BRSF Index | WebMerchants Index | Search


Appendix F:

Travel Cost Methodology


1. Calculation of Non-Financial Recreation Value

2. Shortcomings with Methodology

We calculate the non-financial recreation value of the Buffalo National River (BNR) by a method known as the "Travel Cost Method" (TCM). The "non-financial recreation value" is the additional value that visitors receive, above that which they explicitly pay for (i.e., this recreation value is free value to visitors, above any financial value paid to concessionaires). We estimate their additional valuation as the travel costs that they incur to get to the BNR. The theoretical basis of TCM is that the distance that visitors travel, and the costs incurred in traveling, represent a "willingness to pay," and hence can be used to construct a "demand function" for the BNR can be assigned. The method requires counting the number of visitors by points of origin, multiplying that number by the estimate their travel costs from that origin. The BNR's recreation value estimated from this technique is $37.6 million per year.

1. Determination of Visitor Origins

To estimate the travel costs incurred by recreation visitors to the BNR, we first determined how many visitors come from different regions. We reviewed a sample of invoices from three different concessionaires, over a period of two years, to determine origin points of visitors. We chose the five most common states of origin, and assigned visitors from other states to a general category of "Others." The percent of the total for each state is shown in Table
F-1. We counted both the number of visitors, to determine origin proportions, and the number of vehicles they used, to determine their travel costs later. The number of visitors per vehicle was fairly constant across states, so we use the average (2.63 visitors per vehicle) for all origins.


Visitor Origins                                                   Table F-1           
                    

Arkansas  Missouri Louisiana     Texas  Oklahoma    Others     Total 

No. of                416       109        87        72        44       239       967 
Visitors                                                                              

Percent of          43.0%     11.3%      9.0%      7.4%      4.6%     24.7%           
Total                                                                                 



No. of Cars           157        40        32        29        16        93       367 

Percent of          42.8%     10.9%      8.7%      7.9%      4.4%     25.3%           
Total                                                                                 



Visitors per          2.6       2.7       2.7       2.5       2.8       2.6      2.63 
Car                                                                                   



2. Determination of Visitation Rate

We further broke down Arkansas into four regions, by counting up the home addresses of Arkansas visitors from concessionaire invoices. Texas and Louisiana have major cities at large distances from the BNR, so we broke down those two states into two parts: "within 350 miles" and "beyond 350 miles" from the BNR (a complete discussion of the distance selection appears in Appendix H). We calculated the populations in the two parts by adding up the city populations in each region; 33% of Texas' population was within 350 miles, and 56% of Louisiana's population was within 350 miles. The populations of those two states which are beyond 350 miles are counted as "the rest of the USA." The populations of Oklahoma and Missouri are all counted as within 350 miles, since all of their major cities fall within that distance. We hence have nine regions: four within Arkansas, the four most commonly visiting states, and the rest of the USA (foreign visitors count as "the rest of the USA" also).

We then determined a "visitation rate" for each region. To calculate the total number of visitors from each region, we multiplied the percent of total visitors (from row 2 of Table F-1) by the total number of visitors to the BNR. The total number of visitors comes from the National Park Service's "Monthly Use Reports," as detailed in Appendix C. We used an average of 1992 and 1993 visitation, a total of 1.1 million visitors per year. The populations of each of those regions, divided by the number of visitors from those regions, provides the "visitation rate," shown in Table F-2, which is the basis of the TCM calculations below. The visitation rate indicates what percentage of people from each region visit the BNR each year (over 100% means, on average, each person in the region visits more than once per year).


Visitation rates                                                 Table F-2        

                  Percent         Population      Average No. of       Visitation 

Origin           of Total     within 350 mi.      Visitors 92-93             Rate 

Local AR             8.5%             86,991              92,920             107% 

Fayetteville        10.6%            113,409             115,044             101% 

Little Rock          9.7%            505,600             106,194              21% 

Rest of AR          14.2%          1,693,000             154,866             9.1% 

Missouri            11.3%          5,193,000             119,496             2.3% 

Texas                7.4%          5,897,104              86,635             1.5% 

Oklahoma             4.6%          3,212,000              47,799             1.5% 

Louisiana            9.0%          2,405,007              95,597             4.0% 

Rest of USA         24.7%        222,335,000             277,829            0.12% 



3. Determination of Travel Costs

We estimated the average distance traveled by measuring the distance (on a map) from the largest city in each region to the nearest point of the BNR. Those distances are shown in Table F-3, column 1. The distance for the "Rest of USA" region is necessarily arbitrary; we used 500 miles as the distance traveled by a typical visitor who came from outside of the other eight regions. This region is important because it accounts for 25% of all BNR visitors. The distances will be used again in section 7.4 to predict visitation.

We broke down the costs of traveling to the BNR into two components: the cost of operating a vehicle, and the cost of the time spent traveling. The cost of operating a vehicle for one visitor's trip to the BNR is the cost per mile, times the number of miles, divided by the number of people per vehicle. We use a vehicle operation cost of 15.8¢ per mile, based on the US Department of Transportation's estimate of variable cost per mile of travel, inflated to 1993 dollars. That rate, multiplied by the miles in column 1, divided by the number of people per vehicle (2.63) from Table F-1, times two for a round-trip, yields the "Two-way Car Cost."

We estimated the number of hours required to travel to the BNR, by analyzing the road conditions from each region's largest city to the BNR; our estimate is in column 3. We use an "opportunity cost of leisure time" as one-third of the per capita income for each region. The "Two-way Time Cost" is that per capita hourly income, divided by three for leisure time, times the number of hours spent traveling, times two for a round trip.

The "Total Travel Cost" is the sum of the two components of expenses. We then calculated a weighted average for all travel to the BNR, for later use in our demand curve (section 7.6). The weighted average is the average cost per mile (the total travel cost divided by the number of miles), weighted by the number of visitors from each region (from Table F-2 column 3). We estimate the average cost of travel to the BNR as 22.5¢ per visitor per mile.


Travel Costs                                                    Table F-3         

                 Miles   Per Capita    Hours     Two-way      Two-way       Total 

Origin          to BNR       Income   to BNR    Car Cost    Time Cost      Travel 
                                                                             Cost 

Local AR            40     $10,753         1       $4.78       $3.45        $8.23 

Fayetteville        50     $12,756       1.5       $5.98       $6.13       $12.11 

Little Rock         96     $15,776       2.5      $11.48      $12.64       $24.12 

Rest of AR         150     $12,271         3      $17.94      $11.80       $29.74 

Missouri           252     $18,189         4      $30.15      $23.32       $53.47 

Texas              348     $18,731       7.5      $41.63      $45.03       $86.66 

Oklahoma           222     $17,220       4.5      $26.56      $24.84       $51.39 

Louisiana          300     $15,672         7      $35.89      $35.16       $71.05 

Rest of USA        500     $16,184        10      $59.81      $51.87      $111.68 



4. Calculation of Predicted Visitation

We next came derived an equation to predict visitation rates based on distance, so that we can add distances formulaically to make postulated fees for a demand curve. To make this equation, we fit a regression "best line" to the visitation rate (Table F-2 col. 4 as the dependent variable) and the distance data (Table F-3 col.1 as the independent variable). We used a standard statistical method for TCM, ordinary least squares regression, which gave us this equation:

log(Visitation Rate) = 4.7066 - 0.01376 x Distance "t-statistics": (6.64) (-8.31) R2 = 90.8%

Those numbers demonstrate that there is a very strong correlation between distance and visitation rate (in statistical jargon, well over the 99% confidence level), and distance accounts for 91% of the reason that visitors come to the BNR. We used a "log relationship" because we want to estimate the effect of a one-unit change in distance on the rate (in percentage) of visitation. We considered "left out variable" bias, but for our purposes, this equation is adequate, especially considering the strong confidence levels.

We then used our econometric equation to predict total visitation at the actual distances of each region. Table F-4 compares the predicted rate of visitation from our equation (column 1) with the actual visitation rates (column 2, copied from Table F-2). Our predictions are low for the regions closest to the BNR, but roughly accurate for most regions (Oklahomans, evidently, visit far less than their close distance would imply). We multiply the predicted visitation by the population of the region, to get the predicted number of visitors (column 3), which can be compared to the actual number of visitors (column 4, copied from Table F-2). Our prediction for total visitation is a little high overall, 1.2 million predicted versus 1.1 million actual visitors.


Predicted Visitation                                     Table  F-4            

Predicted          Actual       Predicted         Actual 
Origin               Visitation      Visitation          No. of         No. of 
Rate            Rate        Visitors       Visitors 
Local AR                  63.8%          106.8%          55,525         92,920 

Fayetteville              55.6%          101.4%          63,082        115,044 

Little Rock               29.5%           21.0%         149,339        106,194 

Rest of AR                14.0%            9.1%         237,861        154,866 

Missouri                   3.5%            2.3%         179,285        119,496 

Texas                      0.9%            1.5%          54,335         86,635 

Oklahoma                   5.2%            1.5%         167,563         47,799 

Louisiana                  1.8%            4.0%          42,894         95,597 

Rest of USA               0.11%           0.12%         252,995        277,829 

Total                                                 1,202,879      1,096,380 



5. Calculation of Postulated Visitation

Next, we add distance to the actual distance that visitors from each region travel, and calculate the number of visitors who would still visit at the further distance, based on our formula from section 4. Adding distance to the mileage is directly translatable into adding a fee to the cost of visiting the BNR -- for example, adding 10 miles is the same as adding an entry fee of $4.50 (10 miles, round trip, times 22.5¢ per mile from section 3). Our formula yields a predicted visitation rate (column 2 in Table F-5) for each region at the new distance (column 1), which we multiply by the region's population to get a predicted number of visitors (column 3). The sum of the number of visitors represents a demand at that price -- for example, 1,048,000 visitors would still come if there were a $4.50 entry fee. Repeating the process at higher mileages comes up with other points on the demand curve -- for example, at a postulated distance of 100 miles shown in columns 4 through 6 of Table F-5, 304,000 people would visit with an entry fee of $45.00. We repeated for distances up to 200 miles, resulting in the demand curve shown below.


Postulated Visitation                                              Table F-5           

Distance    Predicted   Predicted        Distance   Predicted  Predicted 
Origin            + 10   Visitation    Visitors           + 100  Visitation   Visitors 
mi.                                      mi.                        

Local AR            50        55.6%      48,387             140       16.1%     14,025 

Fayetteville        60        48.5%      54,972             150       14.0%     15,934 

Little Rock        106        25.7%     130,141             196        7.5%     37,721 

Rest of AR         160        12.2%     207,283             250        3.5%     60,081 

Missouri           262         3.0%     156,237             352        0.9%     45,285 

Texas              358         0.8%      47,350             448        0.2%     13,724 

Oklahoma           232         4.5%     146,022             322        1.3%     42,324 

Louisiana          310         1.6%      37,380             400        0.5%     10,835 

Rest of USA        510        0.10%     220,472             600       0.03%     63,903 

Total                                 1,048,245                                303,831 





See Figure F-5: Demand Curve for Recreational Use of the BNR

6. Total Value of Non-Financial Recreational Demand

The demand curve represented graphically in Figure F-5 is shown numerically in Table F-6. At an added mileage of zero miles, the added cost is $0, and the predicted number of visitors is the total predicted from our formula, in Table F-4. At an added mileage above 200 miles (the equivalent of an entry fee of $90), the demand drops off to zero. Since the actual "entry fee" to the BNR is zero, all of the willingness to pay a fee is a "consumer surplus," since visitors get the benefits of visiting while paying less than they are willing. The entire area under the demand curve, therefore, represents the recreational benefit of the BNR.

To calculate the area, we add up the area of columns on the curve, one for each mileage. (the area under the curve between 10 miles and 20 miles is: a rectangle of height $4.51, by width 1,048,000 minus 913,000; plus a triangle of the same width, by height $9.01 minus $4.51). The area of each column is shown in Table F-6 column 4. Its sum is the area under the entire demand curve, which represents the entire consumer surplus of the recreational value of the BNR. Thus, the non-financial recreational value of the BNR is $37.6 million per year.


TCM-Based Demand Curve                             Table F-6       

  Added         Predicted        Added Cost       Approx. Cost    

Mileage     No. of Visitors      (22.5¢ per       (area under    
                                     mile)            curve)       

     0           1,202,879           $0.00           $348,480      

     10          1,048,245           $4.51           $911,046      

     20           913,490            $9.01          $1,323,214     

     30           796,059           $13.52          $1,614,355     

     40           693,723           $18.03          $1,808,776     

     50           604,543           $22.54         $10,165,203     

    100           303,831           $45.07          $8,514,716     

    150           152,700           $67.61          $5,991,059     

    200           76,744            $90.14          $6,917,958     

   Total                                           $37,594,807     



Technical Notes

Our methodology follows the recommendations of US Water Resource Council, an agency of the federal Department of Interior. Our raw data (for determining the proportions of visitors from each region, for estimating distances and times for each region, and our spreadsheet work to calculate the tables here) are available for inspection but are not included in this report.

Shortcomings with Our Methodology

1) Concessionaires' customers may not be representative of BNR origins. We may have a "selection bias" in our method of choosing particular concessionaires as the source of visitor origins (in Table F-1), since one concessionaire's customers are not a random sample of actual visitors. For example, BOC (one of our concessionaires) advertises in Missouri and hence BOC's customers are more likely to be from Missouri than are BNR visitors in general. We address this bias by using data from multiple concessionaires. More generally, people who use concessionaires may not be representative of BNR visitors as a whole. In particular, the closer visitors lives to the BNR, the more likely is it that they will bring their own equipment, and hence not appear on concessionaire ledgers. We do not address this problem, and it is a potential source of overestimate, since perhaps the actual visitor origin ratios should be at lesser distances.

2) Our total travel cost per mile is based on average estimates. Because our regions of origins are fairly large, we had to estimate the typical time and distance traveled by a person coming to the BNR from that region (in Table F-3). We did this by measuring distances on a map, and by estimating the time by observing the type of highways which connect each region to the BNR. Our accuracy is hence limited by our ability to properly estimate travel time and distance, which is probably accurate only to the nearest 1/2 hour and the nearest 20 miles. We addressed this problem by intentionally estimating low on travel time, so that the resulting travel cost is conservative.

3) The "predicted visitation" formula doesn't account for local preferences. We assume there are some "left out variables" in our formula which determines visitation rate based only on distance from the BNR. We assume that more Arkansans, and more Ozark residents especially, visit the BNR than do residents from other states, just because the Buffalo River is in their "home" (as reflected in the under-prediction of local visitation in Table F-4). We also do not account for the nearness of other similar amenities (e.g., whether there are other rivers nearer to a point of origin), which may the reason for the low representation of Tennessee, for instance. We do not address this problem because it would make our TCM "non-standard." We have no guess as to whether this biases our estimate up or down.