December 2008
The Estimated Economic Impact of
Reduced Recreational Boating Due
to a Deterioration of the Intracoastal
Waterway Channel in Georgia
Wes Clarke, Adam Jones,
Randal Walker, and Paul Christian
The University of Georgia
The Estimated Economic Impact of
Reduced Recreational Boating Due to
a Deterioration of the Intracoastal
Waterway Channel in Georgia
Wes Clarke
Adam Jones
Carl Vinson Institute of Government
Randal Walker
Paul Christian
Marine Extension Service
December 2008
Carl Vinson Institute of Government
The University of Georgia
Athens
Acknowledgments
Georgia’s coastal economy is diverse, and a significant part is made up of marine businesses
including commercial fishing and shrimping; lodging, restaurant, and entertainment enterprises; and various marinas, boat repair shops, and suppliers that serve the more than 21,000
Georgia boaters who use the Atlantic Intracoastal Waterway (ICW). These users rely on
navigable rivers and other waterways as transportation routes for their business operations
and for recreation. The ICW is a major part of that transportation infrastructure, connecting the eastern seaboard from Virginia to the Florida Keys.
In recent years funding for maintenance of the channel along the ICW has been insufficient to maintain its authorized 12-foot depth. The current study addresses the potential loss
to Georgia’s overall economy, and the coastal economy in particular, if lack of maintenance
on the ICW results in further deterioration of the channel. The deferred maintenance has
made portions of the waterway difficult or impossible to navigate at low tide and threatens
businesses in the region.
This research was supported by Rep. Jack Kingston’s office. Primary funding, data, and
other support came from the Georgia Department of Natural Resources. Additional funding and support were provided by the Georgia Marine Business Association, the Georgia
Ports Authority, the Savannah Chamber of Commerce, the Atlantic Intracoastal Waterway
Association, Colonial Oil, and others.
Dr. Rich Clark and his staff at the Vinson Institute assisted with development and
administration of the online surveys. Ms. Valerie Gentry at the Vinson Institute compiled
databases for both the boater and business surveys. Other faculty and staff at both the Vinson
Institute and the Marine Extension Service helped by tracking down information or putting
us in touch with someone who could answer a question. Thanks to the many people at the
Vinson Institute and the Marine Extension Service who contributed to the research.
Wes Clarke
Adam Jones
Carl Vinson Institute of Government
Randal Walker
Paul Christian
Marine Extension Service
Contents
Acknowledgments........................................................................................................ iii
Executive Summary .................................................................................................... 1
Introduction. ................................................................................................................. 2
ICW History and Background................................................................................. 2
Data and Methodology............................................................................................... 4
Boater Survey.............................................................................................................. 5
ICW Navigability in Georgia.................................................................................... 11
Economic Impact Analysis. ....................................................................................... 15
About the GEMS Model. ......................................................................................... 15
Estimated Impacts of Boating on the ICW........................................................... 17
Business Survey........................................................................................................... 20
Conclusion..................................................................................................................... 27
Works Cited................................................................................................................... 27
Appendix A. ..................................................................................................................... 29
Appendix B. ..................................................................................................................... 37
Appendix C...................................................................................................................... 45
Executive Summary
In recent years, the channel along the Atlantic Intracoastal Waterway (ICW) has deteriorated
in many places due to insufficient dredging and maintenance. The Georgia Department of
Natural Resources contracted with the Carl Vinson Institute of Government to conduct a
study to determine the economic benefits of recreational boating on the Georgia portion of
the ICW and to determine the extent of economic loss that might result from a reduction
in recreational boating caused by deterioration of the channel.
Faculty and staff at the Vinson Institute designed two surveys to collect data that could
be used to address these questions. We surveyed owners of registered boats on their current
use of the ICW and the amounts they spent in the coastal area and elsewhere. A separate
survey went to businesses in the coastal counties to determine the extent to which they rely
on the ICW for their business enterprises and the effect that deterioration of the channel
has had on their revenues.
We received responses with usable data for most of the analysis from 1,004 boaters, with
usable data for parts of the analysis ranging from about 800 to 950. Forty-two responses
were received from businesses in the coastal counties. The results of the data analysis are
summarized below.
• Approximately 21,000 of Georgia’s registered boaters with crafts 16 feet and longer
used the ICW in the past 12 months.
• An estimated 1,871 out-of-state boaters used the ICW in Georgia over the same
period.
• Survey results suggest those boaters took more than 137,000 outings on the ICW
in the past 12 months.
• Boaters spent an estimated $213.2 million on those outings (past 12 months).
• Boater spending could fall by nearly $89 million if the ICW channel continues to
deteriorate.
• The total estimated economic impact of that reduction in spending is $124.5 million
annually.
• More than 2,100 jobs with $54 million in personal income could be lost as a result
of reduced use of the ICW.
• Nearly $15 million in state and local government revenue (sales and property taxes
and business licenses) could be lost due to reduced spending by boaters.
• More than 24,000 commercial vessels use the ICW between Virginia and Florida
each year.
• The ICW serves as transportation infrastructure for coastal businesses and for the
harbors at Savannah and Brunswick, where more than 34 million tons of goods were
handled in 2006.
Introduction
Deterioration of the Intracoastal Waterway (ICW) in Georgia threatens both the usefulness
of the waterway for recreational boating and as support infrastructure for the deep harbors
at Savannah and Brunswick. To determine the extent of the economic loss that may occur
if the channel continues to degrade, the Georgia Department of Natural Resources (DNR)
contracted with the Carl Vinson Institute of Government at the University of Georgia to
survey recreational users of the ICW and businesses in the coastal region concerning their
use of and dependence on the waterway and how continued degradation would affect them.
Boaters were asked about their use of the ICW and to indicate how deterioration of the
channel would affect that use. Boaters were asked about the amounts they spent on a typical outing, and the resulting data were used to determine how changes in their use of the
ICW might affect the coastal economy. In the second survey of commercial fishing, marina,
barge, construction, and realty businesses, we asked about the importance of the ICW to
those businesses and the effect that deterioration of the channel might have.
Recreational use of the ICW in Georgia is extensive. Survey data suggest that more
than 21,000 of Georgia’s 160,000 registered owners of boats 16 feet and longer make use
of the ICW in a typical year. In all, these individuals and nearly 2,000 out-of-state boaters
traveling in Georgia’s portion of the ICW made more than 137,000 trips along portions of
the ICW in the past 12 months and spent an estimated $213 million in Georgia’s coastal
counties and at home.
Requests for participation in the survey of businesses in the coastal counties of Georgia were sent to 666 individual businesses, but only 42 completed surveys were received.
Although businesses did not respond at a high rate, those that did respond in the marina
and commercial fishing industries indicate that they and their customers rely heavily on
the ICW as a water transportation route.
In 2006, the ICW handled more than 24,000 commercial vessels between Norfolk,
Virginia, and St. Johns River, Florida, up about 40 percent since 2002 (Institute for Water
Resources 2002, 2006). Those vessels carried more than 160,000 tons of fuel oil, fabricated
metal, machinery, wood products, and other goods. The harbors at Savannah and Brunswick
in Georgia handled more than 34 million tons of goods in 2006. Inbound traffic with a
draft greater than 18 feet at these two ports increased by 20 percent compared with 2002
with more than 2,600 foreign and domestic vessels in 2006. Outbound traffic has increased
similarly. Support service businesses for these two ports, including towing companies, depend
on the ICW as a transportation route in the region.
ICW History and Background
The Intracoastal Waterway is divided into two noncontiguous segments: the Gulf Intra­
coastal Waterway extending from Brownsville, Texas, to Carrabelle, Florida, and the Atlan­
tic Intracoastal Waterway, from Key West, Florida to Boston, Massachusetts. The Cross
Florida Barge Canal in northern Florida, was envisioned to connect the two segments, but
the project was never completed because of environmental concerns. The portion that was
constructed is now a green space. Vessels with a four- or five-foot draft are able to navigate
between the two waterways by traversing the Okeechobee Waterway between Fort Myers
on the Gulf of Mexico and Stuart, Florida, on the Atlantic coast. This waterway was built in
1937 by the Army Corps of Engineers primarily to control flooding at Lake Okeechobee.
2
The Atlantic ICW is a series of natural passageways, rivers, and manmade canals tucked
behind chains of barrier islands that stretch 2,500 miles from Boston to Miami. In the late
1700s and early 1800s, many privately dug canals connecting navigable rivers and bays
were completed to establish local waterways between cities. These efforts increased local
commerce, but there was little thought of creating a continuous passage along the East
Coast until 1808 when Albert Gallatin, a member of Congress, and the longest-serving
U.S. Secretary of the Treasury, realized the advantages of a marine through-route (Parkman
1983). Gallatin noted that there were only four necks of land that stood in the way of an
uninterrupted seaboard passage from Massachusetts to the southern extremity of Georgia.
The four major obstructions were (1) Cape Cod, (2) New Jersey between the Raritan and
Delaware Rivers, (3) the peninsula between the Delaware River and Chesapeake Bay, and
(4) the marshy tract between Chesapeake Bay and the Albemarle Sound in North Carolina.
While Gallatin’s vision of a 10-year, federally funded project did not become a reality, the
waterway was largely fashioned through many local projects over more than a century.
The Georgia ICW has an early history as an important waterway linking Georgia’s rice
and cotton plantations to the ports of Savannah and Charleston, South Carolina. Vessels
engaged in trade in the region used the natural water highway to avoid harsher weather in the
open water of the Atlantic Ocean. Even before the plantation period, the inland waterways
provided a travel route for Spanish traders and Franciscan friars who established missions
in the region during the sixteenth and seventeenth centuries (Sullivan 2005). During the
Civil War, the ICW facilitated commerce and communication between plantations despite
Union blockades off the Georgia coast and in the sounds.
In the late nineteenth and early twentieth centuries, the Army Corps of Engineers began
regular dredging of Georgia’s coastal waterways to enable safer passage of commercial ships
approaching Savannah. Colonial surveyor William Gerard DeBrahm noted the presence
of a narrows on a 1770 map of Amelia Island, which lies just south of Cumberland Island
and is part of northern Florida today (Sullivan 2005). Contemporary examples of areas that
require regular dredging include the Florida Passage–Bear River segment between Ossabaw Island and the Bryan County mainland; North Newport River west of St. Catherines
Island; Buttermilk Sound northwest of St. Simons Island; Jekyll Creek and the Cumberland
Divides (at the southern end of the Brickhill River); Skidaway Narrows, located south of
Savannah and west of Skidaway Island; and Creighton Narrows in McIntosh County between Creighton and Sapelo islands. These areas require constant maintenance dredging
to keep these passages open to navigation (Sullivan 2005).
The federal role during the early development of the ICW was largely one of financial
and engineering assistance, but the River and Harbor Act of 1938 authorized the Atlantic
Intracoastal Waterway and gave responsibility for maintenance to the Army Corps of Engineers (Sullivan 2005). The act also instructed the Corps to establish a mean low-water
depth of 12 feet throughout its length.
Maintenance of the ICW at the authorized depth has not been consistent because appropriations for the purpose have not allowed the Corps of Engineers to perform dredging
operations at regular intervals. The Boat Owners Association (2000) maintains that “the
ICW seldom has been maintained consistently to its authorized depths. A 12-foot channel
is authorized from Norfolk, VA, to Ft. Pierce, FL, and from there to Miami, it’s 10 feet
deep. But in some stretches, the water can be as shallow as five or six feet. Vessel groundings are common in parts of the ICW, creating potentially dangerous situations for both
3
commercial crews and for recreational boaters who must share the narrow channels with
commercial traffic.”
According to Charlie Waller, owner of Isle of Hope Marina and President of the Georgia
Marine Business Association, “There are locations in Georgia where mean low water depth
is 4 feet or less (personal communication, June 2008). Waller notes that four major problem
­areas exist in the Georgia Section of the ICW (from north to south): Hells Gate (mile 602),
Creighton Narrows (mile 642), Little Mud River (mile 654), and Jekyll Creek (mile 680).
Claiborne Young (2007) suggests that “there has been no dredging on the Georgia portion
of the ICW for over 5 ½ years.”
As one travels south on the ICW, the first major shallow area is Hells Gate that leads
from Ossabaw Sound to the Ogeechee River at statute mile 602. Because of the large tidal
amplitude of coastal Georgia and the swift currents of the Ogeechee River, this narrow cut
is prone to shoaling and sand deposition. The pass is narrow (less than 100 feet wide at low
water) and fortunately not very long (less than 1 nautical mile). This location is notorious
as being one of the most troublesome locations in Georgia. During the spring of 2006, the
U.S. Coast Guard (USCG) threatened to remove all the markers from this location, stating
that there was no longer a viable channel left to mark. Only through vigorous letter writing and e-mails to the USCG did ICW cruisers thwart this idea and actually convince the
USCG to add additional temporary markers to this difficult passage.
Creighton Narrows is a passage from Front River to Crescent River in northern McIntosh County just west of Sapelo Island, separating it from Creighton Island. This area has
been a problem location since the nineteenth century, when steamboats making their way
to Savannah from the south would frequently run aground. To eliminate this problem, the
Army Corps of Engineers, with congressional appropriations, dredged Creighton Narrows
in 1908 and has continued to do so since. As the name implies, this passage is very narrow,
especially at low water, and most problematic at the southern end where it joins up with
Crescent River. The passage is about 1 nautical mile in length.
Little Mud River is a long narrow river that connects Doboy Sound with Altamaha Sound
via the South River and the northern portion of the Altamaha River. The river is about 2 ¼
nautical miles in length and very shallow because it floods from both ends (Doboy Sound and
Altamaha Sound), depositing silt from the Altamaha River in the middle and particularly the
southern portion. Water depths along the Little Mud River between markers 194 and 195
are as low as 4 to 5 feet (Young 2007). To date, no dredging is planned for this location.
As cruisers pass to the west of Jekyll Island, they encounter the southernmost of the
four major problem spots on the Georgia ICW, known as Jekyll Creek. While depths are
deteriorating all along the five-mile section of the river, the shallowest section lies about
in the middle, just above and below the Jekyll Wharf Marina. As with Little Mud River,
Jekyll River floods from both ends (St. Simons Sound and St. Andrew Sound), which allows
suspended silt to be deposited where the two flood tides meet. Recorded low-water depths
of 4 feet are common, and the width for passage approaches 100 feet or less. This creek has
not been dredged in over six years, and there are no plans to do so in the near future.
Data and Methodology
Faculty and staff at the Carl Vinson Institute of Government and the Marine Extension
Service at the University of Georgia prepared two surveys to estimate the economic impact
of channel deterioration. These surveys are included as Appendices A and B. From the
boater survey, we estimated the current level of economic activity derived from boaters’
4
use of the ICW and the reduction in activity that could result from further deterioration of
the channel. The business survey data were used to determine the importance of the ICW
for commercial fishing, marina, barge, construction, and realty businesses in the coastal
counties.
Boater Survey
The Georgia DNR registers boats in the state and reports that Georgia has 161,013 registered boats 16 feet and longer. This study is limited to boats at least 16 feet in length since
smaller boats have much shallower drafts and would not be affected as much by a shallower
channel. About one third (54,990) of boaters in this group provided an e-mail address to
DNR with their registration; DNR provided a database of those e-mail addresses for purposes of requesting participation in the survey.
CVIOG faculty and staff prepared a list of e-mail addresses from the database sent
by DNR. The list contained some duplicates and incomplete addresses. After removing
those, a final list of 40,874 e-mail addresses was compiled, and e-mails were sent asking
boat registrants to participate in the survey by accessing the online instrument. Just over 10
percent (4,451) of the e-mail messages were returned as undeliverable. An undetermined
number of the remaining 36,423 e-mail requests may not have reached the intended recipient. Responses were received from 1,004 boaters, and 842 survey responses produced
usable data.
A true response rate cannot be determined since the number of boat owners who actually received the request is unknown. Given the interest that users of the ICW have in its
navigability, we expected a fairly high response rate. One study suggests that surveys about
recreation topics can usually generate a high response rate (Leeworthy et al. 2001).The
usable ICW survey responses represent less than 2.5 percent of the e-mails sent, which is
unusually low. One study of online surveys with e-mail requests for participation suggests
that a number of factors may result in seemingly low response rates (Kaplowitz, Hadlock.
and Levine 2004). In addition to the delivery failure of some messages, some may be screened
by network spam filters or by user-defined filters that reject messages received from any
unapproved sender. Some e-mail addresses may be abandoned as users change Internet
service providers, and others may not be checked on a regular basis.
The map in Figure 1 shows the five regions used for this study. The six coastal counties
are Region 1. Moving in a northwesterly direction away from the coast, the state is divided
into four additional regions, 2–5, grouping counties that are roughly the same distance
from the Georgia coast. We used U.S. Department of Commerce and U.S. Postal Service
ZIP code distances to assign each county to a region. It appears that in some instances, a
county such as Jeff Davis or Toombs could be in either of two regions. We simply let the
distances from the county’s primary ZIP code to the nearest coastal ZIP code determine
the placement.
Respondents to the survey are largely male, White (non-Hispanic), and married (see
Table 1). There is very little variation across the five regions in these demographics. A little
more than 60 percent of respondents have at least a four-year degree except in Region 2,
where the proportion is 50.9 percent. More than 68 percent of out-of-state boaters have at
least a four-year degree. Sixty-two percent of boaters have incomes greater than $75,000
(see Table 2). The proportion with an annual income above $75,000 ranges from 51 percent
in Region 3 to highs of 64 and 65 percent in Regions 1 and 4, respectively.
5
Figure 1. Map of Georgia Depicting Study Regions for Data Analysis
Catoosa
Dade
Towns
Fannin
Murray
Rabun
Union
Habersham
Whitfield
Walker
Gilmer
Stephens
White
Lumpkin
Gordon
Chattooga
Pickens
Dawson
Forsyth
Cherokee
Floyd
Bartow
Region
5
Franklin
Banks
Hall
Jackson
Madison
Gwinnett
Clarke
Paulding
Douglas
Clayton
Morgan
Newton
Taliaferro
Greene
Columbia
McDuffie
Henry
Fayette
Lincoln
Wilkes
Walton
Rockdale
Fulton
Oglethorpe
Oconee
Dekalb
Haralson
Carroll
Elbert
Barrow
Cobb
Polk
Region
4
Hart
Region
3
Warren
Richmond
Coweta
Jasper
Heard
Spalding
Putnam
Hancock
Butts
Glascock
Baldwin
Meriwether
Pike
Troup
Lamar
Monroe
Burke
Jefferson
Jones
Upson
Washington
Jenkins
Talbot
Harris
Bibb
Crawford
Peach
Taylor
Muscogee
Johnson
Wilkinson
Twiggs
Region
2
Screven
Emanuel
Laurens
Houston
Bleckley
Chattahoochee
Marion
Pulaski
Montgomery
Dodge
Dooly
Webster
Bulloch
Effingham
Schley
Stewart
Candler
Treutlen
Macon
Evans
Toombs
Wheeler
Bryan
Sumter
Chatham
Tattnall
Wilcox
Region
1
Telfair
Crisp
Quitman
Terrell
Randolph
Jeff Davis
Lee
Clay
Dougherty
Liberty
Appling
Irwin
Worth
Calhoun
Long
Ben Hill
Turner
Coffee
Wayne
Bacon
McIntosh
Tift
Pierce
Baker
Early
Atkinson
Berrien
Mitchell
Miller
Glynn
Cook
Colquitt
Ware
Brantley
Lanier
Decatur
Camden
Seminole
Thomas
Clinch
Brooks
Grady
Lowndes
6
Echols
Charlton
Chatham
7
2
5
Other
Not Indicated
Total
200
4
Not Indicated
2.0%
0.0%
1.0%
2
0
Widowed
8.0%
3.0%
86.0%
2.5%
1.0%
1.0%
0.0%
1.0%
1.5%
93.1%
(11.6)
2.0%
5.0%
93.0%
16
6
172
Separated
Divorced
Single
Married
Marital Status
202
2
American Indian,
Alaskan Native
Total
0
2
Hispanic
Asian, Pacific
Islander
3
African American
White (nonHispanic)
188
57.2
Average Age
Years (Std. Dev.)
Race
201
Total
4
10
Female
Not Indicated
187
Region 1
Male
Gender
53
0
2
0
5
3
43
53
0
1
1
0
0
0
51
59.3
53
0
0
53
0.0%
3.8%
0.0%
9.4%
5.7%
81.1%
0.0%
1.9%
1.9%
0.0%
0.0%
0.0%
96.2%
(10.9)
0.0%
0.0%
100.0%
Region 2
Table 1. Boater Demographics: Gender, Race, Marital Status
96
1
0
1
3
6
85
95
2
1
1
0
0
0
91
55.8
95
0
7
88
1.0%
0.0%
1.0%
3.1%
6.3%
88.5%
2.1%
1.1%
1.1%
0.0%
0.0%
0.0%
95.8%
(11.1)
0.0%
7.4%
92.6%
Region 3
384
6
0
3
28
29
318
391
6
5
1
3
1
8
367
56.5
389
5
26
358
1.6%
0.0%
0.8%
7.3%
7.6%
82.8%
1.5%
1.3%
0.3%
0.8%
0.3%
2.0%
93.9%
(10.2)
1.3%
6.7%
92.0%
Region 4
60
1
0
1
3
0
55
61
0
1
2
0
2
0
56
58.9
62
0
4
58
1.7%
0.0%
1.7%
5.0%
0.0%
91.7%
0.0%
1.6%
3.3%
0.0%
3.3%
0.0%
91.8%
(11.2)
0.0%
6.5%
93.5%
Region 5
35
1
1
0
2
1
30
35
2
1
1
0
0
0
31
53.9
35
2
2
31
2.9%
2.9%
0.0%
5.7%
2.9%
85.7%
5.7%
2.9%
2.9%
0.0%
0.0%
0.0%
88.6%
(10.6)
5.7%
5.7%
88.6%
Out of State
828
13
3
7
57
45
703
837
15
11
8
3
5
11
784
56.8
835
11
49
775
1.6%
0.4%
0.8%
6.9%
5.4%
84.9%
1.8%
1.3%
1.0%
0.4%
0.6%
1.3%
93.7%
(10.8)
1.3%
5.9%
92.8%
All Boaters
8
200
Total
197
Total
46.2%
91
29
>100,000
Not indicated
17.8%
35
75,001–100,000
14.7%
7.6%
3.0%
6
5.1%
15
10
40,001–50,000
5.6%
60,001–75,000
11
30,001–40,000
0.0%
2.0%
11.5%
2.0%
16.0%
26.0%
50,001–60,000
0
≤30,000
Income (dollars)
23
4
Professional degree
Not indicated
4
Doctorate
30.0%
60
32
12
Associate’s degree
4-year degree
52
Some college
Master’s degree
6.0%
13
H.S. graduate
6.5%
0
0.0%
Region 1
Less than H.S.
Education
51
7
17
13
3
4
4
2
1
53
0
2
2
4
19
5
11
10
0
13.7%
33.3%
25.5%
5.9%
7.8%
7.8%
3.9%
2.0%
0.0%
3.8%
3.8%
7.5%
35.8%
9.4%
20.8%
18.9%
0.0%
Region 2
7
96
15
36
13
11
6
10
4
1
96
1
3
1
19
35
11
19
15.6%
37.5%
13.5%
11.5%
6.3%
10.4%
4.2%
1.0%
1.0%
3.1%
1.0%
19.8%
36.5%
11.5%
19.8%
7.3%
0.0%
Region 3
0
Table 2. Boater Demographics: Education, Income by Region
386
65
182
68
31
17
6
8
9
389
11
25
14
63
136
31
77
31
1
16.8%
47.2%
17.6%
8.0%
4.4%
1.6%
2.1%
2.3%
2.8%
6.4%
3.6%
16.2%
35.0%
8.0%
19.8%
8.0%
0.3%
Region 4
59
13
23
11
3
5
2
1
1
63
0
0
5
11
23
7
10
7
0
22.0%
39.0%
18.6%
5.1%
8.5%
3.4%
1.7%
1.7%
0.0%
0.0%
7.9%
17.5%
36.5%
11.1%
15.9%
11.1%
0.0%
Region 5
35
7
20
1
4
2
1
0
0
35
0
6
2
7
9
3
7
1
0
20.0%
57.1%
2.9%
11.4%
5.7%
2.9%
0.0%
0.0%
0.0%
17.1%
5.7%
20.0%
25.7%
8.6%
20.0%
2.9%
0.0%
Out of State
824
136
369
141
67
40
33
26
12
833
16
59
25
136
282
69
176
69
1
16.5%
44.8%
17.1%
8.1%
4.9%
4.0%
3.2%
1.5%
1.9%
7.1%
3.0%
16.3%
33.9%
8.3%
21.1%
8.3%
0.1%
All Boaters
Requests for survey participation were sent to owners of boats 16 feet and longer. The
average length among all respondents was 21.5 feet with a standard deviation of about 8 feet
(see Table 3). The data in Table 3 indicate that boats transported to the coast from Regions
2–4 may be slightly larger in terms of their draft but not in length. Boats registered outside
Georgia, on average, are longer, at more than 26 feet. The average boat represented in the
survey is about 13 years old, with little variation across the five regions. Nearly 90 percent
of respondents operate either outboard (62.5 percent) or inboard (25.6 percent) boats (see
Table 4). Six percent report operation of sail-powered craft, and another 6 percent report
some other type of propulsion. Most respondents use a trailer to launch their boat, but
a significant percentage use either a marina or private dock to store their craft except in
Regions 2 and 5.
Table 5 shows the number of registered boaters in each region, the number of e-mail
survey participation requests sent, and the number of completed responses. The overall
response rate was about 2.9 percent with a high of 6.7 percent in Region 1 and just under
2 percent in both Region 4 and Region 5. As noted, a true response rate cannot be determined, and the actual response rate is almost certainly higher.
A critical calculation presented in Table 5 is the inference population. Some boaters
do not use the ICW either by choice or because of proximity. In order to produce a conservative estimate of the economic impact of recreational boating on the ICW, we made
two adjustments to the boater population. First, 20 percent of respondents to the survey
told us that they do not use the ICW for their boating activity, so we reduced the population by 20 percent since deterioration of the channel will not make any changes in their
boating activity. A second reduction must be made due to sampling bias from Region 1
through Region 5. Survey sampling bias suggests that boaters in the noncoastal regions
were less likely to respond to the survey if they do not use the waterway and therefore have
no interest in its conditions. Boaters who live in Region 2 are certainly less likely to use the
ICW than those living in Region 1; those living in Region 3 are less likely still, and so on.
To determine if response bias was correlated with distance to the coast, we used the distance
Table 3. Boat Measurements, Age, and Towing Distance by Region
Region 1
Region 2
Region 3
Region 4
Region 5
Out of
State
All
Boaters
Mean Standard Deviation (n)
What is your boat’s
overall length
(in feet)?
22.2
(7.7)
203
19.9
(5.0)
53
20.35
(6.1)
96
21.6
(8.3)
393
19.6
(5.2)
62
26.1
(12.3)
35
21.5
(7.9)
842
What is your boat’s
normal draft (in feet)?
2.5
(2.4)
196
2.5
(3.5)
49
3.5
(4.7)
82
4.0
(8.8)
364
3.8
(5.3)
59
3.4
(2.2)
32
3.4
(6.6)
782
Average age of boat
(in years)
13.2
(9.4)
203
12.1
(8.6)
53
12.1
(8.3)
96
13.7
(10.5)
390
11.0
(8.1)
62
12.2
(9.5)
35
13.0
(9.7)
839
Miles from your home
to usual launch site for
those using trailer
3.6
(4.2)
50
3.75
(5.6)
4
129.5
(70.4)
10
181.8
(118.7)
74
200
(93.5)
5
118.8
(107.8)
13
114.5
(129.6)
151
9
10
203
1.48%
24.63%
22.66%
51.23%
3.5%
53
0
4
5
44
53
2
46
5
0
0.00%
7.55%
9.43%
83.02%
3.8%
86.8%
9.4%
0.0%
Region 2
b
1,967
392
6
76
79
231
392
31
207
127
27
2,342
115.56
17,484
120%
3.8%
189
4,930
21,855
Region 3
3.13%
41.67%
11.46%
75.00%
3.1%
74.0%
20.1%
2.1%
25
5,671
214.77
78,690
120%
1.9%
430
22,059
98,363
0
62
3
5
6
48
62
4
893
279.07
16,107
120%
1.9%
92
4,730
20,134
Region 5
4.84%
8.06%
9.68%
77.42%
6.5%
53.2%
40.3%
0.0%
Region 5
33
Region 4
1.53%
19.39%
20.15%
58.93%
7.9%
52.8%
32.4%
6.9%
Region 4
Response rate is calculated from e-mails sent. The number of e-mail messages actually delivered is unknown.
Based on responses received to the boater survey.
10,178
a
Inference population
44.54
5,659
10,870
16.53
5.3%
120%
6.7%
120%
Average distance to coast
(zip code to zip code)
20% of boaters indicate no
use of the ICW b
Response rate
83
211
a
Number of responses
1,571
3,133
E-mails sent
7,074
13,587
Region 2
96
3
40
11
72
96
3
71
20
2
Region 3
Number of registered boaters
Region 1
Table 5. Boater Population by Region
Note: Percent columns may not add to 100 due to rounding.
Total
50
Private dock
3
46
Marina
Other
104
202
7
Trailer
Storage
Total
Other
14.8%
30
152
Inboard
Outboard
75.2%
6.4%
13
Sail
Type of boat
Region 1
Table 4. Boat Type and Storage Location by Region
35
3
8
11
13
35
3
16
8
8
2.14%
18.19%
18.79%
60.88%
6.0%
62.5%
25.6%
6.0%
1,871
—
—
22,922
670.47
128,810
2.9%
120%
—
1040
36,423
161,013
All Boaters
841
18
153
158
512
840
50
525
215
50
All Boaters
—
35
—
—
Out of State
8.57%
22.86%
31.43%
37.14%
8.6%
45.7%
22.9%
22.9%
Out of State
from the ZIP code in each county that is closest to the coast to the nearest coastal ZIP
code and calculated the average for each region. Distance to the coast and nonresponse rate
were strongly correlated with a Pearson coefficient of 0.959. This is evidence of significant
nonresponse bias. To adjust the inference sample, we calculated the proportional increase
in distance that a boater would need to travel to the coast from each region and used that
proportion as a further reduction of the boating population.
Finally, we received 35 responses from out-of-state boaters. We asked 15 marinas in
the coastal area to estimate the proportion of their customers who have boat registrations
outside Georgia. Using that information, we estimated that 1,871 out-of-state boaters use
the ICW in Georgia. The bottom row of Table 5 shows the inference population by region. Before turning to the economic impact analysis, the next section presents tabulated
responses to questions about the navigability of the ICW in Georgia.
ICW Navigability in Georgia
Boaters were asked questions concerning the current navigability of the Georgia portion
of the ICW and how changes in its navigability would affect their boating activity. About
32 percent of all boaters reported that the navigability of the ICW in Georgia is fair or
poor as shown in Table 6. More than two-thirds reported that navigability is at least good.
One-third of all boaters indicated that the navigability is very good or excellent. When we
look at perceptions of navigability among boaters according to the length of their craft, we
find that those with boats longer than 30 feet are much more likely to rate the navigability
as poor or only fair (see Table 6). Owners of smaller boats are probably less likely to travel
greater distances and may not encounter the four major problem areas along the route of
the ICW in Georgia.
When asked if dredging of the ICW in Georgia were increased to maintain a 12-foot
channel along its length, 27 percent of boaters reported that they would take more trips
on the ICW while almost 60 percent said they would take about the same number of trips
(see Table 7). Boaters indicated that they would take an average of 11 additional trips in the
next 12 months if the channel were maintained at 12 feet. Table 7 reports responses to this
question by boat length. Boaters with crafts longer than 30 feet were more likely to report
that they would take additional trips in the next 12 months if the channel were maintained
at 12 feet, while boaters with craft in the two smaller categories were most likely to indicate
they would take about the same number of trips.
In some parts of the Georgia portion of the ICW, channel depth is less than 4 feet at
low tide, making navigation difficult at those times. Boaters were asked how deterioration of
the channel to 4 feet along the length of the ICW in Georgia would affect their use of the
waterway. Sixty percent indicated that they would take fewer trips or no trips on the ICW
in Georgia over the next 12 months, as reported in Table 8. Nearly 40 percent said they
would take about the same number, and less than 1 percent indicated that they would take
more trips than in the past year. Boat length had some effect on boaters’ response to this
question, as Table 8 shows, but fewer than 10 percent of boaters with vessels over 30 feet
would take the same number of trips, and only about a third of those with boats between
20 and 30 feet would do so. Nearly half of those with boats less than 20 feet reported that
they would take fewer trips or no trips if the channel deteriorated to such an extent.
11
12
47
52
36
27
203
Fair
Good
Very good
Excellent
Total
13.3%
17.7%
25.6%
23.2%
20.2%
53
7
13
20
12
1
13.2%
24.5%
37.7%
22.6%
1.9%
Region 2
62
125
90
119
424
Good
Very good
Excellent
Total
28.1%
21.2%
29.5%
14.6%
6.6%
2
χ = 141.3
Note: Percent columns may not add to 100 due to rounding.
28
Fair
Less than 20 feet
Poor
Navigability
Note: Percent columns may not add to 100 due to rounding.
41
Poor
Region 1
387
99
73
108
67
40
408
66
76
126
92
25.6%
18.9%
27.9%
17.3%
10.3%
Region 4
Boat Length
Responses by Boat Length
19.4%
22.6%
32.3%
19.4%
6.4%
16.2%
18.6%
3.9%
22.6%
11.8%
20 feet to 30 feet
48
93
18
21
30
18
6
Region 3
Responses by Region
80
2
3
12
27
36
21.0%
12.9%
35.5%
17.7%
12.9%
2.5%
2.8%
15.0%
33.8%
45.0%
Over 30 feet
62
13
8
22
11
8
Region 5
35
6
5
9
6
9
912
187
169
263
181
112
17.1%
14.3%
25.7%
17.1%
25.7%
Out of State
Total
833
170
156
241
161
20.5%
18.5%
28.9%
19.9%
12.3%
18.7%
28.9%
19.3%
12.6%
All Boaters
105
Table 6. Considering your own boat, what is your opinion of the navigability of the Georgia portion of the Intracoastal Waterway?
13
1
135
65
203
Fewer
Same
More
Total
32.02%
66.50%
0.49%
0.99%
53
13
40
0
0
24.53%
75.47%
0.00%
0.00%
Region 2
4
286
71
429
Fewer
Same
More
Total
16.6%
66.7%
0.9%
15.9%
χ2 = 107.5
Note: Percent columns may not add to 100 due to rounding.
68
None
Less than 20 feet
Note: Percent columns may not add to 100 due to rounding.
2
None
Region 1
389
101
213
7
68
411
113
238
7
25.96%
54.76%
1.80%
17.48%
Region 4
Boat Length
Responses by Boat Length
26.04%
58.33%
0.00%
15.63%
27.5%
57.9%
1.7%
12.9%
20 feet to 30 feet
53
96
25
56
0
15
Region 3
Responses by Region
80
57
21
0
2
16.13%
58.06%
3.23%
22.58%
71.3%
26.3%
0.0%
2.5%
Over 30 feet
62
10
36
2
14
Region 5
35
12
18
0
5
920
241
545
11
123
34.29%
51.43%
0.00%
14.29%
Out of State
Total
838
226
498
10
104
26.2%
59.2%
1.2%
13.4%
27.1%
59.8%
1.2%
12.5%
All Boaters
Table 7. Suppose that the dredging of the ICW was increased and the average depth of the Georgia portion was about 12 feet. Would you take
more, fewer, or about the same number of trips on the ICW in the next year?
14
82
1
203
Same
More
Total
0.49%
40.39%
55.67%
53
1
21
29
1.89%
39.62%
54.72%
3.77%
Region 2
2
138
209
6
427
Fewer
Same
More
Total
1.4%
49.0%
32.3%
17.3%
2
χ = 58.5
Note: Percent columns may not add to 100 due to rounding.
74
None
Less than 20 feet
Note: Percent columns may not add to 100 due to rounding.
113
Fewer
3.45%
Region 1
7
None
96
0
40
33
23
393
3
152
134
0.76%
38.68%
34.10%
26.46%
Region 4
104
412
1
143
181
87
0.2%
34.7%
43.9%
21.1%
20 feet to 30 feet
Boat Length
Responses by Boat Length
0.00%
41.67%
34.38%
23.96%
Region 3
Responses by Region
80
0
7
49
24
62
0
21
24
17
35
0
13
9
0.0%
8.8%
51.3%
919
7
359
368
185
0.00%
37.14%
25.71%
37.14%
Out of State
13
30.0%
Over 30 feet
0.00%
33.87%
38.71%
27.42%
Region 5
Total
842
5
329
342
166
0.8%
39.1%
40.0%
20.1%
0.6%
39.5%
41.1%
19.9%
All Boaters
Table 8. Suppose that the dredging of the ICW completely stopped and the average depth of the Georgia portion was about 4 feet. Would you
take more, fewer, or about the same number of trips on the ICW in the next year?
Under current conditions, boaters told us that they took an average of 15 (median of
8) outings on the ICW in the past year. Under conditions with the channel at about 4 feet,
­boaters told us that they would reduce that number to an average of 7 (median 2) outings.
Boaters were asked whether they would be willing to pay a fee in addition to their annual
registration if the funds were earmarked for dredging projects on the ICW. We asked boaters
to indicate whether they would be willing to pay a $20 fee, and if they responded no or that
they were not sure, we asked about their willingness to pay a $10 fee. As shown in Table 9,
45 percent of respondents indicated a willingness to pay a $20 to fund dredging projects
along the ICW in Georgia. Boaters in Regions 1 and 2 were much more willing to pay a $20
fee than were boaters in Regions 3–5. A total of 461 respondents indicated that they were
either unwilling to pay a $20 fee or were unsure of their willingness. When asked whether
they would be willing to pay a $10 fee, only 22 percent of these 461 respondents said yes.
Tables 9 also presents data on willingness to pay the additional fee according to boat
length and income level of the respondents. There is a positive association between willingness to pay an additional fee and both boat length and income level. The χ2 statistics in each
table with respect to willingness to pay a $20 fee indicate a statistically significant relationship
between these factors (income and boat length) and willingness to pay the fee. When boaters
who indicated they were not willing to pay a $20 fee or were unsure of their willingness were
asked whether they would be willing to pay a $10 fee, no statistically significant relationship
was found between either boat length or income level and such willingness.
Economic Impact Analysis
About the GEMS Model
Researchers at the Vinson Institute used the Georgia Economic Modeling System (GEMS)
developed specifically for state and local policy analysis and forecasting in Georgia. The
GEMS system assembles the Georgia model using data from the Bureau of Economic
Analysis, the Bureau of Labor Statistics, the Department of Energy, the Bureau of Census,
and other public sources.
GEMS is structural in nature, meaning that it clearly includes cause-and-effect relationships. The model is based on two key underlying assumptions from mainstream economic
theory: (1) households maximize utility and (2) producers maximize profits. Because these
assumptions make sense to most people, lay people as well as trained economists can under­
stand the model.
In the model, businesses produce goods and services to sell to other firms, consumers,
investors, governments, and purchasers outside the region. Output is produced using labor,
capital, fuel, and intermediate inputs. Demand for labor, capital, and fuel per unit of output
depends on relative costs because an increase in the price of any one of these inputs leads
to substitution away from that input to other inputs.
Supply and demand for labor are incorporated into the model to calculate wage rates.
The wage rates, along with other prices and productivity, determine the cost of doing business for every industry in the model. An increase in the cost of doing business causes either
an increase in prices or a decrease in profits, depending on the market for the product. In
either case, an increase in costs would decrease the share of the local and U.S. market supplied by local firms. This market share, combined with the demand previously described,
15
16
Region 1
Region 2
χ2 = 14.4
Note: Percent columns may not add to 100 due to rounding.
6
6
5
17
40.0%
16.8%
43.2%
52.3%
27.3%
20.5%
62
26
67
155
46
24
18
88
8
0
6
14
10
4
21
35
106
34
40
180
130
49
215
394
58.9%
18.9%
22.2%
33.0%
12.4%
54.6%
291
134
115
540
360
179
398
937
57.1%
0.0%
42.9%
28.6%
11.4%
60.0%
Out of State
> $100,000
35.3%
35.3%
29.4%
10.0%
11.2%
78.8%
Over 30 feet
57.1%
18.4%
24.5%
59.7%
19.4%
21.0%
Region 5
$75,001 – 100,000
Responses by Income Level
56.8%
22.3%
20.9%
35.5%
16.9%
47.6%
Boat Length
20 feet to 30 feet
Income Level
≤ $50,000
$50,001–$60,000
$60,001–$75,000
Willingness to pay a $20 fee
No
34
41.5%
24
47.1%
24
32.6%
Not sure
23
58.1%
13
25.5%
23
30.3%
Yes
25
30.5%
14
27.5%
29
38.2%
Total
82
51
76
χ2 = 39.8
Willingness to pay a $10 fee. Asked of those that responded “no” or “not sure” above.
No
27
47.4%
17
56.75
20
42.6%
Not sure
21
36.8%
7
18.9%
14
29.8%
Yes
9
15.8%
13
35.1%
13
27.7%
Total
57
37
47
χ2 = 3.6
Note: Percent columns may not add to 100 due to rounding.
28
9
12
49
56.9%
25.8%
17.3%
8
9
63
80
37
12
13
62
37.9%
19.3%
42.7%
Region 4
Responses by Boat Length
Willingness to pay a $20 fee
No
203
46.5%
149
Not sure
99
22.7%
71
Yes
135
30.9%
200
Total
437
420
χ2 = 62.0
Willingness to pay a $10 fee. Asked of those that responded “no” or “not sure” above.
No
160
52.8%
125
Not sure
79
26.1%
49
Yes
64
21.1%
46
Total
303
220
Less than 20 feet
Note: Percent columns may not add to 100 due to rounding.
42.7%
149
17.7%
76
39.6%
168
393
“not sure” above.
46.6%
128
22.4%
58
31.0%
39
225
Region 3
No
56
27.6%
18
34.0%
41
Not sure
34
16.7%
7
13.2%
17
Yes
113
55.7%
28
52.8%
38
Total
203
53
96
Willingness to pay a $10 fee. Asked of those that responded “no” or
No
42
46.7%
13
52.0%
27
Not sure
28
31.1%
3
12.0%
13
Yes
20
22.2%
9
36.0%
18
Total
90
25
58
Willingness to pay a $20 fee
Responses by Region
Table 9. Would you be willing to pay a fee in addition to the annual registration fee to fund this program?
216
100
93
409
274
134
350
758
53.9%
24.8%
21.3%
38.4%
19.1%
42.5%
53.4%
24.1%
22.6%
36.9%
17.8%
45.2%
52.8%
24.5%
22.7%
36.2%
17.7%
46.2%
All Boaters
Total
246
111
104
461
311
150
381
842
All Boaters
determines the amount of local output. The model has many other feedbacks. For example,
changes in wages and employment affect income and consumption, while economic expansion changes investment and population growth influences government spending.
Within the model, firms produce goods and services that are purchased either by final
consumers or by other firms as inputs to their own production processes. Firms also purchase
labor, capital, and other inputs. Labor and capital requirements depend on both output and
relative costs. Population and labor supply contribute to demand and to wage determination.
Economic migrants, in turn, respond to wages and other labor market conditions. Supply
and demand interact in the wages, prices, and profits block. Prices and profits determine
market shares. Output depends on market shares and the components of demand.
GEMS brings together all of the elements to determine the value of each variable for each
year in the baseline forecasts. Interindustry interactions that are included in input-output
models are used to estimate the values of other regional economic variables. In order to
broaden the model in this way, it was necessary to estimate key relationships. Extensive data
sets covering all areas in the country and two decades worth of research were used to ensure
that the model was theoretically sound and based on all of the relevant data available.
The model has strong dynamic properties; that is, it forecasts not only what will happen but also when it will happen. It enables long-term predictions that have general equilibrium properties, meaning that the long-term properties of general equilibrium models
are preserved, accurate year-by-year predictions are maintained, and key equations can be
estimated by using primary data sources.
Estimated Impacts of Boating on the ICW
We asked boaters about not only the number of outings they have made in the past year and the number made during a typical year but also the amount per trip spent for transportation,
boat launch fees, fuel, lodging, restaurant meals, other food and beverage, fishing supplies,
and other purposes. The survey asked boaters to indicate the total amount spent and the
amount spent in the coastal areas of the state so that economic impacts in each region of
the state could be estimated.
Table 10 presents data on the number of outings boaters reported taking in the past two
months, the past year, and in a typical year. Boaters in each region report a greater number
of anticipated trips in the next year than they took in the current year. The survey also asked
boaters if they would take more or fewer trips annually on the ICW if the channel deteriorated to a depth of 4 feet along most of its length. Each boater was then asked how many
more or fewer trips they would take annually. For each respondent, we calculate the change
in the number of trips. Some boaters reported that they would take more trips. Although
we did not ask for an explanation of such a response, it is possible that some boaters might
anticipate a less crowded ICW and would consider taking more trips.
Total spending by the inference population of boaters is reported in Table 11. We calculate current annual spending by boaters in each region by multiplying the mean number
of trips by the median spending per trip. In order to adjust for a number of extreme outliers
in the data, we use the median per trip spending. Boaters in Region 1 spent an estimated
$14,687 annually while boaters in the other four regions spent much less. Of course boaters
in Regions 2–5 likely spend a significant part of their boating time elsewhere. We estimate
that the inference population spent a total of $213.2 million for outings on Georgia’s portion of the ICW.
17
18
11.5
(22.9)
334
8.0
(17.7)
764
17.4
(25.6)
2.5
(3.7)
11
5.1
(17.3)
31
12.8
(18.3)
34
4.6
(10.9)
30
2.7
(4.9)
56
11.9
(15.4)
59
7.7
(16.2)
167
2.9
(6.3)
344
12.0
(19.3)
372
5.6
(6.6)
45
6.4
(12.7)
85
10.0
(12.5)
94
15.8
(11.8)
18
11.2
(9.6)
53
21.3
(27.6)
53
18.9
(29.0)
195
32.2
(35.7)
199
Number of outings
primarily for fishing
Anticipated outings in the next
12 months
2.0
(6.8)
829
29.8
(39.2)
63
2.7
(3.7)
34
Typical number of
annual ICW outings
in Georgia if past 12
months is not typical
1.4
(2.6)
383
3.3
(12.6)
202
Typical outing
length in days
1.8
(4.4)
62
11.0
(21.1)
842
7.5
(16.5)
358
4.3
(8.3)
62
4.9
(10.4)
393
7.3
(12.6)
96
13.7
(10.6)
53
26.7
(33.4)
203
Number of ICW
outings in Georgia—
past 12 months
1.9
(3.8)
95
0.65
(1.2)
810
0.8
(1.1)
35
0.2
(0.6)
59
0.5
(1.2)
375
0.65
(1.1)
95
0.97
(1.1)
52
0.99
(1.0)
194
Typical outing
length in days
1.35
(1.4)
53
1.5
(3.1)
842
1.29
(2.2)
35
0.2
(0.5)
62
Mean (Standard Deviation) n
0.6
(1.6)
393
All Boaters
Number of ICW
outings in Georgia—
past 60 days
Out of State
1.4
(2.6)
96
Region 5
2.45
(3.6)
53
Region 4
3.6
(4.5)
203
Region 3
Region 2
Region 1
Table 10. Boating Use by Region
19
14,687
149,489,375
Annual spending per
boater
Current annual total
spending
6,532,017
9,729,981
8,929,469
Lodging
Restaurant
Other food and beverage
6,975,080
75,885,863
Other
Total
12,345,278
2,613,010
Boat launch (marina)
Fishing supplies
28,761,028
Transportation/fuel
* 95% trimmed mean
10,178
Boaters
Region 1
Region 3
Region 4
373,274
0
72,354
39,331
0
0
44,526
217,063
22,286,110
11,330
1,967
24,243,525
4,275
5,671
1,611,759
163,165
356,546
56,403
162,410
90,647
87,626
694,962
1,389,797
85,218
111,558
71,893
255,654
181,282
10,970
673,222
Estimated Spending Reduction
8,337,520
3,560
2,342
Estimated Current Annual Spending
Region 2
Table 11. Estimated Spending and Spending Reduction by Region
200,620
4,542
8,176
10,902
32,705
0
1,363
142,932
3,223,730
3,610
893
Region 5
9,341,363
0
0
1,165,258
2,565,707
1,106,461
1,122
4,502,815
5,613,000
3,000
1,871
Out of State
88,802,676
7,228,005
12,893,912
10,273,256
12,746,457
7,910,407
2,758,617
34,992,022
213,193,260
9,300
22,922
All Boaters
GEMS allows us to create inputs in the form of reduced consumer spending at businesses that would typically be used by boaters (marinas, gas stations, restaurants, etc.). Using
the survey and inference population data, we prepared model inputs for each region. For
each category of spending (transportation, boat launch fees, fuel, lodging, restaurant meals,
other food and beverage, fishing supplies, and other purposes), we multiplied the reduction in trips by the mean (or median) amount spent for that purpose. We then multiplied
by the proportion of boaters who report spending in that category. After completing this
calculation for each spending category for each region, we applied those inputs (reductions
to economic activity) to the economic model.
The total reduction in state spending by boaters’ reduction in outings on the ICW
is nearly $89 million. Boaters reported different mixes of spending among the different
components. Table 11 shows the reduction in spending by region. Not all boaters spent in
each category, so we calculated the proportion who spent on each component within each
region. For example, boaters residing in Region 1 were much less likely to spend for lodging and restaurant meals. We then calculated the typical spending for each component for
each region using only those cases reporting such spending. Since outliers in the spending
amount data inflated the means severely, we used median figures for these calculations in
order to produce estimates that are both more conservative and representative of the central
tendency for spending. Using the median rather than the mean results in a zero input for
some items such as lodging and restaurant meals in Region 2. We then estimated spending by component for only the proportion of the inference population suggested by the
proportion within each region who gave a positive response for spending in that category.
This estimation technique results in a conservative estimate of boater spending.
The total estimated economic impact on the State of Georgia from the loss of spending
is $124.5 million (in 2008 dollars). GEMS estimates that the reduction in spending would
result in a loss of 2,136 jobs and about $54 million in personal income (see Table 12). As
expected, most of the impact is in Region 1 with about 85 percent of the lost economic
activity and around 80 percent of the lost jobs and personal income (see Table 13). Tables
14–17 present the estimated impacts in Regions 2 through 5. Most jobs are lost in the retail
trade sector followed by accommodation (lodging) and food services. Jobs are also lost in
the entertainment, health care, and waste management sectors.
In addition to lost jobs, personal income, and economic activity, the reduction in ­boater
activity will result in lower government revenues, primarily in the form of sales taxes.
Overall, state and local government revenues are estimated to fall by nearly $15 million
(see Table 18). Forty-five percent of this loss is at the state level in lost sales and income
taxes. Sales tax, business taxes, and other fees account for the loss at the local level. As with
the other measures of economic activity, the largest impact in government revenue is felt
in Region 1.
Business Survey
The business survey was sent to 666 individual businesses in the coastal counties of Georgia.
Only 42 responses were received, but 55 percent of businesses that responded reported that
they were somewhat or very dependent on the ICW either to serve their customers or for
their customers to have access to their place of business. Responses were received from 10
marinas, 6 barge operators, 7 realtors and construction companies, 6 shrimpers or commercial
fishing companies, and 13 others that included boat storage and boat tour businesses. Eighty
percent of respondents (34 of 42) identified their businesses as largely marine related.
20
21
14
0
12
130
129
116
11,214
121
117
117
119
130
120
145
113
166
180
1455
139
117
0
12,136
9,020
20,573
337,723
472,931
237,553
584,356
227,658
159,913
216,065
197,778
339,888
72,009
386,880
91,216
437,414
79,293
350,721
288,072
734,914
324
5,332,727
b
Employment
ICW
Impacta
88,425
Number of jobs
Thousands of dollars
Source: Georgia Economic Modeling System
a
Agriculture, Forestry,
Fishing, and Hunting
Mining
Utilities
Construction
Manufacturing
Wholesale Trade
Retail Trade
Transportation and
Warehousing
Information
Finance and Insurance
Real Estate and Rental
and Leasing
Professional, Scientific,
and Technical Services
Management of Companies
and Enterprises
Waste Management and
Remediation Services
Educational Services
Health Care and Social
­Assistance
Arts, Entertainment,
and Recreation
Accommodation and
Food Services
Other Services (except
Public Administration)
Government
Non-NAICS Industries
Total
2008
Baselinea
0.00
0.00
10.04
10.01
10.13
10.10
10.02
10.01
10.01
10.03
10.01
10.01
10.01
10.01
10.01
0.00
10.01
10.01
10.01
10.01
10.21
0.00
Percent
38,227,139
23,440,944
624,981,132
12,692,218
15,143,440
3,805,335
29,099,154
3,571,481
16,516,262
20,101,773
32,085,593
27,205,969
55,071,225
44,092,256
33,350,087
837,615
12,007,362
22,454,741
138,052,322
53,969,810
36,308,793
6,947,612
1877
11,020
1124,551
11,611
118,989
13,137
14,097
1446
11,577
13,831
12,047
11,475
13,559
12,678
12,276
124
1999
11,686
16,854
12,925
164,089
1353
0.00
0.00
10.02
10.01
10.13
10.08
10.01
10.01
10.01
10.02
10.01
10.01
10.01
10.01
10.01
0.00
10.01
10.01
0.00
10.01
10.18
10.01
Total Economic Output
ICW
2008
Impactb
Baselineb
Percent
Table 12. State of Georgia, Economic Impact from Reduction in Boater Outings
37,872,059
387
238,729,390
7,516,998
6,403,321
2,055,693
20,604,694
2,985,501
10,858,494
5,253,041
22,357,900
5,487,982
12,880,907
13,212,121
12,669,944
507,244
3,150,177
14,037,746
26,851,516
16,032,163
16,787,964
1,203,538
1864
0
154,242
1944
17,947
11,393
12,937
1373
1954
11,001
11,429
1360
1934
1756
11,010
115
1263
11,054
11,433
1869
129,632
171
0.00
0.00
10.02
10.01
10.12
10.07
10.01
10.01
10.01
10.02
10.01
10.01
10.01
10.01
10.01
0.00
10.01
10.01
10.01
10.01
10.18
10.01
Personal Income
ICW
2008
Impactb
Baselineb
Percent
22
10.5
0.0
11.1
115.1
17.5
17.4
11,126.7
19.7
19.6
17.6
111.8
114.9
115.1
126.9
17.7
137.3
172.0
1419.8
121.4
18.7
0.0
11,820.8
77
1,078
16,385
17,498
7,130
32,301
13,079
4,112
8,349
9,390
10,976
4,049
18,874
4,694
26,067
5,066
28,081
16,518
63,479
15
288,857
b
ICW
Impacta
1,639
Number of jobs
Thousands of dollars
Source: Georgia Economic Modeling System
a
Agriculture, Forestry,
Fishing, and Hunting
Mining
Utilities
Construction
Manufacturing
Wholesale Trade
Retail Trade
Transportation and
Warehousing
Information
Finance and Insurance
Real Estate and Rental
and Leasing
Professional, Scientific,
and Technical Services
Management of Companies
and Enterprises
Waste Management and
Remediation Services
Educational Services
Health Care and Social
Assistance
Arts, Entertainment,
and Recreation
Accommodation and
Food Services
Other Services (except
Public Administration)
Government
Non-NAICS Industries
Total
2008
Baselinea
Employment
10.01
10.04
10.63
10.13
11.49
11.42
10.14
10.17
10.14
10.37
10.14
10.13
10.23
10.09
10.07
10.02
10.11
10.09
10.04
10.10
13.49
10.03
Percent
$3,955,435
$1,119,275
$24,338,613
$718,708
$1,160,835
$179,725
$1,732,617
$167,766
$587,914
$725,667
$700,500
$677,179
$733,589
$1,054,423
$1,154,623
$5,140
$365,695
$938,998
$5,223,579
$1,224,281
$1,741,152
$171,510
1$496
1$480
1$97,376
1$909
1$17,582
1$2,852
1$2,433
1$277
1$874
1$2,704
1$904
1$835
1$1,436
1$1,024
1$892
1$1
1$389
1$863
1$1,790
1$1,267
1$59,308
1$60
ICW
Impactb
10.01
10.04
10.40
10.13
11.51
11.59
10.14
10.17
10.15
10.37
10.13
10.12
10.20
10.10
10.08
10.02
10.11
10.09
10.03
10.10
13.41
10.04
Percent
Total Economic Output
2008
Baselineb
Table 13. Region 1 Economic Impact from Reduction in Boater Outings
$3,677,448
$18
$11,492,171
$415,967
$486,238
$84,641
$1,233,750
$140,240
$378,133
$189,633
$488,272
$167,187
$182,300
$323,888
$623,986
$3,105
$98,028
$587,022
$1,199,128
$363,682
$805,050
$44,454
2008
Baselineb
1$474
$0
1$43,693
1$528
1$7,354
1$1,254
1$1,747
1$231
1$530
1$707
1$628
1$206
1$425
1$284
1$459
1$1
1$104
1$540
1$407
1$376
1$27,422
1$16
ICW
Impactb
Personal Income
10.01
10.04
10.38
10.13
11.51
11.48
10.14
10.17
10.14
10.37
10.13
10.12
10.23
10.09
10.07
10.02
10.11
10.09
10.03
10.10
13.41
10.04
Percent
23
14.95327
11.41245
19.77164
12.53687
16.48949
17.65593
13.34041
10.00276
189.3901
5,332
1,631
12,613
744
9,148
9,867
29,787
9
160,105
12.06407
3,187
10.93114
11.34325
12.23586
1,553
4,249
742
13.66182
8,446
13.26734
10.03373
10.53354
16.71184
16.746
11.79203
122.71575
214
938
11,777
19,814
3,905
20,032
4,223
11.19091
b
Employment
ICW
Impacta
11,898
Number of jobs
Thousands of dollars
Source: Georgia Economic Modeling System
a
Agriculture, Forestry,
Fishing, and Hunting
Mining
Utilities
Construction
Manufacturing
Wholesale Trade
Retail Trade
Transportation and
Warehousing
Information
Finance and Insurance
Real Estate and Rental
and Leasing
Professional, Scientific,
and Technical Services
Management of Companies
and Enterprises
Waste Management and
Remediation Services
Educational Services
Health Care and Social
Assistance
Arts, Entertainment,
and Recreation
Accommodation and
Food Services
Other Services (except
Public Administration)
Government
Non-NAICS Industries
Total
2008
Baselinea
10.01
10.03
10.06
10.08
10.07
10.34
10.08
10.09
10.09
10.13
10.08
10.06
10.09
10.05
10.04
10.02
10.06
10.06
10.03
10.05
10.11
10.01
Percent
$1,123,158
$675,731
$12,265,747
$355,078
$275,870
$17,354
$659,923
$40,157
$173,258
$111,766
$168,641
$126,052
$243,945
$464,456
$673,213
$15,257
$328,277
$501,428
$3,765,672
$541,583
$984,194
$1,020,733
1$124
1$200
1$5,713
1$263
1$195
1$60
1$506
1$35
1$159
1$140
1$131
1$76
1$184
1$269
1$289
1$2
1$187
1$286
1$1,148
1$249
1$1,101
1$110
10.01
10.03
10.05
10.07
10.07
10.35
10.08
10.09
10.09
10.13
10.08
10.06
10.08
10.06
10.04
10.01
10.06
10.06
10.03
10.05
10.11
10.01
Total Economic Output
ICW
2008
Impactb
Baselineb
Percent
Table 14. Region 2, Economic Impact from Reduction in Boater Outings
$1,170,189
$11
$4,791,879
$209,440
$117,390
$8,057
$466,611
$33,569
$82,166
$29,207
$118,917
$37,350
$52,932
$120,319
$292,136
$9,337
$85,829
$313,471
$849,124
$160,882
$455,058
$179,884
1$131
$0
1$2,320
1$157
1$83
1$27
1$358
1$29
1$73
1$37
1$94
1$24
1$43
1$64
1$125
1$1
1$49
1$179
1$241
1$74
1$509
1$23
10.01
10.03
10.05
10.07
10.07
10.33
10.08
10.09
10.09
10.13
10.08
10.06
10.08
10.05
10.04
10.01
10.06
10.06
10.03
10.05
10.11
10.01
Personal Income
ICW
2008
Impactb
Baselineb
Percent
24
11.2
10.1
10.4
12.9
16.2
11.7
134.2
12.6
11.3
12.0
11.7
13.1
10.9
14.3
10.9
19.0
13.6
112.3
14.3
12.8
0.0
195.4
26,553
2,992
3,076
36,430
73,419
17,931
80,292
25,150
8,161
18,721
16,248
24,594
3,670
34,562
6,715
69,929
7,256
43,673
37,707
149,839
38
686,957
b
Number of jobs
Thousands of dollars
Source: Georgia Economic Modeling System
a
Agriculture, Forestry,
Fishing and Hunting
Mining
Utilities
Construction
Manufacturing
Wholesale Trade
Retail Trade
Transportation and
Warehousing
Information
Finance and Insurance
Real Estate and Rental
and Leasing
Professional, Scientific, and
Technical Services
Management of Companies
and Enterprises
Waste Management and
Remediation Services
Educational Services
Health Care and Social
Assistance
Arts, Entertainment,
and Recreation
Accommodation and
Food Services
Other Services (except
Public Administration)
Government
Non-NAICS Industries
Total
0.00
0.00
10.01
10.01
10.03
10.05
10.01
10.01
10.01
10.02
10.01
10.01
10.02
10.01
10.01
0.00
10.01
10.01
10.01
10.01
10.04
0.00
Employment
2008
ICW
Baselinea
Impacta Percent
$7,952,634
$2,733,439
$54,971,326
$1,412,275
$1,377,646
$240,558
$3,879,808
$181,820
$1,122,942
$648,947
$1,555,655
$1,172,483
$1,350,320
$2,212,738
$2,249,112
$310,568
$970,746
$1,832,342
$15,019,712
$2,480,423
$4,191,541
$2,075,619
1$133
1$123
1$6,283
1$157
1$405
1$113
1$488
1$24
1$137
1$156
1$182
1$101
1$196
1$243
1$240
1$11
1$121
1$144
1$1,209
1$238
1$1,767
1$93
0.00
0.00
10.01
10.01
10.03
10.05
10.01
10.01
10.01
10.02
10.01
10.01
10.01
10.01
10.01
0.00
10.01
10.01
10.01
10.01
10.04
0.00
Total Economic Output
2008
ICW
Baselineb
Percent
Impactb
Table 15. Region 3, Economic Impact from Reduction in Boater Outings
$7,713,332
$45
$24,649,641
$832,766
$584,517
$113,559
$2,823,302
$151,988
$656,487
$169,584
$1,019,715
$286,795
$314,028
$616,109
$1,181,786
$190,433
$254,172
$1,145,502
$3,548,267
$736,829
$1,938,027
$372,395
1$133
$0
1$2,648
1$94
1$171
1$50
1$358
1$20
1$79
1$41
1$124
1$28
1$50
1$63
1$119
1$7
1$32
1$90
1$284
1$71
1$817
1$18
Personal Income
2008
ICW
Baselineb
Impactb
0.00
0.00
10.01
10.01
10.03
10.04
10.01
10.01
10.01
10.02
10.01
10.01
10.02
10.01
10.01
0.00
10.01
10.01
10.01
10.01
10.04
0.00
Percent
25
18.6
12.6
18.9
12.0
115.3
15.6
12.3
0.0
1118.3
304,436
73,735
295,217
60,969
245,395
198,588
438,330
234
3,763,078
13.3
154,999
12.8
14.8
15.0
140,744
173,403
62,362
14.9
165,351
18.0
10.1
10.4
15.0
17.0
14.7
126.1
5,078
13,292
238,320
276,182
195,636
401,025
280,278
11.0
39,502
b
Number of jobs
Thousands of dollars
Source: Georgia Economic Modeling System
a
Agriculture, Forestry, Fishing,
and Hunting
Mining
Utilities
Construction
Manufacturing
Wholesale Trade
Retail Trade
Transportation and
Warehousing
Information
Finance and Insurance
Real Estate and Rental
and Leasing
Professional, Scientific, and
Technical Services
Management of Companies
and Enterprises
Waste Management and
Remediation Services
Educational Services
Health Care and Social
Assistance
Arts, Entertainment, and
Recreation
Accommodation and
Food Services
Other Services (except Public
Administration)
Government
Non-NAICS Industries
Total
Employment
2008
ICW
Baselinea
Impacta
0.00
0.00
0.00
0.00
10.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
10.01
0.00
Percent
$23,091,482
$16,928,801
$491,360,437
$9,233,715
$11,517,899
$3,218,254
$20,825,659
$3,055,105
$13,747,732
$18,382,275
$28,619,041
$24,343,670
$51,706,208
$38,995,100
$27,966,038
$449,232
$9,650,840
$17,595,378
$94,665,528
$47,753,058
$26,651,641
$2,963,780
1$115
1$199
1$14,233
1$261
1$757
1$107
1$625
1$107
1$389
1$823
1$808
1$451
1$1,715
1$1,112
1$827
1$9
1$288
1$368
1$2,369
1$1,138
1$1,689
1$75
0.00
0.00
0.00
0.00
10.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
10.01
0.00
Total Economic Output
2008
ICW
Baselineb
Percent
Impactb
Table 16. Region 4, Economic Impact from Reduction in Boater Outings
$23,111,631
$279
$183,058,204
$5,481,827
$4,869,579
$1,780,957
$14,647,160
$2,553,848
$9,251,570
$4,803,698
$19,986,061
$4,788,155
$12,106,425
$11,778,680
$10,031,423
$270,693
$2,529,742
$10,999,880
$17,030,956
$14,185,427
$12,322,822
$527,392
1$117
$0
1$5,233
1$154
1$318
1$59
1$442
1$89
1$261
1$215
1$567
1$98
1$410
1$336
1$296
1$5
1$75
1$230
1$427
1$338
1$781
1$14
Personal Income
2008
ICW
Baselineb
Impactb
0.00
0.00
0.00
0.00
10.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
10.01
0.00
Percent
26
10.2
0.0
0.0
10.6
11.5
10.2
14.2
10.3
10.1
10.2
10.2
10.4
0.0
10.5
10.1
10.8
10.2
11.5
10.5
10.2
0.0
111.9
660
2,188
34,811
86,018
12,951
50,705
15,633
5,344
11,343
13,955
19,817
1,186
23,676
4,441
33,587
5,259
24,423
25,393
53,479
27
433,730
b
0.00
0.00
0.00
0.00
10.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
10.01
0.00
Employment
ICW
Percent
Impacta
8,833
Number of jobs
Thousands of dollars
Source: Georgia Economic Modeling System
a
Agriculture, Forestry, Fishing,
and Hunting
Mining
Utilities
Construction
Manufacturing
Wholesale Trade
Retail Trade
Transportation and
Warehousing
Information
Finance and Insurance
Real Estate and Rental
and Leasing
Professional, Scientific, and
Technical Services
Management of Companies
and Enterprises
Waste Management and
Remediation Services
Educational Services
Health Care and Social
Assistance
Arts, Entertainment,
and Recreation
Accommodation and
Food Services
Other Services (except
Public Administration)
Government
Non-NAICS Industries
Total
2008
Baselinea
$2,104,430
$1,983,697
$42,045,009
$972,442
$811,191
$149,444
$2,001,147
$126,632
$884,416
$233,118
$1,041,755
$886,586
$1,037,163
$1,365,538
$1,307,100
$57,418
$691,803
$1,586,595
$19,377,832
$1,970,465
$2,740,266
$715,971
1$9
1$18
1$946
1$20
1$50
1$5
1$45
1$3
1$19
1$8
1$23
1$13
1$27
1$29
1$27
1$1
1$15
1$26
1$338
1$33
1$224
1$14
0.00
0.00
0.00
0.00
10.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
10.01
0.00
Total Economic Output
2008
ICW
Baselineb
Percent
Impactb
Table 17. Region 5, Economic Impact from Reduction in Boater Outings
$2,199,460
$33
$14,737,495
$576,998
$345,597
$68,478
$1,433,871
$105,855
$490,138
$60,919
$744,935
$208,496
$225,222
$373,125
$540,612
$33,677
$182,406
$991,872
$4,224,041
$585,342
$1,267,007
$79,412
1$9
$0
1$347
1$12
1$21
1$2
1$32
1$3
1$10
1$2
1$16
1$3
1$6
1$8
1$11
$0
1$4
1$16
1$75
1$10
1$104
1$1
0.00
0.00
0.00
0.00
10.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
10.01
0.00
Personal Income
2008
ICW
Baselineb
Percent
Impactb
Table 18. Lost Government Revenue
Fiscal Impactsa
State Government Revenue
Local Government Revenue
Totala
Region 1
$4,780
$5,768
$10,548
Region 2
$869
$1,049
$1,918
Region 3
$340
$411
$751
Region 4
$675
$815
$1,490
Region 5
$56
$68
$124
Total
$6,721
$8,110
$14,831
a
Thousands of dollars
Source: Georgia Economic Modeling System
Two-thirds of respondents (28 of 42) told us that their business was located on or with
easy access to the waterway, and two-thirds of those (19) indicated that their business relied
on easy access. Seven respondents told us that they thought deterioration of the ICW had
affected their business and that their revenues were down by about 28 percent on average in the past year. It cannot be determined whether the reduction in revenue is due to
deterioration of the ICW, but these business owners’ perception is that the loss of channel
depth has affected them.
Conclusion
Continued deterioration of the channel along the length of the ICW in Georgia could harm
the economy of the coastal region if boaters reduce their spending by $89 million as our data
suggest. The reduction in recreational boating in the coastal counties could result in nearly a
$100 million loss to that economy, with an additional $27 million lost elsewhere in the state.
More than 2,100 jobs with $54 million in personal income would be lost as a result.
The loss of economic activity would reduce state and local government revenue by
nearly $15 million annually. Most of the loss would be in state and local option sales taxes,
but less business license and property tax revenue would be produced as well.
Marine-related businesses rely on the ICW as a transportation route for servicing
customers in the coastal region and the deep water harbors at Savannah and Brunswick.
Those harbors handled more than 34 million tons of goods in 2006 and continue to grow
in the volume of both imported and exported goods.
Works Cited
Boat Owners Association. 2000. Boat/US Magazine (May).
Institute for Water Resources. 2002, 2006. Waterborne Commerce of the United States. Alexandria, Va.:
U.S. Army Corps of Engineers.
Kaplowitz, Michael D., Timothy D. Hadlock, and Ralph Levine. 2004. A Comparison of Web and
Mail Survey Response Rates. Public Opinion Quarterly. Vol. 68, No. 1: 94–101.
Leeworthy, Vernon R., Peter C. Wiley, Donald B. K. English, Warren Kriesel. 2001. Correcting
Response Bias in Tourist Spending Surveys. Annals of Tourism Research. Vol. 28, No. 1: 83–97.
Parkman, Aubrey. 1983. History of the Waterways of the Atlantic Coast of the United States. Army Corps
of Engineers publication # NWS 83-10.
Sullivan, Buddy. 2005. “Atlantic Intracoastal Waterway.” The New Georgia Encyclopedia. www.georgia
encyclopedia.org.
Young, Claiborne S. 2007. Cruising Guide to Coastal South Carolina and Georgia. Winston-Salem,
N.C.: John F. Blair.
27
Appendix A
[Note: Both surveys were administered on line and have been converted here to text
format. It is impossible to reproduce the skip patterns that were transparent to respondents as they completed the survey. In some instances certain responses to one question caused one or more questions to be skipped. In the case of the business survey in
Appendix B, there were complete sets of questions for each type of business surveyed.
For example, the on line instrument had different batteries of questions for construction
companies and commercial fishing enterprises.]
Georgia and the Intracoastal Waterway
Welcome to the Carl Vinson Institute of Government’s on-line survey of users of Georgia’s section of the Intracoastal Waterway (ICW). This survey is sponsored by the Vinson
Institute and the Marine Extension Service of the University of Georgia as part of a
study to determine the effects of the federal government’s decision to stop dredging the
ICW. The study will determine the likely loss of use and the economic impact that will
have in Georgia’s coastal counties. You have been invited to participate in this survey
because you are a registered boater. Please be assured that your responses will be kept
in confidence and only aggregate statistics from all responses will be made public.
Enter access code:
Be advised that all Internet communications are insecure and there is a limit to the
confidentiality that can be guaranteed due to the technology itself. However once the
materials are received by the researcher, standard confidentiality procedures will be
employed. Additional questions or problems regarding your rights as a research participant should be addressed to The Chairperson, Institutional Review Board, University of
Georgia, 612 Boyd Graduate Studies Research Center, Athens, Georgia 30602-7411;
Telephone (706) 542-3199; E-Mail Address IRB@uga.edu
Please tell us about your boat.
What is your boat’s overall length? (Please provide answer in feet.)
What is your boat’s normal draft?
In what year was the boat manufactured?
What is the main form of propulsion for your boat?
____ Sail
____ In-board motor
____ Out-board motor
____ Other
During the boating season, is your boat usually . . .
____ Carried by trailer to launch sites
____ Kept at a marina
____ Kept at a private dock at your home
____ Kept somewhere else (please tell us where):
How many miles is it from your home to the place where you usually launch your boat
for outings on the ICW in Georgia?
How many miles is it from your home to the marina where you keep your boat?
What is the annual cost to keep your boat at the marina?
How many miles is the private dock at your home from the ICW in Georgia?
How much does it cost you annually to stow your boat when not in use?
Please tell us about your boating outings.
How many separate boat outings on the ICW in Georgia did you take on your boat in
the past two months?
On average, what was the length in days of one of these boat outings on the ICW in
Georgia that you took during the past two months?
30
How many separate boat outings on the ICW in Georgia did you take on your boat
during the past 12 months?
On average, what was the length in days of one of these boat outings on the ICW in
Georgia that you took during the past 12 months?
Was the past 12 months a typical year in terms of the number of boat outings you took
on the ICW in Georgia?
How many boat outings do you normally take on the ICW in Georgia during a typical year?
How many of these outings were mainly for fishing? How many boat outings on your
primary boat do you think you will take during the next 12 months? (Please take your
best guess.)
The purpose of the following questions is to assess what level of spending on
boating is spent in the coastal counties; those counties include Chatham, Bryan,
Liberty, McIntosh, Glynn, and Camden.
Thinking about your typical outing on the ICW in Georgia during the past 12 months,
please estimate your expenses for the following items.
[Note: the online survey requested inputs for the total amounts spent for the following
purposes and the amounts spent in the coastal counties.]
_______ Car/Truck Transportation
_______ Boat Launch Fee
_______ Boat Fuel Costs
_______ Lodging (Hotel, campsite, etc.)
_______ Restaurant Meals
_______ Other Food and Beverage
_______ Fishing Supplies
_______ Other
Do you own your own boat, or do you rent or lease your boat?
____ Own
____ Rent
____ Lease
31
About how much do you think you could get for your boat if you sold it?
About how much do you spend renting or leasing a boat for a typical outing?
How often do you hire a captain to operate the boat for you?
____ For every outing
____ For more than half of the outings, but not all
____ For about half of the outings
____ For less than half of the outings
____ Never
The Intracoastal Waterway has an authorized navigable depth of 12 feet. It is actually
maintained at depths ranging from 7 to 12 feet. The depth of the Georgia portion of the
intracoastal Waterway can be as shallow as four feet in places. Considering your own
boat, what is your opinion of the navigability of the Georgia portion of the Intracoastal
Waterway?
____ Excellent
____ Very good
____ Good
____ Fair
____ Poor
Suppose that the dredging of the ICW was increased and the average depth of the
ICW in Georgia portion was about 12 feet. Thinking about the number of boat outings
you said you would take during the next 12 months . . . if the depth was consistently 12
feet, would you take more, fewer, or about the same number of boat outings?
____ More
____ The same
____ Fewer
____ None
About how many more outings are you likely to take?
About how many fewer outings are you likely to take?
32
If you take no boating outings because the average depth is 12 feet rather than XX
feet, what would you do with your boat?
____ Sell it
____ Store it
____ Not sure
Suppose that the dredging of the ICW completely stopped and the average depth of
the Georgia portion was about 4 feet. Thinking about the number of boat outings that
you said you would take during the next 12 months… if the average depth were only 4
feet, would you take more, fewer, or about the same number of boat outings?
____ More
____ The same
____ Fewer
____ None
About how many more outings are you likely to take?
About how many fewer outings are you likely to take?
If you take no boating outings because the average depth is 4 feet, what would you do
with your boat?
____ Sell it
____ Store it
____ Not sure
If you were unable to take a boat outing on the ICW due to a decrease in its average
depth to 4 feet, what would you do instead?
Considering your own boat . . . with an average depth of 12 feet for the Georgia portion
of the ICW, what is your opinion of the navigability of the ICW in Georgia?
____ Excellent
____ Very good
____ Good
____ Fair
____ Poor
33
Federal government funds for the dredging and maintenance of the ICW are threatened.
If dredging completely stops, the average depth of the Georgia portion of the ICW would
be about 4 feet. A Georgia dredging and maintenance program would provide enough
funding to maintain an average depth of 12 feet. The program would be funded by a $20
surcharge on your boating registration fee. Each registered boater with a craft longer
than 16 feet using the ICW would be required to display a sticker to be place alongside
the registration number on the boat. Would you be willing to pay a $20 fee in addition
to the annual registration fee to fund this program?
____ Yes
____ No
____ Not sure
Would you be willing to pay $10 each year in addition to the annual registration fee to
fund this program?
____ Yes
____ No
____ Not sure
What is your primary objection to a $10 fee for a Georgia dredging and maintenance fee?
____ I don’t trust the Army Corps of Engineers.
____ I don’t have enough money.
____ Lack of enforcement. I wouldn’t get caught.
____ Commercial boaters should pay.
____ I’m not affected by a shallow ICW.
____ I don’t think it’s fair to make boaters pay.
____ I don’t trust the government.
____ Other
On a scale of 1 to 10 where 1 is “Very Unsure” and 10 is “Very Sure,” how sure are you
that you are willing to pay for the program?
Please tell us about you.
Your gender:
____ Male
____ Female
____ Decline to say
34
Do you consider yourself . . .
____ White, non-Hispanic
____ African American
____ Hispanic
____ Asian, or Pacific Islander
____ American Indian or Alaskan native
____ Other
____ Decline to say
In what year were you born?
In what state is your primary residence?
What is the Zip Code of your permanent residence?
Lastly, a few questions about your household . . .
How many people, including yourself, live in your household?
How many people under 18 years of age live in your household?
Are you . . .
____ Married
____ Single
____ Divorced
____ Widowed
____ Separated
____ Decline to answer
What is the highest level of education that you have completed?
____ Less than high school ____ Masters degree
____ High school graduate ____ Doctorate degree
____ S
ome college but not a ____ Professional degree
college graduate
(e.g. JD, MD, etc.)
____ Associate’s degree ____ Decline to answer
____ 4-year college degree
35
What is your occupation?
As close as you can recall, what is your household’s total annual income for 2007
before taxes?
____ $15,000 or less
____ $15,001 to $25,000
____ $25,001 to $30,000
____ $30,001 to $35,000
____ $35,001 to $40,000
____ $40,001 to $45,000
____ $45,001 to $50,000
____ $50,001 to $60,000
____ $60,001 to $75,000
____ $75,000 to $100,000
____ More than $100,000
____ Decline to answer
Thank you very much for participating in this important study of the Atlantic Intracoastal
Waterway. Your participation will help the state of Georgia move forward in addressing
issues related to the ICW. If you have any questions about this study, please contact
Dr. Wes Clarke at the Carl Vinson Institute of government by phone (706-542-6202) or
email (wclarke@cviog.uga.edu).
36
Appendix B
Intracoastal Waterway Business Survey
The Marine Extension Service and the Carl Vinson Institute of Government at the University of Georgia are conducting a survey to help determine the economic impact of the
Intracoastal Waterway (ICW) in Georgia and the impact that additional maintenance of
the channel would have on businesses in the coastal region that serve persons using the
ICW. Your cooperation will provide input for identifying and quantifying business sales,
personal income, and employment attributable to the ICW in Georgia. Your cooperation
is greatly appreciated and your responses will be held confidential.
What is your name?
What is the name of your company?
What is your title at this company?
What is your ten digit phone number (area code plus phone number)?
What is the physical address of your company?
What is the mailing address of your company?
Now we would like to ask you a few questions about your business. Please remember
that your answers are confidential and will only be reported in group form.
What percentage of your business is: (Total of all categories should be 100%)
% ____ Construction or Real Estate
% ____ Manufacturing
% ____ Transportation
% ____ Wholesale Trade
% ____ Retail Trade
% ____ Finance
% ____ Boat Brokerage
% ____ Services
% ____ Commercial Fishing/ Shrimping
% ____ Other (please specify)
% ____ Total
To what extent is your business dependent on the ICW?
____ Very dependent
____ Somewhat dependent
____ Not very dependent
____ Not at all dependent
Are you located on or with easy access to the ICW?
____ Yes
____ No
Is your business dependent on being located on or with easy access to the ICW?
____ Yes
____ No
What percent of your business volume (sales) do you consider to be marine or water
related?
What percentage of your employees are full-time equivalent? (Full-time equivalent =
2,000 hr/yr; 40 hours per week)
For your operations, please indicate the range of your gross revenue for your most
recent fiscal or calendar year. (Your response will be held confidential and will only be
disclosed in consolidated form.)
____ $0–$500,000
____ $500,001–$1,000,000
____ $1,000,001–$2,500,000
____ $2,500,001–$5,000,000
____ $5,000,001 or greater
38
What is the nature of the business that you operate? (Check all that apply)
____ Marina Operator
____ Barge or transportation
____ Realtor or construction company
____ Shrimper/Commercial fishing
____ None of the above
Please indicate the amenities available at your facility: (Check all that apply)
____ Dry boat storage
____ Boat Sales
____ Boat leasing/rental
____ Boating equipment/Supplies sales
____ Boat hoist/Launch
____ Convenience/Boat Store
____ Bait/Tackle sales
____ Restaurant
____ Fuel sales
____ Snack bar
____ Dive shop
____ Lodging (Motel/Condos)
____ Shower/Laundromat
____ Boat/Propeller repairs
____ Dump Station
____ Engine repairs
____ Other (please specify)
In the previous question, you indicated that your facility had dry boat storage space.
Please indicate the number of dry boat spaces at your facility.
You indicated that your facility leases or rents boats. What is the total value of the inventory you lease and/or rent? (This information will only be presented in group form. The
information will help us estimate the economic resources that are operated on the ICW.)
39
If maintenance on the ICW were decreased and vessel drafts were limited to 3 feet or
less would your business be affected?
____ Yes
____ No
In your opinion, has your business revenue decreased in the last 5 years as a result of
deterioration of the ICW channel?
____ Yes
____ No
By what percentage do you think your business has decreased as a result of the deteri­
or­ation of the ICW channel?
Does your establishment have docking facilities on the ICW?
____ Yes
____ No
Please answer the following questions about the docking facilities at your estabilshment.
What is the length of the largest vessel that can be accommodated at your facility?
What is the draft of the largest vessel that can be accommodated at your facility?
How many transient slips/berths are currently available?
What is the total number of slips/berths available?
Limiting bridge clearance restricting access to your facility, if any:
Do you have the capacity to handle powerboats?
____ Yes
____ No
Do you have the capacity to handle sailboats?
____ Yes
____ No
40
Is use of your facility open to the public or restricted to private/guest use only (such as
condo, hotel, club, etc.)?
____ Private
____ Public
How many barges to you operate?
What is the estimated value of your fleet? (This information will only be presented in
group form. The information will help us estimate the economic resources that are
operated on the ICW.)
How much tonnage do you move annually?
Are you restricted in tonnage due to the current conditions of the ICW?
____ Yes
____ No
Please estimate the percent of capacity your barges normally transport. Do you ever
have to delay transport because of low tide at Hell’s Gate?
____ Yes
____ No
How often do you have to take a route OTHER than the ICW when the ICW would be
your preferred route?
____ Very often
____ Sometimes
____ Rarely
____ Never
In an average year, what is the cost of delays or having to use abnormal routes?
If the ICW were to close to transient traffic (traffic other than local recreational vessels)
due to lack of maintenance or otherwise, to what degree would this affect your ability
to do the following:
41
[The following scale was used for each of the items below. These were asked of each
business.]
____ Does not apply to my business
____ No effect
____ Moderately affect
____ Highly affect
Remove your vessel(s)/equipment from threat of storm or hurricane damage by lateral
coastal relocation?
Provide support to the US government for its supply, war, or security effort?
Maintain your current level of employees?
Maintain your current level of customers?
Maintain your current level of revenue?
Please indicate the type of business you operate. (Please check all that apply)
____ Realtor
____ Residential Construction
____ Commercial Construction
____ Construction Sub-Contractor
____ Other (please specify)
Does your business own equipment that travels on the ICW (boats, cranes, etc.)?
____ Yes
____ No
What is the estimated value of your equipment? (This information will only be presented
in group form. The information will help us estimate the economic resources that are
operated on the ICW.)
What percentage of your business involves property on or adjacent to the ICW?
42
In the past 5 years, has this percentage increased, decreased, or remained the same?
____ Increased
____ Decreased
____ Remained the same
If maintenance on the ICW were decreased and vessel drafts were limited to 3 feet or
less would your business be affected?
____ Yes
____ No
In your opinion, has your business revenue decreased in the last 5 years as a result of
deterioration of the ICW channel?
____ Yes
____ No
By what percentage do you think your business has decreased as a result of the deterioration of the ICW channel?
What is the estimated value of your fleet? (This information will only be presented in
group form. The information will help us estimate the economic resources that are
operated on the ICW.)
During your operational season, how frequently do you use the ICW for your commercial operation?
____ Very frequently
____ Somewhat frequently
____ Seldom
____ Never
If maintenance on the ICW were decreased and vessel drafts were limited to 3 feet or
less would your business be affected?
____ Yes
____ No
43
In your opinion, has your business revenue decreased in the last 5 years as a result of
deterioration of the ICW channel?
____ Yes
____ No
By what percentage do you think your business has decreased as a result of the deterioration of the ICW channel?
If you were unable to navigate the ICW in Georgia, would you be able to use an alternate route?
____ Yes
____ No
____ Not sure
If you were unable to navigate the ICW in Georgia would you move your commercial
operation to another state?
____ Yes
____ No
____ Not sure
Are there any comments you would like to make about the ICW in general, the maintenance of the Waterways, or the need for additional facilities or improvements on the
Waterways?
If you would like a copy of the report emailed to you at the conclusion of the study,
please enter your email address below. If not, please leave the field blank and click
the “next” button.
Thank you so much for participating in this survey. We appreciate your time.
44
Appendix A:
C
The Georgia Economic Modeling System:
A Multisector, Multiyear, Multimodal, County-Level
Computable Geographical Equilibrium Model of the U.S. Economy
INTRODUCTION
Paul Krugman (1998) expressed a hope that the new economic geography research
might one day develop “‘computable geographical equilibrium’ models, which can be
used to predict the effects of policy changes, technological shocks, etc., on the
economy’s spatial structure in the same way that such models are currently used to
predict the effects of changes in taxes and trade policy on the economy’s industrial
structure.” However, he acknowledges that “preliminary efforts in this direction by
several researchers, myself included, have found that such models are not at all
easy to calibrate to actual data.” It is the objective of this paper to unite several
different threads of economic research to develop the framework for just such a
regional “computable geographic equilibrium” model of the U.S. economy. Key tools
and concepts that will be incorporated into the model will include input-output
analysis, Social Accounting Matrices, gravity modeling, and new economic geography. The model framework that is developed is extremely simple, at least by the
standards of most computable general equilibrium models, yet it is capable of
generating a wide range of extremely complex economic behaviors/outcomes, can
model these behaviors at an extremely fine level of geographic and sectoral detail,
and can be calibrated to “real world” data.
THE SECTOR-COMMODITY RELATIONSHIPS IN THE MODEL:
A MERGED IO-SAM FRAMEWORK
The data framework for the model is based on blending the traditional input-output
(IO) tables of Leontief (1941), Stone and Brown (1962), with the closely related
Social Accounting Matrix (SAM) framework as formalized by Pyatt and Round (1985)
based upon the earlier work of Stone that has become widely used in recent
decades. The beauty of the IO framework originally developed by Leontief is its utter
simplicity: each industry sells its output to itself, to other industries, or to final
demanders. Therefore, on a single table, one can capture all the activity in an
economy. Stone and Brown, however, observed that the Leontief IO table implicitly
failed to recognize that every industry uses a mix of commodities and that every
industry makes a mix of commodities. The commodities are a necessary component
to describe accurately and explicitly the system’s behavior. Mathematically, under
the make and use table configuration (a “make” table identifies total spending on
each commodity by each sector in the economy, and a “use” table identifies the total
sales of each commodity by each sector in the economy) of Stone and Brown,
“industries” can be interpreted as a transformation system that converts a menu of
commodities and factor inputs into a menu of commodities. Generally, the Stone and
Brown IO tables can be used to model industry behavior using either Leontief or
Cobb-Douglas production functions. The configuration is particularly well suited to
Cobb-Douglas functions because all cells can be interpreted as the constant budget
share of a Cobb-Douglas production function.
However, these traditional IO tables have very little to contribute when we
attempt to examine or model anything beyond the industry-commodity-industry interactions. SAMs attempt to address these shortcomings by explicitly introducing
household, government, and capital markets, and a host of behaviors such as taxation, intergovernmental transfers, etc. The SAM framework has the advantage of
being absolutely comprehensive, with every transaction type accounted for in some
cell of a SAM matrix. However, while a SAM is comprehensive from an accounting
perspective (every transaction shows up in some cell in the matrix), it is not complete
in an economic sense, in that each cell does not represent a unique exchange of a
commodity for money, as it does in an IO make and use table. This model begins
with an alternative framework that draws on the comprehensiveness of the SAM and
the simplicity and economic cohesion of the IO make and use tables. The proposed
framework involves viewing the economy as a continuous process in which every
sector of the economy is identified according to the menu of commodities they
purchase and the menus of commodities they sell. The resulting merged framework
is presented in Figure 1.
16
46
Financial
Capital
Physical
Capital
Land
Investors
Speculators
Land
Government
Goods
Transfer
Payments
Sectors
Labor
MAKE TABLE
Commodities
Producer
Commodities
Figure 1. A Merged SAM/IO Framework for the Make and Use Tables
Producers
Employed labor
Remittance cohorts
Government
Investors
Speculators
Land
Remittance
Cohorts
Employed
Labor
Commodities
Producers
Sector
Government
USE TABLE
Producer Commodities
Labor
Tran Pmts/taxes/fees
Government goods
Financial capital
Physical capital
Land
Note: The gray cells represent areas that are likely to contain either zeros or insignificantly small
transactions.
It is now possible to merge the IO and SAM methods of conceptualizing an
economy into a unified system. The unified system’s row elements in the make table
include all the various producer industries generally included in make tables. They
also include rows for a labor sector, “remittance cohort” sector (remembering that
unemployed labor, retirees, and other transfer recipients are accounted for explicitly
within this sector), and government. Finally, the make table adds “investor” rows to
47
17
produce financial capital and “speculator” rows to produce physical capital as is
described below.
The unified system also adds several columns to the traditional make table. The
new columns include a “labor commodity” representing the wage bill produced by the
labor sector added above as a make table row; a transfer payments commodity; and
federal, state, and local government commodities. They also include “financial
capital” columns to represent commodities (dividends, interest, and rent) produced
by the investor sector through the savings process and “physical capital” columns to
represent the residential and nonresidential capital commodity outputs of the
speculator industries.
Several columns in the make table require additional discussion. A transfer
payment column is added to represent the “commodity” produced by remittance
cohorts such as unemployed labor and retirees. Conceptually, we are simply saying
that unemployed labor and retirees are producing a commodity because the very fact
that they are being compensated is evidence for the commodity itself. One might
debate the wisdom or rationale behind the transfer payments, but what is beyond
doubt is that unemployed labor and retirees are producing some commodity, which
some entity or entities are purchasing, based upon some decision-making criterion
(optimizing function). This is all that matters from a modeling perspective. Similarly,
additional make table columns include several government commodities, which are
produced by the government “industries” rows added to the make table. Again, we
will infer the presence of the commodity from the presence of the transaction (taxes).
The make table also will include additional columns for residential and nonresidential
physical capital, which will be the commodity produced by the speculator industries
that were added as rows in the make table.
A use table can be constructed along similar lines. As with make table rows, the
use table will add columns for a labor sector, remittance cohorts, government,
investors, and speculators. The use table also will add rows for labor; transfer payments, government taxes, and fees; financial capital; and residential and nonresidential physical capital. The labor sector will use a mix of commodities once
relegated to the use table’s final demand portion. In the same manner, remittance
cohorts and government also will use a mix of commodities from the final demand
portion of the traditional use table.
18
48
The role of the proposed speculator industries deserves a brief explanation. Each
speculator sector will use the mix of commodities identified in the traditional use table
under investment final demand, in addition to the financial capital good, to produce
the physical capital good(s) identified in the make table. The speculator sector is
something of a “ghost in the machine” because it is a mechanism the model will use
to ensure that the presumably quite mobile financial capital commodity flows through
speculator intermediaries to purchase presumably relatively immobile physical capital. As we develop an economic geography model of the United States, it is critical to
model accurately where demand actually occurs, and introducing the speculator
intermediary helps facilitate this. Finally, producer industries, in addition to using the
commodities identified in a traditional IO table, also use labor, government, and
physical capital commodities, which traditionally are identified as value-added components in the use table.
Two industries receive very special treatment in the model as they will both figure
prominently in the behavioral equations and in the ultimate geographic equilibrium:
the “real estate” sector (North American Industry Classification [NAICS] System code
531) and the “owner occupied dwellings” sector, which is not identified in the NAICS
coding system but is rather a constructed sector used in the make and use tables
produced by both the Bureau of Economic Analysis (BEA) and Bureau of Labor
Statistics (BLS) to guarantee compatibility with the U.S. National Income and Product
Accounts. These industries are critical for the model, in that they include land values,
which is the one fixed geographic commodity in our model. Land, as we shall see
shortly, is the only completely immobile commodity in the model, and land prices are
the one factor that will invariably act to disperse economic activity. As such, the
“other value-added” components of these two industries are extracted and are
labeled as a separate land sector, producing a completely immobile land commodity.
The only commodity used by the land sector is financial capital, specifically the rent
(real or imputed) paid to landowners.
Several data sources are used to estimate county-level employment for the
merged IO-SAM at the NAICS five-digit detail level (709 industries). A complete
description of the process used to populate the model can be found in Tanner
(2005). The primary data sources are the County Business Patterns (CBP) from the
Bureau of the Census and the Regional Economic Information System (REIS) from
the Bureau of Economic Analysis (BEA). Wage bill (payroll) data, which will populate
19
49
the regional “labor sector” output in the model and also determine output for many
other industries, are derived with the same techniques and from the same sources
as the employment data. Specifically, the CBP reports the total annual payroll for
each NAICS code up to the five-digit level of detail for the United States and for
every region, state, and county. However, total employment and total payroll data are
subject to suppressions for privacy. Rather than rely strictly on the various RAS and
statistical systems traditionally used to fill all data suppressions, we developed a
unique “range constraining” approach, which uses all information available in the
CBP series and guarantees internal consistency with unsuppressed wage and
employment data (Tanner 2005). All the furnished and estimated CBP wage bill and
employment data are then totaled and scaled to match the wage bill and
employment data reported in the BEA’s REIS, which includes all county and state
wages at the two-digit NAICS level of detail and all employment data at one-digit
NAICS detail. The REIS directly provides wage bill and employment data for the
government and agriculture sectors as well as disposable personal income data by
county.
The process used to build a complete set of historical and forecast IO-SAMs is
also outlined in greater detail in Tanner (2005). Annual IO tables are constructed
using BEA IO make and use tables as well as biennial 11-year IO forecast tables
from the BLS. The very detailed BEA IO make and use tables are extended year-byyear to match the annual changes in make and use composition implied by the
current 10-year BLS IO tables. This generates a detailed annual forecast series of
national IO make and use tables. These national merged IO-SAM tables will serve as
the U.S. national forecast that will drive the model; hence, some key characteristics
of the resulting national merged IO-SAM make and use tables are in order. First, the
national tables explicitly identify international exports of commodities by sector, and
international imports of commodities by sector for each year; these proportions are
held constant across all regions in the model, so regardless of location in the United
States, all industries of the same type will be importing the same proportion of their
inputs and will be exporting the same proportion of their output. This amounts to an
assumption that barriers to international trade in goods and services are sufficiently
large that differences in U.S. regional shipment of goods/services do not generate
any substantial regional price differences for either imports or exports. Second, the
resulting annual IO tables include explicit estimates of total U.S. change in business
20
50
inventories by sector. As with imports and exports, these are held proportional in all
regions in the model, so all industries of a particular type will experience the same
change in business inventory, regardless of region. As such, the profitability variations between regions, which are explicitly calculated in the model, do not manifest
through differences in the annual change in business inventory. Finally, with respect
to the labor sector, the merged IO-SAM is denominated exclusively in terms of
dollars of labor bought/sold and is mute on the point of number of people employed;
hence, it does not say anything about the degree of slack in the national labor
market. As the BLS IO tables that underlie the merged IO-SAM are an element of the
BLS long-term forecast, the roll of labor market dynamics in the forecast is implicitly
embedded in the IO data but is not explicit. However, the regional model will
explicitly estimate the “profitability” of the labor sector in every region, and as such
there will be regional differences in labor market dynamics. Because the purpose of
this model is primarily to estimate how total U.S. economic activity is distributed
across the 3,110 regions in the model and because all of the behavioral equations
are adapted to estimate the proportion of total economic activity in each region, any
U.S. forecast could be embedded in the model structure without need to revise the
allocation equations.
Once the National Merged IO-SAM is constructed, each county’s wage bill by
sector is used to allocate each sector’s national output to counties, the BEA REIS
income data are used to allocate the other sectors (labor, remittance cohorts, government, and investors) to their respective counties, and then the regional output by
sector is allocated to commodities based on the national merged IO-SAM make table
proportions for the years 2000 and 2001.
This assumes that the commodities produced by a sector are truly joint in the
production process as dictated by a nationally uniform production function for all
firms in each industry based on competitive pressures to diffuse advantages quickly
across all firms in an industry. Rather than relying upon the traditional matrix inversion technique used in most IO models (but unwieldy in a model with 3,110 interacting regions), in baseline and simulation forecasting the model will apply the
national IO tables to estimate a complete multiregional supply response to indirect
and induced demand, and to exogenous final demand, in a search cycle that looks
for the suppliers of suppliers across industries and regions. Each cycle in the search
process starts up in every region where the gravity-based production function’s
21
51
previous cycle estimated a supply output response, and so on until the process
reaches a minimum incremental output cutoff point.
THE NEW ECONOMIC GEOGRAPHY BEHAVIORAL ASSUMPTIONS
Regardless of the entity in question, in our model all will face a Dixit-Stiglitx (1977)
constant elasticity of substitution (CES) nested Cobb-Douglas production function of
the form
∏ (g~ )
G
g =1
θ g~st
gmsrt
= E st + qmsrt
(1)
For manufacturer m belonging to sector (industry, labor, government, etc.) s
located in region r at time t , G represents the total number of goods in the
economy.
g~gmsrt is the quantity of composite commodity good g~ used by
manufacturer m in sector s in region r at time t . θ g~st is the share of composite
~ used in sector s at time t . Note that the production function, at
commodity good g
any point in time, is sector and time specific but not region or manufacturer specific.
Est is the fixed cost of production for sector i at time t . Finally, qmsrt is the total
output of manufacturer m in sector s in region r at time t .
This behavioral equation will apply to all sectors, regardless of the “type” of entity
in the traditional sense.
Every sector also faces the traditional constant returns to scale Cobb-Douglas
budget share constraint given by
G
∑θ
g =1
gst
=1
(2)
This is completely consistent with agglomeration economies in the new economic
geography (NEG) framework, which is based on increasing returns at the sector level
but not at the firm level. In addition, a constant returns to scale technology is
consistent with the input-output data structure used throughout the model.
Because we wish to allow for the possibility of joint production, as implied by the
data structure described earlier, we must devise a mechanism for translating between sector production and commodity production. To that end, we specify
22
52
G
qmsrt = ∑ ϑgst qmsrt
(3)
g =1
where
G
∑ϑ
g =1
gst
=1
(4)
where ϑgst is the output share of good g in sector s total output at time t . For joint
production, we shall calculate the U.S. average inputs for commodity g at time t ,
given by
θ g~gt


Qgst
= ∑ θ g~st I

i =1
Qgst

∑
i =1

I






(5)
~ used in the production of commodity
where θ g~gt is the input share of commodity g
g at time t and s is the total number of sectors. To simplify the process of calculating prices across all regions and commodities in the model, we shall use these
input shares in all price and trade calculations. Industries will only reenter the
equation when we allow for sector expansion/contraction in a region in response to
price changes in the various commodities across regions.
The model we are developing will not rely upon traditional iceberg costs. Instead,
we will model the transportation component of the economy as an explicit subset of
inputs into the Dixit-Stiglitz production function. The iceberg transportation cost
assumption is so thoroughly embedded in the NEG literature that it is identified by
Krugman, Fujita and Venables (1999) as one of the three cornerstones of the
literature. At the same time, Krugman (1998) says of iceberg transportation costs,
“It’s too bad that actual transport costs look nothing like that.” Since tractability can
be maintained with a more realistic transportation assumption, for this model,
transportation cost will be given by
Pg~r rt
Pg~r t
∆
= ∏ γ gδt d δ~r rt
θδgt
(6)
δ =1
23
53
where the left-hand side of the equation,
Pg~r rt
Pg~r t
, represents the ratio of the profit-
maximizing price as delivered to region r to the profit-maximizing Ex Works (EXW,
the price at the factory door before any transportation expenses) price for good g
r
produced in region ~
at time t . ∆ represents the number of modes of
transportation. Each mode of transportation, as mentioned earlier, is a commodity in
the overall economy, hence ∆ ∈ G . dδ~r rt represents the effective distance from
r to region r by mode δ at time t . θδgt is the share of transportation
region ~
commodity δ used in production of commodity g at time t , and γ gδt represents the
unit distance cost of shipping commodity g by mode δ at time t . In estimating NEG
models, the concept of dδ~r rt is often approximated inclusively by straight-line distance or an average travel time between two regions.
Under this formulation of prices, and with the CES assumption of our Dixit-Stiglitz
production function, the aggregate profit-maximizing behavior of producers will lead
to a trade relationship for every commodity-county-county combination of
Tg~r rt =
Qg~r t ⋅ Pg~r rt
−σ g
 R
−σ 
 ∑ Qg~r t ⋅ Pg~r rt g 
 ~r =1

⋅ Dgrt
(7)
r to region
where Tg~r rt represents the volume of trade in commodity g from region ~
r , Qg~r t is the aggregate amount of commodity g produced in region ~
r at time t ,
and Dgrt is the aggregate demand for commodity g in region r at time t . Note that
this is a completely traditional gravity model, in that the degree of interaction is a
function of the relative size of the producer, the size of the demander, and the
relative distance (shipping cost) between them. The specification encompasses any
number of regions and commodities and sheds the restrictive iceberg price
assumption.
ESTIMATING PRICE ELASTICITIES AND TRADE FLOWS IN THE MODEL
The gravity model specified above is, by design, demand constrained. If we sum
r , we discover that
across all supplier regions ~
24
54
R
∑T
~
r =1


−σ g
 Q g~r t ⋅ Pg~r rt
= ∑ R
~
r =1 
  ∑ Q g~r t ⋅ Pg~r rt −σ g
 ~
  r =1
R
g~
r rt


R

⋅ D grt  ⇒ ∑ Tg~r rt = D grt ∀g , r , t
~

r =1




(8)
That is, the total trade in commodity g from all regions, terminating in region r ,
is equal to the total demand for good g in region r , an accounting condition that
must be true by definition.
While theoretically complete, accurate empirical estimation of the above model
requires one additional step: The addition of an explicit supply constraint to ensure
that every region in the model sells all output. As we wish to build an applied regional
economic model of the U.S. economy, it is necessary to guarantee that our estimation process also meets the supply constraint that
R
∑T
r =1
g~
r rt
= Q g~r t ∀g , ~
r ,t
(9)
If the model captured all trade perfectly, this would not be a concern, but in the
presence of error in the estimation, we must transform equation (7) into a classic
doubly constrained gravity model following the form developed by Wilson (1970,
1974):
−σ g
Tg~r rt
Pg~r t
Bgrt
∆

θ 
Qg~r t  Pg~r t ⋅ ∏ (γ gδt d δ~r rt ) δgt 
~
δ =1


=
⋅ Dgrt
−σ g
∆
R 



θ
 Q ~  P ~ ⋅ ∏ (γ d ~ ) δgt  
∑
 gr t  gr t δ~ =1 gδt δr rt
~
r =1
 

−σ g
−σ g
−σ g
∆
 R


θδgt 

= ∑ Dgrt  Bgrt ⋅ ∏ (γ gδt d δ~r rt )  
~
 r =1
δ =1

 

−1
−σ g
∆
 R


θδgt 

= ∑ Qg~r t  Pg~r t ⋅ ∏ (γ gδt d δ~r rt )  
~
 ~r =1
δ =1

 

25
55
(10)
(11)
−1
(12)
where Pg~r rt is the profit-maximizing price in region r of commodity g produced in
r at time t , which will drive the distance decay function in the gravity model.
region ~
B grt is a balancing factor that ensures that all output is sold in all regions in the
model; that is, equation (11) is satisfied. As such, the model of trade flows will closely
follow Alonso’s (1973) General Theory of Movement, though applied to trade rather
than migration and built from an explicit microeconomic foundation.
Unfortunately, there is no reliable, comprehensive, and timely data source for
regional trade flows within the United States. However, if we first difference the trade
gravity equation and are willing to make the simplifying assumption that Bgrt = Bgrt −1 ,
then we arrive at the following trade relationship:
∑ D (B
grt −1
⋅ Pg~r rt )
grt −1
⋅ Pg~r rt −1 )
R
Q g~r t
Q g~r t −1
=
r =1
R
grt
∑ D (B
r =1
grt −1
−σ g
(13)
−σ g
where Qg~r t and Qg~r t −1 represent the total quantities of commodity g produced in
r at times t and t − 1 , Bgrt −1 is the demand-balancing term for commodity g
region ~
in region r at time t − 1 , and Dgrt −1 represents total quantity of commodity g
demanded in region r at time t − 1 . Pg~r rt and Pg~r rt −1 are the profit-maximizing prices
r at times t and t − 1 , and σ g is
of commodity g in region r produced in region ~
the elasticity of substitution between individual varieties of commodity g . Derivation
of the trade relationship can be found in Tanner (2005).
The estimated share of each transportation mode devoted to the shipment of
each commodity will be estimated by
θ gδt


ϑgst qst
= ∑ θδst ⋅ I

s =1
ϑgst qst

∑
i =1

S






(14)
where S is the total number of industries, θδst is the budget share of sector s
devoted to the purchase of transportation mode δ at time t (identified by the IO
26
56
table for time t ), qst is the total national output of sector s at time t , and ϑgst is the
share of sector s output that is commodity g at time t . This equation enables the
model to estimate the budget share of commodity g that is devoted to transportation
mode δ as being the average of each sector’s budget share devoted to transportation mode δ , weighted by the sector’s total share of the output of commodity
g . Note that most commodities are produced almost entirely by a single sector;
hence, the commodity share is determined almost entirely by the production function
of that sector.
The distance variables d δ~~r rt , d δ~r rt , d δ~r rt −1 , and d δ~r rt −1 are normally approximated
by some inclusive straight-line distance or time measure, such that
dδ~~r rt = dδ~r rt = dδ~~r rt −1 = dδ~r rt −1 = dδ~r~r t = dδr~r t = dδ~r~r t −1 = dδr~r t −1
(15)
However, rather than using an inclusive straight-line distance or time measure,
this model applies a unique and comprehensive database of transportation impedance measures developed by the Oak Ridge National Laboratories from impedance
information for 1997 (Southworth 1997; Southworth, Peterson, and Chin 1998).
Based on the Oak Ridge impedance database, the impedance in this model can
differ between two regions both with the mode and with the direction of travel, but in
the currently supported analysis
dδ~r rt = dδ~r rt −1
(16)
As additional years of transportation data become available, impedance measures
could be expanded to change over time as well as with the mode and with the
direction of travel.
Under the current assumptions, we can substitute the delivered price equation
into the gravity equation and perform some simple algebra to get
Qg~r t
Qg~r t −1
−σ g
∆

θ δgt 
⋅
⋅
D
B


∑
grt  grt ∏ dδ~
r rt
r =1
δ =1


=
−σ g
∆
R

θ δgt −1 
Dgrt −1  Bgrt ⋅ ∏ dδ~r rt −1

∑
r =1
δ =1


R
(17)
At this point, we have an equation where the only unknowns are the elasticity of
substitution σ g and the balancing factor B grt . Estimates of σ g are calculated for
27
57
each commodity g , using nonlinear least squares. The estimation is made using
data for all 3,110 regions in the U.S. database for the years 1999–2001.
Once σ g has converged, we have effectively estimated the elasticities of
substitution for each commodity in the model, subject to our initial condition that Pg~r t
and Bgrt are 1. These EXW balancing factors Pg~r t and Bgrt are solved iteratively (of
necessity because they enter into the trade flow calculations nonlinearly), and the
iterative estimation of Pg~r t and Bgrt is followed by a re-estimation of σ g . The entire
process is repeated until convergence is achieved.
While trade flows are calculated for every commodity in our conjoined IO/SAM
framework, some restrictions and assumptions will be imposed upon the various
entities in the model to capture specific behavioral limitations. Specifically,
1. No local government commodity can be shipped across county lines. This
effectively prevents the export of local government commodities across
region borders, which means that local government is paid for entirely by
those entities in the region. Because this model will use counties as regions,
this amounts to an assumption that local government does not cross county
borders but is provided uniformly within any given county. This is certainly a
simplifying abstraction from reality, to the extent that some local government
entities cross county borders, while others may have a footprint that does not
cover an entire county.
2. No state government commodity can be shipped across state borders. This
has the same effect for state government as our first assumption did for local
government: State government does not cross state borders but may be
transported within the state, though such shipments are subject to the
explicitly estimated transportation cost for the commodity.
3. Land cannot be shipped across county borders. Recall that the land area in a
region fixes the supply of the land commodities in the region. This means that
any region has a fixed supply of land, and this will act as the fundamental
dispersing force in the model, counteracting any tendency toward catastrophic agglomeration that might occur in the presence of transportation
costs alone.
28
58
CREATING CGE AND DYNAMIC ADJUSTMENT PATHS FOR THE MODEL
Recall from equation (6) that, under our explicit transportation cost assumption, the
r at time t
profit-maximizing price in region r of commodity g produced in region ~
becomes
∆
Pg~r rt = Pg~r t ⋅ ∏ γ gδt dδ~r rt
θ δgt
(18)
δ =1
The next task is to define the vector of EXW profit-maximizing prices for all com-
r at time t :
modities manufactured in region ~
Pg~r t =
σg
Ω g~r t
σ g −1
(19)
where σ g represents the elasticity of substitution between individual varieties of
commodity g and Ω g~r t is the marginal cost function for producing commodity g in
r at time t .
region ~
By working within price space (rather than quantity space), as dictated by the
isomorphic discovery of Robert-Nicoud (2004), the EXW marginal cost function Ω grt
is in turn given by
G −∆
Ω grt = ∏ (Pg~rt ) ggt
θ~
(20)
g~ =1
where G − ∆ is the number of nontransportation commodities, Pg~rt is the price index
~ , in region r at time t , and θ ~ is the share of commodity g~ used
of commodity g
g gt
in production of commodity g at time t . This vastly simplifies the marginal cost
functions used by others (e.g., Fan, Treyz, and Treyz 2000) in developing multisector
NEG models.
The price index Pg~rt is given by
Pg~rt
 R

 ∑ D g~rt

R
Tg~~r rt

Pg~~r rt  ⋅ rR=1
=∑ R


~
r =1
 ∑ Q g~~r t
 ∑ Tg~~r rt
 ~r =1
 ~r =1
(21)
29
59
where R represents the total number of regions in the model. Tg~~r rt is the total trade
~ originating in region ~
r and sold to region r at time t , and Pg~~r rt is
in commodity g
~ produced in region ~
r at time
the profit-maximizing price in region r of commodity g
R
t . The ratio of total demand in all markets,
∑D
r =1
g~rt
, to total supply in all markets,
R
∑Q
~
r =1
g~~
rt
, might seem superfluous. Remember that the national IO tables are
balanced by design, and hence, this ratio should equal 1 and be irrelevant to the
calculation—and indeed, for most commodities, this is the case. However, in the case
of the state and local government commodities and, critically, the land commodity,
markets are not national in scope, and this ratio is likely not going to be 1.
To generate our dynamic NEG model of the economy, it is critical that we unwrap
the concept of the EXW price of good g . Within an NEG framework, the EXW price
can be decomposed as
R
Pgrt =
∑D
r =1
R
∑Q
r =1
grt
G −∆
⋅ ∏ (Pg~rt ) ggt ⋅ Agr
θ~
(22)
g~ =1
grt
That is, the EXW price Pgrt is equal to the demand-to-supply ratio of the
commodity in the market times the production-function weighted price index for all
nontransportation intermediate inputs. The refinement that we must introduce at this
point is the variable Agr , which is the first-nature production cost of commodity g in
region r and is calibrated from the EXW price equation (19). The EXW price
equation (19) is correct, only if there are no location-specific price differences in
production for any region, except those originating from the price of intermediate
inputs. However, in the real world, regions are intrinsically heterogeneous. For
example, coal mining is intrinsically more profitable in Wyoming than in Delaware,
not because market access is better in Wyoming than in Delaware but because
Wyoming is intrinsically different from Delaware—Wyoming has lots of rich coal
deposits, and Delaware does not. Likewise, boat building will tend to be more
profitable when there is a body of water in the region, and agriculture will be more
profitable for regions that have the appropriate soil, etc. In a completely
30
60
homogeneous world, there would be no such first-nature differences; all Agr values
would be expected to equal 1, and the only other force driving the location decision
would be market access. But with our CGE behavioral equations and with our trade
flow calculations from the previous section, we can estimate a completely NEG
model.
r and destination region r for each good g , we calculate
For each origin region ~
the delivered price equation (18) for the last history year using our calculated EXW
price Pg~r t from equations (19) and (20). Once we have calculated the delivered price
for all regions and commodities in the last history year, we can use equation (21) to
calculate the price index for every commodity and region in the last history year.
Finally, the EXW price for every commodity is decomposed into its respective
elements per equation (22), specifically to calibrate the first-nature differences, Agr ,
for each good and region in the last history year. We shall assume that these firstnature differences do not fluctuate over time.
Once these calculations are made, there is certainly no guarantee that profits of
all industries, in all regions will be equal. Given the monopolistic competition
configuration of the model, any potential for profit will be realized in regions that can
produce and deliver output at a low relative price within the various markets they
serve. As such, given the behavioral equations outlined in the previous section, we
can estimate an index of relative profitability for firms in sector i in region r at time
t as
π srt




R
Tg~~r rt
Pg~rt
= ∑ ϑg~st ⋅ ∑  R
⋅


Pg~~r rt
g~ =1
r =1

 ∑ Tg~~r rt
 r =1

G






(23)
where π srt is an index of relative profitability for sector s in region r at time t .
At this point, we must develop an output-adjustment process for the CGE model
in order to recognize that the adjustment to a stable, long-run equilibrium is not an
instantaneous process but rather a series of myopic steps as each sector in each
region makes adjustments over time in response to their profitability signals. An
output adjustment process is estimated by
31
61
Qs~r t +1
R
∑Q
~
r =1
s~
r t +1
=
Qs~r t
R
∑Q
~
r =1
s~
rt
R 
 G 
Pg~rt
+ λs ⋅  ∑ ϑgst +1 ⋅ ∑  Tg~~r rt ⋅

 g~ =1 
Pg~~r rt
r =1 
 
   Qs~r t
  −1 ⋅
  R
   ∑ Qs~r t
(24)
~
r =1
r at times t
where Qs~r t and Qs~r t +1 are the quantity of output in sector s in region ~
and t + 1 , respectively, and λs is the speed of adjustment of sector s to the relative
profitability signal and must be econometrically estimated.
Then, using our historical data, we can use equation (24) to calculate profitability
response λs for each sector by least squares using
Qsrt +1
R
∑Q
~
r =1
s~
r t +1
Qsrt
R
∑Q
~
r =1
= 1 + λs (π srt − 1)
(25)
s~
rt
based upon the calculated profitability π srt and profitability response λs , we can then
calculate the expected market shares for the first forecast year and allocate supply
and demand accordingly. Based upon the new allocation of supply and demand and
the estimated elasticity of substitution, we can calculate a complete and balanced set
of trade flows for the first forecast year.
Then, we calculate the EXW price for each commodity, in each region, in the first
forecast year by using equation (20) and the value of Pg~rt −1 as an estimate of Pg~rt .
Using the EXW price we have just calculated, we use equation (19) to calculate the
r and destination
delivered price Pg~r rt for every good g and for every origin region ~
region r .
Using this estimate of delivered price, we calculate the price index for each good
g and region r in the first forecast year using equation (22). Once all price indices
have been updated, we can recalculate the complete menu of EXW prices to
recalculate a complete set of delivered prices and then recalculate all price indices.
This process is repeated until it converges completely. Because each iteration is
capturing prices across a greater number of regions, the process necessarily
converges very quickly.
With the delivered price and price index data for all regions and goods for the first
forecast year, we can calculate sector i profitability for all industries in all regions
using equation (23). Based upon the calculated profitability π srt and profitability
32
62
response λs , we calculate the expected market shares for the second forecast year
and allocate supply and demand accordingly. The whole process is then repeated for
each and every year of the forecast period to build a complete county-level CGE
model of the U.S. economy that is consistent with the NEG framework.
CHARACTERISTICS AND BEHAVIOR OF THE MODEL
Because of the switch from the Standard Industrial Classification (SIC) to the NAICS
system for coding industries and commodities that took place over the 1997–2000
period, and because the U.S. Bureau of Economic Analysis chose not to collect data in
both formats for a single overlapping year, there exists no technique that will generate
even a remotely useful county-level time series that overlaps the two coding systems
(Tanner and Hearn 2005). Because the model we have developed ultimately is to be
applied to regional planning activity, it has been built entirely in NAICS, which means
that the data series cannot be extended before 1999. As such, the model is constructed using a complete historical database that covers only the years from 1999 to
2001. The major shortcoming of this arrangement is that the model’s forecasting
capability cannot yet be tested against historical data; the estimation of trade flows
requires two years of historical data, and that leaves only one year of historical data
that could be used to test the model. This is clearly insufficient to test a structural
model. So, we are left to explore characteristics of the model forecast while having to
rely upon the integrity of the model logic, as opposed to its historical performance.
Because the model forecasts an enormous number of concepts, identifying data
that will capture the overarching concepts of the NEG framework is a challenge. The
challenge is intensified by the fact that the model forecasts the market share
accruing to each county in every market, so the U.S. aggregate forecast tells us
nothing about the nature of the regional model. Because the NEG model is fundamentally driven by market shares and the amount of land available, it seems the
single metric that best captures the model behavior is “relative total sector output per
acre.” That is, the total amount of output per acre in a county relative to the total
amount of output per acre in the United States. By this metric, a county with a
relative total sector output per acre of 1 is producing exactly as much per acre as the
United States as a whole. A county with a metric greater than 1 is, to some degree, a
core county (a county that has experienced economic agglomeration), and a county
with a metric smaller than 1 is, to some degree, a periphery county (a county that
33
63
has experienced economic dispersion). If the metric for a county is increasing over
time, this would reflect a county that is experiencing economic agglomeration, and a
metric decreasing over time would reflect a county dominated by dispersion forces,
the key features of the NEG literature.
To provide a frame of reference, in 2002 the “most peripheral” county in the
United States was the Yukon-Koyukuk Census Area in Alaska. With a relative output
per acre measure of 0.00031, this region had an “economic density” that was .031
percent of the national average. By this same metric, the five “most peripheral”
counties in the United States in 2001 were Yukon-Koyukuk Census Area, Alaska;
Lake and Peninsula Borough, Alaska; Loving County, Texas; Petroleum County,
Montana; and Yakutat City and Borough, Alaska.
At the other extreme, the most economically dense (or “most core”) county in the
United States was New York County, New York, with a relative economic density of
5803.38, meaning that output per acre in New York County is more than 5,800 times
the national average output per acre. The top five most core counties in the United
States in 2001 were New York County, New York; San Francisco County, California;
Suffolk County, Massachusetts; the District of Columbia; and Arlington, Virginia.
Under this measure of economic density, using what we know of the NEG
structure of the model, we can begin to picture how various counties might be
forecast to behave within this structure. We would expect that periphery regions like
Yukon-Koyukuk are likely to be very stable periphery counties and that they are likely
to see little change in their economic density over time. Likewise, we might expect
that the most core regions like New York County will be relatively stable in their
market share. Between these two extremes, we have an array of regions that might,
over the forecast period, be moving toward “greater coreness” or “greater peripheriness” if they are near their so-called “break point” (the point at which the benefits
of economic agglomeration outweigh the costs, and economic agglomeration/
dispersion occurs). And we might have yet another group of midsize regions that are
losing there “coreness” or “peripheryness” as they pass the sustain point for their
particular equilibrium. If we look at the behavior of these counties in the aggregate,
we expect to see a number of counties that are stable within their core, periphery, or
dispersed equilibrium and some counties that, across the forecast period, will be
making the transition from core or periphery. We have compared our forecast to two
alternative, naïve forecasts, and we see a result that is largely as expected. The first
34
64
alternative forecast assumes the county share of U.S. output will remain constant
throughout the forecast period, and a second assumes that the county share of U.S.
output will grow at the average annual rate exhibited in the 1999–2001 historical
period. Both of these forecasts would be expected to correspond well with the
counties that do not approach a break or sustain point. The constant growth forecast
is expected to perform comparatively well over the short term with counties that are
in transition but will likely perform very poorly as those counties approach their new
core or periphery position. The constant share forecast will not accurately reflect the
counties while they are in transition but will not be wildly incorrect over time as those
counties approach their new equilibrium and settle into a more-or-less fixed output
share. By examination of the correlation coefficients over the forecast period
between our model, the constant shares model, and the constant growth model, we
see results consistent with our intuition (see Figure 2). For the first 15 to 20 years of
the forecast period, the forecasts of county-level relative output per acre are very
tightly correlated among the three forecast types. The correlation of the model
forecast with the constant share forecast then begins to drop off, and by the close of
the forecast period, the correlation between the constant growth forecast and the
NEG model forecast is virtually zero. This is consistent with the idea that counties
that are experiencing share growth are in transition and not exhibiting a permanent
relative growth behavior as suggested by the naïve model.
Figure 2. Correlation of the NEG Model with the Constant Output Share
and Constant Output Growth Models
35
65
The constant share forecast is much more tightly correlated with the NEG model
forecast for a much longer period of time. By the close of the forecast period, there is
still approximately 9 percent correlation between the constant shares forecast and
the NEG model forecast. Once again, this is consistent with our intuition regarding
market behavior in an NEG format.
We can capture this behavior in another way, by looking at the behavior of our
chosen metric, relative output per acre, within deciles. With a total of 3,110 counties,
each year we divide these counties into 10 groups of 311 based upon their relative
output per acre. The 311 counties in the smallest decile are, in a sense, the most
peripheral, and the 311 in the largest decile are the most core. Because our metric is
a county aggregate, it necessarily abstracts from the more in-depth model behavior
since every sector, in every county, can have any degree of coreness or peripheralness. Nonetheless, if we expect that movement toward core and periphery solutions
fundamentally drive the economy, we can expect some specific behaviors to appear
in the data. In an economy moving toward increasing heterogeneity, we would
expect the average growth rate in the very smallest regions to be either constant (if
they are as peripheral as they can get) or shrinking and the growth rate of the very
largest regions to be, in general, either constant (if they have reached a point of
maximum coreness) or growing. Somewhere in the middle of the distribution, we
might expect to see counties that are in transition to a core position or perhaps to a
periphery position. A look at the growth rates by decile in Table 1 reveals some
interesting patterns. First, the relative output of the smallest 311 counties is
shrinking, and it is shrinking slightly faster than it is for any other decile. Deciles 2
through 6 are shrinking slightly as well, although each successive decile is shrinking
slightly less. The 622 regions in deciles 8 and 9 are actually growing in share of U.S.
output, suggesting that they are moving toward becoming cores. The largest 311
regions, however, are exhibiting almost no growth in share of U.S. output, suggesting
that the most core U.S. counties simply cannot get any more core than they already
are. These counties are likely running into the model barrier created by land prices,
which simply precludes further agglomeration.
36
66
Table 1. County Relative Growth in Share of U.S. Output by Decile, 2002–2055
Decile
Average
Growth Rate
Decile
Average
Growth Rate
Smallest
0.9814
6
0.9990
2
0.9883
7
0.9995
3
0.9913
8
1.0045
4
0.9923
9
1.0074
5
0.9950
Largest
1.0002
AGGLOMERATION FROM A HOMOGENEOUS ECONOMY
At this point, we have evidence that the model will maintain core/periphery economies when presented with a heterogeneous economy as a starting point; in this
case, we started the model with our clearly heterogeneous 2001 economy and
allowed the model to go from there. However, it is interesting to test whether the
model can develop a heterogeneous economy from a completely homogeneous
starting point and what characteristics this artificial economy might have. To that end,
the forecasting model was adjusted in a few fundamental ways. First, the inputoutput matrix, which evolves over time in the forecasting model, is “locked down” as
the 2001 input-output matrix, which means that changes in production technology will
not take place, so the economy is evolving toward some fixed equilibrium rather than
an equilibrium that is itself changing due to input-output changes. Secondly, the total
U.S. output for every sector in the model was spread evenly across every county in
proportion to each county’s share of total U.S. land area. So, a county that
represents .1 percent of U.S. land area also was assigned .1 percent of total U.S.
output of every sector. Thus, the model was starting from a truly dispersed “backyard
capitalism” scenario.
With this starting point, a total of five alternative model specifications were built.
In the first model specification, first difference values were set to 1 for all goods in all
regions. That is, the model assumed that there were no first-nature differences for
any production activity in any region (so coal mines, for example, could be located
anywhere). Second, all impedance values, for all modes, for every region-region
combination were set to 1. This means that there was also no transportation-related
37
67
advantage for any region in the model; any region would produce its output and sell
it in every region (including their own) for the same price. All other characteristics of
the model were left unchanged. This model was then allowed to run through 54
simulated years. It should come as absolutely no surprise that under these restrictions no agglomeration whatsoever takes place. The economy at the end of the 54
cycles remains completely homogeneous for the simple reason that with no firstnature price differences and no potential for second-nature differences, there is no
force to encourage any movement from the dispersed equilibrium.
For the second scenario, we reintroduce the first difference values that were
calculated for the model, but we continued to allow all goods to be shipped from any
region to any region for the same price. This model effectively allows for first-nature
differences but removes all second-nature differences. When this model was allowed
to cycle through 54 years, the result was spectacular agglomeration—agglomeration
that is much greater than that actually seen in the U.S. economy in 2001 (as
measured by the standard deviation in county output per acre). The reason for the
spectacular level of agglomeration is simply that with transportation costs not entering into the picture, all economic activity is strongly attracted to the places with the
greatest first-nature advantage in production. Many activities that we intuitively know
are significantly constrained by transportation (e.g., restaurants, gas stations, and
grocery stores) will, nonetheless, cluster in a relatively small number of counties
even if the first-nature price advantage is small, simply because the transportation
effect has been removed.
The next incarnation of the model again removed the first-nature differences, but
this time the impedance values for every mode of transportation were set to equal
the straight-line distance between county centroids. Internal distances for every
region were set equal to the square root of the region’s land area. Under this
configuration, we are removing any first-nature differences among regions and
allowing second-nature differences, but those second-nature differences use the
simplifying assumption that transportation costs are simply proportional to straightline distance. When this model is allowed to continue for 54 years, it generates
economic agglomerations, although the agglomerations are much more modest than
those created by the first-nature difference model. The agglomeration is, of course,
generated strictly through the second-nature differences in this model.
38
68
The next incarnation of the model was very similar except that the straight-line
distances were replaced with the Oak Ridge impedance data. Therefore, this model
included all transportation infrastructure data for second-nature differences but still
included no information about first-nature differences. Not surprisingly, this model
also generated economic agglomeration over the forecast period; the agglomeration
was somewhat more pronounced than that generated by the straight-line distance
model but still much less than the agglomeration generated by the first-nature
differences themselves. The agglomeration in this model is greater than that of the
straight-line distance model simply because the transportation data are much more
heterogeneous than the straight-line distances. Two adjacent counties will face
almost the same menu of straight-line distances and will, therefore, be almost
equally preferable if that is the metric used for transportation costs. However, when a
major highway, a rail line, and a port are located in one county and not the other, the
difference between the two from a profitability standpoint becomes quite dramatic.
The final incarnation of the model included all of the transportation infrastructure
data and all of the first-nature difference data. This version was simply the full model
but run on an initially homogeneous distribution and with a constant IO table. This
model exhibited somewhat more agglomeration than the model with transportation
but not first-order differences. However, the model still showed much less agglomeration than the model of first-nature differences alone.
The purpose of this experiment was not simply to look at the models compared
with one another but also to look at how the models might compare to the actual
2001 U.S. economy. We know that history matters, and that there are a near-infinite
number of potential equilibria in an NEG mode with this many regions and sectors.
However, it seems reasonable that given the distribution of first-nature differences
and given our heterogeneously distributed transportation infrastructure, we might
gravitate toward a similar spatial distribution of economic activity even from very
different starting points. In this case, we are taking our starting point of a homogeneous economy with a fixed 2001 technology and letting each of our alternative
model specifications run for 54 years to see how the resulting economy compares
with the actual U.S. economy in 2001 (which obviously started from a very different
starting point). Once again, we use our metric of relative output per acre for each
county and will see whether any of our model configurations are correlated with the
actual 2001 economy. The summary results are reported in Table 2.
39
69
Table 2. The Degree of Correlation Between the Distribution of Economic Activity
in the United States in 2001 and the Distribution of Economic Activity 54 Years
Removed from a Homogeneous Distribution, for Various Model Configurations
Correlation with
2001 Output
per County
Forecast Method
No first-nature difference
NA
First-nature effect only
.0593
Distance effect only
.1314
Transportation effect only
.5727
Transportation and first-nature effects
.6502
The model with no first- or second-nature differences exhibits no heterogeneity at
the end of 54 years, of course, so there is no correlation to discuss. The model with
first-nature differences but no transportation had a very high degree of agglomeration, but the agglomeration is only minimally correlated with the agglomeration in
the actual economy. Although the first-nature model might perform very well for
some industries (e.g., mining) that are clearly driven by location-specific cost factors,
it tells us little about industries that are more affected by market access rather than
by first-nature differences.
The models that capture transportation (and hence shipping cost) are each much
more strongly correlated with the actual U.S. 2001 data. The model that embeds
impedance data (but without first-nature differences) generates a correlation of more
than 57 percent. Finally, the full model, with first-nature differences and transportation infrastructure, manages to endogenously generate a heterogeneous economy
that is more than 65 percent correlated with the 2001 U.S. economy. These
correlations are surprisingly high and are no doubt driven largely by the fact that
transportation generates economic agglomeration, which drives economic development, so the model is capturing the correlation between level of infrastructure and
the size of the economy. In this way, the model is generating results very similar to
Sutton et al. (1997). They tested the simple correlation between the light levels from
nighttime satellite photos of the United States and the county-level income data for
the United States. Their analysis found a correlation of 84 percent to 93 percent,
which is in line with the numbers found in this analysis.
40
70
Although the exercise of building these alternative models has no immediate
practical application, it is certainly reassuring to note the model’s ability to spontaneously agglomerate a homogeneous economy in a manner consistent with NEG
theory. In examining the degree of correlation between the model and the 2001 data,
it also suggests a certain degree of inevitability in the specific pattern of heterogeneity observed in the U.S. economy.
Although we do not yet have a sufficient historical record against which to test
the model, these results can at least reassure us that the model is behaving as we
would expect given the theory.
CONCLUSION
We have integrated concepts, theories, and data from a number of different areas
into a comprehensive regional economic modeling methodology consistent with the
theoretical NEG literature. The case for using this approach to develop a computable
general equilibrium model appears compelling, and on that basis, we believe the
model takes several important steps forward in the field of applied regional economic
modeling, forecasting, and impact analysis. Although the model development effort
has been significant, what has been built to this point only scratches the surface of
what might be possible as additional data, computing power, and theoretical work
enable making increasingly simple models that can capture increasingly complex
behaviors in an increasingly accurate manner.
REFERENCES
Alonso, William. 1973. National interregional demographic accounts: A prototype. Institute of
Urban and Regional Development Monograph 17. http://www-iurd.ced.berkeley.edu.
Batten, David, and David Boyce. 1986. Spatial interaction, transportation, and interregional
commodity flow models. In Handbook of Regional and Urban Economics, ed. Peter
Nijkamp, Edwin Mills, and P.C. Cheshire, 357–406. Amsterdam: North-Holland.
Beaumont, Paul. 1990. Supply and demand interaction in integrated econometric and inputoutput models. International Regional Science Review 13, no. 2:167–81.
Bröcker, Johannes. 1990. Derivatives of flows in a doubly constrained gravity model. Journal
of Regional Science 30, no. 2:259–268.
41
71
Ciccone, Antonio, and Robert E. Hall. 1996. Productivity and the density of economic activity.
American Economic Review 86, no. 1:54–70.
Dixit, Avinash K., and Joseph E. Stiglitz. 1977. Monopolistic competition and optimum product diversity. American Economic Review 67, no. 3:297–308.
Fan, Wei, Frederick Treyz, and George Treyz. 2000. An evolutionary new economic geography model. Journal of Regional Science 40, no. 4:671–695.
Fotheringham, A. Stewart, and Morton O’Kelly. 1989. Spatial Interaction Models: Formulations and Applications. Dordrecht: Kluwer.
Fujita, Masahisa, Paul R. Krugman, and Anthony Venables. 1999. The Spatial Economy.
Cambridge, MA: MIT Press.
Hewings, Geoffery, Michael Sonis, and David Boyce, eds. 2002. Trade, Networks and Hierarchies: Modeling Regional and Interregional Economies. Berlin: Springer.
Jensen, Rodney. 1990. Construction and use of regional input-output models: Progress and
prospects. International Regional Science Review 13, no. 1:9–25.
Krugman, Paul R. 1980. Scale economies, product differentiation and the pattern of trade.
American Economic Review 70:950–9.
Krugman, Paul R. 1991. Increasing returns and economic geography. Journal of Political
Economy 99, no. 3:483–99.
Krugman, Paul R. 1998. Space: The final frontier. Journal of Economic Perspectives 12, no.
2:161–74.
Leontief, Wassily. 1941. The Structure of the American Economy, 1919-1939. New York:
Oxford University Press.
Neary, J. Peter. 2001. Of hype and hyperbolas: Introducing the new economic geography.
Journal of Economic Literature 39, no. 2:536–561.
Pyatt, Graham, and Jeffery Round. 1985. Social Accounting Matrices: A Basis for Planning.
Washington, D.C.: The World Bank.
42
72
Robert-Nicoud, Frederic. 2004. The structure of simple ‘new economic geography’ models
(or, On identical twins). Journal of Economic Geography 4:1–34.
Round, Jeffrey I. 2003. Constructing SAMs for development policy analysis: Lessons learned
and challenges ahead. Economic Systems Research 15, no. 2:161–183.
Southworth, Frank. 1997. Development of Data and Analysis Tools in Support of a National
Intermodal Network Analysis Capability. Washington, D.C.: U.S. Department of Transportation, Bureau of Transportation Statistics.
Southworth, Frank, Bruce E. Peterson, and Shih-Miao Chin. 1998. Methodology for Estimating Freight Shipment Distances for the 1997 Commodity Flow Survey. Washington, D.C.:
U.S. Department of Transportation, Bureau of Transportation Statistics.
Stone, Richard, and Alan Brown. 1962. A Programme for Growth, Vol.1: A Computable
Model of Economic Growth. London: Chapman and Hall.
Tanner, Thomas. 2005. Applications of new economic geography in regional economic analysis and forecasting. PhD diss., University of Georgia.
Tanner, Thomas C., and Will Hearn. 2005. Economic impact analysis. In The Regional Economic Development Handbook, ed. Robert Pittman, 132–165. Washington, D.C.: International Economic Development Council.
Treyz, George, Dan Rickman, and Gang Shao. 1992. The REMI economic-demographic forecasting and simulation model. International Regional Science Review 14, no. 3:221–253.
Wilson, Alan. 1970. Entropy in Urban and Regional Modeling. London: Pion.
Wilson, Alan. 1974. Urban and Regional Models in Geography and Planning. Chichester:
John Wiley.
43
73
The Carl Vinson Institute of Government has served as an integral part of
the University of Georgia for more than 80 years. A public service and outreach unit of the university, the Institute has as its chief objective assisting
public officials in achieving ­better government and communities, particularly in Georgia. To this end, it draws upon the resources and expertise of
the university to offer an extensive program of governmental instruction,
research and policy analysis, technical assistance, and publications.
Collectively, Vinson Institute faculty and staff design and conduct more than
600 training and development programs per year, in which more than 18,000
public officials participate. Technical assistance takes many forms, including eval­uation of existing facilities and methods, provision of information
for decision makers, and assistance in establishing new programs. Research
helps inform policy decisions at local and state levels.
Research with wide general application is made available through the pub­­lications program. Publications include handbooks for specific governmental
offices, research studies on significant issues, classroom teaching materials,
and reports on practical methods for improving governmental operations.