Approaches to Estimating the Economic Impacts of Tourism; Some Examples

Approaches to Estimating the Economic Impacts of Tourism; Some Examples

Daniel J. Stynes

February 1998

Introduction

The purpose of this bulletin is to present examples of different approaches to estimating the economic impacts of tourism. In a previous bulletin (Stynes 1997), I summarize economic impact concepts and methods as they apply to tourism. Here we apply the methods to illustrative cases in order to demonstrate some practical approaches. Three specific examples are presented. These represent a range of alternatives for estimating the economic impacts of visitor spending. The techniques covered range from methods based largely on judgement, to methods that utilize secondary spending data and published multipliers, to the use of visitor surveys and input-output models. A third bulletin in this series discusses survey methods for measuring visitor spending and includes sample spending instruments. While the construction and operation of tourist facilities also has economic impacts, we will restrict our attention here to the impacts of visitor spending.

Review of Basic Approach and Levels of Analysis

The economic impact of visitor spending is typically estimated by some variation of the following simple equation:

Economic Impact of Tourist Spending = Number of Tourists * Average Spending per Visitor * Multiplier

This equation suggests three distinct steps and corresponding measurements or models:

(1)Estimate the change in the number and types of tourists to the region

(2)Estimate average levels of spending (often within specific market segments) of tourists in the local area.

(3)Apply the change in spending to a regional economic model or set of multipliers to determine the secondary effects.

The three steps and corresponding information typically involve distinct methods, models and information sources. Each component of the equation may be estimated via expert judgement, from secondary sources, through primary data collection, by means of a model, or through some combination of these methods. Table 1 from the earlier report summarizes the alternatives and is repeated here for the readers convenience. The approaches for each step may be mixed in a given study. For example, the number of tourists may be estimated using judgement, spending via a visitor survey, and multipliers from a published secondary source. Or, an input-output model may be applied to spending estimates derived from tourist spending averages and visitation numbers taken from secondary sources.

As one moves from judgement to secondary data to primary data and formal models, the methods become more complex and the time and expense of the study increases. The added cost is hopefully associated with estimates that are both more accurate and more detailed, although this isn’t always the case. In some cases good judgement or existing data may be more accurate than a new visitor survey, particularly if the survey has a low response rate, small sample size, and measurement and sampling procedures that do not guarantee a representative sample or reliable measurements.

Methods based on judgement typically yield highly aggregate estimates, while estimates derived from formal models may estimate spending within several categories and impacts within as many as 500 distinct economic sectors. Which method is preferred depends on the intended uses of the results, the accuracy and detail that are required, and the time, money and expertise available.

Table 1. Approaches to Tourism Economic Impact Assessment

Level / Tourism Activity / Spending / Multipliers
1- Judgement / Expert judgement to estimate tourism activity / Expert judgement or an “engineering approach” [1] / Expert judgement to estimate multipliers
2 / Existing tourism counts for the area or total estimates from a similar area or facility / Use or adjust spending averages from studies of a similar area/market / Use or adjust aggregate tourism spending multipliers from a similar region/study
3 / Estimate tourism activity by segment or revise estimates by segment from another area / Adjust spending that is disaggregated within particular spending categories & segments / Use sector-specific multipliers from published sources
4- Primary data / Visitor survey to estimate number of tourists by segment or a demand model / Survey random sample of visitors to estimate average spending by segment & spending category / Use an input-output model of the region’s economy

Three examples introduced in the previous bulletin are presented here in more complete detail. First we re-introduce them:

• The National Park Service’s “Money Generation Model “(USDI, National Park Service 1990.) is a simple fill-in-form for generating economic impacts. It is an example of a simple approach that relies largely on judgement and available secondary data in a highly aggregate form. While an extremely simple approach, it captures the essential elements of an economic impact analysis. The number of visits, average spending per visitor and an aggregate sales multiplier are entered on a simple worksheet to generate estimates of the direct and total sales effects of visitor spending. Sales effects are converted to income and jobs using ratios of income to sales and jobs to sales. Tax effects of visitor spending can also be estimated by applying local tax rates to sales estimates. With sound judgement in choosing the parameters, the MGM model can yield reasonable ballpark estimates of economic impacts at minimal cost. This approach, however, provides little detail on spending categories or which sectors of the economy benefit from either direct or secondary effects. The aggregate nature of the approach also makes it difficult to adjust recommended spending rates or multipliers to different applications. The example described here is taken from a study of the local economic impacts of Mammoth Cave National Park in Kentucky (Stynes and Rutz 1995).

• The Bureau of Economic Analysis’s (BEA) RIMS II user handbook (USDC, BEA 1992) illustrates how to apply published multipliers to estimate economic impacts. This approach starts with visitor spending (from a survey or secondary sources) divided into a number of spending categories and makes use of sector-specific multipliers to estimate the direct and total sales, income and employment effects. Multipliers from the BEA’s RIMS II model are used to estimate secondary effects. Multipliers are reported for 39 sectors for each state in the second edition of their report (USDC 1992). The BEA method illustrates the proper use of margins to account for retail purchases of goods and makes use of disaggregate sector-specific multipliers for each state. Multipliers for sub-state regions are not as readily available, but can be acquired from BEA or other sources. Secondary effects cannot be disaggregated to individual sectors using the BEA approach. The BEA approach is illustrated using a hypothetical increase in tourism to the State of Illinois.

• The MI-REC/IMPLAN System. Stynes and Propst (1992, 1996) have developed a fairly complete micro-computer-based system for estimating economic impacts of recreation and tourism. The system combines spreadsheets for estimating spending with the IMPLAN input-output modeling system. IMPLAN uses county level data to estimate input-output models for regions down to a county level. IMPLAN generates a complete set of economic accounts within up to 528 sectors for the region, including multipliers and trade flows. MI-REC spreadsheets estimate visitor spending within up to 33 spending categories based on the number and types of visitors attracted to an area. Spending is then bridged to the IMPLAN model sectors to estimate direct, indirect and induced effects in terms of sales, income and employment. Users may estimate spending via visitor surveys or use the MI-REC database of spending profiles, compiled from previous studies. The system also includes price indices to update spending data to a current year. The MI-REC /IMPLAN system is applied here to estimate the statewide economic impacts of tourism to Michigan in 1990.

Example 1. The Money Generation Model

The Money Generation Model (MGM) was developed by the National Park Service (NPS) to generate quick and inexpensive estimates of the economic impact of National Park visitor spending on the region’s economy. In it’s simplest form, the MGM relies on agency records for estimates of visits, American Automobile Association (AAA) estimates of per person per day lodging and meals expenses to estimate spending, and judgement or available sources for multipliers. A pretty good aggregate estimate of impacts can be obtained with this simple method if one has accurate visitation data, spending data that adequately represent the visitors, and multipliers for the local region. The default values suggested in the MGM manual are unlikely to provide accurate estimates for most applications.

The MGM estimates direct and total sales , the income and employment effects of this spending, and state and local government tax revenues. The MGM worksheet (with some minor modifications) is shown in Table 2, with entries completed for an application to Mammoth Cave National Park in Kentucky. All spending and visits is for 1993 and multipliers are for a three county region around the park. Notes explaining the application to Mammoth Cave and other tips for using the MGM approach are provided below by line number on the worksheet. The worksheet is divided into three sections.

Section A. Direct and total sales effects of visitor spending.

Line 1. Only visitors from outside the local region (in this case a three county area around the park) are counted as “new dollars” for an impact analysis. Park visitor counts entered on line 2 are reduced to include only non-local visitors using the percentage of visitors from outside the local area.

Line 2. The number of park visits for a given year are entered on line 2. One must be sure visit estimates are accurate and in units consistent with the spending average entered on line 3. The NPS estimates visits to the park from axle counts on access roads. Vehicle counts are expanded to person visits using a party size factor and adjusted somewhat for re-entries and commercial traffic. In assessing the park’s contribution to tourism in the area, person visits to the park are not as important as party trips to the area. The NPS's report of 2.4 million visitors to Mammoth Cave in 1993 was converted to 430,000 party trips to the area by adjusting for multiple entries to the park (many visitors stay in motels outside the park and may make multiple visits during their stay in the area) and dividing by the average party size per vehicle for each season. In 1993 there were an estimated 430,000 party trips to the area involving a visit to the park, 93% of these trips were from outside the local area.

Line 3. Spending averages are from exit interviews and a mailback survey of a random sample of park visitors throughout the year (422 returns). In this case, spending was itemized in 21 categories covering all spending during the visitor’s stay in the three county region around the park and spending profiles were developed for six visitor segments. Only a single aggregate visitor spending average is used on the MGM worksheet. Visitor spending inside the park for cave tours and camping are not counted here, as they were covered in a separate analysis of impacts of the park’s operational expenses in the area. If local spending data do not exist, the MGM manual suggests using the American Automobile Association recommended average per person per day cost of lodging and meals as an initial estimate of spending. Notice that this approach wouldn’t cover several other kinds of spending by visitors and the average motel rate doesn't apply very well to campers or day users. The AAA spending figures will generally not represent national park visitor spending very well.

Line 4. Multiplying the entries on the first three lines yields total visitor spending of $51 million. This figure includes all spending by non-local visitors in the three county region around the park.

Line 5a. The MGM model doesn’t explicitly handle the capture rate, so we have added an entry to reduce visitor spending to a measure of local final demand. Only 72% of visitor spending shows up as direct effects in the region, after deducting the producer prices of imported items bought by visitors. This deduction would not be necessary if only restaurant and lodging expenses were included as all of this spending is captured. The spending survey measured purchases of “local crafts” to better estimate the portion of manufactured items bought by visitors that were produced locally. Most other goods purchased by tourists, such as gasoline and t-shirts, are not made locally and only the retail margins on these items are captured.

Line 5b. The sales multiplier of 1.93 comes from an input-output model for a three county region around the park. The multiplier includes both indirect and induced effects (Type III). The sales multiplier is only slightly below the MGM suggested default of 2.0, although the effective tourist spending multiplier after taking into account the capture rate is only 1.4 (1.93 * 72%).

Line 6. Total sales impacts are calculated by multiplying visitor spending by the sales multiplier, while also adjusting for the percentage of spending captured as final demand in the local area.

Section B. Tax Revenues. State and local taxes are computed by applying sales and income tax rates to the visitor spending data. The MGM worksheet applies sales tax rates to total sales and an income tax rate to total income in order to estimate tax collections. A better approach is to disaggregate purchases to specific items and apply corresponding tax rates to taxable items. For example, a room use tax would be applied only to the cost of lodging, gasoline taxes only to fuel, and sales taxes only to taxable goods. Many items bought by tourists as well as indirect and induced sales may not be taxed or be taxed at different rates. There are similar problems with estimating income taxes using such an aggregate approach. An accurate estimate of tax effects requires a more detailed itemization of spending within categories that match those to which taxes are applied.

Section C. Income and employment effects are estimated by applying job to sales and income to sales ratios to total sales. Income effects are not calculated explicitly in the original MGM worksheet, but are buried in the tax calculations in section B. I have income effects to the impacts section here, as income is a much better impact measure than jobs and can be computed just as easily. The most commonly used measure of contribution to gross domestic product is value added, which is primarily the income effects (plus indirect business taxes).

Line 2. A job to sales ratio captures the number of jobs required to produce a given amount of sales, usually expressed in jobs per million dollars in sales. These ratios vary considerably from industry to industry. Tourism businesses and retail trade sectors typically have high job to sales ratios (40-50 jobs per million dollars in sales), other services moderate rates, and manufacturing typically has lower rates (less than 25 jobs per million in sales). The appropriate rate for tourism spending applications is probably near the MGM's suggested average of 30 jobs per million in direct sales, as this ratio reflects jobs associated with a mix of direct, indirect, and induced sales. Using an input-output model for Mammoth cave, we were able to accurately estimate this ratio. The ratio for Mammoth Cave visitor spending was 32 jobs per million in sales.

Line 3. The income to sales ratio entered on line 3 captures the total income effects per dollar of total sales. The income figure used here includes wage and salary income and proprietor’s income, rents and profits. Income to sales ratios vary across industries in a similar fashion as the job ratios. Tourism businesses may convert 50-60% of sales directly to income, while ratios for manufacturing can be much lower (20-40%). Based on the sectors receiving direct, indirect and induced effects of tourism spending, the income ratio associated with tourism spending generally falls between 45% and 55%. For Mammoth Cave, the income ratio was 51%.

Readers familiar with multipliers will note that the job and income ratios used in the MGM worksheet are slightly different than job and income multipliers normally used by regional economists. The MGM worksheet estimates total sales first, and then converts total sales to total jobs and income. More traditional multipliers (Keynesian) would convert from direct sales to total income and jobs without the intermediate step. The multipliers (from an IMPLAN model) in this case are .98 for income and 62 total jobs per million in direct sales. These multipliers should be applied to tourist spending that has been adjusted for the capture rate. Equivalently, the multipliers themselves may be adjusted by the capture rate to yield corresponding “tourist spending” multipliers (.71 for income and 44 for jobs per million sales). These adjusted multipliers can be multiplied directly times tourist spending to yield total income and jobs accruing to the region’s economy.