPI were used.
•
Individually Advertised Unit. Any housing unit listed for short-term rental on the Internet sites selected for this study. Individually Advertised Units include individual condominium units, houses, villas, cottages, townhouses, apartments, bungalows, studios, B&B rentals, rooms in residential housing units, and non-standard units such as tents, converted lanais and garages. Definitions used by the four websites for vacation rentals and B&Bs vary from definitions used in the VPI. There were no units advertised on the four websites
4
Note in particular the last several years of the HTA Visitor Plant Inventory for definitions and changes in definitions. See also, The Kauaian Institute (2005) Transient vacation rentals on O‘ahu: A Comparative Analysis of the Geographic and Economic Footprint, September 2005; The Kauaian Institute (2005) Transient vacation rentals on Maui: A Comparative Analysis of the Geographic and Economic Footprint, August 2005; The Kauaian Institute (2004) Transient vacation rentals on Kaua‘i: A Comparative Analysis of the Geographic and Economic Footprint, January 2004.
Page 15 December, 2014
that were identified as apartment hotel, hostel, or “other” units as defined in VPI.
Data Data from Internet Listings For each unit listed in the four websites we reviewed, we attempted to gather the information listed below. Items that were required are marked with an asterisk. Listing that did not include the required items were not included. There were very few omissions. • Website Name* • Listing Identification Number* • Listing Name and/or Description* • Island Location* • Location Indicators: o Area or City* o Street Address (if available) o Zip Code (if available) • Listing Characteristics (when published): o Number of Bedrooms o Number of Bathrooms o Number of Guests that can be accommodated o Rates Most of the items listed above are selfexplanatory and suited to the task at hand. Much depends on the quality of the listing name or description. For some listings the content may be as clear as “Fairways at Maui, oceanfront, condominium hotel unit.” For others, the information may be powerful advertising, but less useful for our research. Housing Data The number of housing units in each area and for each island was obtained from the U.S. Census Bureau’s 2010 Census data.
Data Collection Unit data were obtained directly from listings on each of the four websites. Data were collected electronically using custom-written
programs 5 that mined or scraped html pages on each website, gathering unit characteristics data into subfiles. The data were gathered sequentially by website, island, and region within island (city, village, VDA, or zip code). The raw data were written to Excel worksheets for processing.
Data Processing Data gathered from websites were first arranged in a common format. Separate software was developed for each website to eliminate duplicate entries within the site data. A geographic locator in each file based on whichever data were available at that site -- city, street address, zip code or other information. A unit capacity measure (the maximum number of people that could be accommodated by a given unit) was also generated for each file. Shared assumptions for this process were that there was at least one bedroom per unit listed and that each bedroom had two beds or could accommodate at least two people unless otherwise specified. Regardless of the information supplied by each listing, these assumptions allowed us to calculate a comparable unit capacity for all units. Other programs were written for individual files as they were needed for cleaning, extracting information from larger fields, or other utilities. A sample of the listings obtained where checked manually by SMS professionals in order to confirm that the data obtained electronically matched the information published for the listings. This also served to validate the scripts developed to pull the data from the listing web pages. The manual check consists of a person using a web browser to navigate to the URL of a given listing and then confirming that the 5
VPI Ancillary Study © SMS
Most routines were written in PERL, some in SEQL.
Page 16 December, 2014
details found match the data that was obtained programmatically. A sample of listings identified as duplicates were also checked manually to confirm that they were in fact duplicates in order to validate the process of finding and eliminating duplicates. Finally, data from each of the four websites were merged and a sequential unit identification number was generated for each case. Initial identification numbers from the original files were maintained.
Data Cleaning In most cases, the exact location of the unit or property listed was unavailable. Some records included a ‘neighborhood’, local area, or community name and some listed a city or village name. Only one site, Clearstay.com, included an exact address for its listings. During data cleaning the geographic indicator was standardized across all sources using a list of cities recognized by the USPS and their assigned zip codes. If the listed city or area did not match a city with a zip code, then the closest city with an assigned zip code was assigned as the standardized location. When the location data extracted from the website was insufficient to identify a geographic location, the listing page was visited to determine the location based on descriptions, maps, or images. Listings on the Airbnb website included GPS coordinates, which were used to obtain the zip code which could be translated into island and city codes.
website, and enter the relevant data to the master data file.
Duplicate Listings Duplicate listings were identified by comparing case identifiers within a given geography. Within each of the four component files, it was possible to identify duplicates by their ID numbers. In each of the four files we compared ID numbers and “listing names” of two records. The listing name is a property name or short descriptive phrase or sentence describing the rental unit. For example, if in the VRBO site two listings had the same name and listing ID, then they were considered duplicates. Duplicate records were removed and saved for further analysis. A sample of duplicate listings was manually reviewed in a browser to confirm that they were in fact listings for the same unit. In the larger file, we compared the “listing names” and other information for two records. Across records from all four sources/websites, the identification of duplicate listings was conducted by island using the listing name. When needed additional details of the listings were used such as number of bedrooms and bathrooms. Using these methods, more than 1,500 duplicate listings were identified and removed from the database.
Similar processes were applied to important variables such as the capacity indicator and unit type (VR-condo, VR-house, or B&B). Raw data extracted from the websites is used to develop an indicator that works for all cases. In cases where the raw data are insufficient to generate a comparable value for the variable under consideration, staff members get the listing URL, visit the
VPI Ancillary Study © SMS
Page 17 December, 2014