Accuracy of the MGM2 estimates depends mainly on the accuracy of the three inputs:
(1) the number and types of visitors,
(2) spending averages and
(3) economic multipliers
-- roughly in this same order of importance.
The official NPS Public Use Statistics are the basis for visit estimates.
Errors may be introduced in adjusting recreation visit figures for park
re-entries and converting visits from a person to a party basis. The
NPS visitation reporting system measures entries to a national park,
not days/nights spent in the area or the number of distinct visitors
or trips. For some parks, visitors may be counted multiple times during
a stay in the area. Limited information is available for most parks
to estimate re-entry factors or length of stay in the area.
Many parks also lack accurate information to estimate the percentage
of visitors who are local, on day trips or staying overnight outside
the park. Estimates of segment shares for parks with recent visitor
surveys are likely more accurate than those based on manager judgment.
or figures at similar parks. Where recent visitor surveys exist, segment
share estimates are prone to errors due to non-response, sampling locations
and dates, and length of stay bias. Local residents are often under-represented
in visitors surveys due to all three of these factors. Overnight visitors
and especially those camping in developed campgrounds inside the park
tend to be over-represented, based on comparisons of survey results
with park overnight stay data. Backcountry campers also tend to be under-represented.
VSP studies typically are conducted during a 1-2 week period, which
can introduce significant seasonal biases in the results. Some adjustments
for these potential biases were made in our analysis of VSP study data.
Spending averages are estimated for four primary visitor segments (local
day trips, non-local day trips, overnight stays in motels, overnight
stays in campgrounds) that help explain variations in spending. Sampling
errors (95% confidence interval) for spending averages estimated in
park visitor surveys generally are around 5% overall and 7-20% for individual
segments. In applying the spending averages to different parks, the
issue is one of generalizability of spending figures from one park or
situation to another rather than sampling error. It is difficult to
estimate the potential generalization errors, although comparisons of
spending averages within segments across different parks, indicates
that segment shares capture much of the variation in spending. This
supports the basic assumption of the MGM2 model that spending profiles
of individual segments are more readily applied across parks than an
overall visitor spending average.
Multipliers mainly affect the estimates of secondary effects. In most
cases, the regional multipliers for tourism-related sectors can be predicted
quite accurately based on the population of the area. Use of MGM2 generic
multipliers, which are grounded in input-output models estimated with
the IMPLAN system, does not appear to introduce significant errors.
It should be noted, however, that the MGM2 multipliers are considerably
lower than exaggerated multipliers that are in fairly widespread use
within recreation and tourism. Many analysts are using outdated, statewide,
non-recreation and other available multipliers inappropriately to estimate
local economic impacts of recreation spending. MGM2 Type II sales multipliers
generally range from 1.3 for rural areas to 1.6 for statewide regions
or metropolitan areas.
Perhaps the greatest source of variations across published studies
and economic impact figures involves decisions about which visitors
and spending should be attributed to the park. MGM2 estimates will be
conservative in that we only include spending in the local area and
do not attempt to capture spending that would not be lost to the region
in the absence of the park. For many areas, however, identifying which
trips of spending would be lost to the area requires a more complete
understanding of visitor motivations and behavior.
Economic impact analysis measures the changes in economic activity
resulting from some action. Analysts generally prefer a "with vs
without" analysis that sorts out what spending would be lost to
the region in the absence of the proposed action. The action being evaluated
in the park impact estimates reported here is the closure or unavailability
of a given national park to visitors. It is difficult, however, to isolate
which trips to the area and spending would not occur in the absence
of the park. Many visitors come to the region for a combination of attractions,
facilities, activities and events, some on business or to visit friends
or relatives. In our approach, only the additional expenses for a park
visit is counted if the trip to the area is primarily for other reasons,
be it visiting friends/relatives, staying at a seasonal home, or visiting
another attraction. Trip purpose information, however, is only available
for parks with recent visitor surveys that have asked these questions.
Spending by visitors who live in the local area is normally excluded
in an impact analysis, on the grounds the spending would not be lost
in the absence of the park. Inclusion of local resident spending depends
on whether the substitute for the park visit is a different local activity
or a trip outside the region. The MGM2 results include local visitor
spending, assuming more trips would be made outside the local area if
the park were unavailable. As locals are treated as a distinct segment
in the MGM2 model, their contribution to spending is readily subtracted,
if desired.
Impacts provided here only cover spending by park visitors in the
local region and exclude park entry fees paid to the NPS. The economic
impacts of park payroll, operations and construction on the local economy
may be estimated using MGM2Operate.xls.
MGM2 only includes spending that occurs within the local region, typically
defined by a 30-100 mile radius depending on the park type and regional
characteristics.
Finally, the MGM2 impact estimates are based on regional input-output
(I-O) models, and are therefore subject to the general limitations of
I-O models. In particular, the model is linear so a doubling of spending
or sales in a given sector will double jobs, income and value added
-- the model does not take into account any economies or diseconomies
of scale. Firms in a given industry are assumed to use a common production
function (the national average). One must also assume that the categories
in which visitors report spending are correctly matched with the sector
producing the given commodity in order to properly represent the production
function. Production functions of smaller firms in rural areas can differ
considerably from national averages and many goods and services purchased
by visitors do not fit neatly into standard economic sectors. Campgrounds
and B&B's, for example, are grouped with hotels, and the amusements
sector is a catch-all for a host of distinct types of firms. Nevertheless,
I-O models have proven to be fairly robust, and errors due to the I-O
part of the MGM2 model are generally small relative to the visit and
spending pieces.
Top
MGM2 Home