On this page we respond to questions about the MGM2 model. Please send any questions to Daniel Stynes.  Answers to questions that seem to be of general interest will be posted here.

General Questions

G1. Who developed MGM2?

MGM2 was developed at Michigan State University by Daniel Stynes and Dennis Propst. The work was funded through the NPS Social Science Research Program. The development of the MGM2 model was completed in June, 2000. In June 2001, the contract was extended to maintain the model and to help selected parks in applying the model.  The original MGM model was developed by Ken Hornback.

G2. How does MGM2 differ from MGM?

MGM2 is a multi-page Excel spreadsheet that provides considerably more detail and  flexibility than MGM. The MGM2 model disaggregates the analysis to handle distinct subgroups of visitors (segments), itemizes spending by segment within 12 spending categories, and uses sector-specific multipliers derived from IMPLAN models for regions around National Park units. Default spending data for park visitors and multipliers for regions around parks are built into the model. Uses may choose from sets of defaults, adjust these to fit a particular application, or enter their own spending data. The original MGM model was used primarily to estimate overall impact of all visitors to a particular park. MGM2 can also be used to evaluate specific management and policy decisions. To do so, you must estimate the change in the number and types of visitors resulting from a particular action. You may also enter all park visitors in MGM2 to obtain overall park visitor spending impacts.
 

Technical Questions

T1. Where does data in MGM2 come from?

MGM2 requires three types of data (1) park visits, (2) spending averages, and (3) regional economic multipliers. Park visit data is entered by the user. NPS public use statistics can be used but must be converted to a party-night basis. Spending data comes from park visitor surveys. MGM2 includes "generic" spending profiles for parks and historic sites that were assembled from recent park visitor surveys. See  Appendix C  of MGM manual for details. MGM2 Multipliers are estimated from IMPLAN models for regions around selected National Parks using 1996 IMPLAN data. See  Appendix E of MGM2 manual for details.

T2. How are impacts computed in MGM2.

Chapter 4 of the MGM2 manual (MGM2.pdf ) contains technical details about the model. MGM2 estimates spending by visitors in the local region by multiplying the number of visitors (in party nights) by an average spending profile (per party night). Visitors are divided into distinct segments to capture different spending patterns of day users vs overnight visitors and those staying in campgrounds, motels, or with friends and relatives. Spending is then applied within spending categories to sector-specific multipliers for the local region. The multipliers convert spending to the associated sales, income and jobs. Multipliers are also used to calculate secondary effects from the circulation of this spending within the local economy.

T3. Why are MGM2 multipliers lower than others I've seen.

MGM2 multipliers are type SAM multipliers estimated from IMPLAN models for local regions around National Park units. The original MGM model used RIMS II multipliers, often for state level regions. Multipliers for a small local area will be much smaller than for larger regions. Many parks are in predominantly rural settings, where most goods and services are imported from outside the local economy. Park visitor spending is mostly for services and retail items. These sectors are labor intensive with limited linkages to backward linked industries. Therefore the induced effects of visitor spending are generally equal to or greater than the indirect effects. Induced effects of tourism spending have frequently been exaggerated. The Type SAM multipliers in IMPLAN do not recirculate payroll benefits that go to retirement programs and they also adjust for worker commuting patterns. MGM2 uses sector-specific multipliers which vary quite a bit across economic sectors. The MGM2 aggegate "generic" tourism sales multipliers range from about 1.3 for rural areas to 1.6 for statewide regions. These are substantially lower than the commonly used tourism multiplier of 2.0. MGM2 provides much more detail on the direct effects of visitor spending. We recommend focusing first on these direct effects and downplaying the "multiplier effects".

T4. How do I estimate impacts of park operations and payroll with MGM2?

A separate spreadsheet has been developed to estimate local economic impacts of park operations. Download the spreadsheet and brief documentation
 MGM2Operate.xls  MGM2Operate.doc .  If estimating impacts of both park operations and visitor spending, you must be careful not to double count receipts to the park. The best approach is to omit any spending that accrues to the NPS when estimating the visitor spending impacts with MGM2 (e.g., entrance and camping fees). You should include spending within the park that accrues to concessions, as these are treated the same as any other local business.  The impacts on the local economy of park operations depend on what the park spends in the local area, not how much it takes in through user fees.
When visitor admission and camping fees in the park are included in MGM2 analysis of visitor spending, we are assuming the production function for the park is similar to a private campground or recreation facility. This will not be the case if park operations are subsidized.

T5. How do I include the value of volunteers, benefits to park visitors, preservation benefits, impacts on surrounding property values ?

Economic impact analysis traces the actual flows of money within the local economy. Non-market transactions that do not explicitly entail purchases of goods or services are not covered. While volunteers add value to park experiences and reduce costs to NPS, the donated time should not be counted within economic impact accounts. We recommend treating these other "benefits" outside of the economic impact analysis. Note that an economic impact analysis  doesn't measure the value of the park to visitors - only what they spend and the value of this to local economy. Other methods estimate these other  values of parks.

T6. Why does MGM2 use party nights vs recreation visits as the unit of analysis?

The NPS measure of recreation visits counts the number of people entering the park. Economic impacts depend on how long a person stays in the local region moreso than how many times they enter the park.  NPS recreation visit data will generally count a visit each time someone enters the park. So a camper who leaves the park and re-enters six times during their stay will count as six visits. For economic analysis,  the best measure is the length of stay in the area, expressed in days or nights. We generally favor estimating spending on a per party rather than per person basis. The "party" is a group of people in a single vehicle or in some cases staying in the same room or campsite. When spending is estimated on a per person basis, there is a tendency to double count shared expenses (lodging, gasoline) and not handle children correctly. With spending estimated on a party-night basis, the lodging expense is easily interpreted as the average cost for a room or campsite. MGM2 can handle other units of analysis as long as spending and visits are entered in the same units.

T7. Why are direct sales less than total spending?

Not all visitor spending is captured as direct sales. The difference is due to the handling of retail purchases in regional economic models. Retail purchases are broken up into a retail margin, wholesale margin, transportation margin and the cost of the item at the factory. Retail margins are put in the retail trade sector and accrue to the local economy. If the good is not made locally, the producer portion of the price is not captured. The wholesaler and shipper may also be located outside the local area. Gasoline, groceries, clothing, souvenirs, and all other goods bought by visitors are handled this way. For example, if a visitor buys a $100 camera in the local area and retail margin is 40%, then $40 accrues to the retailer. If the camera is made elsewhere, the other $60 is not captured by the local economy. The direct sales impact is $40, not $100 and if the sales multiplier for retail trade is 1.5, the total impact is $60. It is clearly incorrect to multiply 1.5 times $100 and estimate total sales impact of $150. Many simple aggregate impact models make this error and inflate the impact estimates.
 
 T8. How accurate are the MGM2 estimates?

Short answer: Figure errors of 10-20% if the input choices/data are reasonable.

Longer answer: Accuracy of the estimates rests largely on the accuracy of the model inputs (GIGO) - that is visits, segment shares, spending averages and multipliers. Our experience is that the largest errors are usually due to the visit data. Next in importance are the spending averages if they don't accurately reflect your visitors. Multipliers mostly affect the estimates of the secondary effects and if properly chosen are the least important source of errors. The NPS Park visit estimation protocols are some of the best of any resource management agency (thanks to Butch Street), but counting visitors is inherently messy and MGM2 requires additional information to sort out park re-entries, determine days/nights spent in the local area, and estimate segment shares.  If the spending averages come from your own park visitor survey, then sampling errors for the study provide an estimate of one source or error. However, "non-sampling" errors can be much higher due to poor question design, unrepresentative samples, failure to weight the data for different probabilities of selection, and handling of missing data and outliers. The basic model is multiplicative (visit *spending*multiplier), so errors in visits, spending and multipliers will multiply the overall error,  if all errors are in the same direction, but may cancel if in different directions. For example, if visits, spending and the sales multiplier are each inflated by 10%, total sales will be inflated by 33%.  If visits are overestimated by 20% and spending is underestimated by 20%, the errors will roughly cancel out in total spending .

Better approach: Instead of the statistical approach we recommend establishing the validity of the impact estimates by demonstrating the face validity or reasonableness of the inputs or showing the consistency with other accepted data/sources. Spending averages for a typical day visitor or camper should be consistent with common sense and reflect the current prices in the area. Spending per party per night for lodging should reflect the local room or campsite rates. MGM2 reports spending by segment on a party night basis, because these averages are more readily evaluated against common sense and published rates. As spending rests heavily on the number of visitor nights in area motels and campgrounds, we strongly recommend attempting to validate these figures against local data sources. Obtain counts of available rooms (or campsites) in the area and average occupancy rates from local tourism organizations. Compute total room nights for the area (number of rooms * average occupancy rate * 365 day season)  and determine what share of these nights you are claming as due to park visitors. Make sure it seems reasonable and, in particular, isn't more than 100%.