VARIATIONS IN MULTIPLIERS AND RELATED ECONOMIC RATIOS FOR
RECREATION AND TOURISM IMPACT ANALYSIS

BY

Wen-Huei Chang

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Park, Recreation and Tourism Resources

2001

ABSTRACT

VARIATIONS IN MULTIPLIERS AND RELATED ECONOMIC RATIOS FOR
RECREATION AND TOURISM IMPACT ANALYSIS

BY

Wen-Huei Chang

Economic impact analysis estimates the changes in economic activity
within a region resulting from some action. Managers and decision-makers can
justify tourism’s economic significance by tracing the effects of tourist spending
on sales, income and jobs in a region. Economic multipliers can be used to
capture the secondary effects of tourist spending on a region’s economy.

Despite the growing use of economic multipliers in recreation and tourism
impact studies, limited guidance is available for choosing multipliers suitable for a
given application. This study describes regional variations in economic
multipliers for tourism-related sectors, identifies key factors associated with these
variations, and evaluates procedures for predicting tourism multipliers for a given
region. One hundred and fourteen regions varying in size and economic
development were selected. Input-output models were estimated for each region
using the IMPLAN system and multipliers for tourism sectors were extracted for
analysis.

Multipliers were compared across sectors and regions. The Coefficients
of Variation for most multipliers were between 5 and 20 percent across the 114
regions. All tourism multipliers are positively correlated with the natural log of

population except for job multipliers, which are negatively correlated. The natural

log of population was identified as the best predictor of tourism multipliers,
explaining 50 to 80 percent of the variation. Regions were formed into four
groups with distinct multipliers. A multiplier lookup table was developed with
guidance for choosing sector-specific multipliers for a given type of region. The
regression models and lookup procedures were evaluated and compared. Errors
from using generic multipliers are generally within 2 to 9 percent, similar to the
errors from using regression models.

The results of the study will be used to refine procedures for estimating
the economic impacts of recreation and tourism, including the National Park

Service’s Money Generation Model.

ACKNOWLEDGMENTS

First and foremost, I want to thank my Ph.D. advisor, Dr. Daniel J. Stynes,
for his support and advice. He was always available and instrumental in all
stages of my Ph.D. program at Michigan State University. I am especially
thankful for his patience and constructive criticism in the completion of this
dissertation.

Secondly, I wish to express my profound appreciation to Dr. Dennis B.
Propst for his ongoing help in the success of my Ph.D. program. Along with Dr.
Stynes, Dr. Propst provided innumerable contributions to my educational
endeavors and continued financial support throughout my study at Michigan
State University.

In addition, I would like to extend my appreciation to Dr. Bruce W. Pigozzi
of Geography Department and Dr. Larry. A. Leefers of Forestry Department for
their input in shaping this manuscript and for serving on my graduate committee.
I would also like to thank Dr. Joseph Fridgen, Chair of the Park, Recreation and
Tourism Department, for his generous advice and help toward my research and
career.

Special thanks to my colleagues at the US. Army Corps of Engineers
Waterways Experiment Station for their moral support and for sharing my
workload during the final stage of writing my dissertation. Notable are R. Scott
Jackson and M. Kathleen Perales, who were always ready for discussion and

ideas, and were never short of support and encouragement.

iv

An acknowledgement cannot be complete without recognizing my parents
and family back in Taiwan for their love, understanding, and support through this
long process. Also to our son LeeYung for the spirit and happiness he has
brought in the first year of his life. Last, but never least on my list, is to thank my
dear wife, Lorna, for all the sacrifices she has made while I pursued my degrees.

Without her, and all the people above, this dissertation would not have been

possible.

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................ viii
LIST OF FIGURES .............................................................................................. xi
CHAPTER 1
INTRODUCTION .................................................................................................. 1
Research Objectives ....................................................................................... 3
An Overview of Economic Impact Analysis ..................................................... 3
Definitions of Economic Multipliers ................................................................. 5
Tourism Sectors .............................................................................................. 9
Variations in Economic Multipliers .................................................................. 9
Approaches to Economic Impact Analysis .................................................... 10
Misuse and Abuses of Economic Multipliers ................................................. 12
An Application for Tourism and Recreation Impact Study ............................. 14
CHAPTER 2
LITERATURE REVIEW ...................................................................................... 15
Techniques for Estimating Economic Multipliers ........................................... 15
Economic Base Models ............................................................................ 16
Input-Output Models ................................................................................. 18
Comparisons Between I-0 and Economic Base Models .......................... 21
Microcomputer-Based l-O Systems .............................................................. 23
IMPLAN (IMpact analysis for PLANning) .................................................. 24
Variations in Economic Multipliers ................................................................ 29
Variations Across Economic Models and Changes Over Time ................ 35
Approaches to Estimating the Economic Impacts of Recreation and Tourism
...................................................................................................................... 37
Money Generation Model - Aggregate Multipliers ................................... 37
RIMS lI — Sector-Specific Multipliers ........................................................ 39
IMPLAN - l-O Modeling System .............................................................. 44
Problems in Applying Multipliers to Tourism Impact Studies ......................... 46
Generalization Errors ............................................................................... 46
Aggregation Errors ................................................................................... 47
Misapplication .......................................................................................... 48
Failure to Margin Retail Purchases .......................................................... 48
CHAPTER 3
RESEARCH METHODS ..................................................................................... 51
Select Study Regions .................................................................................... 51
Develop l-O Models and Extract Economic Multipliers ................................. 53
Choice of Industry Sectors ............................................................................ 54
Defining a Tourism Multiplier ......................................................................... 57
Describe Variations in Multipliers .................................................................. 58

vi

Identify Key Factors that Explain Variations in Multipliers ............................. 59

Propose and Evaluate Procedures for Predicting Multipliers ........................ 61
CHAPTER 4
RESULTS ........................................................................................................... 64
Objective 1. Describe Variations in Multipliers ............................................. 64
Variations in Regional Characteristics Across Regions ............................ 64
Variations in Economic Ratios and Multipliers Across Sectors ................. 65
Variations in Economic Ratios and Multipliers Across Regions ................ 69
Sales Multipliers ....................................................................................... 71
Job Multipliers .......................................................................................... 73
Income Multipliers .................................................................................... 75
Value Added Multipliers ........................................................................... 77
Objective 2. Identify Key Factors that Explain Variations in Multipliers ........ 79
Correlation Analysis ................................................................................. 79
Regression Analysis ................................................................................. 82
Objective 3. Propose and Evaluate Procedures for Predicting Multipliers ...89
Develop a Multiplier Lookup Procedure ................................................... 90
Evaluate Procedures for Selecting] Predicting Multipliers ........................ 95
Handling Regions with Low Job Multipliers ............................................ 103
CHAPTER 5
DISCUSSION AND CONCLUSIONS ................................................................ 107
Discussion of the Results ............................................................................ 108
Objective 1. Describe Variations in Multipliers ...................................... 108

Objective 2. Identify Key Factors that Explain Variations in Multipliers .109
Objective 3. Propose and Evaluate Procedures for Predicting Multipliers

............................................................................................................... 110
Applications of the Proposed Multiplier Lookup Procedure ......................... 111
Conclusions ................................................................................................ 1 15
Recommendations for Future Study ........................................................... 116

APPENDIX A

STUDY REGIONS AND CORRESPONDING PARKS AND CITIES ................. 122

APPENDIX B

VARIATIONS IN MULTIPLIERS FOR EIGHT OTHER SECTORS ................... 126

APPENDIX C

REGIONAL CHARACTERISTICS AND MULTIPLIERS BY TYPE OF REGION
....................................................................................................................... 130

APPENDIX D

SECTOR SPECIFIC MULTIPLIERS FOR THE FOUR TYPES OF REGIONS.135
BIBLIOGRAPHY ............................................................................................... 140

vii

LIST OF TABLES

Table 2-1. Basic Input-Output Transactions Table ............................................. 19
Table 2-2. Sample IMPLAN Multiplier Report- Employment (1996, US) ............ 28
Table 2-3. Sample Tourism and Recreation Economic Impact Studies ............. 30
Table 24. Sample Worksheet for Money Generation Model for a Rural Area
National Park in the Rocky Mountain Region ...................................................... 38
Table 2-5. RIMS ll State Multipliers for Michigan—Output Earnings, and
Employment by Industry Aggregation ................................................................. 41
Table 2-6. Sample Visitor Impacts Estimated Using RIMS II Multipliers ............ 43
Table 2-7. Sales Effects of CE Marina Slip Renters' Trip Spending to the Natio‘r;5
Table 3-1. Economic Multipliers and Ratios Used in this Study ......................... 54
Table 3-2. Tourism Sectors and Corresponding Spending Categories .............. 55
Table 3-3. Michigan Statewide Multipliers — 1996 .............................................. 56
Table 3-4. Weights for Combining Multipliers into a Tourism Multiplier .............. 58
Table 4-1. Variations in Regional Characteristics Across 114 Regions (1996)..65
Table 4-2. Direct Effect Ratios for 12 Tourism and Recreation Related Sectors66
Table 4-3. Economic Multipliers for 12 Tourism and Recreation Related Sector;8
Table 4-4. Pearson Correlation Coefficients for Multipliers Between Tourism
Sectors ................................................................................................................ 70
Table 4-5. Regional Variation in Sales Multipliers, Four Primary tourism Sector:2
Table 46 Regional Variation in Job Multipliers, Four Primary tourism Sectors 74
Table 4-7. Regional Variation in Income Multipliers, Four Primary tourism
Sectors ................................................................................................................ 76

viii

Table 48 Regional Variation in Value Added Multipliers, Four Primary tourism

Sectors ................................................................................................................ 78
Table 4-9. Pearson Correlation Coefficients Between Tourism Multipliers and

Regional Characteristics ..................................................................................... 80
Table 4-10. Least Squares Regression Results for Sales Multipliers ................. 83
Table 4-11. Least Squares Regression Results for Income Multipliers .............. 85

Table 4-12. Least Squares Regression Results for Value Added Multipliers ..... 86

Table 4—13. Least Squares Regression Results for Job multipliers .................... 88
Table 4-14. Errors in Regression Predicted Multipliers ...................................... 89
Table 4-15. Characteristics of Four Types of Regions ....................................... 91

Table 4-16. Descriptive Statistics and ANOVA Table for Population by the Four
Region Types ...................................................................................................... 92

Table 4-17. Ranges of Multipliers for Regions Within Four Primary Groups ...... 93

Table 4-18. Multipliers for the Lodging Sector by Type of Region ..................... 94
Table 4-19. Ranges of Errors by Using the Multiplier Lookup Approach ............ 96
Table 4-20. Ranges of Errors by Using the Regression Models ........................ 97
Table 4-21. Evaluation of the Low Job Multiplier Subgroup ............................. 104
Table 4-22. Average Personal income for the Lodging Sector within each Group

of Regions ......................................................................................................... 105
Table 5-1. Sample Computation of Total Visitor Spending, Lake Mendocino Are:
Table 5-2. Multipliers for Lake Mendocino Area ............................................... 113
Table 5-3. Sample Computation of Visitor Spending Impacts .......................... 114
Table A. Study Regions and Corresponding Parks and Cities ......................... 122
Table B-1. Regional Variation in Sales Multipliers, Other Sectors ................... 126
Table B-2. Regional Variation in Job multipliers, Other Sectors ...................... 127
Table B-3. Regional Variation in Income Multipliers, Other Sectors ................. 128

Table 34.
Table C-1.
Table 0-2.
Table C-3.
Table 0-4.
Table 0-1.
Table D-2.
Table 0-3.
Table D-4.

Regional Variation in Value Added Multipliers, Other Sectors ........ 129

Rural Regions (N=19) .................................................................... 130
Small Metro Regions (N= 34) ......................................................... 131
Large Metro Regions (N= 44) ......................................................... 132
States and Other Large MSA's (N= 17) .......................................... 134
Multipliers for Tourism-Related Sectors — Rural Areas ................... 135

Multipliers for Selected Tourism-Related Sectors — Small Metro 1 36
Multipliers for Selected Tourism-Related Sectors — Larger Metro .. 137

Multipliers for Selected Tourism-Related Sectors — State .............. 138

LIST OF FIGURES

Figure 4-1. Distribution of the Type II Tourism Sales Multipliers Across Regions
............................................................................................................................ 71

Figure 4-2. Distribution of the Type II Tourism Job multipliers Across Regions .73

Figure 4-3. Distribution of the Type II Tourism Income Multipliers Across
Regions ............................................................................................................... 75

Figure 44. Distribution of the Type II Tourism Value Added Multipliers Across
Regions ............................................................................................................... 77

Figure 4-5. Distribution of Sales Multipliers Predicted by the Lookup and
Regression Approaches ...................................................................................... 99

Figure 4-6. Distribution of Income Multipliers Predicted by the Lookup and
Regression Approaches .................................................................................... 100

Figure 4-7. Distribution of Value Added Multipliers Predicted by the Lookup and
Regression Approaches .................................................................................... 101

Figure 48 Distribution of Job Multipliers Predicted by the Lookup and
Regression Approaches .................................................................................... 102

xi

Chapter 1

INTRODUCTION

Economic impact analysis is carried out by both public and private
agencies to evaluate the economic effects of proposed projects or existing
facilities and programs. For many government agencies, there is a strong
demand for economic analysis at the national, state, and local levels (Jackson et
al., 1992). Planners and decision-makers can utilize the information to evaluate
facilities and programs and to justify the importance of a project to the local
community (Loomis and Walsh, 1997-p 242).

Economic impacts of recreation and tourism activity are generally made by
first estimating visitor spending and then applying a regional economic model to
trace the effects of this spending on the local economy. Many early studies of
tourism's economic effects stopped with estimates of consumer or visitor
spending, as input-output and related economic models were too expensive and
complex for most recreation and tourism analysts. Some early studies employed
economic base models to estimate an overall "tourism multiplier" (Archer and
Owen, 1971; Archer, 1973). With the advent of microcomputer based economic
modeling systems like IMPLAN,1 recreation and tourism studies began to make
considerable use of input-output models and to use multipliers derived from

these models (Rickman and Schwer, 1995b).

 

I lMpact analysis for PLANning, an l-O modeling system currently maintained by Minnesota
IMPLAN Group, Inc.

Multipliers capture the impacts of a given change in final demand (e.g.,
visitor spending) on sales, income, and jobs in a region. Most importantly, they
capture the secondary effects resulting from the circulation of visitor spending
within a local region. Armed with a set of multipliers for the local region, tourism
analysts can readily compute the economic impacts of a given change in
spending.

Growing use of economic multipliers and regional economic models in
recreation and tourism has been accompanied by many abuses and misuse.
Few tourism analysts and recreation managers have formal training in regional
economic methods and most are not very familiar with input-output models or
multipliers. Many tourism studies use "off-the-shelf" multipliers or borrow
multipliers from previous studies, often without understanding that there are
many different kinds of multipliers or that they vary by region, sector, and over
time (Archer, 1984; Beattie and Leones, 1993; Holland, 1994). Limited guidance
is available for choosing multipliers suitable for a given application.

The purposes of this study, therefore, are to describe variations in
economic multipliers for recreation and tourism related sectors and to explain
how multipliers vary across regions. If variations in multipliers across regions
can be explained and predicted, a simple procedure for selecting multipliers can
be developed. This study will propose and evaluate procedures for predicting
tourism multipliers that can be used to refine procedures for estimating the

economic impacts of recreation and tourism.

Research Obiectivg
There are three research objectives for this study:
1. To describe variations in tourism multipliers across sectors and regions of
varying economic development.
2. To identify key factors that explain variations in these multipliers.
3. To propose and evaluate procedures for predicting tourism multipliers for

a given region.

Variations in economic multipliers for tourism related sectors will be
thoroughly examined to better understand the multiplier effects for regions of
different population, geographical sizes, and economic development. This study
will also develop a procedure that can provide recreation managers and tourism
analysts, especially for those who are non-economists, a means for estimating
economic impacts. The rest of this chapter will introduce economic impact

concepts and research relevant to the problem being studied.

An Overview oLEconomic Impact Analvsis
Economic impact analysis (EIA) estimates the changes in economic
activity within a region resulting from some action. EIA can produce estimates of
the total economic impacts of holding a sport event, closing a power plant,
passing an environmental bill, relocating a military base, opening an amusement
park, and other actions that will influence a region’s economy. There are two

components to an economic impact analysis; to directly convert the action into

monetary values such as sales, income, and jobs, and to estimate the secondary
effects that are associated with the action (Pleeter, 1980 p-7).

Economic impact analysis traces changes in economic activity through the
economy to measure the cumulative economic effects of an action. For example,
visitors who purchase goods and services in a region will directly contribute to
businesses such as hotels, restaurants, and retail stores. These businesses will
pass the money to their employees as wages and salaries and their employees
will spend the money they receive to purchase goods and services from other
businesses in the region. These businesses in turn make additional purchases in
the region, thereby creating a chain effect. The cumulative result is the total
economic impacts of visitors’ spending in the region (Frechtling, 1994b).

Economic impact analysis helps policy analysts and decision makers to
evaluate current and proposed projects by providing estimates that are
measurable and comparable. Tourism industries need support from the local
community, as tourism activities affect the entire community. Recreation and
tourism development are regarded as attractive investments because they can
lure new businesses and visitors to the region. Quantifying tourism's economic
significance helps build support among the business community, government
officials, and the general public. Economic impact analysis provides tangible
estimates of tourism’s economic contributions to the region's economy. These
economic contributions often result in public policies or decisions that are
favorable to tourism development (Clawson 8. Knetsch, 1966-p 230; Stynes,

1 9993).

Activities associated with tourism and recreation involve monetary
transactions. These activities include visitor spending on trips and durable
goods, expenditures on development and construction of tourism and recreation
facilities, and operational costs of these facilities and programs. Both public and
private agencies are interested in the economic impacts of tourism and
recreation for the following two reasons. First, economic impact analysis
answers the fundamental questions raised by government legislators, regional
developers, and the general public— “What monetary benefits will tourism bring
to the community?” “How much income and how many jobs are supported by
visitor spending?” Second, EIA is important in terms of decision making for
evaluating a new or existing project and allocating budget funds. The information
gathered by an EIA can be used to determine the relative benefits and costs of
alternative tourism and recreation development strategies (Hastings and Brucker,
1993; Archer, 1982). ElA’s can also help to assess the degree of dependence of
the local economy on tourism and the potential economic growth from tourism

(Stynes, 1999a; Frechtling, 1994a).

Defflons of Economic Mulmrs
Economic impacts may be categorized into direct, indirect, and induced
effects. The summation of indirect and induced effects is also called “secondary
effect.” Multipliers capture the size of the secondary effects, usually expressed
as a ratio of total effects to direct effects (Miller and Blair, 1985-p 101). The

larger the multiplier, the greater the impact a dollar of visitor spending will have

on the region’s economy. For example, the sales multiplier for the lodging
sector was 1.74 for the State of Florida in 1996. This means that a visitor
spending $100 on lodging will have a total effects of $174 in sales within the
state; that is, $100 received by the hotel as direct sales effects and another $74
received by other related industries in the region as secondary effects.

Direct effects are changes in the industries associated directly with visitor
spending. In the previous example, $100 spent on lodging in the region will
directly increase sales in the hotel sector. This is the direct sales effect of the
visitor spending. The hotel will also hire employees and pay wages and salaries,
which are the direct job and income effects.

Indirect and induced effects are the secondary effects resulting from the
initial visitor spending. Indirect effects are sales, income, or jobs resulting from
various rounds of the purchases the hotel made to other “backward-linked”
industries in the region. For example, a hotel buys linen supply and utilities from
other industries to deliver the services to its customers. The linen supply
industry, on the other hand, also buys raw materials and equipment such as
cotton and machinery from other industries. The sales of these backward-linked
industries and the associated income and jobs generated from these sales are
indirect effects.

Induced effects are the sales, income, or jobs resulting form household
spending of income earned as a result of visitor spending- either directly or
indirectly. The employees of hotels, linen suppliers, utility companies, etc., for

instance, will spend their wages and salaries in the region and generate new

rounds of sales, income, and jobs. Several iterations (rounds) may occur before
dollars from indirect and induced effects leak entirely from the region. As a
result, money spent by visitors will impact not only tourism industries, but also
related industries in the region.

The Type I multipliers capture only the indirect effects, while the Type II
multipliers include both indirect and induced effects (Richardson, 1972-p 23;
Minnesota IMPLAN Group, 2000). Multipliers may be expressed in terms of
sales, income, value added or jobs, the most frequently used measures of
economic impacts (Bull, 1995; Miller and Blair, 1985).

Economic multipliers are expressed as a ratio of the total effects relative to
the direct effects. For instance, an income multiplier is the ratio of total income
effects (direct, indirect, and induced) to the direct income effects. This type of
multiplier has been called a "ratio multiplier" (Baaijens, et al., 1998; Archer,
1984), "direct-effect multiplier" (U. S. Department of Commerce Bureau of
Economic Analysis [USDC BEA], 1997; 1992), or simply ”multiplier" (Minnesota
IMPLAN Group, 2000). Ratio multipliers, like these, should be used as indicators
of a region's economic self-sufficiency and should not be applied to visitor
spending (or direct sales effects) without proper justification (Propst, 1991 ).
Ratio type multipliers do not directly convert sales from visitor spending into
income or jobs in the region and can be confusing to people who are not familiar
with multipliers (Archer, 1984; Frechtling and Horvath, 1999).

Since both the income earned by households and jobs created in the

region are caused by sales (e.g., visitor spending), income and job multipliers are

better expressed as ratios of total income or job to direct sales or spending
(Frechtling and Horvath, 1999; Frechtling, 1994b-p 383; Crompton, 1999; Stynes
et al., 2000). This type of multiplier has been called a "normal multiplier"
(Frechtling and Horvath, 1999; Baaijens, et al., 1998), ”final-demand multiplier"
(USDC BEA, 1997; 1992; Frechtling and Horvath, 1999), or "Keynesian-type
multiplier” (Archer, 1984; Propst, 1995). In this study, this type of multiplier will
be termed an economic multiplier or simply multiplier.

It is also useful to distinguish economic ratios from economic multipliers.
Economic ratios convert between various economic measures, while economic
multipliers capture the secondary effects (Stynes et al., 2000). For example, the
statewide job to sales ratio for Florida for the hotel sector in 1996 was 18 jobs per
million dollars in sales. This ratio can be used to convert hotel sales directly to
hotel employment. The state Type II job multiplier for the hotel sector in 1996
was 28 jobs per million dollars in sales (IMPLAN estimates). This means that for
every $1 million in sales in the lodging industry, 18 direct jobs are created in the
hotel sector, and 10 (28 minus 18) jobs are created through secondary effects.
These secondary jobs are created in industries that receive the secondary
effects.

Multipliers can represent an aggregation of industries or a specific sector.
An aggregate tourism multiplier, for example, uses one number to represent the
tourism sectors within a region’s economy. Visitor spending is applied as a
whole to the aggregate multiplier for total effects. For example, economic base

models (e.g., Archer's “Tourist Regional Multiplier") estimate an overall

aggregate multiplier for the study region’s tourism industries (Archer and Owen,
1971 ). Sector-specific multipliers provide multipliers for different industry sectors.
Visitor spending on different items can be applied to different ratios to capture
multiplier effects across sectors. The use of l-O model is the most common way

to generate sector-specific multipliers.

Tourism Sectors

Recreation and tourism involves a number of different industries. There is
no single “tourism” sector in the Standard Industry Classification (SIC) system
(Johnson et al., 1989). The US. Travel and Tourism Satellite Accounts have
identified 15 tourism-related industries for the US. economy. Sales to tourists
account for more than 20 percent of the total industry sales for eight of these 15
industries (Okubo and Planting, 1998).

IMPLAN generates multipliers for up to 528 industries in a given region.
This study will focus on the four primary tourism sectors that receive the bulk of
visitor spending: Hotels and Lodging Places, Eating and Drinking, Amusements
and Recreation, and Retail Trade. A complete list of multipliers selected for this

study is provided in chapter 3.

Variations in Economic Multipliers
Multipliers vary across regions, sectors, and over time. The Type II job
multipliers for the State of Michigan, for instance, ranged from 30 jobs per million

dollars in sales for the Local Transportation sector to 50 for the Eating and

Drinking sector (USDC BEA, 1992). A million dollars spent on meals and
beverages would therefore create 20 more jobs than the same spending on local
transportation.

Multipliers also vary across regions with distinct characteristics. One of
the most common errors in tourism impact studies is the applications of state
level multipliers to sub-state regions. Multipliers indicate the interdependence of
industry sectors within a region's economy and are influenced by the size of the
region and population (Tooman, 1997; Baaijens et. al., 1998; Detomasi, 1987;
Propst and Gavrilis, 1987; Chang et al., 1999). The size of a multiplier for a
given region depends on how the study region is defined and its economic
characteristics. Propst et al. (1998), for instance, computed sales multipliers for
regions surrounding 108 Corps of Engineers lakes. The Type II sales multipliers
for these regions varied from 1.37 (a rural region around Dworshak Lake in
Idaho) to 1.88 (Metro-Nashville, TN region around Lake J. Percy Priest).

This study will describe and explain variations in multipliers across sectors
and regions. Multipliers can also vary across different economic models and
they can change over time. Sources of variations in multipliers are discussed

further in Chapter 2.

Approaches to Economic Impact Analvsis
The most common approaches for estimating economic impacts of

tourism are multiplier methods and input-output (I-O) models. The multiplier

10

approach uses either aggregate (e.g. MGM modelz) or sector-specific multipliers
(e.g. RIMS ll, MGM23) to estimate changes in economic activity due to visitor
spending. The l-O approach makes use of an input-output model for the region.
For example, IMPLAN estimates multipliers for local areas and also computes
impacts of visitor spending. As many recreation and tourism analysts do not
have immediate access to input-output models or systems like IMPLAN, the
multiplier approach is more common in applied world. The multiplier approach is
the focus of this dissertation. When using the multiplier approach, the key issues
are selections of suitable multipliers for the given region and proper applications
of these multipliers.

The original MGM model recommended a sales multiplier of 2.0 (US.
Department of Interior National Park Service [USDI NPS], 1990). Some MGM
applications employed state level multipliers from RIMS II that were "adapted" to
local areas. RIMS II is an l-O model that is maintained the Bureau of Economic
Analysis (BEA). BEA has published state level multipliers in its 1992 publication
for 39 sectors and illustrated how to apply them to a typical tourism application
(USDC BEA, 1992). BEA’s example illustrates the proper handling of retail
purchases by applying multipliers for the retail trade sector to the margins on

goods purchased at retail stores. Tourism analysts using aggregate multipliers

 

2 The National Park Service's Money Generation Model is a simple one-page pencil and paper
worksheet for estimating economic impacts of visitor spending using aggregate multipliers. A
detailed discussion of this model is in chapter 2.

3 MGM2 is an update of the National Park Service’s Money Generation Model. MGM2 was built
as an electronic spreadsheet that automates many routine calculations and provides a wide array
of options. The multiplier lookup procedure proposed in this study has been used in the MGM2.
The MGM2 software and manual can be found at http://www.prr.msu.edu/mgm2/MGM2web.htm.

11

have tended to apply them to all spending, even if the goods bought by tourists
are not locally made.

The most readily available (published) multipliers tend to be for state level
regions. While multipliers may be purchased for local regions from BEA (RIMS
ll), IMPLAN, REMI and other sources, most recreation and tourism organizations
are either unaware of these sources or lack the funds to acquire their own
multipliers. Hence, state level multipliers have been widely applied to local
regions, usually yielding inflated estimates of impacts. Perhaps the greatest
source of multipliers is simply borrowing them from past studies. This practice
has tended to reinforce the misconception that there is a single “tourism
multiplier" and has established 2.0 as the most popular figure (Beattie and

Leones,1993)

Misuse and Abuses of Economic Multipliers
The third research objective entails the development of procedures for
selecting tourism multipliers for a given region. Such systems can help to reduce
errors that are commonly made in selecting and applying multipliers in tourism
applications. A brief review of the most common problems provides some

guidance for a multiplier selection system.
1. Using state level multipliers for local applications. This is indicative of a

more general problem of choosing multipliers that do not represent the region of

interest. This problem stems from less familiarity with economic methods and

12

concepts among tourism analysts and a lack of understanding on how multipliers
vary across regions and sectors. The most common abuse noted by researchers
is the applications of national or state level tourism multipliers to sub-state
regions (Archer and Owen, 1971; Archer, 1984).

2. Applying aggregate multipliers. Because the total sales, jobs and
income to direct sales ratios vary from sector to sector, the same amount of
money spent in different sectors can result in very different impacts. For
example, $100 spent on hotels and $100 spent on souvenirs will have different
impacts to the local economy. Since the value of an aggregate multiplier
depends on the distribution visitor spending, applying aggregate multipliers that
are estimated from one application to another will not capture the variations in
multiplier effects across sectors.

3. Failure to margin retail purchases. It is common to see analysts apply
multipliers to spending that will not be captured by local businesses, such as
spending on goods that are not locally made. Since the multipliers are ratios of
total effects to the direct sales effects, they should only be applied to sales that
will accrue to the local region (Archer, 1984; Stynes, 1999a).

4. Incorrect application of multipliers, such as confusing sales and income
multipliers, and ratio vs. Keynesian versions. Researchers have pointed out that
on many occasions, inappropriate multipliers were used (i.e., use sales
multipliers to compute income effects) due to ignorance or a lack of
understanding of multipliers (Archer, 1984, Propst and Gavrilis, 1987; Propst,
1995)

13

Mplication for Toquism ang Recreation Impact SIM

Recreation and tourism analysts need to understand how multipliers vary
across regions, and have an approach that can simplify the procedures for
estimating economic impacts while providing accurate and detailed information.
A lack of accuracy and guidance on choosing multipliers for a given region was
one major weakness cited by Duffield et al. in reviewing the National Park
Service’s Money Generation Model (Duffield et al., 1997; Stynes and Propst,
1999)

This study will develop and evaluate procedures for choosing multipliers
for tourism applications. The approach is to simplify procedures for choosing
appropriate multipliers for a given region, .without sacrificing accuracy of the
economic impact estimates.

The multiplier analysis is part of the revisions to the National Park
Service’s Money Generation Model (Stynes et. al., 2000). The results of this
study will also be used in a Michigan Tourism Impact Model (Stynes and Chang,
2000), and the Recreation Economic Assessment System for the US. Army

Corps of Engineers (Chang et al., 2001).

14

Chapter 2

LITERATURE REVIEW

This chapter reviews research on economic multipliers and their
applications in tourism and recreation studies. The chapter is divided into five
sections. The first two sections discuss regional economic methods and
applications in general, while the last three sections focus particularly on
applications to recreation and tourism. The first section reviews the two most
common methods for estimating economic multipliers- economic base and input-
output models. The second section discusses systems for estimating multipliers
and focuses particularty on the IMPLAN system, as this is the source of all
multipliers used in this dissertation. Section three reviews previous studies that
have examined variations in multipliers across regions, sectors and time.
Section four summarizes the most common approaches for conducting economic
impact analysis of recreation and tourism and section five reviews common

errors and abuses in applying multipliers within recreation and tourism studies.

Technigpes for Estimating Economic Mm
Two commonly used techniques for estimating economic multipliers are
economic base models and input-output methods (Eadington and Redman,
1991; Richardson, 1985; Pleeter, 1980). This section will introduce both

approaches and show how multipliers are derived from them.

15

Economic Base Models

Economic base models divide economic sectors into basic and non-basic
sectors. The basic sector consists of all firms that serve markets outside the
region (exports). The non-basic sector, on the other hand, consists of the firms
that serve markets inside the region (non-exports). Income of the region can be
partitioned into basic and non-basic components:

Y = B + NB
where

Y = total income

B = basic income

NB = non-basic income

The economic base multiplier (M) is expressed as the ratio of total income

(Y) to basic income (B).

M=YIB

It can also be defined as:

where

e = propensity to purchase locally

The equivalence of the two definitions can be shown as follows:

M=1/(1-e)

l6

M = Y / B Substitute B with (Y-NB)
M = Y I (Y - NB) Substitute NB with eXY

M = Y / (Y - eXY) Factor out Y

One major task for using the economic base model is to identify exports
(B) so the non-basic (NB) sector can be estimated. Once N8 is known, the
propensity to purchase locally (9) can be estimated from NB = eXY(non-basic
sector represents local consumption and is assumed as a function of total
income) (Tiebout, 1962; Richardson, 1985; Hinojosa and Rios, 1991; Kendall and
Pigozzi, 1994).

The economic base multiplier shows the change in total income in a
region as a function of exports. The economic base multiplier will be larger if
residents buy more from local producers (larger 9 results in larger M in the
previous formula). One method for measuring the economic base (B) is the
Location Quotients (LQ) method (Richardson, 1985). This method assumes the
nation is self-contained and the productivity and consumption per employee are
the same in the region and in the rest of the country. L0 is also an option in
IMPLAN l-O model for constructing social accounts to estimate multipliers.
There are several other methods for measuring the economic base, such as
survey, judgment, and minimum requirements. Detailed descriptions of these
approaches are beyond the scope of this study. Readers can refer to
Richardson's "Input-Output and Economic Base Multipliers: Looking Backward

and Forward" (1985) for details on this topic.

17

Input-Output Models

Input-output (l-O) models are based on the pioneering development by
Wassily Liontief in the late 30's, for which he later received the 1973 Nobel Prize
in Economic Science (Leontief, 1986). The l-O model has been further
developed and refined by economists and regional scientists such as lsard,
Tiebout, Miemyk, Richardson, Bulmer-Thomas, and Miller and Blair in the 50's to
803 (Richardson, 1985, 1969; lsard, 1975; lsard and Langford, 1971; Hewings,
1985; Miemyk, 1965).

I-O models identify the monetary transactions between an industry and a)
other industries (intermediate sales), b) labor, capital payment, rents, etc. (value
added), and c) ultimate consumers such as tourists (final demand). The region’s
economic activities are presented in mathematical matrices called transactions
tables. A simplified l-O transactions table for a three-sector region is illustrated in
Table 2-1.

In this l-O matrix, each column and each row represents a single industry.
Each column shows the amount of input that is used by a single sector. A row
shows the amount of output a single sector provides to all other sectors. The
three by three matrix in the top left corner of the table represents the intermediate
transactions. Industry 1 sells goods and services to other industries in the region
(x11 to industry 1, x12 to industry 2 and x13 to industry 3). Industry 1 also buys
goods and services from other industries in the region (x11 from industry 1, x21

from industry 2 and x31 from industry 3). Households (C1) and other institutions

18

(I1) purchase goods and services from industry 1 as final demands. Industry 1

also makes payments such as labor (L1), rents, imports, etc. (V1) to households

and other institutions as value added.

Table 2-1. Basic Input-Output Transactions Table

 

 

To Purchasing Final Demand Total
sectors Output
1 2 3 households other
From institutions
Producing 1 X11 X12 X13 C1 I1 X1
SBCtOl'S 2 X21 X22 X23 C2 I2 X2
3 X31 X32 X31 C3 I3 X3
Value labor L1 L2 L3 LC L. L
added other
value V1 V2 V3 V0 V1 V
added
Total outlays X 1 X 2 X 3 C l X

 

 

Source: “Input-Output and Regional Economics” (Richardson, 1972).

The first row of the table can be simplified as a mathematical equation:

X11+X12+X13+C1+I1=X1

where C1 + l1 equals total final demand.

Since the idea of multiplier analysis is to estimate the changes in total

industry output (X) resulting from changes in final demand (Y), it is essential to

know the mathematical relationship between X and Y. The above equation can

be rewritten as:

X1-a11*X1-a12*X2-a13*X3=Y

I9

where a), = X I ,-/ X j, the coefficient that specifies the amount of input industry i
needs to produce a unit of j. Now the simplified transactions of the three-sector

economy can be presented in matrix form as:
X1 311 812 313 X1 Y1
X1 - 821 822 323 X2 Y2
X1 331 832 833 X3 Y3

X-AX=Y

or

where X and Y are vectors of output and final demand and A is the matrix of
coefficient at). By restoring an identity matrix I to the equation, it can be written
as:

x *(I - A) = Y
or

x = (I - A)" Y
where (I - A)‘1 is the Leontief inverse matrix. The elements of the matrix
represent the purchases from one industry to others in order to produce another
unit of output for the final demand. Since multiplying this matrix by a vector of
final demand Y will produce the output X, this matrix also represents the
multiplier effects. The multipliers for each sector can be calculated from this
matrix. The summation of each column in the matrix is the multiplier for the

matching industry of that column.

20

Comparisons Bitumen l-(gnrficonomicJBase Mow

l-O methods have become the primary tools for tourism economic impact
analysis. The use of l-O models to estimate economic impacts of recreation and
tourism has increased considerably in the last couple of decades because of the
ability to provide accurate and detailed information as well as the ease of
interpreting the results (Stevens and Rose, 1985; Summary, 1987; Fletcher,
1989). One major advantage of the l-O model is that it provides detailed
information on direct, indirect and induced effects of visitor spending on all
economic measures for different industries in the local economy (Loomis and
Walsh, 1997 p-254).

Economic base models, on the other hand, have not received as much

attention by regional economists as the l-O models. Richardson claims that,

“Economic base models have had a long and checkered history, going
back to the 1940’s and even earlier. They have not been quite
academically respectable, and the revival of research on measuring the
economic base in the 1970's was unexpected.”

However, he also comments that,

“(economic base models) have staged something of a revival since the
late 1960’s because of their incorporation in regional econometric
models ....... they offer a clear link from the national economy to the region
within a standard macroeconometric (income determination) framework”
(Richardson, 1985).

Multipliers derived from l-O models capture how sectors of the economy

are linked together within a region, but to develop l-O models require

considerably more data and efforts than to develop economic base models.

21

Multipliers derived from economic base models, on the other hand, are relatively
simple to develop in terms of data gathering and are much less expensive to
apply, but these multipliers provide less detailed estimates because of the high
level of aggregation. There is usually only one aggregate multiplier for a region
(Archer, 1996; Eadington 8. Redman, 1991; Kottke, 1988; Richardson, 1985;
Pleeter, 1980). Economic base models are not as widely used in tourism
research. One reason is because tourism is part of many different sectors and it
is difficult to allocate tourism-related industries into basic and non-basic sectors.
The assumption that all types of export sales have the same multiplier effects
and the assumption that all the regional economic growth is attributable to export
sales only are also impractical for tourism and recreation applications (Krikelas,
1992)

Archer’s “tourist regional multiplier" approach is one of the few
applications of economic base models to tourism and recreation. While adapting
the economic base approaches, Archer’s tourism regional multiplier approach
includes two important components for tourism applications- the propensity of
consumption by different visitor segments, and shares of tourist spending in
different industries (Archer and Owen, 1971; Archer, 1973). Archer’s concept of
a tourist regional multiplier is that the multiplier not only represents the region’s
economy for tourism sectors, but also reflects tourists’ spending profiles. The
tourist regional multiplier approach is equivalent to an aggregation of l-O
multipliers when multipliers for different sectors are weighted in proportion to

tourist spending they receive (aggregate multipliers).

22

Micmcomputer-MIOQSystems

l-O models have been used in the past mainly by academic economists.
However, because of the ability of advanced microcomputers, I-O models have
been packaged as ready-to—use computer software and database (Hastings and
Bmcker, 1993). Among these ready-to-use l-O models, IMPLAN and RIMS II are
the most popular systems for recreation and tourism applications (Rickman and
Schwer, 1995a, 1995b; Brucker, Hastings and Latham, 1990). The Bureau of
Economic Analysis maintains an l-O model of the entire US. economy called
RIMS II and can provide sector-specific multipliers for the US. or sub-regions.
IMPLAN, on the other hand, provides software and database for estimating local
models and impacts in a microcomputer WindowsTM environment.

Besides RIMS II and IMPLAN, other ready-to-use systems were
developed beginning in the mid 80’s, e.g., ADOTMATR by Lamphear et al.
(1983), RSRI by Stevens et al. (1983), SCHAFFER by Schaffer and Davidson
(1985), and GRIMP by West (1983) (Brucker, Hastings and Latham, 1987, 1990;
Deller et al., 1993). However, these models have not drawn as much attentions
as RIMS II and IMPLAN and are not as widely used.

Another commercially available economic impact model is REMI (Regional
Economic Models, Inc.). REMI is an economic simulation model that uses
econometric and general equilibrium models to trace the total effects over time of
changing economic conditions in a study area. REMI includes some l-O model

embedded functions such as the computation of inter-industry relationships, plus

23

additional functions to forecast effects of future changes. However, economic
simulation models involve greater analytic sophistication and cost than l-O

models (Weisbrod and Weisbrod, 1997).

IMELAN (IMpact ELM/sis forjEANninm

IMPLAN is the source of multipliers used in this study. IMPLAN is an
input-output modeling system that was originally developed by the US
Department of Agriculture Forest Service as a tool to assist the Forest Service in
land and resources management planning. IMPLAN began as a mainframe-
computer application in the 80's. The Minnesota IMPLAN Group (MIG) began to
work on IMPLAN in 1987, and IMPLAN has since migrated from the mainframe
version to a DOS application (Olson and Lindall, 1993), and then to the current
IMPLAN Pro WindowsTM version (Minnesota IMPLAN Group, 2000; Propst,
2000).

Like most non-survey l-O modeling systems, IMPLAN uses the national l-
0 matrices with regional data to create regional models (Chamey and Leones,
1997; Bushnell and Hyle, 1995). IMPLAN generates detailed sector-by-sector
reports that can include as many as 528 industry sectors for a given region.
lMPLAN's sectoring scheme is based on the Standard Industry Classification
(SIC) code system and the Bureau of Economic Analysis I-O sectoring, and for
the most part, is similar to a 3 and 4 digit SIC code system (Minnesota IMPLAN
Group, 2000).

24

The original DOS version of IMPLAN (before 1995) calculates Type I and
Type III multipliers while the current Windows version calculates Type I, II and
Type SAM multipliers. The Type I multipliers capture indirect effects while the
Type II, Type III and Type SAM multipliers capture both indirect and induced
effects resulting from the changes of final demands.

Type I multiplier =
direct effects + indirect effects

 

direct effects

Type II (Type III, Type SAM) multiplier =

direct effects + indirect effects + induced effects

 

direct effects

IMPLAN Pro 2.0 generates Type SAM multipliers to capture the total
effects. The SAM framework tracks both market and non-market flows. The
non-market flows are transactions between non-industrial institutions such as
households to government, government to households and so on. These flows
are called inter-institutional transfers (Alward and Lindall, 1996). Since total
personal income is income from all sources, including employment income and
transfer payments that are based on both place of work and place of residence,
some of this income may not be related to personal consumption expenditures in
the region. The SAM multiplier approach enables the model to account for
commuting, social security tax payments, household income tax payments and

savings and hence adjusts the Type I I multipliers for income that is not normally

25

respent immediately within the region, such as commuting workers who live
outside the region and retirement benefits (Minnesota IMPLAN Group, 2000).
The Type SAM multipliers are more conservative than the traditional Type II
multipliers for tourism and recreation applications as the induced effects are
smaller and are likely more realistic for tourism and recreation applications
(Stynes et. al., 2000).

Social accounts have to be constructed in IMPLAN before economic
multipliers can be computed. Social accounts are the trade flows that specify the
transfers of goods and services between the region and the rest of the world.
There are several possible assumptions regarding imports, exports, and the
propensity that the final demand can be met locally with the generation of social
accounts. IMPLAN Pro provides three options for constructing social accounts
(Minnesota IMPLAN Group, 2000-p 141 ):

1. Regional Purchase Coefficients (RPC’s)“. This approach uses
an econometric equation to predict the percentage of goods that
are purchased locally based on a region’s characteristics.

2. Supply/Demand Pooling. This approach assumes everything
that can be purchased locally will be purchased locally. This
approach will maximize the multipliers since it assumes
consumers will not buy imports unless the local supply cannot

meet the local demand.

 

‘ See “IMPLAN RPC's” (Minnesota IMPLAN Group, 2001) for details on how RPC’s are used in
IMPLAN models. See “Use of IMPLAN to Estimate Economic Impacts Stemming from Outdoor
Recreation Expenditures in the Upper Lake State" (Pederson, 1990) for details on how RPC's will
affect the estimates of economic multipliers using IMPLAN.

26

Location Quotients (L0). This approach measures an industry’s
relative concentration compared to a base area. LO assumes
that the commodity will be purchased more locally if more

production exists in the region (relative to the national average).

IMPLAN allows users to modify production functions, regional purchase

coefficients and other base trade flows that can alter the estimates of multipliers.

Like other l-O modeling systems, IMPLAN is based on five key assumptions

(Minnesota IMPLAN Group, 2000-p 103):

A

“Constant returns to scale. This means the production functions
are considered linear; if additional output is required, all inputs
increase proportionally.

No supply constraints, supplies are unlimited. An industry has
unlimited access to raw materials.

. A fixed commodity input structure implies that price changes do not

cause a firm to buy substitute goods.

Homogenous sector output: An industry won’t increase the output
of one product without proportionally increasing the output of all its
other products.

Industry technology assumption assumes that an industry uses the
same technology to produce all its products. An industry has a
primary product and all other products are byproducts of the
primary product.”

IMPLAN calculates economic ratios and multipliers for output (sales),

income, value added, and jobs. IMPLAN’s industry output data at the national

level is from the BEA’s output series and the Annual Survey of Manufacturers.

The Bureau of Labor Statistics (BLS) growth model is also used in cases where

the census or survey data are not available. IMPLAN uses the national output

per worker multiplied by state or county employment to get the total industry

27

output in the region. The regional industry output is then adjusted based on how
the value added to employment ratios in the region deviate from the national
averages (Minnesota IMPLAN Group, 2000-p 255).

IMPLAN’s job effects (employment) include wage and salary employees
and self-employed jobs in the region. They are not full-time equivalents, but
count any full-time or part-time job as one job. The employment data comes
from three sources: The US. Department of Labor’s ES-202 data, the US.
Department of Census’ County Business Patterns (CBP), and BEA’s Regional
Economic Information System (REIS). IMPLAN’s employment multiplier report
illustrates the structure of IMPLAN’s multiplier reports. IMPLAN's job multiplier
report includes direct, indirect, induced and total effects based on $1 million of

sales in the matching industry (Table 2-2).

Table 2-2. Sample IMPLAN Multiplier Report— Employment (1996, US)

 

Direct Indirect Induced Total Typel Type SAM

 

Sector effects“ effectsIi effectsa effects“ multiplier” multiplierc

463 Hotels And Lodging Places 17.22 7.65 9.77 34.64 1.44 2.01
454 Eating and Drinking 28.93 7.73 9.79 46.45 1.27 1.61
488 Amusement And Recreation 26.44 7.15 9.48 43.07 1.27 1.63
432 Manufacturing 9.34 8.18 8.72 26.23 1.88 2.81

 

 

a. Per million dollars of output.
b. Type I multiplier = (Direct effects + Indirect effects) / Direct Effects
c. Type SAM multiplier = (Direct effects + Indirect effects + Induced effects) / Direct Effects

Ratio multipliers are reported in the two multiplier columns. The numbers

in the “Total Effects” column are economic multipliers in this study (ratios of total

28

 

job to direct sales), while the numbers in the “Direct Effects” column represent
economic ratios (direct job to direct sales). The numbers in the “Type I Multiplier”
and “Type SAM Multiplier” columns are the ratio type multipliers.

IMPLAN's value added data consists of four components, employee
compensation, proprietor income, other property income, and indirect business
taxes. The first two components are also called personal income (or labor
income) in the IMPLAN WindowsTM version. Value added data are controlled to
match the National Income and Produce Accounts (NIPA) published by BEA.
The sources of IMPLAN's value added data are the same as the employment
data. However, if income information is not disclosed at the county level, the
state level income per worker ratios are used. lMPLAN’s income and value

added multiplier reports are similar to the employment multiplier reports.

Variations in Economic Multipliers

Based on a review of previous recreation and tourism studies, the type II
sales multiplier is the most frequently reported multiplier and IMPLAN is the most
widely used system. In tourism studies Type II sales multipliers vary from 1.2 to
2.4 for local areas and is normally larger than 2.4 for state and national regions
(Table 2.3). The national tourism sales multiplier for the US. in 1996 was 2.95
from the Corps of Engineers marina slip renter trip spending study (Chang et al.,

2000). Tourism sales multipliers for state or local areas should be smaller.

29

 

$.00 u 30..

 

 

 

 

 

 

 

 

 

 

who u 2.8:. 68m.
.uwtoac. .VN u 3.3 5.5682
2m 96.3.2.2: .225: 2:0 ...mcozmz .. ws=m 9.6:on 2250... 355 new .wj 232.. .99...
..= .8.
3d I IN u 3.8
.2203 $28 33.9 c>> .2.
.8859 89.9 .2252 SN u 21 u 8.8 __ 22:... 2.358 268 ..= a .3. 8.3m .82. _o~mo
couch}? 5233:: oEoontSoom
238:: 6:0 :2: A 5m 5
.652 92.92.35 ho omen. EoEmoi 9.6:on 9.5: 2805.2. mmocv. .ucm
m $083 $053 one a; r m; umcfim .mco_mm22n. .mcozmococm 22w 522250
9.65%
._2_m_> 32%
2.5.“. as... as. 2.28. 88: 3.822..
oN u 668:. coaotom “mom 595.5. new ohms. new .3530
32¢;
E23355:
3:82 22232 to 2%.
m. 9.28% ..2.m.> 2 .2552”. cc $3292.. :3 C c950
2623 668:. .22 So cam”. mmd u 058:. E50... {3292. 9%:on $.50... coca .23. new 5:24.

 

 

 

 

 

 

£23.5st .2333: seguu< Soto! omen 252.com

 

 

3.02

 

auoa_a> 3.35:2

 

22.32:!
we «350w

 

£233

 

.332 £55

 

.33. 35.2

 

 

3.2.5 .038. 282.com. cosmetom cam 82.30... 0353 .m-~ 0.9m...

3O

 

a... lo... "moon.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

09.82 «a I 0.. u 2.85 8.8.. .58 c. .82.
05 0.2.2.5... 25mm. MN I me n 00.0w Zia—2. 9.30% 52.28. 02.500 00...... ...m .0 22.95
Bran.
um. I 5.. u 80.. .98 $2 25m «.980
8:82 EN I B. u 058... 5.; 8.29 m 2.285.... .82. .._m
2m 0.2.2.5... 05mm 54 I 2: u 020m 2.50. z<..n.s._ 9......on 52.3... 20.3.. mcofiom .0 E0593
222.3... 32.2.,
03 u 30.. E 83 0.5. 38.32
52.80. 3.? n 0600... 5.3 ..2m.o> o>mo 52.....22 .6. c. 00.0 .05. 3mm... 53..
ea 92.23... 22. 8.. u 8.8 won. 25%: .3 9......on 258 09.: can 8.50
...m u 32. 9......on a...
8.. u 2.85 2.58 .295. .88.
mad u 00.0w 23.0.2. 2% 02.2.. mo ..w.: ...m .0 9.9.0
2.2%. can
2.2.3.33 2.0 .0220 £25.58
om... I 3... u 0E8... 2.2.0. maotm> 9......on 52.20... 5200.0w 8me .0522“.
Enos. 0..
0.: I 3 u 82. .0...
.2208 08.00 003.9 «Nd I to u 2.50... 629.233. 88: 59.2..
2.30.00. 0095. 2.05.32 ov... I mwé u 023 __ 92$. 9......on .258. no.0 050.2 a 92.202”.
c0022; 3.35:2 050253.200m
0.2.2.3:
0202 .uo:_u> 3.3.2:! .0 000.30..." 3.22 :29. 20:5 .22: 35.2

 

 

 

 

 

 

 

6.28. .3 2an

31

00.0: :. 00.0.0
000.:3 ._ 00>... 0.0 000000. 90:00.3... .2. 00.0: :. 000.0000 00.2050 000.:3 0:060. 000.00 003_0> 0.:000.00. 00:0. 0:00.22 .0

 

 

 

 

 

 

 

 

 

 

 

 

00.60 0.00 .000:
0.... u 0:00 w: 0:0 00.60 0.0. u 000.. .0800 .000... .05.0.0
:25... F .00 0. 0:005... 000... 00.0 n 0.50:. 0.. 0..0:0.< 050:000 .0030... 00_.0:0>0w 0:0 .05....
05.. I um... u 00.0w
0.8%
0.3 u 000.. .000:
mm... H 0800:. 3.20m
wad u 00.0w .0: 0:. 0....0..>> 0:0
...0:o...02 2505.. 0:20.03 0.._0_.>> 0:0 00.0.0 on ...0.0 .59.
0.0.. I 0.00 n 000.. 0.:0>0 .0620

cm... I mod n 0800:. 0:0 0.:0E0E3o. 00...0 2.0:. Sam...
Ed I an; n 00.0w 2505.. 00000 .w... :0>0m 3.0.00.0

0.00.0.0

0.00505

00 I an n 82. 0.2.2.3 .o 0&8
0.03:2... 0.8.0 0830... 2.. I 00.: u 0685 z. 83 80:39.3 .000:
0:. 00.03.00 0.0530 0:... 00.. Inn... u 00.0w 00030.. z<.._0.2_ 050:000 8:05 00.2300 ...0 .0 .0090
.000: 0..
20:03.35 .80...
00.02 .003.0> 3:03.32 .0 000.500 3.2.0.. c200. >03m .053<

 

 

 

 

 

 

 

8.28. 0-0 2an

32

Economic multipliers reported in these studies demonstrate considerable
variations across regions, sectors and sources. Three studies cited here use an
aggregate multiplier. Archer and Owen's “Tourist Regional Multiplier” was
estimated with the authors' own economic model, one study uses the “so-called”
standard multiplier of 2.0 and one is based on professional judgment.

Many studies have used RIMS ll multipliers. Most of the studies reviewed
here estimate state-level impacts or adjust state-level multipliers to local areas.
Mak (1989), for example, applied RIMS ll multipliers to tourist spending in each
of the 50 states and the District of Columbia to estimate economic impacts of
tourism spending (Mak, 1989). Frechtling and Horvath extended the analysis to
Washington 0.0., where the RIMS ll sales multipliers ranged from 1.24 for the
Local Transportation sector to 1.38 for the Hotel and Lodging Places sector.
Total job to sales ratios in this study ranged from 17 to 7 for these two sectors,
respectively (Frechtling and Horvath, 1999). RIMS II sales multipliers for tourism
sectors at the state level are generally around 2.0.

Input-output models are applied more at sub-state regions compared to
the multiplier approaches because the availability of data. The Type II sales
multipliers estimated from IMPLAN varied from 1.37 for regions surrounding CE
recreation projects (Propst et al., 1998) to 2.3 for Dade County in Florida (English
et al., 1996). Note that some earlier estimates with IMPLAN tend to be larger,
with regional sales multiplier approaching 2.0 at sub-state level (Stynes and

Rutz, 1995). This is due to the upward bias IMPLAN has with induced effect

33

estimates for its Type III multipliers. This problem will be discussed later in this
chapter.

Fletcher compiled income multipliers estimated from l-O models for 30
countries, cities, and regions around the world. The ranges of these multipliers
varied from 0.19 for the City of Winchester, UK to 1.96 for Turkey. When he
ranked these multipliers in order he found that the multipliers are larger for
regions with larger and more developed economies (Fletcher, 1989).

Research has shown that multipliers are influenced by the geographic
size, population, and economic development of the region (T ooman, 1997;
Detomasi, 1987; Propst and Gavrilis, 1987; Fletcher, 1989; Becker, 1997; Olfert
and Stabler, 1994). Baaijens et al. took income multipliers from 11 studies and
estimated regression models by using population, area size, number of tourist
arrivals, and other regional characteristics to predict income multipliers. They
found that there is a positive relationship between the natural log of population
and the tourist income multiplier (Baaijens et. al., 1998). Chang et al. found
similar results using regional characteristics to predict sales multipliers for
regions surrounding 50 Corps of Engineers Lakes. They concluded that sales
multipliers could be best predicted by using a combination of area size and a
logarithmic variable representing the region’s economic activity (Chang et al.,

1999).

34

yaL'ations AcrossfiEgonomic Models and Changes Over Time

Although this study focuses on variations in multipliers across sectors and
regions, multipliers can also vary between different models and can change over
time. Borgen and Cooke (1990) compared the output multipliers of the 1977
RIMS model with the earlier version of IMPLAN (1982) for the State of Idaho.
They found that the IMPLAN Type III multipliers were consistently lower than the
RIMS II Type II multipliers in 29 out of 35 comparable industry sectors. The RIMS
ll multipliers were greater than lMPLAN's by an average of 7 percent, with
differences ranging from 2 to 34 percent. The six sectors where IMPLAN
multipliers were higher than RIMS II are hotels/lodging and amusements, eating
and drinking places, retail trade, wholesale trade, insurance, and rubber and
leather products. All but one of these six sectors are service oriented and are
labor intensive. Brucker et al. (1990) compared five Input-Output models to
estimate economic impacts of seven economic scenarios from petroleum refining
in Texas to poultry processing in North Carolina. While RIMS ll output estimates
were greater than IMPLAN for six of seven scenarios, IMPLAN income and
employment estimates were higher on six of seven scenarios.

Multipliers estimated with the IMPLAN DOS version were compared with
RIMS II and REMI (Regional Economic Models Inc.) models by Rickman and
Schwer (1995a, 1995b). They compared output and employment multipliers for
Clark County, Nevada for the nine largest sectors in this county. Their findings
indicate that the IMPLAN Type III multipliers are generally larger than the other

two model's Type II multipliers in these selected sectors. Note that as Clark

35

County is a highly recreation and tourism oriented region (Las Vegas), most of
the selected sectors (hotel, amusement and recreation, eating and drinking, etc.)
are highly labor intensive and would have lower than average wages and salary.

Researchers have concluded that the IMPLAN method for generating
Type III multipliers overestimates multipliers in industries with lower than average
wages and underestimates multipliers in higher paying sectors (Chang, 1999;
Chamey and Leones, 1997). IMPLAN’s Type III multiplier approach assumes
that income per job and household respending of this income does not vary
across industry sectors. Tourism-related sectors generally pay below average
wages and salaries due in part to the number of seasonal and part-time
employees. The IMPLAN Type III multipliers are therefore overestimated for
these sectors. Newer versions of IMPLAN (Windows versions, after 1995)
generate Type II and Type SAM multipliers that are more comparable to the
RIMS II Type II multipliers. Tourism analysts should use caution when
comparing economic impacts of recreation and tourism from different versions of
IMPLAN systems, and be aware of the upward bias in the IMPLAN Type III
multipliers for tourism sectors (Chang, 1999).

Multipliers may also change over time because of price changes or
structural changes in the economy (e.g., new industries move in to the region).
Since sales, income, and value added multipliers are expressed as ratios of
money to money, these multipliers may not change significantly over time
(Loomis and Walsh, 1997). Job multipliers, when expressed as a ratio of number

of job to direct sales, will be affected by changes in general price levels. The

36

IMPLAN manual suggests that users price adjust spending to the year of the
model and multipliers.

Another approach is to price adjust job multipliers over time to match the
year of the spending data (Propst et al., 1998; Stynes et al., 2000). This
approach assumes that when there is no major change in the region’s economic
structures, multipliers will be reasonably stable. The sales and income
multipliers won’t change significantly and the job to sales ratios will change

based on changes in general price levels over time.

Moheg to Estjmating_th_e_E_conomic Impacts of Recreation and Togfim
The National Park Service’s Money Generation Model (MGM), Bureau of
Economic Analysis’s RIMS II system, and Minnesota IMPLAN Group’s IMPLAN l-
0 modeling systems are reviewed in this chapter. The MGM is a simple
approach to estimating impacts of visitor spending while the other two are

complete l-O modeling systems.

Money Generation Mofidel — Aggregate Multipliers

The National Park Service’s Money Generation Model is a simple one-
page pencil and paper worksheet for estimating sales, tax (and income), and jobs
effects of park visitor spending. This model first uses total visitation, average
spending per visit, and the percentage of visitors from outside the region to
estimate total non-resident visitor spending in the region. Total visitor spending

is then applied to an aggregate sales multiplier to estimate total sales effects.

37

The MGM estimates total sales effects first, and then converts total sales to total

tax and job effects. Note that the spending data used in MGM was the average

per person per day rates for lodging and meals (USDI NPS, 1990). The MGM

worksheet illustrates how to apply aggregate multipliers (Table 2-4).5

Table 2-4. Sample Worksheet for Money Generation Model for a Rural Area
National Park in the Rocky Mountain Region

 

A. Sales Benefits from Tourism

(”#de

O)

B
1
2
3
4
5
6
7
8

Estimated non-local percent of park use
Total recreation visits

Average daily expenditures per person
Calculate direct sales (1) X (2) X (3)

Enter estimated sales multiplier (Type II) (range 1.2 — 2.8,
average 2.0)
Calculate total sales benefits (4) X (5)

.T_a_x Revenue Benefits from Tourisg

Estimated total sales from A.6

Enter combined state and local retail sales tax rate
Calculate increased sales tax revenue ( 1) X (2)

Estimated total sales from A.6

Enter the taxable income ratio (range .20-.60, average 30%)
Enter combined state and local income tax rate

Calculate income tax revenue (1) x (5) x (6)

Compute total tax revenue (3) + (7)

C. Job Benefits from Tourism

1

Estimated total sales from A.6 (in millions)

2 Estimate job to sales ratio (range is 10-50 jobs per million in

sales, average = 30)
Calculate new jobs created by tourism (1) X (2)

50%

 

1,457,100

 

66.47

 

699?

48,426,719

 

2.00

 

96,853,437

96,853,437
7%

 

999’

6,779,741
96,853,437

 

25%

 

17%

 

4,116,271

 

9993

10,896,012

 

96.85

 

30

 

2,906

 

 

Source: “Money Generation Model,” USDI National Park Service (1990)

 

5 For step-by-step instruction for the MGM worksheet, refer to “Money Generation Model” (USDI,
National Park Service, 1990), or “Approaches to Estimating the Economic Impacts of Tourism;
Some Examples,” (Stynes, 1999b).

38

The MGM worksheet requires estimates of three aggregate multipliers: a
Type II sales multiplier to compute secondary sales effects, a “taxable income
ratio” to estimate income from sales, and an employment multiplier to estimate
jobs from total sales. These multipliers appear on Line A5, B5 and C2 of the
MGM worksheet (Table 2-4). A significant problem in applying the MGM model
is choosing values of these three multipliers for a given region. National Park
Service personnel are generally not familiar with economic multipliers and do not
have access to l-O models or multipliers for their regions (Duffield et al., 1997;
Stynes and Propst, 1999).

To implement the MGM model, the National Park Service made use of
published sate level multipliers from RIMS II. In some cases these multipliers
were “adapted” to local areas using judgment. In other cases state level
multipliers were used in local applications. As RIMS ll does not report multipliers
for tourist spending, an average of two multipliers for the Hotel and Lodging

Places and Eating and Drinking was used (Duffield et al., 1997).

RIM; ll - Sector-Specific Multipliers

Sector-specific multipliers enable the impact analysis to capture variations
in spending in different sectors. The Bureau of Economic Analysis’s "Regional
Multipliers: A User Handbook for the Regional Input-output Modeling System”

demonstrates how to apply published sector-specific multipliers in an economic

39

impact analysis (USDC BEA, 1992).6 The publication includes RIMS ll (Regional
Input-output Modeling System) 1989 output, income and employment multipliers
for 39 industry sectors for each of the 50 states in the United Sates. State
multipliers after 1989 and multipliers for sub-state regions are not available in
published form but can be purchased from the BEA.

Table 2-5 is a sample table of RIMS Il multipliers for the State of Michigan
from the 1992 handbook. The multipliers are reported in two forms- as a ratio of
total effects to direct sales effect (termed "final demand multipliers”) and as a
ratio of total effects to direct sales (termed “direct-effect multipliers").

Visitor spending on different items can be applied to multipliers for
different sectors using this approach. Total economic impacts can be more
accurately captured with the sector-specific multipliers since spending on
different items will result in different impacts on the region’s economy. For
example, one million dollars spent by visitors on lodging/would have different
impacts on the region compared to the same amount of money spent on
souvenirs. Applying one aggregate multiplier would not capture this difference
since the $1 million would be applied to the same multiplier. However, when
using sector-specific multipliers, as in BEA’s RIMS II approach, the estimated
secondary effects are across all industries rather than for any specific industry

(Table 2-5).

 

6 See “Regional Multipliers: A User Handbook for the Regional Input-output Modeling System,”
(USDC BEA, 1992), or “Approaches to Estimating the Economic Impacts of Tourism; Some
Examples,” (Stynes, 1999b) for step-by-step instructions for applying RIMS II multipliers.

4o

 

 

 

 

 

 
 

Emma mammd mém 5mm 0 gmod mmezuE ctaoflacmE maoocm=oom=2
FFmFN 88F F.mN 8N3 8me 2885 86.9 new £85.52.
NmowN mFoNN N.NN 88¢ mmNFN 8.2%; Laos .898 F8853 83585:
SNE N53 3F F83 ENFN 29858 Em 8629, .902
88N NNNNN EN 536 NEFN 5:558 28:8.» 95 252m
ovaN NoN F.oN mFNNd FmoFN 8:820 385 5228.2
mnmmd mmmvd mdw mmofio wommd $9.605 .maoE Emerson.
FFmoN FoomN N. FN 88¢ mFmFN 8:335 .92: meE
otoN «SN 4. FN 350 33 N £885 8% 2m .50 .956
mvmvN ammNN EN 85 o NooF N 93.5. ucm $885 89s .25 .883
SFFN BEN FNN 38¢ 38 N 30:85 553 as... 523.
88a 88N .02 $85 9.8 F 9E2 23.28 EN 28?ch
owe—aw nmmme NKN 9mm 0 came N @5553 new mczcrm
ommmN BSN 2F FNovd 88 N 9883 8% 2m .88
BEN 8FoN 3F BN3 88 F .282
ESN N8FN NNF named 88 F £885 an. 2:me
3de NBFN cm 886 88 F 0822 95 £883 85%. Ba noon.
”9.2303352
BEN 83 F. FN m8; $NF N 88.588 F82 95 8:22:52
FFovN NoNN N8 82.0 m FmF N 85.588 262
"cozuauwcoo
085 NFSN FNN 85¢ 89 N BEE. 888.822
FvovN 33 E NNmFd :3. F mg 932 Em 5:298 8.5
008.0 0086 0.0 88.0 88 F m:_c_E .moo
BEES.
354.. 993 mm mmde 88 F 90:85 Ewan: new $88".
FmeF FNNoN F8 83.0 83 F 822% 5%: ucm Ease .9232me cam $885 .9339?
52.23: ucn $53.8 .8:F_=o_._u<
omnoq. mmEEmm 032. mmEEmm «5930 .998
22.335 625-695 92 22: 23865

 

 

 

 

cozmmoamxx 926:. E EoE>oEEw new .mmEEmm F39:O|:mmEo_s_ .8 99.2%: 9me = ws=m .m-~ 03m...

41

..NmmF 6.8.0.2. 0.E0:00w .0 =005m coma ..Ewam mc..m60.2 6930.53. .0560”. 05 :0. 0.00008: :00... < ”0.0.3.535. .0853... ”0050.0.

.58 :08 8

96:88:00 506:. 05 c. 00.— 655660 :000 :0. 8506:. 30.. ..0 :. 80.20 :SEsc :. 090:0 .80. 05 3:08:00. 5:28 0.5 :_ 5:0 :00m .0
, $58 :08 8 96:88:00 50.36:. 05 >0 00>0_0E0 8.9.0030: 2 500.6 6.00 8:58 .0
0:06 6:05.660 :08 .0. 8586:. 30: ..0 >0 60>0_0E0 8.9.080: .0 85500 :. 080:0 6.60 .052 05 9:08:00. 5:28 0.5 :. 5:0 200.... 6

$58 :08 2 96:88:00 586:. 05 >0 0:0E06 .05 2

60.02.06 6930 .0 6:06 6:05.660 :08 :0.— 850365 30. ..0 :_ 80.. 50 80:5: :. 090:0 6:06 .32 05 3:08:00. 5:28 0.5 :. 5:0 :00m .0
03:0 2000 2 96:88:00 586:. 05 >0 6:0E06 .05. 0. 00.02.00 59:0 .0
6:06 6:05.660 :08 .0. 8586:. 30: ..0 .3 606.060 8.9.080: .0 8:500 :. 096:0 6:06 .058 05 3:08:00. 5:28 0.5 :. 5:0 :00m .0

.58 :08 0. 96:88:00 506:. 05 >0 6:0E06 .05. 0. 60.02.06

5930 .0 6:06 6:05.600 :08 .0. 00506:. 30: ..0 :. 05000 «05 59:0 :_ 090:0 0:06 .050. 05 8:08.00. :E:_00 0.5 :. 5:0 :00m. .0

 

 

 

 

 

 

................. m NF ~me0 mmwoF 8.0880...
ommmF mNNmF oNFV mend FwwoN 80.200 88:088....
3mm: 9 BF Nam noomd NNFoN 803.00 5.00...
wmmmF 58F 0.9. oowmd mmmmF 800.: 90.5.6 6:0 0:.me
omonF 3:: Don 0590 am: 80.28 0850%.
080. F mvmm. F 0.8 Kmfio mmmm. F 80.28 .0comL0n.
oonF 0va 0.00 043... 88F $888350 0:0 800.: 9.502 0:0 86:
”80.20...”
Fmaoé wwwvh Em 83d oven. F 0.0.8 .00m
8va wood mdm owmwd mmFNN 00:058.
wooed wave. F v. Fm mind ommm. F 00:05“.
.288 .00.. 0:0 60:058. 60:05”.
omvv. F omom. F :6 N Fwnd 38. F 060.. :08”.
NF FoN owmm. F «.3. nmmwd onNF 060.. 0.08.9.3
”060... :08: 6:0 0.08.055
080.0 manna ms 9 Fwd omeF 803.60. 502:8 6:0 :903 .8m .0...F00_m
moooN mm E. F Fd F R55 3%. F :0..00.::EE00
003. F game mam Fm Ed mmmw. F 8:08:80...
"8.5.5: 0.330 6:0 co=0toawc0._._.
080.. 8:53 082. mm:.E0m. 05050 .208
0.0.. 5.3:. 6050-625 90.. 5.2: 20800.05“.

 

 

 

 

 

8.28. .8 0.an

42

Table 2-6 illustrates sample visitor spending impacts estimated using
sector-specific multipliers. Total visitor spending by visitors is listed for each
category in column two. All spending on local services like lodging and
recreation will accrue as direct sales (column 3). Purchases of goods are divided
into retail margins and producer prices. Only the retail margins and the local
production portion of the goods will be captured by the region’s economy as
direct sales in column three. The direct sales estimated from visitor spending for
each category are then applied to multipliers for the matching sectors in column

four to estimate the total sales effects (column five).

Table 26 Sample Visitor Impacts Estimated Using RIMS ll Multipliers

 

 

Sector Visitor Direct Salesa Sales Total Sales

Spending (000's) Multiplier” Effects (000’s)
(000’s)

Eating and Drinking $500 $500 1.94 $970

Hotels And Lodging $400 $400 1.86 $744

Places

Amusement And $100 $100 1 .88 $186

Recreation

Groceries (local $50 $40b 1.80 $72

supplies)

Groceries (imports) $450 - - -

Retail Trade _-_ $100° J._9_5_ 11%

Total $1,500 $1,150 188" $2,167

 

a. Direct sales are the local production portion of visitor spending and retail margins.
b. Total spending on locally produced groceries less retail margins.

c. Retail margins of purchase on groceries. A 20% retail margin is used and applied to
the $50 spending on local produced and the $450 spending on imported groceries.

d. From Column 2, Table 2-5.

0. Computed as the ratio of total sales ($2,167) to direct sales ($1,150).

43

MN — l-O Modeling System

l-O models analyze the flow of monetary transactions from one industry to
another and from the consumer to the producer. The transactions tables used in
an l-O model include many industry sectors in the region and show the linkages
among industry, households, government, and exports (Bulmer-Thomas, 1982;
Minnesota IMPLAN Group, 2000). IMPLAN, for instance, generates a complete
set of multipliers for up to 528 industry sectors for a given region at the county
level.

l-O models enable the allocation of visitor spending to different industry
sectors in a detailed manner. Some ready-to-use l-O models such as IMPLAN
also handle the margins and local production issues. In addition to the detailed
information provided by l-O models, the l-O approach also provides a number of
advantages compared to the other two approaches. The researcher can
customize an l-O model and focus on individual industry sectors. For example,
analysts can single out the hotel sector and see which other sectors will be
affected by spending on lodging and by how much. It also provides direct and
secondary effects for individual sectors in which the researcher is interested.

Table 2-7 is a sample economic impact analysis report that was generated
using IMPLAN. Economic impacts of Corps of Engineers marina slip renter
spending on trips were estimated by applying visitor spending to an l-O model of
the US. economy. The indirect and induced effects reported in Table 2-7 are the

sales effects to individual row industries.

44

Table 2-7. Sales Effects of CE Marina Slip Renters' Trip Spending to the Nation

 

 

 

Direct Indirect Induced Total
SALES EFFECTS ($MM)
Manufacture 302.34 335.67 243.06 881 .07
Transportation & Services 33.18 235.56 466.30 735.04
Recreation 39.10 10.00 15.99 65.08
Hotel and Lodging 54.05 6.96 9.30 70.30
Eating and Drinking 176.30 4.41 30.53 211.24
Retail/ Wholesale 188.50 50.32 1 14.67 353.49
Government M 19_._4_4_ 18.16 30._44_
Total $795.30 $653.37 $898.00 $2346.67

 

Source: “Visitor expenditures and economic impacts by marina slip renters
associated with the US Army Corps of Engineers” (Chang et al., 2000).

When applying tourism spending to an I-O model, spending must be
allocated to production sector (industry) in the l-O model (Propst and Siverts,
1990). For example, visitor spending on groceries involves more than 50 sectors
in the IMPLAN system. With a proper "bridging" scheme to match visitor
spending to industry sectors in the model, more targeted and potentially
meaningful results can be derived. The Micro-lmplan Recreation Economic
Impact (Ml-REC) system was used to bridge the total spending into the
appropriate industrial sectors of the l-O model (Stynes and Propst, 1998; Chang
et al., 1998). However, this level of detail is only necessary if the goods are
locally produced and the production functions vary significantly. If visitors spend
only a small percentage of their overall spending on a given item, detailed
attention to the corresponding sector will not significantly change the results,
unless one is particularly interested in the impact of tourists on a particular

industry.

45

Problems in Applying Multipliers to Tourism Impact Studies
There are four main categories of problems in applying multipliers to
tourism impact studies, generalization errors, aggregation errors, incorrect

application of multiplier, and failure to margin retail purchases.

Generalization Errors

Generalization errors occur when a multiplier for a region, sector, or time
period is applied to another situation that is not the same. Many tourism analysts
apply national or state level multipliers to sub-state regions or apply the largest
multiplier that is available (Archer and Owen, 1971). Since the economic
leakage in a sub-region is substantially higher than a larger region, this practice
results in overestimated economic impacts. A similar abuse was also observed
by Beattie and Leones (1993) as

”...the 'all purpose' or 'standard' multiplier. These are multipliers that

people pull out of the air because they don't have access to multipliers

estimated using a full blown model or because they don't know any better.

The most common ones we hear, are output multipliers of 3 or 3.5. These

are large for state economies and even more so for county and city

economies."

Burress (1989) notes that numerous Kansas reports quoted multipliers
from a study by Heins (1982) contracted by the Institute for Economic and
Business Research (IEBR) on the behalf of Kansas Chamber of Commerce and

Industry. Although the Heins study was released but not endorsed by IEBR for

its implausible methods and results, the multipliers reported in the Heins study

46

were heavily quoted in Kansas because they were much higher than results from
other reports (Burress, 1989). The use of inappropriate multipliers for the tasks,
especially to apply multipliers for a larger region to a smaller region, has been
one of the most common abuses of economic multipliers (Beattie and Leones,
1993).

The key point for applying borrowed multipliers is whether the borrowed
multipliers can adequately reflect the study region’s economy (Holland, 1994;
Chappelle, 1985). Even for the same economic activity, multipliers still vary
across regions, sectors, and times. Borrowing multipliers from other regions
without comparing the regional characteristics and proper adjustment can lead to

significant errors in estimates.

Aggregation jnors

Aggregation errors occur when applying aggregate multipliers that fail to
capture the variations of visitor spending. Visitors spend money on many items,
and the money goes to many different sectors. Because economic ratios and
multipliers vary from sector to sector, the same amount of money spent in
different sectors can result in very different impacts. For example, $1 million
spent in restaurants yields 50 jobs while the same amount spent on
manufactured goods yields only 25 jobs in Michigan (based on BEA’s RIMS ll
multipliers, see Table 2-5). The use of aggregate multipliers in studies of visitor

spending cannot capture these variations.

47

Misapplication

Using inappropriate multipliers is also a common problem in tourism and
recreation studies. There are many kinds of economic multipliers and ratios. On
many occasions, inappropriate multipliers have been used due to ignorance or a
lack of understanding of multipliers (Archer, 1984, Propst and Gavrilis, 1987). It
is not uncommon for the analysts to fail to specify the kind of multiplier being
used, to use the wrong multipliers (i.e., use sales multipliers to compute income
effects), or to mix the “direct-effect” with the "final-demand” multipliers. For
example, the final demand income multiplier is 0.57 for the Hotel and Lodging
Places and the Amusement sector for the State of Michigan, while the direct-
effect income multiplier is 1.85 (Table 2-5). The final demand income multiplier is
based on direct sales effects (ratio of total income to direct sales), while the
direct-effect income multiplier is based on direct income effects (ratio of total
income to direct income). A common mistake in applying RIMS ll multipliers is to
apply the direct-effect income multiplier to the final demand (i.e., visitor spending
in lodging) (Frechtling and Horvath, 1999). In this case, the total income will be
overestimated by a factor of 3 if the direct effect multiplier is applied to direct

sales on lodging.

Failure to Ma_rgin Retail Purchases
Another common error in recreation and tourism studies is to apply
multipliers to total visitor spending, even though the manufactured goods

purchased by visitors are not locally made. The retail prices paid by visitors are

48

called purchaser prices. The purchaser price includes the producer price (price
of good at the factory) and retail, wholesale and transportation margins.
Purchaser Price = Producer Price + Margins (Retail, wholesale and

transportation)

If the goods bought by visitors are not made locally, only the retail margin
and possibly wholesale and transportation margins will be captured by the local
economy as direct sales. For example, $100 worth of souvenirs purchased
from a local shop may be broken down into two components: Producer price and
margins.

Purchaser Price = Producer Price + Margins

$100 $60 ($15 local) $40 ($30 local)

Only the portion of local productions ($15 of the $60 in producer prices)
and the local margins ($30 of the $40 in margins) will accrue to the local
economy as direct sales.

Direct sales effects = $45 ($15 Local Production + $30 Local Margins)

Since multipliers are ratios of total effects to the direct sales effects, they
should only be applied to direct sales that will accrue to the region (Archer, 1984;
Stynes, 1998). Therefore, if goods purchased by the visitor are not
manufactured in the local region, there will be no direct effect for the

manufacturing sector in the region. Only the margins will stay in the region. In

49

analyzing the economic impacts of tourist spending, all purchases of services will
accrue to the local region, but only the local margins (retail and locally operated
wholesale and transportation) on goods should be treated as direct effects when

the goods are not manufactured in the region.

This chapter has reviewed studies of economic multipliers and their
applications in tourism and recreation studies. Economic multipliers vary across
industry sectors and regions. Despite the growing use of economic multipliers
and regional economic models in recreation and tourism, limited guidance is
available for choosing multipliers suitable for a given application. l-O models
provide detailed estimates of both direct and secondary effects. However, users
must have access to I-O modeling systems such as IMPLAN and have some
knowledge of regional economics. Few recreation and tourism analysts have
formal training in regional economic methods and most are not very familiar with
input-output models or multipliers. Sector-specific multipliers provide a

compromise between easy and accuracy for recreation and tourism applications.

50

Chapter 3

RESEARCH METHODS

The research objectives for this study are to describe variations in
economic multipliers for tourism sectors, to identify key factors that explain the
variations, and to propose and evaluate procedures for predicting tourism
multipliers for a given region. To achieve these objectives, 114 regions varying
in size and economic development were selected. Input-output models
estimated with IMPLAN were developed for each of these regions and multipliers
were extracted. Variations in multipliers were examined by comparing multipliers
for recreation and tourism-related sectors across these regions. Regression
models were estimated to identify regional characteristics that explain the
variations in multipliers. A simple “multiplier lookup” procedure was developed
for choosing multipliers for a given region. The lookup and regression

approaches for predicting multipliers were both evaluated and compared.

Select Study Regions

One hundred and fourteen regions in five states (California, Colorado,
Florida, Michigan and Massachusetts) were selected. These five states were
selected mainly based on the availability of data. Michigan State University has
purchased the datasets from the Minnesota IMPLAN Group for a project to revise
the National Park Service’s Money Generation Model (Stynes and Propst, 1999).

Regions from these five states include a wide variety of regional characteristics,

51

which is an essential component for this study. Regions were selected to cover
different population sizes, geographical sizes, and degree of economic
development. These regions were formed into one of the following four
categories:

1. State regions for each of these five states (California, Colorado,
Florida, Michigan and Massachusetts) to represent the state level
economies. .

2. Metropolitan Statistical Areas (MSA's) in these states (about 70
models). The MSA is defined by USDC Bureau of Economic Analysis
as “a geographic area consisting of a large population nucleus together
with adjacent communities having a high degree of economic and
social integration with the nucleus” (USDC BEA, 2001). Each MSA
consists of one or several contiguous counties. MSA’s are semi-self-
sufficient economic units that include places for people to live, work,
and consume. They are also called functional economic areas and are
recommended to serve as economic impact study regions (Minnesota
IMPLAN Group, 2000).

3. Regions surrounding National Parks (30 to 50 mile radiuses) that
include gateway cities to National Parks to represent economies of
tourism regions (about 20 models, some overlap with regions in
category 2).

4. Single or multiple county non-metro regions (about 20 models).

Regions that consist of single or multiple counties were formed to

52

represent low economic development areas. These are regions in
rural areas with low populations that are not included in the other three

categories.

A complete list of all of the 114 regions selected for this study is given in

Appendix A.

Develmo Models and Extract Economic Multigflefi

Input-output models were built for each of the 114 regions using IMPLAN
Pro version 2.0 with the 1996 IMPLAN database (the most current year available
at the time). Economic multipliers (using IMPLAN Pro’s Type SAM multiplier
method) were estimated within IMPLAN. The default Regional Purchase
Coefficients approach was used to estimate trade flows for estimating multipliers.

IMPLAN stores all model information including multipliers in a Microsoft
AccessTM database file. An Excel file with Visual Basic Macros was developed to
extract the relevant data from these files. Information that was extracted
included the 1996 base year data (industry output, income, value added, and
jobs), multipliers (for sales, income, value added, and jobs), regional purchase
coefficients (RPC's), and population and area sizes of the study regions.

Table 3-1 summarizes the economic ratios and multipliers that were
extracted for this study. Eight different economic ratios and multipliers for four
economic measures were extracted. The income data extracted for this study is

IMPLAN's personal income (also called labor income). IMPLAN’s personal

53

income consists of employee compensation (all income to workers paid by

employers) and proprietor income (payments received by self-employed

individuals as income).

Table 3-1. Economic Multipliers and Ratios Used in this Study

 

 

 

 

 

 

Economic
measures Sales Income Value Jobs
Added (VA)

Multipliers
Direct effect --- direct income direct VA direct [obs
ratio direct sales direct sales direct sales
Type I direct + indirect sales --- --- ---
multiplier direct sales
Type II total sales8 total income3 total VA“ total jobs"
multiplier direct sales direct sales direct sales direct sales

 

 

 

 

 

a. Total effects = direct + indirect + induced effects

Choice of Industry Sectg_r§

Multipliers and ratios for 12 tourism-related sectors were selected for this

study (Table 3-2). These 12 sectors were selected to match spending categories

used in the National Park Service’s Money Generation Model 2 (Stynes et. al.,

2000). There are ten spending categories in the MGM 2 to identify the kinds of

goods and services a typical visitor will buy. Besides multipliers for the ten

sectors that match the ten spending categories in MGM 2, multipliers for the retail

and wholesale trade sectors were also extracted to accommodate the margins of

visitor spending on merchandise (Table 3-2).

54

 

Table 3-2. Tourism Sectors and Corresponding Spending Categories

 

 

Spending Category IMPLAN Sector Name Sector #
Lodging Hotels and Lodging Places 463
Restaurant or food on-site Eating and Drinking 454
Recreation and entertainment Amusement and Recreation 488
Retail margins on goods Retail Trade 448-453, 4557
Whole sale margins on goods Wholesale Trade 447

All auto expenses other than gas and oil Auto Repair and Services 479
Local transportation (taxis, buses, etc.) Local Transportation 434
Gas and oil Petroleum Refining 210
Sporting goods Sporting Goods 421
Grocery, or food off-site Food Processing 103
Clothing Apparel 124

All other miscellaneous goods General manufacturing 432

 

 

 

 

 

Table 3-3 is an example of the multiplier table for one of the 114 regions.
Eight different economic ratios and multipliers plus regional purchase coefficients
were extracted for each of the12 industry sectors. All Type II multipliers are
ratios of total effects to direct sales effects. These Type II multipliers were
extracted from the “total effects” columns in IMPLAN’s multiplier reports. RPC’s
represent the proportion of local demand satisfied by local producers. RPC’s are
applied to the producer prices of goods purchased by visitors to allocate the

direct sales effects to the local economy.

 

7 There are seven IMPLAN retail trade sectors: 448 Building Materials 8. Gardening Supplies,
449 General Merchandise Stores, 450 Food Stores, 451 Automotive Dealers & Service Stations.
452 Apparel & Accessory Stores, 453 Furniture & Home Furnishings Stores, and 455
Miscellaneous Retail. The average of these seven sectors was used in this study.

55

Table 3-3. Michigan Statewide Multipliers — 1996

 

 

 

 

 

IMPLAN Sector Direct Effect Ratio Tvne ll multiplier Type I
Sector Name NO- Jobs/ Value Jobs/ Value
$MM lncome/ Added $MM Income Added/ IModel
Sales Sales /Sales Sales Sales / Sales Sales Sales RPCa
Hotels and
Lodging Places463 22.79 0.34 0.52 1.70 32.72 0.60 0.96 1.40 44%
Eating and
Drinking 454 30.91 0.35 0.49 1.66 38.97 0.57 0.87 1.38 89%
Amusement
land Recreation 488 30.72 0.35 0.58 1.65 39.67 0.60 0.98 1.35 66%
Retail Trade N/Ab 28.07 0.51 0.80 1.52 35.16 0.70 1.12 1.17 95%
Wholesale
Trade 447 8.39 0.41 0.69 1.53 15.68 0.61 1.02 1.23 63%
Auto Repair
and Services 479 11.97 0.32 0.50 1.60 19.30 0.53 0.85 1.33 49%
Local
Transportation 434 27.09 0.58 0.69 1.61 35.22 0.81 1.06 1.20 45%
Petroleum
Refining 210 0.61 0.05 0.14 1.38 3.96 0.14 0.35 1.30 20%
Sporting
Goods 421 8.28 0.27 0.52 1.53 14.65 0.47 0.83 1.30 3%
Food
Processing 103 5.34 0.15 0.30 1.54 11.84 0.34 0.60 1.37 57%
Apparel 124 11.33 0.33 0.39 1.51 17.80 0.51 0.69 1.25 2%
Manufacturing 432 9.05 0.26 0.45 1.56 15.80 0.46 0.77 1.32 3%

 

 

 

 

 

 

a. Regional purchase coefficient.
b. Retail is an average of seven retail trade sectors.

Local Purchase Coefficients (LPC) are used to estimate the portion of

tourist spending that accrues to the local economy as direct sales. LPC’s are

100% for visitor spending on services (e.g., Eating and Drinking Places,

Recreation and Amusement, etc.) and retail trade, since all these expenditures

will accrue to the region.

56

 

IMPLAN's RPC's are used to estimate local production for wholesale
trade, transportation and all manufacturing sectors. Use of LPC's may be
illustrated using visitor spending on gasoline as an example. The national
average retail margin on gasoline is 22% and the wholesale margin is 8% (USDC
Census bureau, 1998a; 1998b). For each $100 that visitors spend on gasoline,
$30 goes to the retail and wholesale sectors as margins and the remaining $70 (
the producer price) is allocated to the Petroleum Refining sector. The RPC for
the Petroleum Refining sector is 20% for Michigan (Table 3-3). Thus, only 20%
of this $70, or $14, will accrue to the Michigan Petroleum Refining sector. The
other $54 represents imports to the region and does not accrue as direct sales to

the local economy.

Defining a Tourism Multiplier

To explain general finding for the four primary tourism sectors, lodging,
eating and drinking, recreation and amusement, and retail trade, a “tourism
multiplier" is defined as:

Tourism multiplier = 0.32 X Lodging Multiplier +

0.29 X Eating and Drinking Multiplier +
0.13 X Recreation Multiplier+

0.26 X Retail Trade Multiplier

Tourism multipliers were computed as weighted averages of the

multipliers for the four primary tourism sectors. These four tourism sectors

57

account for more than 80 percent of the sales from a typical visitor spending and
are the focuses of this study (Table 3-4). Multipliers for these four tourism
sectors were weighted into a tourism multiplier in proportion to the percentage of

the direct sales they receive.

Table 3-4. Weights for Combining Multipliers into a Tourism Multiplier

 

 

Sector Percent of Percent of Weight = Percent among
SpendigrL Direct Salesa top four categories

Lodging 22% 27% 32%

Eating and Drinking 21% 25% 29%

Recreation 9% 10% 13%

Retail Trade - 20% M

Local Production 12% 18%

Imports 361/9 _-_

Total 100% 100% 100%

 

Source: “Money Generation Model 2” (Stynes et al., 2000).

a. Spending on goods is divided into retail margins and producer prices. Only the retail
margins and the local production portion of the goods are captured by the region’s
economy as direct sales.

One of the study objectives is to explain how multipliers vary across
regions. As multipliers for the primary tourism sectors are highly correlated
(pages 69-70), variations may be largely captured by the aggregate tourism

multiplier, defined above.

Describe Variations in Multipliers
The first study objective is to describe variations in economic multipliers
and ratios across sectors and regions. Descriptive statistics were first computed

for the population, area size, and population density across all 114 regions.

58

Means, medians, minimums, maximums, standard deviations, and coefficients of
variations were computed to illustrate the variations in regional characteristics
across regions.

Averages of multipliers, ratios, and RPC's among all regions were then
computed for the 12 tourism-related sectors. The differences in multipliers and
ratios across sectors were compared to evaluate the variations of multipliers
across sectors.

Means, medians, minimums, maximums, ranges, standard deviations, and
coefficients of variations of multipliers and ratios among all regions were then
computed and compared for each sector. These basic descriptive statistics show
how multipliers vary across regions. Values of Type II tourism multipliers were
also plotted for sales, income, value added, and jobs to illustrate the distributions

of multipliers across regions.

lgentifv Kev Factors that Explain Variations in Multipliers

The second study objective is to identify key factors that explain variations
in economic multipliers and ratios across regions. Correlation coefficients were
first computed for all multipliers and regional characteristics to examine
relationships among these variables. Regression models were then estimated to
identify the best set of regional characteristics for explaining the variations in
multipliers. The factors identified are used to help characterize groups of regions
in the next section, and the results of the regression models are used as a

benchmark to evaluate the “multiplier lookup” procedure in the next section.

59

Tourism multipliers for four economic measures, sales, income, value
added, and jobs, were used as dependent variables in four regression models.
The four linear regression models (ordinary least squares) hypothesized for this
study are:

1) Type II Sales Multiplier = a0 + a1R1 + a2R2 + a3R3 + a4R4 + a5R5

2) Type II Income Multiplier = a0 + a1R1 + asz + a3R3 + a4R4 + a5R5

3) Type II Value Added Multiplier = a0 + a1R1 + asz + a3R3 + a4R4 + a5R5
4) Type II Job multiplier = a0 + a1R1 + a2R2 + a3R3 + a4R4 + a5R5

where

R1 = Population

R2 = Area size

R3 = Population density

R4 = Natural log of population

R5 = Natural log of area size

Since one of the research objectives is to develop a tool to simplify the
selection of multipliers for recreation and tourism applications, only regional
characteristics that are readily available to recreation and tourism managers
were tested as independent variables. Based on the literature review, all
independent variables were introduced in their original values and as logarithmic

values.8 A stepwise regression procedure was used to help define the best

 

8 The natural log of population density was not included since it is a function of the natural log of
population and the natural log of area. For any given region, In(popu|ation density) =
In(popu|ation) — In(area).

6O

subset of regional characteristics to explain the variations of multipliers (Hauser,
1974).

The mean absolute percent errors (MAPE) and the maximum of APE’s
were computed for all four models by comparing the multipliers predicted by
regression model to the multipliers estimated by individual l-O model to evaluate

the possible range of errors of the regression approach.

2 I predicted value - individual multiplier
MAPE =

 

2 individual multiplier

Propose and Evaluate Procedures for Predicting Multipliers

The third study objective is to propose and evaluate procedures for
predicting multipliers for a given region. Regression models in the previous
section provide one approach to predicting multipliers. A “multiplier lookup”
procedure that can lead the users to choose multipliers for a given class of region
based on the regional characteristics identified in Objective 2 was also
developed. Regression models yield continuous predictions of multipliers while a
lookup approach will identify a small number of discrete points from which users
may select. A regression approach requires measures of the independent
variables to make predictions, and allows only quantitative variables. A lookup
approach can capture factors that are not easily quantified (e.g., central place

theory, similarities between regions, etc.) in a more subjective manner.

61

Procedures for developing and evaluating the multiplier lookup approach
were to 1) group regions with similar multipliers, 2) characterize these groups
based on regional characteristics, 3) use the group averages as generic
multiplier values for the group, and 4) evaluate the magnitude of errors when
generic multipliers are used instead of actual IMPLAN multipliers for each region.

The set of 114 regions were first sorted by the aggregate Type II tourism
sales multipliers. Several cut-off points were identified to form the initial groups.
The cut-off points were first selected based on the multipliers’ distribution
examined in Objective 1. The cut-off points were then adjusted to minimize the
errors between the individual region’s sales multiplier and the group average.
Tourism income, value added, and job multipliers for each region were then
compared with group averages. Differences were examined and adjustments
were made to ensure reasonable homogeneity within each group.

After regions have been classified into different groups, means and
variations of the key regional characteristics identified in Objective 2 were also
computed for each group. The research hypothesis to be tested is that there are
significant differences in the key regional characteristics between groups of
regions formed by multipliers. Regions grouped by economic multipliers will also
have distinguishable regional characteristics. ANOVA and descriptive statistics
were used to measure variations in regional characteristics within and between
these groups.

Groups of regions were characterized based on key regional

characteristics identified in Objective 2. The idea is to develop simple rules for

62

selecting which group a given region falls into, where the group average
multipliers can be used as generic multipliers for that type of region. For each
group of regions that a distinctive set of multipliers can be identified, there are
descriptions on the region’s characteristics that can serve as a “lookup table.”
The “multiplier lookup” procedure was evaluated by computing errors in
multipliers estimated by this approach and comparisons with regression models.
For each region, IMPLAN generated multipliers were used to test the accuracy of
the multiplier lookup procedure and the regression method. To evaluate the
multiplier lookup procedure, errors were computed using the multipliers for each
individual model and then comparing them with the group multipliers (using the
group averages to which the region belonged). The mean absolute percent
errors (MAPE) and the maximum APE’s were computed for all regions in same
group to demonstrate the possible range of errors of this “multiplier lookup”

approach.

2 group multiplier - individual multiplier
MAPE =

 

2 individual multiplier

The same comparison was also made between multipliers for each
individual model and multipliers estimated from regression models. Both mean
and maximum APE’s were also computed for the regression model approach for
each group of regions. The results for both the “multiplier lookup” procedure and
the regression models were examined and compared to evaluate these two

approaches.

63

Chapter 4

RESULTS

Results are presented for each of the three objectives—to describe
variations in economic multipliers for recreation and tourism sectors, to identify
key factors associated with the variations, and to propose and evaluate

procedures for predicting tourism multipliers for a given region.

Objective 1. Describe Variations in Multipliers
The first research objective is to describe how economic multipliers vary
across industry sectors and regions of different levels of economic development.
Variations in regional characteristics across 114 regions, variations in multipliers
across sectors, and variations in multipliers across regions are presented in this

section.

Variations in Regional Characteristics Across Regions

The average population for all 114 regions was 1.3 million and the median
was about 300,000. The average population was inflated by a few large values
since only 20 percent of the 114 regions (23 regions) had populations over 1.3
million (average of all regions). The largest population was about 32 million for
the State of California, while the smallest was just above 5,000 for Lake County
in Michigan. The average area was slightly less than 6,000 square miles for all

regions. The median was just about 1,600 square miles. Because of the wide

64

variety in the regions’ populations and areas, the population density ranged from
a high of almost 16,000 people per square mile for San Francisco to a low of 2

people per square mile for the Death Valley National Park area.

Table 4-1. Variations in Regional Characteristics Across 114 Regions (1996)

 

 

Statistics Population Area (square Population
miles) Density (per

square mile)

Mean 1 276,964 5,890 597
Standard Deviation 3,595,786 18,670 1,828
Coefficient of Variation“ 282% 317% 306%
Median 313,151 1,626 189
Maximum 31,878,234 155,973 15,746
Minimum 5,478 47 2
Sample size 114 114 114

 

a. Coefficient of Variation= Standard Deviation / Mean.

Variations in Economic Ratios a_nd Multipliers Across Sectog

Economic ratios vary across industry sectors (Table 4-2). Average job to
sales ratios (number of direct jobs per million dollar sales) varied from 30 for the
Eating and Drinking sector to less than one for the Petroleum Refinery sector
(jobs created from sales received by the petroleum industry, not sales at gas
stations). Average income to sales ratios varied from 0.57 for the Local
Transportation sector to 0.05 for the Petroleum Refining sector, while value
added to sales ratios varied from 0.80 for the Retail Trade sector to 0.14 for the

Petroleum Refining sector.

65

Table 4-2. Direct Effect Ratios for 12 Tourism and Recreation Related Sectors

 

 

 

 

 

 

  

 

 

Sector
IMPLAN Sector Name No. Averge" Sample Model
Jobsb/ lncomel Valuel Size“ RPCd

$MM Sales Added

Sales lSaIesI
Hotels and Lodging Places 463 21.62 0.35 0.53 114 62%
Eating and Drinking 454 30.21 0.36 0.50 114 88%
Amusement and Recreation 488 30.05 0.35 0.58! 113 68%
Retail Trade WA“ 28.87 0.51 0.80 114 91%
Wholesale Trade 447 11.45 0.40 0.69 114 67%
Auto Repair and Services 479 12.52 0.31 0.49 114 85%
Local Transportation 434 28.81 0. 57 0.68 109 51 %
Food Processing 103 5.58 0.14 0.28 78 39%
Apparel 124 13.39 0.25 0.29 90 4%
Petroleum Refining 210 0.61 0.05 0.14 40 13%
Sporting Goods 421 10.85 0.23 0.44 85 2%
Manufacturi 49.2.. 10.55 0.23 0.3 87 2%
Mean (12 sectors) 17.04 0.31 0.48I
Maximum 30.21 0.57 0.80
Minimum 0.61 0.05 0.14
Range 29.60 0.52 0.66

 

 

 

 

a. Average across all regions.
b. Not full-time equivalent. Any part-time or full-time job is counted as one job.
c. Sample sizes are different because not all sectors existed in all regions.

d. Regional purchase coefficient. RPC's are available in all 114 regions for all sectors.
e. Retail is an average of seven retail trade sectors.

On average, the Eating and Drinking and the Amusement and Recreation

sectors generated the most jobs directly from $1 million sales in that industry.

Each had more than 30 jobs per $1 million sales in 1996. The Local

Transportation and the Retail Trade sectors, on the other hand, had the highest

66

direct income and value added per sales ratios. Both sectors converted more
than half of the direct sales into personal income and converted 70 and 80
percent of the direct sales into value added.

The four primary tourism sectors (Lodging, Eating and Drinking,
Recreation and Retail) had higher than average ratios for jobs, income and value
added. The higher jobs and income per sales ratios indicate that these tourism
sectors are labor intensive (more employees are required to deliver a certain
amount of sales and higher percent of sales are passed to employees as
income). In contrast, the manufacturing sectors (food processing, apparel,
petroleum refining, sporting goods, and general manufacturing) had relatively low
direct effects ratios. None of the five manufacturing sectors converted $1 million
sales into more than 15 direct jobs. Only a quarter or less of the direct sales
were converted into direct income and the highest value added per sales ratio
was 0.44 (for the Sporting Goods sector).

The regional purchase coefficients (RPC's) were also lower for the
manufacturing sectors, ranging from 2 to 39 percent. The low RPC means that
the portion of total demand that is met by local production is low. Manufacturing
sectors are less important to tourism economic impact analysis as little visitor
spending goes directly to manufacturing sectors in the region, and the linkages

between primary tourism sectors to local manufacturers are limited.9

 

9 Most of the indirect (backward-linked) purchases made by the primary tourism sectors in a
region go to service sectors. For example, more than 80% of lodging purchases are from utilities,
banking and finance, transportation, and other service sectors.

67

The means for total effects multipliers (ratios of total effects to direct sales)
for different sectors are shown in Table 4-3. The average Type II sales multiplier
varied from 1.56 for the Hotel and Lodging sector to 1.34 for the Petroleum
Refining sector, while the Type I sales multipliers varied from 1.34 for the Food
Processing sector to 1.15 for the Retail Trade sector. The average Type II
multipliers for jobs varied from 37 to 3, while the income and value added

multipliers varied from 0.76 to 0.14 and 1.06 to 0.34, respectively.

Table 4-3. Economic Multipliers for 12 Tourism and Recreation Related Sectors

 

 

 

 

 

IMPLAN Sector Name Sector Tvoe II Total Effect Mufiipliera Type I Sample
No. Sales Jobs”/ Income! Value Salesa Size
$MM Sales Added
Sales Salesr
Hotels and Lodging Places 463 1.56 29.90 0.56 0.88 1.33 114
Eating and Drinking 454 1.48 36.78 0.53 0.79 1.26 114
Amusement and Recreation 488 1.51 37.45 0.54 0.90 1.28 113
Retail Trade N/Ac 1.42 34.91 0.66 1.06 1.15 1 14
Wholesale Trade 447 1.43 17.70 0.57 0.96 1.20 114
Auto Repair and Services 479 1.43 18.49 0.46 0.74 1.24 114
Local Transportation 434 1 .49 35.79 0.76 0.98 1.17 109
Food Processing 103 1.48 11.69 0.31 0.56 1.34 78
Apparel 124 1.45 19.33 0.41 0.55 1.28 90
Petroleum Refining 210 1.34 3.37 0.14 0.34 1.28 40
Sporting Goods 421 1.50 17.17 0.42 0.74 1.32 85

   

Manufactunn 432 1.48 16.73 0.41 0.68 1.30
Mean (12 sectors) 1.46 23.28 0.48 0.76 1.26
Maximum 1.56 37.45 0.76 1.06 1.34
Minimum 1.34 3.37 0.14 0.34 1.15
Range 0.22 34.08 0.62 0.72 0.19
a. Average across all 114 regions.

b. Not full-time equivalent. Any part-time or full-time job is counted as one job.
c. Retail is an average of seven retail trade sectors.

 

 

 

 

 

 

68

 

Similar to direct effects ratios, service sectors had higher total jobs,
income, and value added per $1 million direct sales than manufacturing sectors
except for the Auto Repairs and Services sector. The differences between
service and manufacturing sectors, however, are not as significant for sales
multipliers. This is also because of the fact that service sectors are labor
intensive, and less money is spent on purchases of materials to other backward-
linked industries compared to manufacturing sectors, which results in lower

indirect effects and thus lower Type I multipliers.

Variations in Econgmic Ratios and Multipliers Across Regions

Variations in multipliers are presented for each of the four economic
measures, sales, jobs, income, and value added, that were analyzed in this
study. Results are presented for each of the four economic measures using
distribution plots of aggregate tourism multipliers and tables of sector-specific
multipliers. The aggregate tourism multipliers as defined in the previous chapter
(page 57) are used to illustrate distribution of multipliers across regions in this
section. Multipliers for the four primary tourism sectors are all highly correlated
across regions (Table 4-4). Correlation coefficients for the four sectors are
higher than 0.95 for sales, income and value added multipliers. Correlation
coefficients are mostly higher than 0.7 for job multipliers except for the recreation
sector. The distributions of multipliers for the four primary tourism sectors are all
similar and the aggregate tourism multipliers capture the general patterns of the

variations in multipliers across regions.

69

Table 4-4. Pearson Correlation Coefficients for Multipliers Between Tourism
Sectors

 

Type II Sales Multiplier

Sector Lodging Restaurant Recreation Retail Aggregate
Lodging 1.00 0.95 0.97 0.98 0.99
Restaurant 1 .00 0.96 0.95 0.98
Recreation 1 .00 0.97 0.98
Retail 1.00 0.99
Aggregate 1.00
Type II Income Multiplier
Lodging 1.00 0.97 0.96 0.95 0.99
Restaurant 1.00 0.96 0.93 0.99
Recreation 1 .00 0.95 0.98
Retail 1.00 0.97
Aggregate 1.00
Type II Value Added Multiplier
Lodging 1.00 0.97 0.96 0.96 0.99
Restaurant 1 .00 0.95 0.94 0.99
Recreation 1 .00 0.95 0.97
Retail 1.00 0.97
Aggregate 1.00
Type II Job Multiplier
Lodging 1.00 0.86 0.25 0.74 0.94
Restaurant 1.00 0.32 0.71 0.93
Recreation 1 .00 0.01 0.39
Retail 1.00 0.85
Aggggate 1.00

 

Four sample regions, representing typical rural (Modoc County, CA), small
metro (Redwood National Park area), large metro (Springfield, MA) and state
regions (Florida), were selected to illustrate the sizes of multipliers for different
types of regions. Descriptive statistics are presented for each of the four primary
tourism multipliers.1o The coefficient of variation (CV) is used as the primary

indicator of variation.

70

Sales Multipliers

The Type II tourism sales multipliers varied from 1.32 for Modoc County in
California to 1.67 for the State of Florida (Figure 4-1). Values of the Type II
tourism sales multipliers increased as the regions’ economic development
increased (from rural areas to small metro, large metro, and to state regions).
Three cut-off points, 1.4, 1.5, and 1.6, roughly divide the distribution into four
groups. About 40 percent of the regions had sales multipliers that were between
1.5 and 1.6, and 30 percent of the regions were between 1.4 and 1.5. The other

30 percent of the regions were equally distributed with multipliers above 1.6 or

 

 

   
    

    

  

 

 

 

 

 

under1.4
Type II Sales 8
I
l 8. I
16% 1 g
: :3 e
o g 3 T.
12/o 8 g
8 8. 1.5
I 8% - g a: g.
c
t
4% 8 3
3
E
0% T T , T
g 1.2 1.3 1.4 1.5 1.6 1.7

:
I I
Figure 4-1. Distribution of the Type II Tourism Sales Multipliers Across Regions

 

1° Statistics for all other eight tourism-related sectors are given in Appendix B.

71

The Type II sales multipliers varied from 1.17 to 1.77 across all regions for
the Eating and Drinking sector (Table 4-5). The coefficient of variation (CV) for
the Eating and Drinking sector was 8 percent of the mean, i.e., a 95 percent
confidence interval for these multipliers is within plus or minus 16 percent of the
mean. The CV for the Retail Trade sector was the lowest of the four primary
tourism sectors at 6 percent. The Type II sales multipliers for the Retail Trade
sector varied from 1.14 to 1.58 across all regions. The CV’s for the Type I sales
multipliers for all four sectors were smaller than for the Type II multipliers,
ranging from 3 percent for the Retail Trade sector to 5 percent for the Eating and

Drinking and the Amusement and Recreation sectors.

Table 4-5. Regional Variation in Sales Multipliers, Four Primary tourism
Sectors

 

Sector Mean Std. Coef. of Median Maximum Minimum Range
Dev. Variation

 

Tm ll Sales Multiplier

Hotels and

Lodging Places 1.56 0.11 7% 1.58 1.76 1.24 0.53
Eating and .

Drinking 1.48 0.12 8% 1.49 1.77 1.17 0.59
Amusement and

Recreation 1.51 0.12 8% 1.53 1.73 1.16 0.58
Retail Trade 1.42 0.09 6% 1 .43 1 .58 1.14 0.44

Tm I Sales Multiplier

Hotels and

Lodging Places 1.33 0.05 4% 1.34 1.42 1.17 0.25
Eating and

Drinking 1.26 0.06 5% 1.27 1.43 1.12 0.32
Amusement and

Recreation 1 .28 0.06 5% 1 .29 1 .39 1 .09 0.30
Retail Trade 1.15 0.03 3% 1.15 1.19 1.05 0.14

 

72

Job Multipliers

In contrast to the sales multipliers, the Type II tourism jobs multipliers
were larger for regions with less economic development and smaller for areas
with greater development (Figure 4-2). Job multipliers varied from 32 for Florida
to 43 for Modoc County. Four cut-off points, 28, 33, 36, and 41, roughly divide
the distribution into five groups. About 40 percent of the regions had job
multipliers that were between 33 and 36, and 30 percent of the regions were
between 36 and 41. About 15 percent of the regions had job multipliers between
28 and 33, and the other 15 percent of the regions were equally distributed

among the groups that had job multipliers larger than 41 or smaller than 28.

-_ _ _ _. __.__ ____.__-____.__1

Type II Jobs

 

   

   

     

 

 

 

20%
16% ~
3
12% ~ 2
3 °.
-- a
8% . g 5 §
5 3 g
0 .4 C
4" “a ‘2’ s
“’ E
0% . i .
25 30 35 4O 45

 

 

 

Figure 4-2. Distribution of the Type II Tourism Job multipliers Across Regions

73

The Coefficients of variation were higher for the Type II job multipliers than
for sales multipliers, ranging from 9 to 16 percent (Table 4-6). The Type II job
multipliers for the Retail Trade sector varied the most from 22.45 to 61.85 across
all regions with a range of 40 jobs. The CV was 16 percent of the mean for the
Retail Trade sector. The CV’s for the direct job ratios were even higher than for
the Type II multipliers, ranging from 11 percent for the Eating and Drinking sector

to 20 percent for all the other three sectors.

Table 4-6. Regional Variation in Job Multipliers, Four Primary tourism
Sectors

 

Sector Mean Std. Coef. of Median Maximum Minimum Range
Dev. Variation

 

Type II Job multiplier

Hotels and

Lodging Places 29.90 4.27 14% 30.12 40.48 16.18 24.30
Eating and

Drinking 36.78 3.18 9% 37.30 44.80 27.11 17.69
Amusement

and Recreation 37.45 6.13 16% 37.47 51.13 21.93 29.20
Retail Trade 34.91 5.50 16% 34.70 61.85 22.45 39.40

Direct Effect Ratio

Hotels and

Lodging Places 21.62 4.42 20% 21.22 32.95 11.47 21.48
Eating and

Drinking 30.21 3.23 11% 30.41 38.86 22.89 15.96
Amusement

and Recreation 30.05 6.06 20% 29.86 45.02 16.65 28.37

Retail Trade 28.87 5.69 20% 28.57 57.83 18.05 39.79

Note: Job multipliers are not full-time equivalent. Any part-time or full-time job is counted
as one job.

74

Income Multipliers
The Type II tourism income multipliers varied from 0.47 for Modoc County

to 0.66 for Florida. Values of the income multipliers also increased as the
regions’ economic development increased (from rural to state regions). Three
cut-off points, 0.50, 0.57, and 0.63, roughly divide the distribution into four
groups. About 40 percent of the regions had income multipliers that were
between 0.57 and 0.63, and 25 percent of the regions had multipliers that were
between 0.5 and 0.57. About 23 percent of the regions had income multipliers

that were larger than 0.63 and 12 percent of the regions had multipliers that were

lower than 0.50.

 

Type II Income

 

l
l
15% i
i
I

12% ~

  
 
 

9%~

Modoc County, CA; 0.47

6%4

    

  

 

3%-

 

Redwood NP; 0.54
Springfield, MA; 0.61

 

 

0% l l T !
0.43 0.48 0.53 0.58 0.63 0.68 .

..W_. J

Figure 4-3. Distribution of the Type II Tourism Income Multipliers Across
Regions

 

 

75

The CV’s for the direct income and value added ratios were all 0 percent
for Retail Trade Sector, as IMPLAN uses national income to sales ratios for this
sector for all regions (Tables 4-7 and 4-8). The lowest coefficient of variation for
the Type II income multipliers was for the Retail Trade sector at 6 percent (Table
4-7). However, this number may be underestimated since there is no variation in
the direct income ratio. The highest CV was 14 percent for the Hotel and
Lodging sector. The CV’s for the direct income ratios were lower than for the
Type II multipliers, ranging from 0 percent for the Retail Trade sector to 10

percent for the Hotel and Lodging sector.

Table 4-7. Regional Variation in Income Multipliers, Four Primary tourism
Sectors

 

Sector Mean Std. Coef. of Median Maximum Minimum Range
Dev. Variation

 

Type II income Multiplier

Hotels and

Lodging Places 0.56 0.08 14% 0.57 0.67 0.37 0.30
Eating and

Drinking 0.53 0.06 12% 0.53 0.65 0.36 0.29
Amusement and

Recreation 0.54 0.05 10% 0.55 0.65 0.41 0.24
Retail Trade 0.66 0.04 6% 0.66 0.73 0.56 0.17
Direct Effect Ratio

Hotels and

Lodging Places 0.35 0.03 10% 0.35 0.42 0.26 0.16
Eating and

Drinking 0.36 0.03 7% 0.36 0.41 0.28 0.13
Amusement and

Recreation 0.35 0.02 5% 0.36 0.39 0.32 0.07
Retail Trade 0.51 0.00 0% 0.51 0.51 0.51 0.00

 

76

Value Added Multipliers

The Type II tourism value added multipliers varied from 0.76 for Modoc
County to 1.04 for Florida (Figure 44). Values of the Type II tourism value
added multipliers also increased as the regions’ economic development
increased. Three cut-off points, 0.80, 0.90, and 0.97, roughly divide the
distribution into four groups. About 37 percent of the regions had value added
multipliers that were between 0.90 and 0.97, and 25 percent of the regions were
between 0.80 and 0.90. About 25 percent of the regions had value added

multipliers that were larger than 0.97 and 13 percent of the regions had

multipliers that were lower than 0.80.

 

Type II Value Added

 

20%

 

 

  

16% ~

 

12% 4

   

 

8%1

odoc County, CA; 0.76

4% «

Springfield, MA; 0.96

Florida; 1 .04

 

 

 

0% l I F l T l '
0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05

 

L
Figure 4-4. Distribution of the Type II Tourism Value Added Multipliers Across
Regions

 

77

Variations in the value added multipliers are similar to the income
multipliers. The lowest CV was 5 percent for the Retail Trade sector11 and the
highest was 13 percent for the Hotel and Lodging sector (Table 4-8). The Type II
value added multipliers varied from 0.88 to 1.16 for the Retail Trade sector
across all regions and varied from 0.57 to 1.06 for the Hotel ad Lodging sector.
The CV’s for the direct value added ratios were lower than for the Type II
multipliers, ranging from 0 percent for the Retail Trade sector to 10 percent for

the Hotel and Lodging sector.

Table 48 Regional Variation in Value Added Multipliers, Four Primary tourism
Sectors

 

Sector Mean Std. Coef. of Median Maximum Minimum Range
Dev. Variation

 

Type il Value Added Multipliers

Hotels and

Lodging Places 0.88 0.11 13% 0.91 1.06 0.57 0.49
Eating and

Drinking 0.79 0.10 12% 0.80 0.98 0.51 0.46
Amusement and

Recreation 0.90 0.08 9% 0.91 1 .05 0.69 0.36
Retail Trade 1 .06 0.06 5% 1 .06 1 .16 0.88 0.28
Direct Effect Ratios

Hotels and

Lodging Places 0.53 0.05 10% 0.53 0.64 0.39 0.25
Eating and

Drinking 0.50 0.04 7% 0.49 0.58 0.39 0.18
Amusement and

Recreation 0.58 0.03 5% 0.58 0.64 0.52 0.12
Retail Trade 0.80 0.00 0% 0.80 0.80 0.80 0.00

 

78

Economic multipliers varied across different industry sectors. Service
sectors generally had higher multipliers than manufacturing sectors in all four
economic measures. Economic multipliers also varied across regions varying in
economic development. Overall, sales, income and value added multipliers
increased as the regions’ population increased, while job multipliers decreased

as the regions’ population increased.

Mtive 2L. ldentjfv Lev Factors that Explain Variations in Multipliers
The second research objective is to identify regional characteristics that
explain variations in tourism multipliers across regions. Results are presented
first for the correlation analysis and then the regression analysis. The aggregate
tourism multipliers as weighted averages of multipliers for the four primary

tourism sectors are used in this section.

gum AfilVSiS

Tourism sales, income, and value added multipliers are highly correlated
with each other (Table 4-9). The correlation coefficients are higher than 0.95
among all three multipliers. The coefficients of determination (CD’s, computed
as r2) are larger than 0.90 among these three variables, which means that more
than 90 percent of variation in one multiplier can be explained by the other

(Griffith, 1997). The relationships between job multipliers and the other three

 

1' This number may also be underestimated since there is no variation in the direct effect ratio for
the Retail Trade Sector.

79

multipliers are not as strong, where the correlation coefficients range from -0.46

to -0.70 (CD’s range from 21 to 48 percent).

Table 4-9. Pearson Correlation Coefficients Between Tourism Multipliers and
Regional Characteristics

 

 

 

Type II Multiplier Natural log of

Sales Income VA Jobs Pop. Density Area Pop. Densfly Area
Sales 1.00 0.95 0.96 -0.46 0.46 0.05 0.35 0.86 0.59 0.41
Income 1.00 0.99 069 0.43 0.21 0.27 0.90 0.72 0.281
Value added 1.00 —0.6fl 0.43 0.19 0.28 0.88 0.68 0.30
Jobs 1.00 —0.25 -O.51 -0.07 -0.63 -0.69 0.07
Population 1.00 0.05 0.82 0.58 0.22 0.54
Pop. Density 1.00 -0.07 0.23 0.54 -0.42
Area 1 .00 0.40 -0.06 0.67
In(Popln) 1.00 0.76 0.38
In(Density) 1.00 -0.31
In(Area) 1.00

 

 

 

 

 

Note: Correlations are across 114 regions and all multipliers were Type II aggregate
tourism multipliers.

The relationships between regional characteristics and multipliers are
stronger when logarithmic values are used for population and population density.
Among all regional characteristics, the natural log of population has the highest
correlation coefficients with tourism multipliers ranging from 0.90 for income to
0.63 for jobs. The coefficients of determination between the natural log of
population and multipliers are about 75 to 80 percent for sales, income, and
value added multipliers but less than 40 percent for job multipliers. The natural
log of population density has the highest CD with job multipliers among all

regional characteristics at 47 percent. Area size is weakly associated with

80

tourism multipliers compared to other regional characteristics (all the CD’s are
under 20 percent).

It should be noted that the relationships among multipliers are also
influenced by IMPLAN’s assumptions and may not reflect the true variations.
IMPLAN’s state and county level indirect business taxes (IBT) and other property
type income (OPTI) are estimated using the national IBT and OPTI to labor
income ratios (MIG, lnc., 2000-p 251 ). Since the value added effect is a linear
function of personal income, they are supposed to be perfectly correlated.12 The
reason that income and value added multipliers are not perfectly correlated is
because minor adjustments to state and county level estimates of IBT and OPTI
are made so they can be added up to match the next level's estimates.13 The
high correlation between sales multipliers and value added and income
multipliers is also influenced by IMPLAN’s assumptions. The county (or state)
level personal income (or IBT, or OPTI) to employment ratio is compared with the
national average. The county or state level total industry output (sales) is then
adjusted if the county or state income to employment ratio is higher or lower than

the national average (MIG, Inc., 2000—p 255).

 

‘2 Value added = personal income (Pl) + IBT + OPTI, where IBT =j(Pl) and OPTI =f(Pl).
‘3 That is, the summation of total counties’ estimates is equal to the state value, and the
summation of total states’ estimates is equal to the national value.

81

' Regression Analysis

Separate regression models for aggregate tourism sales, income, value
added, and jobs multipliers were estimated using SPSS 10.0’s stepwise
procedure. Five independent variables were tested in each model: population,
natural log of population, area size, natural log of area size, and population
density. In the SPSS stepwise regression procedure the p-value of the partial F
statistic was set at 0.05 to include a variable and at 0.10 for removal. At each
step, the SPSS stepwise regression algorithm selects an independent variable
with the highest correlation with the dependent variable. The variable is included
in the equation if the partial F statistic is significant at the 95% level (p-value
<0.05). The process continues until no variable is significant enough to be
included. During each step the algorithm also checks the partial F statistic for
each included variable in the equation and removes variables that have p—values

greater than 0.10 (SPSS, 1999).

Model 1: Tourism sales multipliers

Following the stepwise regression procedure, the best prediction equation
for the Type II tourism sales multipliers was identified as:
. Tourism sales multiplier = 1.566 + 0.053 x in(POP) — 0.010 x POPDEN
where
In(POP) = natural log of population (in millions)

POPDEN = population density (thousand people per square mile)

82

Two predictors were identified for this model and the model explains 76
percent (adjusted R2 = 0.76) of the variation in the Type II tourism sales
multipliers (Table 4-10). Based upon the standardized coefficients (Betas), the
natural log of population is the most significant determinant of the Type II tourism
sales multipliers. This means that there is a linear relationship between the

multiplier and population in logarithmic form.

Table 4-10. Least Squares Regression Results for Sales Multipliers

 

Dependent Variable: Type ii Tourism Sales Multiplier

Multiple R 0.876
R Square 0.767
Adjusted R Square 0.763
Standard Error 0.051
Observations 1 14

Sum of Mean
df Squares Square F Sig. F

Regression 2 0.940 0.470 180.964 0.000
Residual 1 10 0.286 0.003
Total 1 12 1 .225

Standard Standardized

 

Variable Coef. Error Beta t Stat P-vaiue Tolerance VIF
(Constant) 1 .566 0.006 245.797 0.000
In(POP)‘ 0.053 0.003 0.899 18.994 0.000 0.946 1.057

POPDENb -0.009 0.003 -0.159 -3.359 0.001 0.946 1.057
a. In(POP) = natural log of population (in millions)
b. POPDEN = population density (thousand people per square mile)

The partial coefficients for the natural log of population can be interpreted

as, holding other variables constant, a one percent increase in population is

accompanied by an increase of 0.00053 (one hundredth of the coefficient for the

83

independent variable) in tourism sales multiplier (Gujarati, 1995). The partial
coefficient for population density indicates that each unit (thousand people per
square mile) change in population density lowers the tourism sales multiplier by
0.009.

The tolerances for both independent variables are very high (the range of
tolerance is from 0 to 1). A high tolerance means that the percent of an
independent variable that is not explained by other independent variables is high,
which indicates multicollinearity with other independent variables is less

problematic.

M: Totflm income multipliers

The best prediction equation for the Type II tourism income multipliers was
identified as:
0 Tourism income multiplier = 0.617 + 0.032 x in(POP) - 0.002 x POP

where

In(POP) =natural log of population (in millions)

POP = population (in millions)

Two predictors were identified for this model and the model explains more
than 80 percent (adjusted R2 = 0.82) of the variation in the Type II tourism
income multipliers (Table 4-11). Based upon the standardized coefficients, the
natural log of population in the region is again the most significant determinant of

the regression model. Note that although population was statistically significant

84

enough to be included in the model as a negative predictor, the standardized
beta is very small. The tolerance is 0.66 for both independent variables, which
indicates the presence of multicollinearity (between In(POP) and POP).
However, multicollinearity is not serious enough to be problematic as the

tolerance is still larger than 0.6.

Table 4-11. Least Squares Regression Results for Income Multipliers

 

Dependent Variable: Type II Tourism Income Multiplier

Multiple R 0.906
R Square 0.821
Adjusted R Square 0.817
Standard Error 0.025
Observations 1 14

Sum of Mean
df Squares Square F Sig. F

Regression 2 0.305 0.153 251.751 0.000
Residual 1 10 0.067 0.001
Total 1 12 0.372

Standard Standardized

 

Variable Coef. Error Beta tStat P-vaiue Tolerance VIF
(Constant) 0.617 0.004 170.785 0.000

In(POP)‘ 0.032 0.002 0.987 19.782 0.000 0.655 1.528
POPb -0.002 0.001 -0.153 -3.060 0.003 0.655 1.528

 

a. In(POP) = natural log of population (in millions)
b. POP = population (in millions)

Mpgel 3: Tflrism value added multipliers
The best prediction equation for the Type II tourism value added
multipliers was identified as:

0 Tourism value added multiplier = 0.968 + 0.048 x In(POP) — 0. 003 x POP

85

where
In (POP) = natural log of population (in millions)

POP = population (in millions)

Based on the near perfect correlation between income and value added
multipliers, it is not surprising to see similar regression results for these two
variables. The two predictors identified for this model were the same as for the
previous income model. This model explains almost 80 percent (adjusted R2 =

0.78) of the variations in Type II tourism value added multipliers (Table 4-12).

Table 4-12. Least Squares Regression Results for Value Added Multipliers

 

Dependent Variable: Type II Tourism Value Added Multiplier

Multiple R 0.887
R Square 0.786
Adjusted R Square 0.782
Standard Error 0.041
Observations 1 14

Sum of Mean
df Squares Square F Sig. F

Regression 2 0.674 0.337 202.050 0.000
Residual 1 10 0.184 0.002
Total 1 12 0.858

Standard Standardized

 

Variable Coef. Error Beta tStat P-vaiue Tolerance VIF
(Constant) 0.968 0.006 161 .498 0.000

In(POP)a 0.048 0.003 0.960 17.616 0.000 0.655 1.528
POPb -0.003 0.001 -0.137 -2.521 0.013 0.655 1.528

 

a. In(POP) = natural log of population (in millions)
b. POP = population (in millions)

86

Based upon the standardized coefficients, the log of population in the
region is again the most significant determinant of the dependent variable. The

tolerance is also 0.66 for both independent variables.

Mpdel 4: Toprism iob multipliers
The best prediction equation for the Type II tourism job multipliers was
identified as:14
0 Tourism job multiplier = 33.216 - 1.116 x in(POP) - 0.762 x POPDEN
where
In(POP) =natural log of population (in millions)

POPDEN = population density (thousand people per square mile)

Two predictors were identified for this model and all of them have negative
coefficients, i.e., the size of job multipliers is negatively correlated with the log of
population and population density. This model explains 52 percent (adjusted R2
= 0.52) of the variations in Type II tourism job multipliers (Table 4-13). Based
upon the standardized coefficients, the log of population is still the most
significant determinant of the dependent variable, but not as predominant as the
previous three models. The tolerances for these two independent variables are
very high, which indicates multicollinearity between these two independent

variables is less problematic.

 

1‘ One independent variable, the natural log of area, was dropped from the equation because of
multicollinearity (tolerance < 0.6). Dropping this independent variable decreases the model’s R
square from 53 to 52%.

87

Table 4-13. Least Squares Regression Results for Job multipliers

 

Dependent Variable: Type M Tourism Job multiplier

Multiple R 0.729
R Square 0.531
Adjusted R Square 0.522
Standard Error 2.530
Observations

Sum of Mean
df Squares Square F SQ F

Regression 2 796.792 398.396 62.226 0.000

 

Residual 1 10 704.259 6.402
Total 112 1501.051
Standard Standardized
Variable Coef. Error Beta tStat P-vaiue Tolerance VIF
(Constant) 33.216 0.316 104.992 0.000
In(POP)a -1.116 0.139 -0.538 -8.016 0.000 0.946 1.057

POPDENb -0.762 0.134 -0.382 -5.687 0.000 0.946 1.057
a. In(POP) = natural log of population (in millions)
b. POPDEN = population density (thousand people per square mile)

All four models are statistically significant at 0.01 level. Models for sales,
income and value added multipliers explain more than 75 percent of variation
while the job multiplier model explains 52 percent of variation in multipliers. The
outputs of SPSS’s stepwise regression models indicate one single independent
variable alone can explain almost all variation in the dependent variables. For
the sales, income, and value added multiplier models, adding one more
independent variable in addition to the log of population only increased the R

square by 1 to 2 percent. For the job multiplier model, the two independent

88

variables explain 52 percent while the natural log of population alone explains 40
percent of variation in the job multipliers.

On average, the mean absolute percent errors (MAPE) range from 3 to 4
percent for sales, income and value added multipliers and is the highest for job
multipliers at 6 percent (Table 4-14). The maximum errors range from 10 to 12
percent for sales, income and value added multipliers and is 26 percent for job

multipliers.

Table 4-14. Errors in Regression Predicted Multipliers

 

 

 

 

Type II Type II Type II Type II Tourism
Tourism Sales Tourism Tourism Value Added
Jobs Income
[MAPEa 3% 6% 3% 4%
[Max APE” 10% 26% 10% 12%

 

 

 

 

 

 

a. MAPE =Mean absolute percent error
b. Max APE= Maximum absolute percentage error

Objective 3. Propose and Evaluate Procedures for Predicting Multipliers

The final research objective is to propose and evaluate procedures for
predicting tourism multipliers for a given region. One procedure is to use the
regression models identified in the previous section. The second approach is to
use a “multiplier lookup” procedure to choose multipliers for a given region. The
lookup approach involves identifying a small numbers of types of regions and
using multipliers for these types to represent a class of region. The lookup and
regression approaches for predicting multipliers were both evaluated and

compared.

89

Develop a Multiplier Lookup Procedure

The 114 regions were first sorted based on the Type II tourism sales
multipliers. Cut-off points at 1.40, 1.51 and 1.58 were identified based on the
distribution of multipliers across regions (Figure 4-1) to form four primary groups.
The average Type II multipliers within each group were 1.30, 1.45, 1.55, and
1.65.

As income and value added multipliers were highly correlated with sales
multipliers (Table 4-9), the resulting groups are also homogeneous across sales,
income and value added multipliers. Regions with multipliers that were close to
the cut-off points and were better off by relocating to another region were shifted.
Eight regions were shifted from one group to another to reduce errors in income,
value added and job multipliers. The cut-off points and means were slightly
changed as a result of this adjustment. Job to sales ratios do not correlate as
well with sales multipliers as some regions experienced job to sales ratios that
were more than 15 percent below the average for their group.

All 114 regions were formed into four groups, each representing a type of
region with distinct values of multipliers. These four types of regions were then
classified according to regional characteristics and were categorized as rural,

small metro, large metro, and state regions (Table 4-15).

90

Table 4-15. Characteristics of Four Types of Regions

 

Rural

0 Smaller rural regions with low population (below 30,000).

. Low sales multipliers and high job to sales ratios.

. Representative regions: Modoc County (CA), Antrim County (MI) Pictured Rock NL,
Dinosaur NP, Bents NM.

 

Small metro

- Larger rural regions or small metro areas with population up to 500,000. Regions
with smaller populations that serve as population centers of the surrounding areas
may fit into this category.
Low to medium sales multipliers and medium to high job to sales ratios.
Representative regions: Redwood NP, Mesa Verde NP, Gainesville MSA (FL),
Lansing MSA (Ml), Pueblos MSA (CO).

 

Large metro

0 Medium to larger metro areas with population up to 1,000,000. Regions with
smaller populations that serve as population centers of the surrounding areas may
fit into this category.
Medium to high sales multipliers and medium to low job to sales ratios.
Representative regions: Rocky Mt. NP, Lassen Volcanic NP, Springfield MSA (MA),
Santa Barbara MSA (CA), Grand Rapids MSA (Ml).

 

 

State
a State level regions or regions including larger metro areas (1 ,000,000 and more).
0 High sales multipliers and low job to sales ratios.

0 Representative regions: State models, Everglades NP, San Diego MSA (CA),
Denver MSA (CO), Detroit MSA (Ml).

 

 

Note, MSA: Metropolitan Statistical Area; NP: National Park; NM: National Monument;
NL: National Lakeshore.

The rural area represents single or multiple county non-metro regions
where populations are below 30,000 and have limited economic development.
The small metro area is for larger rural regions or small metro areas with
populations up to 500,000, while the large metro area is for regions with
populations up to 1,000,000 and the state area is for regions with populations of

1 million and more. There was some overlap in population size between groups.

91

For example, regions with smaller populations that serve as the economic center
of the surrounding area may be included in a group with higher population. A
complete list of all regions within each group is available in Appendix C.
Population size was compared across the four types of regions. Table 4-
16 reports the descriptive statistics and ANOVA results for population across four
groups of regions. Populations are significantly different across these four
groups of regions and there was little overlap of ranges of population between
groups. The results of the ANOVA and the regression analysis from the previous
section show that population size can be used to explain variations in multipliers
across regions. Population size is therefore the primary variable for classifying

regions into groups that explain differences in multipliers.

Table 4-16. Descriptive Statistics and ANOVA Table for Population by the Four
Region Types

 

Descriptive
Mean Std. Error Minimum Maximum N
Rural 21,413 2,733 5,478 54,565 19
Small metro 210,789 27,391 21 .999 645,068 34
Large metro 734,124 106,904 108,371 3,015,783 44
State 6,217,572 1,861,033 1,008,633 31,878,234 17
Total 1,276,964 336,776 5,478 31,878,234 1 14
ANOVA
Sum of
df Squares Mean Square F Sig.
Between Groups 3 4.97E+14 1.66E+14 18.88 0.000
Within Groups 110 9.65E+14 8.77E+12
Total 113 1 .46E+15

 

Ranges of multipliers for the four primary groups are shown in Table 4-17.

These four sets of "generic" multipliers can be used to represent regions of

92

different populations and levels of economic diversity. These aggregate
multipliers would correspond to what might be used in the old MGM model, when

only a single multiplier is applied to visitor spending.

Table 4-17. Ranges of Multipliers for Regions Within Four Primary Groups

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Muitipiief Sales ll Jobs ll Income ll Value Added II
Group
Rural Mean 1 .31 37.58 0.48 0.76
[N= 19 [Minimum 1.18 33.01 0.42 0.65
[Maximum 1.40 43.08 0.53 0.85
cvb 5% 9% 8% 6%
Small metro [Mean 1.45 35.66 0.55 0.86
(N= 34 Mnimum 1.38 24.42 0.51 0.79
'Maximum 1.51 38.68 0.58 0.91
cv 2% 8% 4% 3%
Large metro [wean 1.54 32.51 0.60 0.95
'N= 44 Minimum 1.43 24.58 0.57 0.90
Maximum 1.60 38.22 0.65 1.01
cv 2% 10% 3% 3%
State [Mean 1.63 31.75 0.65 1.01
'N= 17 [Minimum 1.58 29.13 0.62 0.97
IMaximum 1.71 36.08 0.67 1.06
cv 2% 6% 3% 3%

 

 

 

 

 

 

 

a. All multipliers are aggregate tourism multipliers as defined in chapter 3 (p.57).
b. CV= Coefficient of variations = standard deviation/mean

Both income and value added multipliers are positively correlated with
sales multipliers while job multipliers are negatively correlated with all other
multipliers. Rural regions had the lowest sales multipliers and the highest job
multipliers, while state and other large metro regions had the highest sales

multipliers and the lowest job multipliers. The coefficients of variation ranged

93

from 2 to 6 percent for sales, income and value added multipliers across groups
of regions and ranged from 6 to 10 percent for job multipliers.

The generic tourism multipliers perform reasonably well in explaining
variations across regions of different degrees of economic development and
population. Variations in multipliers for individual sectors across the four types of
regions were also examined and similar results are obtained. The lodging sector

is used to illustrate how generic multipliers vary across regions (Table 4-18).

Table 4-18. Multipliers for the Lodging Sector by Type of Region

 

 

Multiplier Rural Small Metro Large Metro State
Direct effects

Jobs/ MM sales 27.84 23.41 19.09 18.02
Personal inc/sales 0.30 0.33 0.37 0.37
Value Added Isales 0.46 0.51 0.56 0.57
Total effects

Type I Sales 1.24 1.32 1.34 1.38
Type II Sales 1.36 1.52 1.61 1.70
Type II jobs 33.59 31.73 27.84 27.41
Type II income 0.43 0.53 0.60 0.64
Dpe Il value added 0.69 0.84 0.95 1.01

 

The Type II sales multipliers increase from 1.36 for rural regions to 1.70
for state and large metro areas. All sales, income and value added multipliers
and ratios increase with increasing economic development. The job multipliers
and ratios, however, decline with increasing economic development (from rural to

state areas). The Type II job multipliers decrease from 34 for rural regions to 27

94

for state and large metro areas. Detailed sector-by-sector values for multipliers

of all 12 sectors for each primary group are given in Appendix D.

Mate ProceQures for Selecting/ Predictipg Multipliers

A “multiplier Lookup” approach was presented in the previous section.
Four sets of sector-specific multipliers were developed, representing typical rural,
small metro, large metro, and state regions. Table 4-15 can be used as a
“lookup table” to select the type of region. Users can then apply generic
multipliers from Tables D-1 to 04 for that type of region. Criteria provided for
selecting a type of region are 1) the study region’s population, 2) whether the
region serves as a population center of the surrounding area, and 3) the similarity
of the region with other regions that are included in the lookup table.

To evaluate this multiplier lookup procedure, mean absolute percent errors
(MAPE) were computed by comparing the “generic” multipliers (predicted values)
with IMPLAN model multipliers for each region (observed values) (Table 4-19).
On average, the MAPE’s range from 2 to 4 percent for different groups of regions
for the aggregate tourism sales multipliers. The MAPE's are higher for job
multipliers at about 5 to 9 percent. All MAPE's for income multipliers are less
than 4 percent within each group and are less than 5 percent for value added

multipliers.

95

Table 4-19. Ranges of Errors by Using the Multiplier Lookup Approach

Sales ll Job ll Income ll Value Added II

5

APE 4 1 5
a. All multipliers are aggregate tourism multipliers as defined in chapter 3.
b. MAPE =Mean absolute percent error (comparing group average multiplier with
multipliers for each individual region in the group).
c. Max APE: Maximum absolute percentage error, the largest error in the group.

 

The maximum APE represents the greatest errors within each group.
Except for a couple of regions, the maximum errors are less than 10 percent for
regions within each group for sales, income and value added multipliers. The
errors from using generic sales, income and value added multipliers will be less
than 10 percent when the appropriate set of multipliers is selected and in most
cases will fall in the 2 to 5 percent range. The prediction of job multipliers is the
weakest among all multipliers. The MAPE’s range from 5 to 9 percent while the
maximum APE’s range from 12 to 46 percent for different types of regions. The

weak prediction is caused partly by the fact that many regions had lower than

96

group average job multipliers. Effects of regions with lower job multipliers will be
evaluated later in this chapter.

Absolute percent errors were also computed by comparing multipliers
predicted by regression models with each individual region’s multipliers. On
average, the MAPE’s range from 2 to 4 percent for the aggregate tourism sales
multipliers for different types of regions (Table 4-20). The MAPE's are higher for
job multipliers at about 5 to 7 percent. All MAPE's for income and value added
multipliers are less than 6 percent. The maximum APE’s are generally less than
10 percent for regions within each group for sales, income and value added
multipliers. The maximum APE’s are higher for job multipliers, ranging from 11 to

26 percent.

Table 4-20. Ranges of Errors by Using the Regression Models

Sales ll Job II Income II Value Added ll

 

APE 4 1

a. MAPE =Mean absolute percent error (comparing regression predicted values with
multipliers for each individual region in the group).
b. Max APE= Maximum absolute percentage error, the largest error in the group.

97

To illustrate the possible ranges of errors for different types of regions,
multipliers estimated by l-O models and predicted by both lookup and regression
approaches were plotted against population (Figures 4-5 to 4-8). The aggregate
tourism multipliers for sales, income, value added, and jobs are presented. The
horizontal dashed bars in these figures represent values of generic multipliers for
the four different types of regions. Population is used to illustrate the types of
regions because of two reasons. First, population is one of the three criteria
used in the multiplier lookup approach and is the only quantifiable criterion
among the three. Second, population (in logarithmic value) is the most
significant determinant identified in all regression models. Adding other
independent variables only slightly [increases the R Squares of these regression
models.

The comparisons between predicted and l-O estimated multipliers for the
sales, income, and value added multipliers are similar (Figures 4-5 to 4-7). Both
approaches perform better for the large metro and state areas than for the small
metro and rural areas. The regression approach performs better than the lookup
approach for a few large regions with very high multipliers and populations (i.e.,
state areas). The lookup approach, on the other hand, performs better than the
regression approach for the large metro and small metro areas. The distribution
of l-O estimated multipliers for these two types of regions (small and large metro
areas) indicate the advantage of using the lookup approach where multipliers do

not vary a lot within a type of regions.

98

Errors in predicted multipliers for the rural areas are higher than for the
other three types of regions for both approaches. Since natural resource-based
tourism areas are mostly located in rural areas, the results also suggest that
more rural regions should be included in the analysis and maybe to further break

up rural areas into more types of regions for the lookup approach.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1.8
a
1 7 o I o
0
State- ° .1! .. ..
1.6 1 o 0 0 ° 0
.9 Large Metro "0'
g 1.5 0 g
'5 o
(5“ Small Metrogm'L Q2" 0 i
. l
a; 1.4 000 o o o
' ' 8 [o l-O Estimates.
0 8 ° I Regression '
1.2 0 -Lookup J
1.1 . 1 T .
1,000 10,000 100,000 1,000,000 10,000,000 100,000,000
Population (in Logarithmic Scale) [
L

 

 

 

Figure 4-5. DistributiorT of Sales Multipliers Predicted by the Lookup and
Regression Approaches

99

 

 

 

0.7

 

 

06— _-

lncome Multiplier

0.5 1

 

o l-O Estimates [
I Regression '
- Lookup

 

 

 

i

 

 

 

 

0.4 . . . .
[ 1,000 10,000 100,000 1,000,000 10,000,000 100,000,000
Population (in Logarithmic Scale) J

 

 

 

 

Figure 4-6. Distribution of Income Multipliers Predicted by the Lookup and
Regression Approaches

100

 

 

  

 

 

 

 

° [
‘ 0 ° 8 9 . I
[ 1 _4 State a- - - - - [
1 8“" ° °
%_ Large Metro (b -'; [
‘-5 0.9 1 ° ° 1
2 s n M tr - [
3 ma e 00 go- o .
'0 o
2 0° 00 o
g 0.8 - o o
go - °° - '
[ Rural I 0° 8 00° Io l-O Estimates[ [
[ 0-7 ‘ I Regression
0 L-Lookup
0.6 T T i I
1,000 10,000 100,000 1,000,000 10,000,000 100,000,000

 

 

Population (in Logarithmic Scale) '
[

 

Figure 4-7. Distfibution of ValueAdded Multipliers Predicted by the Lookup and

Regression Approaches

101

Job multipliers do not have the same distribution as sales, income and

value added multipliers (Figure 4-8). In general, both lookup and regression

approaches do not predict job multipliers as well as sales, income, or value

added multipliers. Errors in predicted job multipliers are higher for the state and

rural areas than for the small and large metro regions. Both regression and

lookup approaches yield similar errors in job multipliers. The comparison

between job and sales multipliers indicates that regions may be grouped

differently if they were grouped solely based on job multipliers.

 

 

 

 

 

 

 

   

 

 

l 0 O
0
° 0
40 1
I o o
e 0 0° 005 °
,9 Rural - - 8
:3, o 0&9 o
g 35 4 ° 0 W co
2 19' —5' Small Metro
.0 ° 00
o o <b
5 o
1 6° 0
32 ° _ Large Metro
4 [ ' Q - I - -
1 o I-O Estrmates[ go WI- “ 0 State
. I Regression j ° ° 0 . I 8 .
7 -Lookup ___J ° 0 ° °I f - I
i o °°°
28 . . e A .
1,000 10,000 100,000 1,000,000 10,000,000 100,000,000

 

l

Population (in Logarithmic Scale)

Figure 4-8. Distribution of Job Multipliers Predicted by the Lookup and

Regression Approaches

102

 

 

Handlinlmions with Low Jop MuMaLers

When forming regions into four primary groups, some regions experienced
job to sales ratios that were more than 15 percent below the average for their
group. These regions had comparable sales, income, and value added
multipliers but lower job multipliers when compared with group averages. To
evaluate the effects of regions with lower job multipliers, thirty-two regions were
split out from the rural, small metro, and large metro groups.

Job multipliers for the 32 split out regions, on average, were 10 percent
lower across all groups relative to the parent groups that were computed with all
114 regions. Job multipliers for the 32 split out regions were about 15 percent
lower than the remaining 82 groups (Table 4-21). By splitting out the 32 regions,
job multipliers for the remaining 82 regions are more homogeneous within each
group. The MAPE’s for job multipliers are down from 6 to 9 percent for all 114
regions to 3 to 5 percent for the remaining 82 regions across all groups. The
maximum APE’s for job multipliers are less than 10 percent for all groups when
computed with the remaining 82 regions. This suggests that regions could be
formed into subgroups to better capture variations in job multipliers. Note that
splitting out these 32 regions in the generic multiplier analysis does not alter the
results for sales, income, and value added multipliers. The means and MAPE’s
are almost identical to results for these 32 split out regions and to the 82

remaining regions.

103

Table 4-21. Evaluation of the Low Job Multiplier Subgroup

T ll Job
32 low multiplier Remaining 82 All 114

34. 39. 37.
33.01 36. 33.01

35. 43. 43.

APE 1

of 1
a. All multipliers are aggregate tourism multipliers as defined in chapter 3.
b. No region was split out from the state group.
c. CV= Coefficient of variations = Standard deviation/mean
d. MAPE =Mean absolute percent error (comparing regression predicted values with
multipliers for each individual region in the group).
e. Max APE= Maximum absolute percentage error, the largest error in the group.

 

These 32 regions with lower than average job multipliers typically were
large metropolitan regions, rural areas with substantial tourism development, and
regions with a strong year-round tourism market. Using hotels as an example,

the lower job to sales ratios could be a result of higher room rates, or fewer part

104

time and seasonal jobs (resulting in higher average wages). Table 4-22 shows
the average personal income for hotel employees within each of the four primary
groups for regions that were split out from the groups and the remaining regions.
It indicates that in those 32 regions where lower job to sales ratios were found,
the average personal income in the hotel sector was much higher than the
remaining 82 regions. Since all 32 regions with low job multipliers are
consistently over-estimated by the regression model, this also suggests that, for
future research, job multipliers may be better predicted by including other

independent variables in the regression model.15

Table 4-22. Average Personal income for the Lodging Sector within each
Group of Regions

 

 

 

 

 

 

 

 

 

 

 

 

Average Personal

Income for the 32 Average Personal

Low Job Multiplier Income for the t-
Group wins“ RemaininLRegionsa statistics df SiLnificance
Rural $17,912 $13,649 3.272 17 0.004
Small metro $23,137 $18,105 4.835 31 0.000
Eye metro $33,750 $23,063 7.952 43 0.000
State WA $28,394 N/A N/A N/A

 

a. Computed using IMPLAN 1996 personal income and employment data for the hotel
and lodging sector.

This chapter has presented findings for the three study objectives.
Multipliers vary across industry sectors and across regions with different

characteristics. Service sectors had higher total jobs, income, and value added

 

‘5 For example, including average personal income for the hotel sector as an independent
variable in the job multiplier regression model increases the model’s adjusted R Square from 52%
to 82%.

105

per $1 million direct sales than manufacturing sectors. Sales, income and value
added multipliers increased as the region’s population increased, while job
multipliers decreased as the region’s population increased. Regions were
classified into groups with distinct multipliers. Errors in the multiplier lookup
procedure are similar to the regression model approach, if the appropriate region
is selected. The results indicate that the lookup procedure delivers similar

accuracy compared to the regression models.

106

Chapter 5

DISCUSSION AND CONCLUSIONS

The major goal of this study was to explain variations in multipliers for the
tourism and recreation sectors. The three study objectives are to describe
variations in economic multipliers, to identify key factors associated with these
variations, and to propose and evaluate procedures for predicting multipliers for
recreation and tourism applications. The development of a multiplier lookup
procedure is to allow analysts to select a type of region that matches the regional
characteristics of his or her study area and use the “generic” multipliers for that
type of region. A total of 114 regional input-output models were built for this
study. Multipliers were extracted from these models and regions were formed
into four groups based on regional multipliers within each group. A lookup table
was developed to help analysts select a set of sector-specific multipliers. Errors
introduced by the multiplier lookup approach are on average between 2 and 9
percent and are comparable with results from regression models. This chapter
interprets the findings and explains how these results fit each research objective.
A sample application of the multiplier lookup procedure is provided. Conclusions

and recommendations for future research are also included in this chapter.

107

Discussion of the Results

Objective 1. Describe Variations in Multipliers

Economic multipliers vary across different industry sectors. The values of
multipliers in one sector could be several times higher than in another sector
(Tables 4-1, 4-2). Overall, multipliers for service sectors were higher than for
manufacturing sectors. This is caused by the fact that service sectors are more
labor intensive than manufacturing sectors. More workers are required to deliver
a fixed amount of sales in service sectors compared to manufacturing sectors,
where a majority of the direct cost goes to purchases of raw materials and
equipment. As a result, more income will result from sales in service sectors.
The higher worker income is in a region, the more purchases employees will
make in the local region. And these local purchases will result in more money
being re-circulated in the region, higher induced effects, and higher multipliers.

To capture spending on different items and to reflect the variations in
multipliers across sectors, this study recommends the use of sector-specific
multipliers. Sector-specific multipliers can reduce aggregation errors and
increase flexibility in terms of applying multipliers to spending in different
categories. For example, day visitors who spend the most money on souvenirs
would have very different multiplier effects than overnight visitors who spend the
bulk of their money on lodging and restaurants. By applying an aggregate
multiplier to visitor spending, the variations in multipliers across sectors won’t be

captured.

108

Economic multipliers also vary across regions. The largest multipliers
were mostly 50 to 100 percent higher than the smallest multipliers across regions
for the same sector (Tables 4-3 to 4-6). Therefore, by applying the state
multipliers to a rural region, the impacts can be overestimated by as much as 100
percent. The coefficients of variations were between 5 and 20 percent for most
multipliers, which means the ranges for all multipliers were plus or minus 10
percent to 40 percent of the means in 95 out of 100 times. The variations in
multipliers across regions show the importance of using multipliers that are
appropriate and can reflect the region’s characteristics. The results from
objective 1 show why there is a need for better understanding of the different
types of regions and for a system to adopt multipliers for a particular type of

regions.

Objective 2. Identihr prfldormt Explain Variations in Multipliers
Population and population density explain more than 75 percent of the
variations in multipliers for sales, income and value added. The model explains
about 52 percent of the variations in job multipliers. For all four models,
population (in logarithmic value) is the most significant predictor. Multipliers will
increase with population (except for job multipliers, which are negatively
correlated with population) in a nonlinear relation. As the population grows, the
multipliers will increase at a slower speed and will stop at certain thresholds,
even when the region’s population continues to grow. This finding supports

Mulligan and Gibson’s study (1984) in which they found that the economic base

109

multiplier grows at a decelerated rate and stops at a certain value as the region’s
employment increases.

The results of this analysis are encouraging based on the fact the all
models explain variations in multipliers well, and the average errors are 3 to 6
percent for different multipliers. These results provide two useful pieces of
information. First, population alone can explain variations in multipliers well. As
regions are formed into groups based on multipliers, population within each
group should be homogeneous. These regional characteristics can be used to
describe types of regions within each group and used as guidelines for choosing
multipliers for a given region. Second, the results of regression models can be
used as benchmarks to evaluate the accuracy of the multiplier lookup approach
as proposed in this study. A comparison between the results of the multiplier
lookup procedure and the regression model approach can help evaluate whether

the multiplier lookup procedure is an accurate means of choosing multipliers.

Objective 3. Propose and Evaluate Procedures for Predicting Multipliers

All regions were formed into four groups each with distinct multipliers.
Multipliers for three economic measures (sales, income and value added) were
highly correlated. Multipliers for these three measures show reasonable
homogeneity within each group. All regions (except for one or two from each
group) fell within 10 percent of the group’s average.

The results from this analysis show that the multiplier lookup approach for

estimating economic impacts may be adequate for most recreation and tourism

110

applications. Errors introduced by “generic” multipliers are on average between
2 and 5 percent for sales, income and value added multipliers and at most 10
percent compared to multipliers that were estimated from l-O models for the
regions. The average errors are also similar to multipliers that are estimated
from regression models. The possible errors that may result from using the
proposed generic sector specific multipliers are considered to be less than using
state multipliers in a rural area or borrowing multipliers from one region to
another.

Some regions experienced job multipliers that were more than 15 percent
lower than their primary groups because of inconsistency in job multipliers with
other economic measures. Problems in grouping job multipliers were expected
because job estimates are confounded by part-time and seasonal workers and
thus would not be as highly correlated with sales as other economic measures.
This problem can be reduced by splitting out a separate sub-group of regions

with low job multipliers.

Applications of the Proposed Multiplier Lookup Procedure

Three components are needed to estimate economic impacts of visitor
spending on a given region- total visitation, average spending per visit, and
economic multipliers for the region. This study proposes a multiplier lookup
procedure for selecting economic multipliers for a given region. Four sets of
"generic" multipliers each representing a type of region of certain size and

economic diversity are provided. There are three steps for carrying out the

111

impact analysis of visitor spending on a local area. Estimate of economic

impacts of visitor spending to Lake Mendocino in California is used to illustrate

the procedure. Lake Mendocino is a US. Army Corps of Engineers recreation

project and is located about 100 miles north of San Francisco.

1. Estimate total visitor spending. Lake Mendocino attracts 520,000 day

visits and 40,000 overnight visits a year (in person trips). Visitors to

this lake spend an average of $15 per person per trip for day visitors

and $65 for overnight visitors in the local area.16 By multiplying an

average visitor spending for each segment by total number of visits for

each segment, total spending for these 560,000 visits is estimated at

$10.4 million a year (Table 5-1).

Table 5-1. Sample Computation of Total Visitor Spending, Lake Mendocino Area

 

 

 

Day Visitor Overnight Visitor Total
Number of Visits 520,000 40,000 560,000
(person trip)
Spending Category
Lodging 5 0 $ 20 $800,000
Eating and Drinking 3 12 $2,040,000
Recreation 2 3 $1,160,000
General Merchandise 10 30 $6,400,000
Total $ 15 $ 65 $10,400,000

 

 

‘6 The average party size is 2.8 people for day visitors and 3 for overnight visitors. The average
length of stay is 2.5 nights for overnight visitors. Thus, the corresponding per party per night
spending is $42 for day users and is $78 for overnight users.

112

2. Select multipliers for the region based on regional characteristics
shown in the lookup table. Table 5-2 shows sample multipliers for
Lake Mendocino areas that were estimated from the lookup approach17
and from an actual IMPLAN l-O model for this region. Using Table 4-
15, users can select the type of region represented by the study area.
In this case, Lake Mendocino is located in Mendocino County in
California. The county’s population is 85,216 (in 2000 US. Census)
with no city over 20,000 people. The Lake Mendocino area is

therefore classified into the small metro area category.

Table 5-2. Multipliers for Lake Mendocino Area

 

 

 

 

Generic Multipliers for Actual IMPLAN
Small Metro Area“ ModeiMuItipIiersb

Direct Direct

Jobs/ Type II Type II Jobsl Type II Type II
Sector Sales sales Jobs Sales sales Jobs
Hotels And Lodging Places 23.41 1.52 31.73 22.60 1.51 30.87
Eating and Drinking 31.51 1.43 37.89 31.11 1.42 37.66
Amusement And
Recreation 32.07 1 .46 39.26 29.73 1 .46 37.22
Retail Trade 30.23 1 .38 36.20 12.30 1.40 36.13
Manufacturing 10.93 1.40 16.56 13.07 1.38 18.72

 

a. Complete sets of sector-specific multipliers for the four types of regions are given in
Tables D-1 to D4 in Appendix D.

b. Multipliers were estimated using IMPLAN with 1996 data.

 

‘7 Complete sets of sector-specific multipliers for the four types of regions are given in Tables D-1

to D4 in Appendix D.

113

3. Compute economic impacts of total visitor spending. Economic
impacts can be estimated by multiplying the direct sales effects by
corresponding sector-specific multipliers. Spending on manufactured
goods from Table 5-1 is first separated into margins and producer
prices, and the producer price is further divided into local production
and imports (column two in Table 5-3).18 Only local margins and local
production are included as direct sales effects (column three in Table
5-3). These direct sales effects are then multiplied by the generic
multipliers for small metro area in Table 5-2 to estimate jobs and sales

effects (columns 4 to 6 in Table 5-3).

Table 5—3. Sample Computation of Visitor Spending Impacts

 

 

 

 

Sector Total Direct Sales Direct Jobs Total Sales Total Jobs
Spending ($MM) ($MM)
($MM)
Lodging $ 0.80 $ 0.80 19 $ 1.22 25
Eating and Drinking $ 2.04 $ 2.04 64 $ 2.92 77
Recreation $ 1.16 $ 1.16 37 $ 1.69 46
Retail Trade $ 2.20 $ 2.20 67 $ 3.04 80
Local Production $ 1.30 $ 1.30 14 $ 1.82 22
Immrts § 2.90 § - - § - -
Total $10.40 $ 7.50 201 $10.68 249

 

In the above example, 72% of the $10.4 million visitor spending is

captured as direct sales by the region’s economy (column three). These sales

 

1° An example of extracting retail margins on purchases of good is given in Table 2-6 in Chapter
2. See ‘Estimating National Park visitor spending and economic impacts; The MGM2 Model,”
(Stynes et a., 2000-p 4.10) for details on settings for local production and retail, wholesale
margins.

114

support 201 direct jobs (column four). With multiplier effects, visitor spending in
this region results in $10.68 million total sales (column five) and supports 249
jobs (column six). Should multipliers computed from an l-O model for the region
have been used (columns 5 to 7 in Table 5-2), the same spending would have
resulted in $10.8 million total sales and supported 249 total jobs. The errors
made by using generic multipliers instead of multipliers specific to the region are

around 2 to 3 percent in this case.

Conclusions

Sector-specific multiplier approaches are the compromise between
accuracy and practical issues when conducting EIA of visitor spending. By using
the sector-specific multipliers, analysts can get reliable results of economic
impacts of recreation and tourism without directly using an l-O model.

Multipliers vary across regions and the magnitude of multipliers increase
with increasing economic diversity and size. Sales, income and value added
multipliers are highly correlated with each other and are positively correlated with
population in the region while job multipliers are negatively correlated with the
region’s population. Four sets of generic sector-specific multipliers were
developed to capture variations in multipliers across regions and sectors.

The evaluation of the multiplier lookup procedure shows this approach can
provide reasonably accurate estimates of multipliers for a given region. The
“multiplier lookup” approach developed in this study will serve as a tool to help

tourism analysts choose appropriate multipliers for a given region and

115

application. Although the proposed “multiplier lookup” approach and generic
multipliers have been developed for the National Park Service’s Money
Generation Model, the multiplier lookup approach can be used in estimating
economic impacts of recreation visitor spending for many applications. Currently,
this approach is also used in the Michigan Tourism Economic Impact Model and
in the US Army Corps of Engineers’ Recreation Economic Assessment System.
With modification of the sectors, the proposed multiplier lookup approach could
also be used by tourism analysts who are interested in special market segments
(I.e., add a boat building and repair sector for boaters). In addition, this quick and
cost effective approach for selecting multipliers will allow tourism analysts to
conduct economic impact analysis without directly using an l-O model while still
benefiting from the detailed direct effects and reliable aggregate secondary
effects. This approach can substantially reduce costs of public and private sector
agencies’ efforts to assess economic implications of any current and proposed

policies and plans.

Recommendations for Future Study
This study examined multipliers that were generated from IMPLAN models
and evaluated the feasibility of a simple “multiplier lookup” procedure. The
analysis and findings suggest five related topics for further research.
1. Include multipliers for different sectors to accommodate other
recreation and tourism analysis. This study has focused on variations

in multipliers for the four primary tourism sectors, lodging, restaurant,

116

recreation, and retail trade. However, tourism analysts also need to
assess the economic impacts from visitors of special market segments
or of constructions and operations. Similar analysis can be done for
other sectors such as the Water Transportation sector (for spending on
boating and cruising), the Maintenance and Repair of Facilities sector
(for maintenance and operational costs), and the Boating and
Repairing sector (for purchases of new boats) to accommodate other
types of recreation and tourism impacts. To include multipliers for
other sectors can also help understand how these multipliers vary

across regions for different sectors.

. Extend analysis to additional regions. The selections of study regions
for this study were limited by the availability of datasets. Michigan
State University has purchased the datasets from the Minnesota
IMPLAN Group for a project to revise USDI National Park Service’s
Money Generation Model. Further study can include additional regions
from other states to test the generalizability of the methods used in this
study. Multipliers can be extracted from other regions to see if
variations in multipliers across regions have similar patterns, and to
see whether regions can be formed into similar types of groups as

found in this study.

117

3. Study how multipliers change over time. Multipliers may change over
time because of price changes or structural changes in economy (e.g.,
new industries move in to the region). This change may be small and
the multipliers may not need to be updated very often. The USDC
Bureau of Economic Analysis, for instance, only publishes RIMS ll
multipliers every five years. This is the case for sales, income, and
value added multipliers. Job multipliers, however, will probably need to
be updated more frequently since the increase of wages and salary
over time will directly result in lower job per sales ratios. To know how
multipliers change over time can reduce the efforts and frequency on
updating multipliers and will allow adjustments to be made when

multipliers and spending data are from different years.

4. Disaggregation of tourism sectors used in current l-O models. l-O
models such as IMPLAN generally do not have sectors for many
different recreation and tourism businesses. The Hotel and Lodging
Places sector in IMPLAN, for instance, includes large and small hotels,
bed and breakfasts, camp grounds and all other commercial lodging
places. As the operation for a small bed and breakfast will not be the
same as for a nationwide-chain hotel, the production functions of these
two businesses will be very different (proportions of commodities used
to produce one dollar of sales). These differences will result in

different indirect effects and thus different multipliers. Therefore, the

118

industry average ratios and multipliers may not reflect the recreation
and tourism part of the industry well. The new North American
Industry Classification System (NAICS) is in the transition of replacing
the old Standard Industry Classification (SIC) system. There are nine
new service sectors and 250 new service industries being recognized
in the new NAISC, which Include casinos, casino hotels and other
tourism and recreation industries (USDC Census Bureau, 2001). To
incorporate the new NAICS industries or other recreation and tourism
sectors into l-O models can help identify the relationships between
these industries and better estimate the indirect effects of recreation

and tourism.

. Evaluate the accuracy of induced effects estimated by IMPLAN and
other l-O models. Since recreation and tourism industries are labor
intensive, a high percentage of sales goes into wages and salaries that
makes induced effects very important. However, high numbers of part-
time and seasonal employees may distort the estimates of induced
effects. IMPLAN is one of the most used input-output systems to
estimate economic impact analysis of tourism and recreation. Chapter
2 reviews several studies that compare IMPLAN multipliers with other
systems, and one of the findings is that IMPLAN tends to overestimate
induced effects for service sectors. However, all of those comparisons

were made based on the old IMPLAN Type III multipliers and IMPLAN

119

has since changed its multipliers to Type II and to the current Type
SAM multipliers for capturing the induced effects. Although the Type
SAM multipliers are considered an improvement over IMPLAN’s earlier
Type II and Type III multipliers, it would be particularly useful to know
how accurate are the estimated induced effects. It is also useful to
know how the current IMPLAN multipliers compare to the old
multipliers and multipliers estimated by alternative systems so analysts
can make adjustments when comparing findings from different reports

or want to borrow multipliers from other study.

120

APPENDICES

121

APPENDIX A

STUDY REGIONS AND CORRESPONDING PARKS AND CITIES

Table A. Study Regions and Corresponding Parks and Cities

 

 

Region Number NPS Parks/MSA'S in the region“ Population Area
Name” of (000’s) (square
counties miles)
Caiifomia
CA96 - (State model) 31878 155973
SantaMNRA 2 Santa Monica NRA/ Los Angeles MSA, 9842 5906
Ventura MSA
LA-MSA 1 Los Angeles MSA 9128 4060
RiversideMSA 2 Joshua Tree NM/ Riverside-San 3016 27270
Bernardino MSA
SanDiegoMSA 1 San Diego MSA 2655 4205
OrangeCyMSA 1 Orange County MSA 2637 790
OaklandMSA 2 Oakland MSA 2210 1458
SanFranciscoMSA 3 Golden Gate NRA, etc/ San Francisco 1655 1016
MSA
SanJoseMSA 1 San Jose MSA 1600 1291
SacramentoMSA 3 Sacramento MSA 1482 4082
KingsCanyonNP 2 Kings Canyon NP 1101 10788
FresnoMSA 2 Fresno MSA 862 8102
SanFranciscoCo 1 Golden Gate NRA, etc 735 47
VenturaMSA 1 Ventura MSA 715 1846
BakersfieIdMSA 1 Bakersfield MSA 623 8142
StocktonMSA 1 Stockton MSA 533 1399
NapaMSA 2 Vallejo-Fairfield-Napa MSA 482 1582
SantaRosaMSA 1 Santa Rosa MSA 421 1576
ModestoMSA 1 Modesto MSA 416 1495
SantaBarbaraMSA 1 Santa Barbara MSA 386 2739
\ftsaliaMSA 1 \flsalia-Tulare-Porterville MSA 350 4824
SalinasMSA 1 Salinas MSA 339 3322
LassenV-NP 4 Lassen Volcanic NP/ Redding MSA 268 13848
SantaCruzMSA 1 Santa Cruz MSA 238 446
SanLuisMSA 1 San Luis MSA 229 3305
ChicoMSA 1 Chico—Paradise MSA 193 1640
MercedMSA 1 Merced MSA 192 1929
YosemiteNP 4 Yosemite NP 189 8870
ReddingMSA 1 Whiskeytown Shasta-Trinity NRA/ 162 3786
Redding MSA
RedwoodNP 2 Red Wood NP 150 4581
YoloMSA 1 Yolo MSA 150 1012
YubaMSA 2 Yuba CityMSA 137 1233

 

122

Table A. (cont’d)

 

 

Region Number NPS Parks/MSA's in the regionb Population Area
Name' of (000's) (square
counties miles)
MendocinoCo 1 (single county region) 83 3509
LavaBedsNM 2 Lava Beds NM 54 10232
SiskiyouCo 1 (single county region) 44 6287
DeINorteCo 1 (single county region) 27 1008
GlennCo 1 (single county region) 26 1315
DeathVaIleyNP 1 Death Valley NP 18 10192
ModocCo 1 (single county region) 10 3944
Colorado
0096 - State model 3823 103729
DenverMSA 5 Denver MSA 1867 3761
RockyMtNP 3 Rocky Mountain NP 489 5194
COSpringsMSA 1 Florissant Fossil Beds NM/ Colorado 473 2127
Springs MSA
BoulderMSA 1 Boulder MSA 258 743
FtCollinsMSA 1 Fort Collins MSA 222 2601
GreeleyMSA 1 Greeley MSA 152 3993
PueblosMSA 1 Pueblos MSA 131 2389
GDJunctionMSA 1 Colorado NM/ Grand Junction MSA 108 3328
CurecantiNRA 2 Curecanti NRA, Black Canyon of the 42 5480
Gunnison NM
MontroseCo 1 Black Canyon of the Gunnison NM 30 2241
MesaVerdeNP 1 Mesa Verde NP 22 2037
AlamosaCo 1 Great Sand Dunes NM 14 723
DinosaurNP 1 Dinosaur NP 12 4743
BentsNM 1 Bent's Old Fort NHS 5 1514
Florida
FL96 - State model 14400 53937
EvergladesNP 3 EvergladesNP, Biscanye NP, Dry 2345 4967
Tortugas NP/ Miami MSA
TampaMSA 4 Tampa-St. Pertersburg-Clear Water MSA 2199 2555
MiamiMSA 1 Miami MSN Biscanye NP 2076 1945
FtLauderdaleMSA 1 Ft Lauderdale MSA 1438 1209
OrlandoMSA 4 Orlando MSA 1417 3491
JacksonvilleMSA 4 Ft. Caroline NM, Castillo de San Marcos 1009 2636
NM, Ft. Matanzas NM/ Jacksonville MSA
WPalmB—MSA 1 West Palm Beach-Boca Raton MSA 993 1974
SarasotaMSA 2 DeSoto NM/ Sarasota-Bradenton MSA 529 1313
DaytonaMSA 2 Daytona MSA 456 1591
MelbourmeMSA 1 Melbourme-Titusville-Palm Bay MSA 454 1019
LakelandMSA 1 Lakeland-Winter Haven MSA 441 1875
PensacolaMSA 2 Gulf Islands NS/ Pensacola MSA 386 1679
FtMyersMSA 1 Fort Myers-Cape Coral MSA 380 804
FtPierceMSA 2 Fort Pierce-Port St. Lucie MSA 287 1128

 

123

Table A. (cont’d)

 

 

Region Number of NPS Parks/MSA's in the region2 Population Area
NameII , counties (000's) (square
miles)
TallahasseeMSA 2 Tallahassee MSA 259 1183
OcalaMSA 1 Ocala MSA 230 1579
GainesvilleMSA 1 Gainesville MSA 197 874
NaplesMSA 1 Naples MSA 188 2026
FtWaItonB-MSA 1 Ft. Walton Beach MSA 166 936
PanamaMSA 1 Panama MSA 145 764
PuntaGordaMSA 1 Punta Gorda MSA 130 694
JacksonCo 1 (single county region) 45 916
OkeechobeeCo 1 (single county region) 31 774
HolmesCo 1 (single county region) 18 483
TaylorCo 1 (single county region) 18 1042
DixieCo 1 (single county region) 12 704
Massachusetts
MASS96 - State model 6092 7838
GrBoston° 4 Massachusetts part of Boston MSA"/ 2695 1282
Boston area NHS’s and NHP’s
MiddlesexCo 1 Longfellow NHS, Lowell NHS, Minute 1413 824
Man NHP
Boston 1 Boston African American NHS, Boston 645 59
NHP
NorfolkCo 1 Adams NHS, F L Olmsted NHS, J F 637 400
Kennedy NHS
WorcesterCo 1 (single county region) 720 1513
EssexCo 1 Salem Maritime NHS, Saugus Iron Works 687 498
NHS
SpringfieldMSA 2 Springfield MSA/ Springfield NHS 592 1148
SpringfieIdNHS 1 Springfield NHS 442 619
CapeCodNS 1 Cape Cod NS/ Bamstable-Yarmouth 202 396
MSA
PittsfieIdMSA 1 Pittsfield MSA 135 931
FranklinCo 1 (single county region) 71 702
Michigan
Ml96 - State model 9594 56809
DetroitMSA 6 Detroit MSA 4318 3897
GrRapidsMSA 4 Grandrapids-Muskegon-Holland MSA 1015 2759
AnnArborMSA 3 Ann Arbor MSA 530 2029
LansingMSA 3 Lansing-East Lansing MSA 448 1707
Kzo-BtCk-MSA 3 Kalamazoo-Battle Creek MSA 444 1882
FlintMSA 1 Flint MSA 436 640
SaginawMSA 3 Saginaw-Bay City-Midland MSA 403 1775
MuskegonCo 1 (single county region) 165 509
BentonHarborMSA 1 Benton Harbor MSA 161 571
1

JacksonMSA

Jackson MSA

155

707

 

124

Table A. (cont’d)

 

 

Region Number of NPS Parks/MSA’S in the regionb Population Area
Name“ counties (000's) (square
miles)

SIeepBD-NP 3 Sleeping Bear Dunes NL 125 1612
Straits 4 (Mackinac Strait area counties) 99 3766
E-UP 3 (Eastern Upper Peninsula counties) 55 3486
lsleRoyaleNP 2 Isle Royale NP, Keweenaw NHP 38 1553
AlpenaCo 1 (single county region) 31 574
DickinsonCo 1 (single county region) 27 766
RoscommonCo 1 (single county region) 23 521
AntrimCo 1 (single county region) 21 477
PicturedRocksNP 1 Pictured Rocks NL 10 918
LakeCo 1 (single county region) 10 568

 

a. All models were named to indicate which of the above four categories they belonged
to (i.e., model names ended with “MSA" indicate metropolitan statistical areas; model
names ended with “NP” indicate National Park regions; model names ended with “CO”
indicate single or multiple county regions). .

b. MSA: Metropolitan Statistical Area; NP: National Park; NM: National Monument; NRA:
National Recreation Area; NHS: National Historical Site; NHP: National Historical Park,
NL: National Lakeshore; NS: National Seashore.

c. The BEA defined Boston-Worcester-Lawrence-LowelI-Brocktn MSA model was not
constructed because data sets for counties in New Hampshire were not available.

125

VARIATIONS IN MULTIPLIERS FOR EIGHT OTHER SECTORS

Table B-1. Regional Variation in Sales Multipliers, Other Sectors

APPENDIX B

 

 

Sector Mean Std. Coef. of Median Maximum Minimum Range
Dev. Variation

Type ll Sales Multiplier

Wholesale Trade 1.43 0.10 7% 1.45 1.63 1.15 0.49

Auto Repair and

Services 1.43 0.11 8% 1.44 1.88 1.13 0.74

Local

Transportation 1 .49 0.1 1 8% 1.50 1 .74 1.16 0.58

Food Processing 1.48 0.09 6% 1.47 1.69 1.26 0.43

Apparel 1 .45 0.1 1 8% 1.44 1 .72 1.15 0.57

Petroleum

Refining 1.34 0.13 10% 1.28 1.61 1.15 0.46

Sporting Goods 1.50 0.11 7% 1.51 1.73 1.18 0.54

Manufacturing 1 .48 0.08 6% 1.48 1 .67 1.28 0.39

Type I Sales Multiplier

Wholesale Trade 1.20 0.05 4% 1.20 1.29 1.08 0.21

Auto Repair and

Services 1.24 0.07 5% 1.24 1 .54 1.08 0.46

Local

Transportation 1.17 0.05 4% 1.17 1.28 1.07 0.21

Food Processing 1.34 0.07 5% 1.34 1.52 1.20 0.32

Apparel 1 .28 0.08 6% 1.26 1 .45 1.09 0.36

Petroleum

Refining 1.28 0.1 1 9% 1.24 1.53 1.12 0.41

Sporting Goods 1.32 0.08 6% 1.33 1.44 1.11 0.34

Manufacturing 1.30 0.05 4% 1.30 1.42 1.18 0.24

 

126

Table B-2. Regional Variation in Job multipliers, Other Sectors

 

 

Sector Mean Std. Coef. of Median Maximum Minimum Range
Dev. Variation

Type II Job multiplier

Wholesale Trade 17.70 2.80 16% 17.45 24.24 10.77 13.47

Auto Repair and

Services 18.49 1.49 8% 18.58 21.36 14.03 7.33

Local

Transportation 35.79 5.94 17% 34.31 58.58 24.52 34.06

Food Processing 1 1.69 1.62 14% 1 1.39 16.53 8.32 8.21

Apparel 19.33 2.01 10% 19.55 25.80 13.68 12.12

Petroleum

Refining 3.37 0.97 29% 3.24 5.24 1.72 3.52

Sporting Goods 17.17 3.40 20% 17.07 28.22 10.65 17.57

Manufacturing 16.73 1.89 1 1% 16.96 21.94 10.48 1 1 .46

Direct Effect Ratio

Wholesale Trade 11.45 2.95 26% 10.94 19.69 6.28 13.40

Auto Repair and

Services 12.52 1.31 10% 12.47 17.22 9.99 7.23

Local

Transportation 28.81 6.08 21 % 27.24 52.57 20.98 31 .59

Food Processing 5.58 0.75 14% 5.60 7.61 3.68 3.93

Apparel 13.39 1.97 15% 13.51 21.47 7.66 13.80

Petroleum

Refining 0.61 0.03 5% 0.61 0.67 0.54 0.13

Sporting Goods 10.85 2.97 27% 10.70 21.19 5.70 15.49

ManufacturinL 10.55 1.50 14% 10.64 15.13 6.71 8.43

 

127

Table B-3. Regional Variation in Income Multipliers, Other Sectors

 

 

Sector Mean Std. Coef. of Median Maximum Minimum Range
Dev. Variation

Type ll income Multiplier

Wholesale Trade 0.57 0.04 8% 0.57 0.65 0.45 0.20

Auto Repair and

Services 0.46 0.05 12% 0.47 0.66 0.31 0.36

Local

Transportation 0.76 0.07 9% 0.77 0.86 0.55 0.31

Food Processing 0.31 0.05 17% 0.31 0.42 0.17 0.25

Apparel 0.41 0.10 24% 0.40 0.66 0.18 0.47

Petroleum

Refining 0.14 0.04 27% 0.13 0.21 0.08 0.13

Sporting Goods 0.42 0.06 15% 0.43 0.55 0.23 0.31

Manufacturing 0.41 0.05 12% 0.41 0.51 0.28 0.23

Direct Effect Ratio

Wholesale Trade 0.40 0.00 1% 0.41 0.41 0.39 0.01

Auto Repair and

Services 0.31 0.02 6% 0.31 0.34 0.24 0.10

Local

Transportation 0.57 0.03 5% 0.58 0.61 0.44 0.18

Food Processing 0.14 0.04 27% 0.14 0.24 0.04 0.20

Apparel 0.25 0.07 30% 0.24 0.48 0.03 0.45

Petroleum

Refining 0.05 0.01 26% 0.05 0.08 0.02 0.06

Sporting Goods 0.23 0.05 20% 0.24 0.31 0.08 0.24

Manufacturing 0.23 0.03 15% 0.23 0.31 0.12 0.20

 

128

Table B-4. Regional Variation in Value Added Multipliers, Other Sectors

 

Sector Mean Std. Coef. of Median Maximum Minimum Range
Dev. Variation

 

Type ii Value Added Multiplier

Wholesale Trade 0.96 0.07 7% 0.96 1.08 0.77 0.32
Auto Repair and

Services 0.74 0.08 11% 0.76 0.87 0.51 0.36
Local

Transportation 0.98 0.09 10% 0.99 1 .14 0.70 0.45
Food Processing 0.56 0.10 17% 0.56 0.75 0.29 0.45
Apparel 0.55 0.13 23% 0.55 0.86 0.28 0.58
Petroleum

Refining 0.34 0.10 29% 0.31 0.53 0.17 0.36
Sporting Goods 0.74 0.11 15% 0.76 0.96 0.42 0.54
Manufacturing 0.68 0.08 12% 0.68 0.85 0.47 0.39
Direct Efiect Ratio

Wholesale Trade 0.69 0.00 1% 0.69 0.69 0.67 0.02
Auto Repair and

Services 0.49 0.03 6% 0.49 0.54 0.38 0.16
Local

Transportation 0.68 0.04 5% 0.68 0.72 0.52 0.21
Food Processing 0.28 0.08 27% 0.28 0.46 0.07 0.39
Apparel 0.29 0.09 30% 0.28 0.57 0.03 0.54
Petroleum

Refining 0.14 0.04 26% 0.14 0.22 0.06 0.16
Sporting Goods 0.44 0.09 20% 0.45 0.59 0.14 0.45

Manufacturing 0.39 0.06 15% 0.39 0.54 0.21 0.34

 

129

APPENDIX C

REGIONAL CHARACTERISTICS AND MULTIPLIERS BY TYPE OF REGION

Table C-1. Rural Regions (N=19)

 

 

 

 

Population Multiplierc

Model“ Population densityb Sales II Jobs ll Income ll VA II
BentsNM 5,478 4 1 .18 36.28 0.42 0.65
DixieCo 12,352 18 1.24 37.05 0.45 0.71
GlennCo 26,202 20 1 .25 33.06 0.46 0.72
HolmesCo 18,174 38 1.25 34.53 0.46 0.73
PicturedRocksNL 9,971 11 1.26 41.51 0.45 0.71
TaylorCo 18,173 17 1.26 35.16 0.47 0.74
LakeCo 9,874 17 1.27 38.94 0.44 0.71
AntrimCo 20,595 43 1 .32 37.50 0.47 0.76
ModocCo 9,693 2 1 .32 43.08 0.47 0.76
DelNorteCo 26,947 27 1 .33 35.43 0.49 0.77
Eastern Upper

Peninsula 54,565 16 1 .33 33.88 0.50 0.79
RoscommonCo 22,847 44 1 .34 39.46 0.48 0.77
DickinsonCo 27,285 36 1.35 41.16 0.48 0.76
AlpenaCo 30,746 54 1 .36 42.1 1 0.47 0.74
lsleRoyaleNP 38,240 25 1 .36 42.62 0.47 0.74
OkeechobeeCo 30,894 40 1 .37 33.24 0.53 0.82
DeathVaIleyNP 18,433 2 1 .39 33.01 0.53 0.85
AlamosaCo 14,300 20 1 .40 38.29 0.51 0.81
DinosaurNP 12,086 3 1 .40 37.72 0.49 0.81
Average 21,413 23 1.31 37.58 0.48 0.76

 

a. Sorted by Type II sales multipliers.

b. People per square mile.
c. All multipliers are aggregate tourism multipliers as defined in chapter 3.

130

Table C-2. Small Metro Regions (N= 34)

 

 

 

 

Population Multiplierc
Model“ Population density” Sales II Jobs II Income II VA II
Muskegon 164,913 324 1.38 36.79 0.51 0.79
Boston 645.068 1 1027 1 .39 24.42 0.58 0.91
JacksonCo 44.728 49 1 .42 37.39 0.51 0.83
LavaBedsNM 53,886 5 1 .42 38.26 0.52 0.84
MercedMSA 192,311 100 1.42 36.11 0.52 0.82
MesaVerdeNP 21.999 1 1 1.42 36.18 0.51 0.82
Mackinac Strait area 99,248 26 1 .42 34.71 0.55 0.85
F lintMSA 436,128 682 1 .43 36.08 0.54 0.84
FranklinCo 71 ,209 101 1.43 35.54 0.54 0.85
JacksonMSA 154.563 219 1.43 38.05 0.54 0.84
SyskiyouCo 44,193 7 1 .43 38.08 0.53 0.86
YosemiteNP 189,043 21 1 .43 32.86 0.55 0.87
YubaMSA 136,555 111 1.44 35.19 0.55 0.86
MelbourmeMSA 453.998 446 1.44 33.17 0.56 0.88
GainesvilleMSA 196,525 225 1 .45 37.00 0.54 0.86
MendocinoCo 83,298 24 1 .45 35.04 0.55 0.86
FtWaltonB-MSA 165,873 177 1 .45 33.55 0.56 0.88
YoloMSA 149,925 148 1.45 33.01 0.55 0.86
CurecantiNRA 41,749 8 1.46 36.51 0.53 0.86
RedwoodNP 149,970 33 1.46 36.54 0.54 0.85
VisaliaMSA 349.922 73 1 .46 33.96 0.55 0.87
AnnArborMSA 529.898 261 1 .47 33.72 0.57 0.89
GreeleyMSA 152,189 38 1 .47 38.48 0.54 0.84
Kzo—BtCk-MSA 444.428 236 1 .47 37.31 0.56 0.87
MontroseCo 29,601 13 1 .47 38.04 0.53 0.86
PueblosMSA 131,217 55 1.47 38.61 0.55 0.86
PuntaGordaMSA 130.426 188 1 .47 33.73 0.57 0.90
BentonHarborMSA 161 .434 283 1 .48 38.68 0.55 0.87
TallahasseeMSA 259,380 219 1 .49 36.38 0.56 0.89
BoulderMSA 258.234 348 1 .49 33.08 0.58 0.91
PanamaMSA 144,637 189 1.49 33.81 0.57 0.91
SanLuisMSA 229,437 69 1 .50 36.26 0.58 0.90
LansingMSA 447,538 262 1.51 37.54 0.56 0.89
gginawMSA 403,301 227 1 .51 38.30 0.56 0.89
Average 210,789 477 1 .45 35.66 0.55 0.86

 

a. Sorted by Type II sales multipliers.

b. People per square mile.
0. All multipliers are aggregate tourism multipliers as defined in chapter 3.

131

Table C-3. Large Metro Regions (N= 44)

 

 

 

Population Multiplierc
Model“ Population density” Sales II Jobs II Income II VA II
SanJoseMSA 1 .599.604 1239 1 .43 27.01 0.59 0.91
SanFranciscoCo 735.315 15746 1 .45 24.58 0.60 0.94
NapaMSA 482.048 305 1 .48 31 .56 0.58 0.91
ModestoMSA 415.786 278 1.51 34.66 0.57 0.90
NaplesMSA 188,187 93 1 .51 28.68 0.62 0.97
COSpringsMSA 472,924 222 1 .52 34.53 0.59 0.92
DaytonaMSA 456,464 287 1 .52 35.09 0.59 0.93
StocktonMSA 533.392 381 1.52 35.1 1 0.58 0.91
RiversideMSA 3,015,783 1 11 1.52 31.70 0.60 0.93
SalinasMSA 339.047 102 1.52 29.99 0.61 0.95
BakersfieldMSA 622.729 76 1 .53 35.40 0.58 0.92
LassenV—NP 267,876 19 1 .53 34.80 0.58 0.93
OcalaMSA 230.068 146 1 .53 35.64 0.58 0.92
SantaCruzMSA 237.821 534 1 .53 32.90 0.60 0.94
CapeCodNS 201.970 510 1.53 30.94 0.61 0.96
LakelandMSA 440.954 235 1 .54 32.01 0.61 0.95
PensacolaMSA 385.820 230 1 .54 36.07 0.59 0.93
PittsfieldMSA 134.788 145 1 .54 33.69 0.60 0.95
RockyMtNP 489.495 94 1 .54 34.53 0.59 0.93
FtMyersMSA 380,001 473 1 .54 30.91 0.62 0.97
San FranciscoMSA 1 ,655,454 1630 1 .54 24.94 0.64 1.00
FtCollinsMSA 221.725 85 1.55 37.31 0.58 0.92
FtPierceMSA 287.255 255 1 .55 33.38 0.60 0.96
ReddingMSA 161,740 43 1.55 34.88 0.59 0.94
SleepBD-NL 125.134 78 1.55 36.39 0.59 0.93
VenturaMSA 714.733 387 1 .55 32.62 0.61 0.96
Worcester 719.545 476 1 .55 33.77 0.61 0.95
MiddlesexCo 1 .412,561 1715 1.55 28.13 0.63 0.99
ChicoMSA 192,507 1 17 1.56 37.28 0.58 0.93
FresnoMSA 861 .753 106 1 .56 34.26 0.60 0.94
GrRapidsMH-MSA 1 .015.099 368 1.56 36.68 0.58 0.91
KingsCanyonNP 1,101,194 102 1.56 34.29 0.59 0.94
SantaBarbaraMSA 385.573 141 1.56 32.02 0.62 0.97
GrBoston 2,695,017 2103 1.56 26.89 0.65 1.00
NorfolkCo 637,388 1 595 1 .57 29.13 0.64 0.99
SarasotaMSA 528.803 403 1 .57 31 .37 0.63 0.99
WPaImB-MSA 992.840 503 1.57 28.07 0.64 1.01

 

132

Table C-3. (Cont’d)

 

 

 

 

Population Multiplier’=
Model“ Population (19051th Sales II Jobs ll Income ll VA ll
MiamiMSA 2,076,175 1068 1.57 28.85 0.63 0.99
GDJunctionMSA 108.371 33 1.58 38.22 0.59 0.94
SpringfieldMSA 591,804 516 1.58 34.15 0.61 0.96
SpringfieldNHS 442.194 715 1.58 34.01 0.61 0.97
EssexCo 686,774 1379 1 .59 31 .61 0.63 0.99
OrangeCyMSA 2,636,888 3339 1 .59 29.08 0.64 1 .00
SantaRosaMSA 420.872 267 1 .60 33.23 0.63 0.98
Average 734,124 878 1 .54 32.51 0.60 0.95

 

a. Sorted by Type II sales multipliers.
b. People per square mile.

c. All multipliers are aggregate tourism multipliers as defined in chapter 3.

133

Table C-4. States and Other Large MSA’s (N= 17)

 

 

 

 

Population Multiplierc
Model’ Population dansityb Sales II Jobs II Income II VA II
OrlandoMSA 1 .417,291 406 1 .58 29.88 0.64 0.99
DetroitMSA 4,318,145 1 108 1.59 33.19 0.62 0.97
EvergladeNP 2,345,092 472 1.59 29.13 0.64 1.01
JacksonvilleMSA 1 ,008.633 383 1 .59 33.16 0.62 0.97
FtLauderdaleMSA 1 .438.228 1 190 1 .61 30.77 0.65 1 .01
SacramentoMSA 1,482,208 363 1 .61 33.43 0.63 0.99
SanDiegoMSA 2,655,463 632 1 .61 31.41 0.64 1.01
OaklandMSA 2,209,629 1516 1.63 30.00 0.65 1.01
TampaMSA 2,199,231 861 1.63 33.17 0.64 1.01
Michigan 9,594,350 169 1.64 36.08 0.62 0.98
LA-MSA 9,127,751 2248 1.65 29.66 0.66 1.03
DenverMSA 1 .866.978 496 1 .66 32.71 0.65 1.03
Massachusetts 6,092,352 777 1 .66 30.24 0.67 1.05
SantaMNRA 9,842,484 1667 1 .66 29.97 0.66 1 .04
Florida 14,399,985 267 1 .67 31.99 0.66 1.04
Caiifomia 31,878,234 204 1.71 31.06 0.67 1.06
Colorado 3,822,676 37 1 .71 34.00 0.66 1.04
Average 6,217,572 753 1.63 31.75 0.65 1.01

 

a. Sorted by Type II sales multipliers.
b. People per square mile.

c. All multipliers are aggregate tourism multipliers as defined in chapter 3.

134

APPENDIX D

SECTOR SPECIFIC MULTIPLIERS FOR THE FOUR TYPES OF REGIONS

Table D-1. Multipliers for Tourism-Related Sectors — Rural Areas

 

 

 

 

Direct effect raLio Total effect multiplier Type I

jobs value jobs value

lMM income Added IMM income added Model
Sector sales /sales lsales[ sales sales /sales lsaleSI sales RPC"
Hotels and
Ilodging places 27.48 0.30 0.46 1.36 33.59 0.43 0.69 1.24 67%
Eating and
drinking 33.88 0.33 0.46 1.30 38.87 0.43 0.64 1.18 86%
Amusement and
recreation 30.63 0.35 0.58 1.32 35.98 0.46 0.77 1.18 59%
Retail tradeb 37.11 0.51 0.80 1.27 41.81 0.60 0.96 1.10 78%
Wholesale trade 14.84 0.40 0.68 1.26 19.43 0.49 0.84 1.12 43%
Auto repair and
[service 14.21 0.28. 0.45 1.26 18.74 0.37 0.61 1.16 82%
Local
transportation 35.64 0.54 0.64 1.29 40.98 0.64 0.81 1.10 26%
Food processing 5.74 0.13 0.26 1.29 10.33 0.23 0.43 1.22 14%
Apparel 15.47 0.16 0.19 1.21 18.74 0.23 0.31 1.16 2%
Petroleum
refining 0.62 0.05 0.12 1.51 4.24 0.16 0.44 1.45 3%
Sporting goods 9.11 0.26 0.49 1.24 13.06 0.34 0.63 1.15 1%
[Manufacturing 10.74 0.22 0.39 1.30 15.66 0.33 0.56 1.20 0%

 

 

 

 

 

 

a. Regional purchase coefficient.

b. Retail is an average of seven retail trade sectors.

Note: All multipliers are computed as averages for regions in the group.

135

Table D-2. Multipliers for Selected Tourism-Related Sectors — Small Metro

 

 

 

 

 

 

Direct effect ratio Total effect multiplier Type I

jobs value jobs value

IMM income Added IMM income added Model
Sector sales Isales lsalest sales sales Isales Isales sales RPC’I
Hotels and
Ilodging places 23.41 0.33 0.51 1.52 31.73 0.53 0.84 1.32 56%
Eating and
drinking 31.51 0.35 0.48 1.43 37.89 0.50 0.75 1.24 88%
Amusement and
recreation 32.07 0.35 0.57 1.46 39.26 0.52 0.86 1.26 65%
Retail tradeb 30.23 0.51 0.80 1.38 36.20 0.65 1.04 1.14 92%
Wholesale trade 12.33 0.40 0.69 1.39 18.44 0.55 0.93 1.18 56%
Auto repair and
[service 12.66 0.31 0.48 1.40 18.63 0.45 0.72 1.23 84%
Local
transportation 30.25 0.57 0.67 1.43 37.04 0.73 0.94 1.15 45%
Food processing 5.95 0.12 0.24 1.41 11.63 0.26 0.47 1.31 39%
Apparel 14.06 0.22 0.26 1.36 19.33 0.35 0.47 1.23 4%
Petroleum
refining 0.62 0.05 0.12 1.26 3.17 0.11 0.28 1.22 7%
Sporting goods 12.45 0.21 0.40 1.41 18.32 0.36 0.64 1.27 3%
[Manufacturing 10.93 0.22 0.38 1.40 16.56 0.37 0.62 1.26 2%

 

 

 

a. Regional purchase coefficient.

b. Retail is an average of seven retail trade sectors.
Note: All multipliers are computed as averages for regions in the group.

136

Table D-3. Multipliers for Selected Tourism-Related Sectors - Larger Metro

 

 

 

Direct effect ratio Total effect multipliers Type I

jobs valueI jobs value

IMM income Added IMM income added Model
Sector sales Isales Isales: sales sales Isales IsaleSI salesl RPCaI
Hotels and
Ilodging places 19.09 0.37 0.56 1.61 27.84 0.60 0.95 1.34 63%
Eating and
drinking 28.71 0.37 0.51 1.53 35.68 0.56 0.84 1.29 88%
Amusement and
recreation 29.49 0.36 0.58 1.56 37.41 0.57 0.94 1.31 71%
Retail trade” 25.86 0.51 0.80 1.46 32.24 0.68 1.09 1.16 94%
Wholesale trade 10.20 0.41 0.69 1.48 16.88 0.59 0.99 1.22 76%
Auto repair and
Iservice 11.89 0.32 0.50 1.47 18.16 0.49 0.79 1.25 87%
Local
transportation 26.89 0.58 0.69 1.53 34.18 0.78 1.02 1.18 59%
Food processing 5.45 0.15 0.29 1.48 11.61 0.32 0.58 1.34 44%
Apparel 13.12 0.26 0.30 1.47 19.23 0.43 0.58 1.29 5%
Petroleum
refining 0.60 0.05 0.15 1.34 3.20 0.14 0.35 1.28 15%
Sporting goods 10.45 0.24 0.45 1.53 16.92 0.44 0.77 1.33 2%
Manufacturing 10.50 0.23 0.40 1.49 16.76 0.41 0.69 1.31 2%

 

 

 

 

 

a. Regional purchase coefficient.

b. Retail is an average of seven retail trade sectors.

Note: All multipliers are computed as averages for regions in the group.

137

 

Table 04. Multipliers for Selected Tourism-Related Sectors - State

 

 

 

Direct effect ratio Total effect multipliers Type I

jobs value jobs valuel

IMM income Added IMM income added Model
Sector sales Isales lsales sales sales lsales [sales sales RPC“
Hotels and
[lodging places 18.02 0.37 0.57 1.70 27.41 0.64 1.01 1.38 66%
Eating and
drinking 27.37 0.38 0.53 1.64 35.05 0.61 0.91 1.34 89%
Amusement and
recreation 26.87 0.36 0.59 1.66 35.47 0.62 1.01 1.35 77%
Retail tradeb 24.73 0.51 0.80 1.53 31.54 0.71 1.13 1.18 95%
Wholesale trade 9.10 0.41 0.69 1.57 16.40 0.63 1.04 1.26 91%
Auto repair and
service 11.95 0.31 0.50 1.57 18.80 0.53 0.84 1.31 84%
Local
transportation 24.96 0.59 0.70 1.64 32.98 0.83 1.09 1.23[ 71%
Food processing 5.37 0.15 0.30 1.57 12.11 0.36 0.63[ 1.39 53%I
Apparel 12.70 0.27 0.32 1.58 19.72 0.48 0.66 1.34 7%
Petroleum
refining 0.61 0.05 0.13 1.37 3.67 0.15 0.35 1.30 29%I
Sporting goods 9.97 0.25 0.47 1.62 17.11 0.48 0.84 1.39 4%
Manufacturing 10.08 0.24 0.41 1.59 17.02 0.46 0.76 1.36 3%

 

 

 

 

 

 

a. Regional purchase coefficient.
b. Retail is an average of seven retail trade

sectors.

Note: All multipliers are computed as averages for regions in the group.

138

BIBLIOGRAPHY

139

Bibliography

Alward, G.. & Lindall, S. (1996). Deriving SAM multiplier models using IMPLAN.
Presented at the 1996 National IMPLAN Users Conference, Minneapolis.

Archer. B. H. (1973). The impact of domestic tourism. 2”" impression. Bangor.
Wales: University of Wales Press.

Archer. B. H. (1982). The value of multipliers and their policy implications.
Tourism Management, 3: 236-241.

Archer, B. H. (1984). Economic impact: Misleading multipliers. Annals of Tourism
Research, 11: 517-518.

Archer. B. H., & Owen. C. B. (1971). Toward a tourist regional multiplier.
Regional Studies, 5(4): 289-294.

Archer. B. H., & Fletcher, J. (1996). The economic impact of tourism in the
Seychelles. Annals of Tourism Research, 23(1). 3247.

Archer. B. H. (1996). Economic impact analysis. Annals of Tourism Research.
23(3): 704-707.

Baaijens. S. R.. Nijkamp. P., & Montfort, K. V. (1998). Explanatory meta-analysis
for the comparison and transfer of regional tourist income multipliers.
Regional Studies, 32: 839-849.

Beattie. B. R.. & Leones, J. (1993). Uses and abuses of economic multipliers.
Community Development Issues. 1(2): 1-4.

Becker. D. R. (1997). Classification of Corps of Engineers Projects for economic
impact assessment. M.S. thesis. East Lansing. MI: Michigan State
University, Department of Park. Recreation 8 Tourism Resources. 126 pp.

Bergstrom. J. C., Cordell, H. K., Ashley, G. A., & Watson. A. E. (1990).

Economic impacts of recreational spending on rural areas: A case study.
Economic Development Quarterly. 4(1), 29-39.

140

Borgen, H.. & Cooke. S. (1990). The comparison of IMPLAN and RIMS ii output
multipliers for the State of Idaho. Departmental working paper.
Department of Agricultural Economics and Rural Sociology. University of
Idaho. Boise. ID.

Brucker. S. M.. Hastings. S. W.. & Latham, W. R. Ill. (1987). Regional input-
output analysis: A comparison of five ready made model systems. Review
of Regional Studies. 17(2): 1-16.

Brucker. S. M., Hastings. S.W.. & Latham, W. R. Ill. (1990). The variations of
estimated impacts from five regional input-output models. intemational
Regional Science Review. 13: 119-139.

Bull. A. (1995). The economics of travel and tourism. Melbourne, Australia:
Longman.

Bulmer—Thomas. V. (1982). Input-output analysis in developing countries:
Sources, methods, and applications. Chichester, New York: Wiley.

Burress, D. (1989). Economic impact multipliers for Kansas. Kansas Business
Review. 12(3).

Bushnell, R.. & Hyle, M. (1985). Computerized models for assessing the
economic impact of recreation and tourism. In Propst, D. B. (Ed.).
Assessing the economic impacts of recreation and tourism (pp. 46-51).
USDA. Forest Service, Southeastern Forest Experiment Station,
Asheville, NC.

Chang. W. H. (1999). Potential bias of using iMPLAN Type iII multipliers for
assessing economic impacts of tourism spending. [Online] Available
http://www.msu.edu/user/changwe4/implan/compare.htm. [July, 10. 2000].

Chang, W. H.. Jackson, R. S., & Stynes, D. J. (1999). Using regional
characteristics for development of effective tourism spending multipliers.
Unpublished manuscript. Department of Park. Recreation and Tourism
Resources. Michigan State University. East Lansing. MI.

Chang. W. H.. Perales, M. K.. & Propst, D. B. (2000). Visitor expenditures and
economic impacts by marina slip renters associated with the US Army

141

Corps of Engineers. In Proceedings of the 2000 Travel and Tourism
Research Association Conference. San Fernando Valley, CA.

Chang. W. H.. Stynes. D. J., 8 Propst. D. B. (1998). Running tourism impact
analysis in IMPLAN Pro using Mi-REC utilities. [Online] Available
http://www.msu.edu/user/changwe4/iprol, [June 10, 1998].

Chang. W. H.. Stynes. D. J., Jackson. R. S.. & Propst. D. B. (2001). US Army
Corps of Engineers Recreation Economic Assessment System. Poster

presentation at the National Natural Resources Management Conference,
2001. Portland.

Chappelle, D. E. (1985). Strategies for developing multipliers useful in assessing
economic impacts of recreation and tourism. ln Propst, D. B. (Ed.).
Assessing the economic impacts of recreation and tourism. (pp. 1-6).
USDA. Forest Service, Southeastern Forest Experiment Station.
Asheville, NC.

Chamey. A. H.. & Leones. J. (1997). IMPLAN’s induced effects identified through
multiplier decomposition. Joumal of Regional Science. 37(3): 503-517.

Clawson, M.. 8 Knetsch, J. L. (1966). Economics of outdoor recreation.
Baltimore: The John Hopkins University Press.

Crompton, J. (1999). Measuring the economic impact of visitors to sports
tournaments and special events. Ashbum, VA: The National Recreation
and Parks Association.

Deller, S. C.. Sumathi, N. R., 8 Marcouiller, D. (1993). Regional economic
models for the State of Wisconsin: An application of the Micro-IMPLAN
Modeling System. Staff Paper 93.6. Center for Community Economic
Development, Department of Agricultural Economics, University of
Wisconsin, Madison, Wisconsin.

Detomasi, D. D. (1987). The use of multipliers in planning. Plan Canada, 27: 126-
127. 143.

Duffield. J., Power, T.. Neher. C.. & Morgan. B. (1997). Review of the National
Park Service Money Generation Model. Report to National Park Service.

142

Eadington. W. R., 8 Redman, M. (1991). Economics and tourism. Annals of
Tourism Research. 18: 41-56.

English. D. B. K., Kriesel. W.. Leeworthy. V. R., 8 Wiley, P. C. (1996). Economic
contribution of recreating visitors to the Florida Keys/Key West. Final
report to Monroe County Tourist Development Council, FL.

Fletcher. J. E. (1989). Input-output analysis and tourism impact studies. Annals
of Tourism Research. 16: 514-529.

Frechtling, D. C. (1994a). Assessing the impacts of travel and tourism --
Introduction to travel impact estimation. Chapter 31 in Ritchie. J. R. B., 8
Goeldner. C. R. (Eds.), Travel tourism and hospitality research: A
handbook for managers. New York: John Wiley.

Frechtling, D. C. (1994b). Assessing the impacts of travel and tourism --
Measuring economic benefits. Chapter 32 in Ritchie. J. R. B.. 8 Goeldner.
C. R. (Eds.). Travel tourism and hospitality research: A handbook for
managers. New York: John Wiley.

Frechtling, D. C.. 8 Horvath. E. (1999). Estimating the multiplier effects of tourism
expenditures on a local economy through a regional input-output model.
Joumal of Travel Research. 37: 324-332.

Gartner. W. C., 8 Holecek. D. F. (1983). Economic impact of an annual tourism
industry exposition. Annals of Tourism Research, 10(2):199-212.

Gazel. R. C. (1998). The economic impacts of casino gambling at the state and
local levels. Annals of the American Academy of Political and Social
Science. 556: 66-84.

Griffith. D.. 8 Amrheim. C. (1997). Multivariate Statistical Analysis for
Geographers. New Jersey: Prentice Hall.

Gujarati. D. N. (1995). Basic econometrics. McGraw-Hill, Inc. New York.

Gunderson. J. 8 Kreag, G. (1991). Estimated economic impact of recreational
fishing on Minnesota waters of Lake Superior. Minnesota Sea Grant
Research Report. Minnesota Sea Grant, Duluth, MN.

143

Hastings, S. E., 8 Brucker. S. M. (1993). An introduction to regional input-output
analysis. In Otto. D.M.. 8 Johnson, T.G. (Eds.), Microcomputer-based
input-output modeling: Applications to economic development (pp. 1-27).
Boulder: Westview Press. Inc.

Hauser, D. P. (1974). Some problems in the use of stepwise regression
techniques in geographical research. Canadian Geographer, 18(2): 148-
158.

Heins, J. (1982). The Impact of 100 New Manufacturing Jobs on an Average
County in Kansas. Report prepared for the Kansas Association of
Commerce and Industry, and the Kansas Department of Development,
under the contract to the Institute for Economic and Business Research.
University of Kansas.

Hewings, Geoffrey J. D. (1985). Regional input-output analysis. Beverly Hills.
CA: Sage Publications, Inc.

Hinojosa. R. C., 8 Rios. A. J. (1991). Constructing economic base models for
developing countries: Lessons from a case study in Panama. international
Regional Science Review. 14: 149-162.

Holland. D. (1994). Coping with change: Economic impacts, what an impact
statement says. Western Rural Development Center, Utah State
University. Logan, Utah.

lsard, W. (1975). Introduction to regional science. Englewood Cliffs. NJ: Prentice
Hall.

lsard. W.. 8 Langford, T. (1971). Regional input-output study: Recollections.
reflections, and diverse notes on the Philadelphia experience. Cambridge.
MA: The MIT Press.

Jackson, R. S., Stynes. D. J., Propst. D. B.. 8 Siverts, E. (1992). Economic
impacts analysis as a tool in recreation program evaluation. Instruction
Report R-92-1. Vicksburg. MS: US. Army Engineer Watenrvays
Experiment Station.

Johnson, R. L., Obermiller, F.. 8 Radtke. H. (1989). The economic impact of
tourism sales. Joumai of Leisure Research. 21(2): 140-154.

144

Kendall. J., 8 Pigozzi, B. W. (1994). Nonemployment income and the economic
base of Michigan counties: 1959-1986. Growth and Change, 25: 51-74.

Kottke, M. (1988). Estimating economic impacts of tourism. Annals of Tourism
Research. 15: 122-133.

Krikelas, A. C. (1992). Why regions grow: A review of research on the economic
base model. Federal Reserve Bank of Atlanta Economic Review, 77: 16-
29.

Lamphear, F.. Charles. R. J., Supalla. D, Jewell, R. K.. 8 Montgomery. C. (1983).
The ADO TMA TR software program for developing and applying regional
input-Output models for economic analysis. Lincoln. Nebraska: University
of Lincoln.

Leontief. W. (1986). input-output economics. 2"" ed. New York: Oxford University
Press.

Loomis. J. 8.. 8 Walsh. R. G. (1997). Recreation economic decisions: Comparing
benefits and costs, 2"” ed. State College. PA: Venture Publishing. Inc.

Mak, J. (1989). The economic contribution of travel to state economies. Joumai
of Travel Research. 28(2): 3-5.

Miemyk, W.H. (1965). The elements of input-output analysis. New York: Random
House.

Minnesota IMPLAN Group. (2000). iMPLAN professional software, analysis, and
data guide. 2"” edition. Stillwater. MN: Minnesota IMPLAN Group, Inc.

Minnesota IMPLAN Group. (2001). IMPLAN RPC’s. IMPLAN technical paper.

[online]. Available: httpzllwww.implan.com/support/implan_documents.htm.
[August, 24, 2001].

Miller. R. E. and Blair. P. D. (1985). input-output analysis: Foundations and
extensions. Englewood Cliffs, NJ: Prentice Hall.

145

Miller, R. E., Polenske. K. R., 8 Rose. A. Z. (1989). Frontiers of input-output
analysis. New York: Oxford University Press.

Milon, W. J., Mulkey. D. W.. 8 Ellerbrock. M. J. (1982). Regional impact analysis
and recreation multipliers. The Review of Regional Studies. 12(3): 11-21.

Mulligan. G. F ., 8 Gibson. L. J. (1984). Regression estimates of economic base
multipliers for small communities. Economic Geography, 60: 225-237.

Okubo. S.. 8 Planting. MA. (1998). US. Travel and Tourism Satellite Accounts
for 1992. Survey of Current Business. July: 8-22.

Olfert, M., 8 Stabler, J. (1994). Community level multipliers for rural development
initiatives. Growth and Change, 25(Fall): 467-486.

Olson, D., 8 Lindall, S. (1993). Micro IMPLAN user’s guide version MI91-F.
Stillwater. MN: Minnesota IMPLAN Group. Inc.

Pederson, L. D. (1990). Use of iMPLAN to estimate economic impacts stemming
from outdoor recreation expenditures in the upper lake states. Doctoral
dissertation. Michigan State University. East Lansing. MI.

Pleeter, S. (1980). Methodologies of economic impact analysis: An overview. In
Pleeter, S. (Ed.), Economic impact analysis: Methodology and
applications. chapter 1 (pp. 7- 31). Boston: Maritinus Nijhoff Publishing.

Propst. D. B. (2000). Use of IMPLAN to assess economic impacts of recreation
and tourism: Chronology and trends. In Proceedings of the Fifth Outdoor
Recreation and Tourism Trends Symposium. East Lansing. MI.

Propst. D. B. (1995). Understanding multipliers and how to interpret them.
Appendix E in Stynes, D. J., and Propst. D. B., Micro-lmplan Recreation
Economic Impact estimation system; MI-RE C user’s manual. Version 2. 0.
Department of Park, Recreation and Tourism Resources, Michigan State
University. East Lansing, MI.

Propst. D. B. (1991). Economic dimensions. In Fridgen, J. D.. Dimensions of
tourism, chapter 6 (pp. 146-155). East Lansing. MI: Educational Institute of
the American Hotel 8 Motel Association.

146

Propst, D. B., 8 Gavrilis, D. G. (1987). The role of economic impact assessment
procedures in recreational fisheries management. Transactions of the
American Fisheries Society. 116, 450-460.

Propst, D. 8.. Gavrilis, D. G.. Cordell, H.K.. 8 Hansen, W. (1985). Assessing the
Secondary Economic Impacts of Recreation and Tourism: Work Team
Recommendations. In Propst. D. B. (Ed.). Assessing the economic
impacts of recreation and tourism, (pp. 52-63). USDA. Forest Service.
Southeastern Forest Experiment Station, Asheville, NC.

Propst. D. B. and Siverts, L. E. (1990). Economic impacts of recreation:
Refinements in expenditure to input-output sector allocations. In
Proceedings. National Outdoor Recreation Trends Ill Symposium.
Indianapolis. IN.

Propst. D.B.. Stynes. D.J., Chang, W.H., 8 Jackson, RS. (1998). Estimating the
local impacts of recreation at Corps of Engineers Projects - 1996.
Technical Report R98-1. US. Army Engineers Watenrvays Experiment
Station. Vicksburg, MS.

Regional Economic Models. Inc. (2001). REMI: Policy Insight, user guide version
3.0. Amherst. MA: Regional Economic Models. Inc.

Richardson, H. W. (1969). Regional economics: Location theory, urban
structure, and regional change. New York. N.Y.: Praeger Publishers.

Richardson. H. W. (1972). Input-output and regional economics. New York: John
Wiley.

Richardson. H. W. (1985). Input-output and economic base multipliers: Looking
backward and forward. Joumai of Regional Science. 25(4): 607-661.

Rickman. D.. 8 Schwer. R. (1995a). Multiplier comparisons of the IMPLAN and
REMI models across versions: illuminating black boxes. Environment and
Planning A. 27: 143-151.

Rickman. D.. 8 Schwer. R. (1995b). A comparison of the multipliers of IMPLAN,
REMI. and RIMS ll: Benchmarking ready-made models for comparison.
The Annuals of Regional Science. 29: 363-374.

147

Schaffer. W. A.. 8 Davidson, L. S. (1985). Economic impacts of the Falcons on
Atlanta: 1984. Atlanta: The Atlanta Falcons.

SPSS Inc. (1999). Statistical analysis using SPSS 9.0. Chicago: SPSS Inc.

Stevens, 8.. 8 Rose, A. (1985). Regional input-output methods for tourism
impact analysis. In Propst, D. B. (Ed.). Assessing the economic impacts of
recreation and tourism (pp. 16-22). USDA. Forest Service. Southeastern
Forest Experiment Station, Asheville, NC.

Stevens, B. H.. Treyz. G. l.. Erlich. D. J., 8 Bower. J. R. (1983). A new technique
for the construction of nonsurvey regional input-output models and
comparisons with two survey-based models. international Regional
Science Review, 8: 271-86.

Stynes, D. J. (1999a). Economic impacts of tourism. [Online]. Available:
http:i/www.msu.edu/course/prr/840/econimpact. [July 12. 2000].

Stynes, D. J. (1999b). Approaches to estimating the economic impacts of
tourism; some examples. [Online]. Available:
http://www.msu.edu/course/prr/840/econimpact. [July 12. 2000].

Stynes, D. J., 8 Chang. W. H. (2000). Michigan Tourism Economic Impact
Model. [Online]. Available:
http:/lwww.msu.edu/course/prr/840/econimpact. [August 20. 2000].

Stynes, D. J., 8 Propst. D. B. (1998). Micro-implan Recreation Economic impact
Estimation sztem; MI-REC user's manual. Version 4. Department of
Park. Recreation and Tourism Resources. Michigan State University, East
Lansing, MI.

Stynes, D. J., 8 Propst. D. B. (1999). Advancing the Money Generation Model:
Research project #1. Proposal to National Park Service (RFP #Ul99-
101 H). Department of Park. Recreation and Tourism Resources.
Michigan State University. East Lansing. MI.

Stynes, D. J., 8 Rutz, E. A. (1995). Regional economic impacts of Mammoth
Cave National Park. East Lansing. MI: Department of Park, Recreation
and Tourism Resources. Michigan State University.

148

Stynes. D. J., Propst. D. 3.. Chang. W. H.. 8 Sun. Y. Y. (2000). Estimating
National Park visitor spending and economic impacts; the MGM2 model.
Final report submitted to National Park Service Cooperative Park Studies.
Department of Park, Recreation and Tourism Resources. Michigan State
University. East Lansing. MI.

Summary. R. M. (1987). Tourism’s contribution to the economy of Kenya. Annals
of Tourism Research. 14: 531-540.

Tiebout, C. M. (1962). The community economic base study. New York:
Committee for Economic Development.

Tooman. L. A. (1997). Multipliers and life cycles: A comparison of methods for
evaluating tourism and its impacts. Joumai of Economic Issues. 31(4):
91 7-932.

Travel Industry Association of America. (2000). Impacts of travel on state
economies 1998. Washington. DC: The Travel Industry Association of
America.

US. Department of Commerce. Census Bureau. (2001). North American Industry
Classification System (NAICS). [Online]. Available:
http://www.census.gov/epcd/wwwlnaics.html, [July 6, 2001].

US. Department of Commerce. Bureau of Economic Analysis. (1992). Regional
multipliers : A user handbook for the regional input-output modeling
system (RIMS Ii). 2nd edition. Washington, DC: US. Government Printing
Office.

US. Department of Commerce, Bureau of Economic Analysis. (1997). Regional
multipliers : A user handbook for the regional input-output modeling
system (RIMS Ii). 3rd edition. Washington. DC: US. Government Printing
Office.

US. Department of Commerce. Bureau of Economic Analysis. (2001).
Metropolitan Statistical Area component county list. , [Online]. Available:
httpzllwww.bea.doc.gov/bea/regional/docs/msalist.htm. [July 20. 2001].

US. Department of Commerce. Census Bureau. (1998a). Annual benchmark
report for retail trade, January 1988 through December 1997, [Online].

149

Available: http:l/www.census.gov/econlwww/re0100.html [March 20.
2001L

US. Department of Commerce, Census Bureau. (1998b). Annual benchmark
report for wholesale trade. January 1988 through December 1997.
[Online]. Available: http://www.census.gov/svsd/wwwlwhltable.html [March
20.2001j

US. Department of Interior. Fish and Wildlife Service. (1998). 1996 national and
state economic impacts of wildlife watching. Washington, DC: US.
Government Printing Office.

US. Department of Interior, National Park Service. (1990). Money Generation
Model. Denver, CO: Office of Social Science, Socio-economic Studies
Division.

US. Department of Interior, National Park Service. (1999). Request for proposal:
Advancing the Money Generation Model.

Weisbrod, G.. 8 Weisbrod, B. (1997). Measuring economic impact of projects
and programs. Boston: Economic Development Research Group.

West. G. (1988). GRIMP: Input-Output analysis computer program. St. Lucia:
University of Queens land.

150