NATIONAL ASSOCIATION OF STATE PARK DIRECTOR S REPORT FY 2012 OUTLOOK AND ANALYSES. Prepared By: Christos Siderelis and Yu-Fai Leung

Transcription

1 2012 Outlook and Analyses of U. S. States Park Systems 1 NATIONAL ASSOCIATION OF STATE PARK DIRECTOR S REPORT FY 2012 OUTLOOK AND ANALYSES Prepared By: Christos Siderelis and Yu-Fai Leung Department of Parks, Recreation & Tourism Management North Carolina State University May 3, 2012 (revised version) It has been reported that Arizona eliminated the state funding for parks, Colorado, Georgia, and Massachusetts reduced state spending by $3 million, $10 million, and $23 million, respectively (Shinkle, 2011). We have begun monitoring the monthly state coincident index (Philadelphia Federal Reserve Bank, 2012). The coincident index summarizes a state s economy by combining four variables into a state s index--nonfarm payroll employment, average hours worked in manufacturing, the unemployment rate, and wage and salary disbursements deflated by the consumer price index. We investigate the influences of producer behaviors with the states coincident indexes in determining their influence on the states expenditures for outdoor recreation later. For now, we display the annual means of the indexes that span the current and the past five election cycles (Figure 1). The trend is relatively flat from 1988 to 2007, and then the trend line slopes downward in 2007, coinciding with the recent economic recession '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 '10 80 Figure 1. Trend of the annual means of the states' coincident indexes, which are measures of each state's economic activity in time (Philadelphia Federal Reserve Bank, 2012).

2 2012 Outlook and Analyses of U. S. States Park Systems 2 Against this weak economic backdrop, we begin the report with time-trend analyses of the revenue, capital and operating expenditures, labor, and parkland. We describe their long-run directions and movements and display the step-ahead forecasts for We next focus on the allocation decisions by the states officials in providing the quality and quantity of outdoor recreation that the state residents demand, commensurate with the charges levied on them by the state at the different points in time. We perform this task by relating the production of outdoor recreation and the operators annual deployment of three inputs capital, labor, and parklands in determining the capacity utilization rates of the parks systems. We are well aware that in the case of the producer, the producer regards intuitively the allocation of resources as causal in its effect on production because the market is separable from production and the producer can determine optimal market conditions for production. For the states parks, the linkages between the production of annual attendance for consumers and the allocations of the factors of production may not be causal at all. While the operator may want additional labor to serve an increase in attendance, it is not clear that the hiring of more labor will generate an increase in attendance, as presumably would be the case with the producer wanting to gain a greater product market share. Nevertheless, the question of how operators who are in charge of the direct government spending for parks allocate resources between the factors should provide a clearer understanding of how these factors affect the production of outdoor recreation. We then characterize the states parks as price-excludable public goods and follow with an analysis of the influences of the states economic conditions on the expenditure constraints faced by the operators. We investigate the associations between the average expenditures, average revenues, and capacity utilization rates and conclude with our speculations about the current direction and the prospects for the self-sufficiency by operators. In producing this report, we left the appendices containing the data collection process, the remaining information about the statistical methods, and the reporting of results in support our findings in a separate document for downloading by the reader. Where appropriate, we direct the reader to a particular appendix.

3 2012 Outlook and Analyses of U. S. States Park Systems 3 Time-Trends We began with the assumption that trend in the time series data for a supply variable is a random walk where the original variable is equal to its most recent value plus a random error. We take this position because of the uncertainty surrounding the economic, legal, and environmental effects on the states legislative outcomes and appropriations during the past 26 fiscal years on the states parks. Consequently, we are unable to relate the prior events occurring in the 50 states to the directions of the trends. For example, if the aggregate amounts of annual revenues seemingly trend higher, we then leave it to the reader to posit whether increased marketing, accessibility, or other actions could possibly explain the increase or decrease in annual revenues. We do however comment on the direction of the trend lines over the span in time. Our one-step projections apply dynamic panel estimators of the forecasting equations to the AIX panel data (50 states 26 year) for the time series variable. The non-linear trend-following method generates forecasts by extrapolating past sequences of movements in the time series data into the future. The variables are serially dependent over time. This allows us to estimate the equations that describe the annual series conservatively from the specified, time-lagged (previous) values of the variables. Adhering to a necessary statistical assumption of the time-series estimator, we aggregate the annual operating and capital expenditures. The forecasting equations therefore include only the lagged variables and are special cases of linear regression equations. We next sum the time series data for the supply variable by each year spanning FY 1988 to FY 2011 to coincide with the current and the past five election cycles, although we worked the AIX panel-data from FY The step-ahead forecast for 2012 is the product from the dynamic panel estimator. In addition, we provide the relative changes in the series to provide readers with insights of the annual adjusts by operators in producing and supplying outdoor recreation. A relative change results from calculating the natural logarithm of the current value divided by the past value at each point in time. We list in Appendix B (Table B2) the 2012 expected values for the supply variables by the parks systems.

4 2012 Outlook and Analyses of U. S. States Park Systems 4 3,800 3,300 Millions 2,800 2,300 1,800 1,300 '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 '10 '12 Figure 2. Annual total expenditures adjusted for inflation (2011 dollars) and 2012 forecast. 20.0% 10.0% 0.0% -10.0% -20.0% '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 '10 '12 Figure 3. Relative changes of annual total expenditures. During FY 1993 to FY 2008, capital and operating spending increases with a subsequent decline in spending in FY 2009 (Figure 2).This appears to coincide with the downturn in the states coincident measures (Figure 1). Since FY 1992, the relative changes in spending appear stable to FY 2008 with a rebound in growth reported in FY It is difficult to identify the sources of growth, which may or may not relate to federal stimulus capital improvement projects (Figure 3). The one-step forecasting model projects an increase in spending for FY 2012.

5 2012 Outlook and Analyses of U. S. States Park Systems 5 60,500 55,500 50,500 45,500 40,500 35,500 '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 '10 '12 Figure 4. Trend of the annual labor (workers) and 2012 forecast. 20.0% 10.0% 0.0% -10.0% -20.0% '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 '10 '12 Figure 5. Relative changes of the annual labor (workers). Since FY 1988, the employment of labor for the states parks steadily increases or remains stable until FY 2008 (Figure 4). The relative changes in labor decline slightly in FY 2009 and since then are returning to stable employment levels (Figure 5). The one-step forecasting model projects that park labor nationwide will be returning to the FY 2008 level in FY 2012.

6 2012 Outlook and Analyses of U. S. States Park Systems '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 '10 '12 Figure 6. Trend of annual parkland (acres) and 2012 forecast. Millions 10.0% 0.0% -10.0% '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 '10 '12 Figure 7. Relative changes of annual parkland (acres). Since FY 1988, the states added parklands to the national inventory (Figure 6). Following the accumulations of parklands in the early 1990 s to FY 2005, the trend in the relative changes suggests a stable inventory of parklands even with the slight FY 2006 decline in parkland (Figure 7). The one-step forecasting model projects no change in the parklands under management for FY 2012.

7 2012 Outlook and Analyses of U. S. States Park Systems '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 '10 '12 Figure 8. Annual revenues adjusted for inflation (2011 dollars) and 2012 forecast. 1,100 1,000 Millions 20.0% 10.0% 0.0% -10.0% -20.0% '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 '10 '12 Figure 9. Relative changes of annual revenues. Revenues have risen steadily since FY 1988 (Figure 8) although the relative changes in revenues fluctuate irregularly, varying from a relative high of 15% in FY 1991 to a relative low of -10% in FY 2010 (Figure 9). The one-step forecasting model projects a decrease in revenues in 2012 from the reported high FY 2011 amount.

8 2012 Outlook and Analyses of U. S. States Park Systems 8 Production of Outdoor Recreation Landrum s (2004) historical examination of the states parks concludes that the collective choices and mandates of public officials guide the systems of state parks, and not necessarily market demand and supply conditions. We illustrate Landrum s view with a conventional description of a state s park. We employ a single state park and a visitor with a fixed utility U(Y, X) for the illustration, although the consideration of many visits would not alter this form. Utility is the total satisfaction from consuming visits. Y is the single output from the consumption of recreation opportunities from visits to the state park. X are the factors of production that are empirically observable in the normal process of production capital investment, labor, and parkland. The operator is producing the highest amount of outdoor recreation opportunities with the maximum amount of expenditures, C = f(x), allowed. I include the symbol f as an argument because the operating labor wages and rental rates of capital are variable. Applying the theory of production from economics, Seneca and Cicchetti (1969) propose a user response function to gain an understanding of the relationship between the counts of visits at a cross-section of 154 parks in the Appalachian Region and the factors of production required to produce those visits. In this regard, we propose the operator solves the problem of maximizing U(Y, X) by producing outdoor recreation for the visitor with the response function Y = f(x) (Siderelis, Moore, Leung, & Smith, 2012). Operating in the best welfare of the visitor, the operator substitutes the response function into the visitor s utility until the marginal utility equals the marginal disutility of providing additional outdoor recreation. Conceptually, the operator attempts to satisfy this objective in the long run by incrementally appropriating more factors of production from the direct government spending on operations and capital improvements efficiently, until the attendance increases to an optimum quantity of outdoor recreation at a break-even point in the supply and demand and at the user s fixed utility. A theoretical aspect in applying production theory is that the user response function should be homogenous of degree one or exhibit constant returns to scale, which implies simply that the multiplication of all factors by the same constant results in the multiplication of a park s attendance by that constant. In this application, the user response equation is homogenous of degree 0.84, or less than one, so the response function exhibits decreasing return to scale. If the operator doubles the number of factors of production, the operator obtains less attendance. Given Landrum s (2004) observation about the nonmarket nature of the states parks and the varying political actions influencing the 50 states production processes, we question whether the constant returns to scale argument is a reasonable assumption for the cross-sectional supply of 50 states park systems. We remark at the time that in our opinion it is an empirical question (Siderelis, et al., 2012). In addition, we report the inefficiencies in the relationships between attendance and the factors of production might be contingent on a wide variety of unobservable actions and decisions taken by the state officials in supporting outdoor recreation. To begin with, the states parks system budgets are not set through regional planning commissions or coordinated efforts. In fact, the states have adopted various land-management orientations typically bound by past legislative actions and directives that influence appropriations. It is also plausible to us that states parks agencies may under-invest in the recreation facilities and their continuing operations if the adjacent states offered outdoor recreation opportunities of a higher quality as perceived by visitors. The logic of this spillover effect is that state policy-makers would be less likely to invest in their own outdoor recreation services if adjacent states have a high level of financial commitment to the provision of those services. A colleague points out that this effect might be especially notable in the eastern United States where interstate travel is less of a barrier to recreation participation. In the east, this scenario might be a valid option for state agencies to conserve their own fiscal resources and still meet their constituents outdoor recreation needs. Other impediment to efficient production may include distributional equity

9 2012 Outlook and Analyses of U. S. States Park Systems 9 issues involving having to produce of outdoor recreation among dispersed geographical locations, or the legal restrictions imposed on the salary levels of park workers, or even the restrictions on permissible debt levels imposed on states. In specifying the user response equation, we assume that the annual visits to the park systems are of constant average quality. Even though, we are aware that the park systems must fulfill mandates to serve all users, and operators are more likely to vary the quality of services and not likely to ration the number of users. The user response equation describes the contributions of the state operator s decisions to allocate (a) non-recurring capital investments in park improvements, (b) required labor, and (c) acres of parkland. The operators control the factors, which are necessary in the production of outdoor recreation and common in all the 50 of the parks systems (Table 1). Table 1. Average Products for the 50 Park Systems for FY 1986 to FY 2011 (see note) Production Factor Mean Standard Deviation Capital improvement $3.69 $14.08 Labor 17,757 18,466 Parkland Note. An average product is the annual attendance divided by the amount of the production factor. The sample size is 1,300 observations (50 states 26 years). Extending the statistical analysis from FY 1986 to FY 2011, involves us fitting the user response equation to the panel data (see Appendix C). Overall, the equation is statistically significant with each of the factors of production significantly influencing the counts of annual attendances. The expected utilizations are 75% correlated to the 1,300 reported attendance counts. Of the significant factors, labor influences the expected utilization the greatest, followed by the inventory of parkland, then the capital improvement. The user response equation predicts the expected utilization over the past 26 year and performs well in capturing the dispersion in the annual attendance counts (Table 2). Although we continue to refer to the results as the expected utilization from the production process, we define the expected utilization as the amount of outdoor recreation opportunities that the operator is producing for visitor consumption. Table 2. Utilization Statistics for the 50 Park Systems for FY 1986 to FY 2011 Mean Standard Deviation Reported attendance 14,707,556 16,658,253 Expected utilization 14,500,238 12,474,815 Note. The sample size is 1,300 observations (50 states 26 years). As we did in the time-trend analyses, we sum the annual attendance counts and the expected utilizations for the parks systems by year and plot this information in Figure 10. The annual attendance counts

10 2012 Outlook and Analyses of U. S. States Park Systems 10 increase annually prior to FY 2000, but the attendance growth experiences a decline since then. Alternatively, the same changes in capital improvements, additional labor, and the added inventory of parkland result in the expected utilizations trending upward until FY 2008, then declines Millions '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 ' Figure 10.Trends of annual attendance and expect utilization (broken line) for the national supply. Not shown in Figure 10 is the attendance projection for FY We estimate an increase of 2.9% in national attendance of 742,130,627 when inserting the 2012 product values into the user response equation. Since the user response equation is consistent across the 50 states, we calculate the capacity utilization rates for states from the factors of production. The capacity utilization measures the rate at which the reported annual attendance meets the expected utilization amount. The rate is the ratio of the reported attendance to the expected attendance times 100. At capacity is a utilization rate of 100, while slack capacity is less than 100 and over-capacity is greater than 100. We list this information for FY 2011 by the parks systems (Table 3). Table 3. Mean Attendance and Capacity Utilization Rates for the States Park Systems (FY 1986 to FY 2011) Park System Reported Attendance Expected Utilization Rate Alabama 5,430,163 9,820, Alaska 4,888,640 9,346, Arizona 2,372,166 6,200, Arkansas 7,637,435 12,969, California 76,687,040 61,181, Colorado 10,710,043 17,201, Connecticut 7,498,986 12,859,298 59

11 2012 Outlook and Analyses of U. S. States Park Systems 11 Delaware 3,572,948 8,298, Florida 15,889,737 26,036, Georgia 13,432,831 14,904, Hawaii 12,892,969 3,292, Idaho 2,805,016 6,175, Illinois 39,796,469 20,206, Indiana 14,868,450 18,797, Iowa 13,100,728 6,863, Kansas 6,460,227 5,539, Kentucky 13,478,194 21,561, Louisiana 1,561,961 7,763, Maine 2,212,816 6,297, Maryland 9,862,001 17,745, Massachusetts 16,642,249 25,448, Michigan 23,295,402 20,869, Minnesota 7,820,592 13,844, Mississippi 3,569,754 5,946, Missouri 15,972,805 13,491, Montana 2,979,092 4,492, Nebraska 9,563,518 12,430, Nevada 3,144,892 5,758, New Hampshire 3,004,919 8,775, New Jersey 13,993,228 19,347, New Mexico 4,561,931 6,308, New York 57,323,659 66,562, North Carolina 11,431,822 13,311, North Dakota 1,020,793 3,263, Ohio 57,174,623 21,454, Oklahoma 14,843,114 11,647, Oregon 40,741,446 10,868, Pennsylvania 36,267,923 24,335, Rhode Island 5,254,378 5,519, South Carolina 8,314,502 11,764, South Dakota 6,951,715 8,932, Tennessee 27,952,723 20,303, Texas 17,459,896 22,331, Utah 5,676,843 7,584, Vermont 938,091 5,822, Virginia 5,332,531 11,461, Washington 44,873,943 16,694, West Virginia 8,154,015 16,980,942 50

12 2012 Outlook and Analyses of U. S. States Park Systems 12 Wisconsin 13,674,560 11,489, Wyoming 2,284,034 4,908, Grand mean 14,707,556 14,500, As a statistical measure of the state of affairs for the national supply, we believe the capacity utilization rates to be an effective means for park system comparisons to at capacity (100) only and not optima for each system. This is because an optimal allocation is not consistent with the legislative priorities and decentralized decision-making of the states officials that may influence the production of outdoor recreation opportunities. Historically, the nation s supply exhibits an over-capacity to FY 1990, and then declines to capacity until FY 2003 (Figure 11). The period of slack capacity continues although since FY 2009 is returning to at capacity with the current mix of the product factors '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 '10 70 Figure 11. States' parks average capacity utilization rates where at capacity = 100, slack capacity < 100, and over-capacity > 100. Overall, the current mix of product factors would suggest otherwise. Operators are applying the appropriate mix in producing outdoor recreation opportunities for the public, as shown in Figure 11. We continue to assume the outdoor recreation opportunities produced are of constant average quality. Earlier, we mentioned that the park operators have dealt with increased economic pressures on state budgets from FY 2008 to FY The transfer of the legislatively appropriated amounts of money occurs annually. The amount of the appropriation coupled with the revenue from user charges, if any, is a constraint that describes the consumption options available to allocate this limited amount among a system s operations, park maintenance, and capital improvement programs. In this regard from our historical FY 1986 to FY 2011 analysis of the systems annual expenditure constraints, we estimate that at capacity (100) the average annual expenditure constraint will be $39,579,867 to maintain and operate a park system that produces 14,500,238 outdoor recreation opportunities (Appendix C).

13 2012 Outlook and Analyses of U. S. States Park Systems 13 Not mentioned is the disutility of the increased capacity utilization rate of a park s system to the visitor stemming from increases in congestion of facilities (e.g., parking), noise, and conflicts in use. If the actual value of the capacity rate is higher than the value of 100, at capacity, we might then infer from the overcapacity rate that the congestion costs to the visitor caused by the increased attendance may contribute to a reduction in the nature of their outdoor recreation experiences. This information can only come from the collection of onsite data. Parks Systems Are Price-Excludable Public Goods Total revenues contribute to both the operating and capital expenditure budgets providing constant sources of income toward the self-sufficiency of the park s systems over the past decades. We examine the revenues and charges from park admissions, golf courses, resorts, ski areas, aquatic facilities, marinas, and other sources that are becoming increasingly important in covering the shortfalls of direct government spending on parks (Figure 12). As of FY 2011, annual revenues as a percent of total expenditures (2011 constant dollars) increase 21% on a relative basis, while attendance at the states parks declines by 2.8%. 45% 40% 35% 30% 25% '88 '90 '92 '94 '96 '98 '00 '02 '04 '06 '08 '10 20% Figure 12. Trend of revenues as a percent of total expenditures in deflated 2011 dollars. A state s park becomes a price-excludable public good when adopting different market-based pricing strategies that make it possible to prevent people from attending who have not paid the necessary fees while retaining the characteristic of non-rival consumption (Burns & Walsh, 1981; Pearce, 1995). We attempt to explain the revenue-maximizing behaviors of parks operators as follows. Information on the aggregate demand of all visitors to a parks system is inadequate because all the visitors can fully and equally consume the production of outdoor recreation. Technically, the operator then produces outdoor recreation never to exceed what the operator thinks is required to satisfy the highest demand at any price. Meanwhile, visitors face prices that are generally less than the cost of production at the margin. Operationally, this situation may necessitate the rationing of access to parks and facilities of the highdemand users (e.g., closure of California s state parks) to offset annual losses in expenditures rather than by price as would occur in the marketplace. We argue then that many operators do not concern themselves with demand, that is, the relationship of the price (costs visitors pay to visit the parks) and the

14 2012 Outlook and Analyses of U. S. States Park Systems 14 aggregate consumption of outdoor recreation. Rather, the operator with knowledge of the net direct government spending, the annual total expenditures minus the total park-generated revenues, and the production costs for providing outdoor recreation is more concerned about identifying the maximum revenue obtainable from the given level of outdoor recreation produced. In fact, we speculate that net spending will decrease slightly or remain constant while income from user fees will continue to rise incrementally in the future. We acknowledge that the operator may possess information about the composition of park visitors, enough to determine the number of visitors who would purchase a certain quantity of visits when having to pay a particular revenue-maximizing fee. The user fee is just another form of a tax directly borne by the visitor for the admittance and the use of facilities and represents some portion of the value that a visitor receives from a state park visit. User fees once imposed by operators change infrequently unlike the market place where prices fluctuate over time. Technical Measures of Performance In our attempt at further understanding the operations of parks systems, we examine two popular metrics from a recent 2012 poll of the states operators in tracking the comparative performances of parks systems. We perform these tasks by first relating the long run, direct annual spending for operations and capital improvements with annual attendances to compute the average total expenditure per visit or simply the average expenditure. Next, we relate the total annual revenue to the annual attendance in calculating the revenue per visit as the measure of average revenue. The revenue per visit is a popular metric among the operators even though the park visitor is not required to pay a user fee for access or use of the facilities in every instance. In evaluating the technical indicators on systems operations, we are primarily interested in examining the influences of the capacity utilization rates (Table 4). Table 4. Mean Values of Popular Technical Indicators and the Capacity Utilization Rates by Park Systems (FY 1986 to FY 2011) Park System Capacity Utilization Rate Average Expenditure Average Revenue Alabama 62 $8.40 $5.11 Alaska 53 $2.53 $0.37 Arizona 43 $10.60 $2.80 Arkansas 64 $6.37 $2.01 California 128 $4.62 $1.13 Colorado 64 $4.20 $1.62 Connecticut 59 $3.38 $0.53 Delaware 45 $10.18 $2.23 Florida 61 $6.02 $1.96 Georgia 94 $5.40 $1.92 Hawaii 402 $2.98 $0.37 Idaho 49 $5.82 $1.24 Illinois 200 $5.57 $0.40 Indiana 83 $2.81 $1.83 Iowa 192 $1.57 $0.27

15 2012 Outlook and Analyses of U. S. States Park Systems 15 Kansas 122 $2.04 $0.72 Kentucky 65 $11.15 $5.55 Louisiana 23 $20.06 $1.61 Maine 36 $4.63 $1.07 Maryland 59 $5.99 $1.38 Massachusetts 65 $5.16 $0.54 Michigan 119 $2.33 $1.35 Minnesota 57 $5.44 $1.39 Mississippi 61 $6.78 $2.77 Missouri 119 $2.57 $0.36 Montana 67 $4.57 $0.71 Nebraska 77 $2.19 $1.21 Nevada 55 $4.15 $0.63 New Hampshire 38 $6.62 $4.50 New Jersey 72 $3.93 $0.53 New Mexico 74 $5.62 $1.00 New York 93 $4.25 $1.07 North Carolina 90 $3.80 $0.32 North Dakota 32 $4.95 $1.12 Ohio 271 $1.60 $0.46 Oklahoma 149 $3.11 $1.42 Oregon 399 $1.16 $0.37 Pennsylvania 152 $2.72 $0.41 Rhode Island 95 $2.96 $0.72 South Carolina 72 $4.08 $2.04 South Dakota 78 $2.39 $1.08 Tennessee 138 $3.23 $1.32 Texas 80 $5.56 $1.70 Utah 78 $7.54 $1.73 Vermont 16 $10.21 $7.55 Virginia 48 $6.29 $1.27 Washington 276 $1.64 $0.32 West Virginia 50 $5.14 $2.45 Wisconsin 124 $2.11 $0.89 Wyoming 48 $4.45 $0.40 Mean 100 $5.20 $1.56 Average Expenditure We must mention here that the officials of the parks systems in the U. S. use the term expenditure rather than costs. Expenditure relates to the external payments made by the states parks that will only cost the state money. The spending by the states operators may or may not have economic impacts in creating

16 2012 Outlook and Analyses of U. S. States Park Systems 16 additional employment and incomes for the local economies surrounding the parks from the spending generated locally by visitor from outside the region. As an aside, we would like to remind the reader that the purposes of the states parks are to contribute to the public well-being by offering accessible outdoor recreation opportunities, which can contribute to recreationists physical conditioning and mental wellbeing, in performing their traditional roles of protecting natural resources. The states parks provide welfare or economic benefits in the form of consumer surplus to the users of the states parks. We notice that the states operators tend to confuse the concept of welfare benefits with estimations of economic impacts from visitor spending studies. We relate the long run, direct annual spending for operations and capital improvements with annual attendances to compute the average total expenditure per visit or simply the average expenditure. An examination of the average expenditures with the varying capacity utilization rates depict the case of a continuous, downward sloping curve with a point elasticity of a negative 0.82% (SE = 0.015). A point elasticity of less than zero is evidence of the decline in average expenditures at a 1% increase in capacity as is the case where the operating efficiency is less than constant return to scale (Figure 13), while a 1% increases in the coincident index results in a 0.64% increase in the average expenditure. At capacity (100), the average expenditure is $4.76. Regarding the magnitudes of the significance values of the two determinants, the operator effort in apportioning expenditures efficiently continues to influence the annual funding for a park system more so than the state s economic condition (Appendix D). $22.00 $20.00 $18.00 $16.00 $14.00 $12.00 $10.00 $8.00 $6.00 $4.00 $2.00 $ Capacity Utilization Rate Figure 13. Average expenditures (total expenditure per visit) and predicted expenditure curve (circles) at varying capacity utilization rates. The average expenditure at capacity (100) is $4.76 with the remaining determinants held constant at their means.

17 2012 Outlook and Analyses of U. S. States Park Systems 17 Average Revenue We associate the coincident index and the ecological regions with the average revenue. A systematic examination of the long-run influence of average revenues at varying capacity rates depicts a continuous, downward sloping curve with a point elasticity of a negative 0.77% (SE = 0.026) (Appendix D). A point elasticity of less than zero is evidence of a decline in average revenues for a 1% increase in capacity (Figure 14). We posit that this outcome maybe the opposite of what the operator s expect. An operator implementing market-based pricing practices may believe revenues will increase as attendance increases to a break-even point with the average expenditure. This is simply not the case here. User fees infrequently change unlike the marketplace where prices fluctuate with demand and supply conditions. $8.00 $7.00 $6.00 $5.00 $4.00 $3.00 $2.00 $1.00 $ Capacity Uitilization Rate Figure 14. Average revenues (revenue per visit) and predicted average revenue curve (circles) at varying capacity utilization rates. The average revenue at capacity (100) is $1.04 with the remaining determinants held constant at their means. Theoretically, absent of prices for park entrance, rentals of facilities, permits, and so forth in the AIX data, the prevailing onsite cost to the visitor or price could be the ratio of total annual park-generated revenue and the annual attendance, which is the average revenue. Interpreted in this manner, the price may include the entrance fee with the remaining costs, if any, dependent upon the endogenous quality choices the visitor makes. Examples include sleeping accommodations, food services, facility rentals, and other complementary outdoor recreation activities consumed onsite. The price then is actually just the onsite pecuniary cost the visitor pays, and is not the visitor s cost for the length of time spent onsite or travel cost as is the case in behavioral demand studies. In studying the relation to price to the capacity utilization rate, the average revenue function is an inverse demand function because we set price equal to the average revenue. We can then solve for price as a function of the capacity utilization rate as shown with the average revenue curve in Figure 14.

18 2012 Outlook and Analyses of U. S. States Park Systems 18 Ecological Conditions in the U.S. In analyzing the average expenditures and average revenues, we include the ecological regions in the statistical analyses to provide the environmental context of the parkland locations that may contribute to variations in the patterns of spending and the collections of revenues per visit among operators. Ecological regionalization is a common tool for natural resource assessments, planning, and management at broader spatial scales. The Commission on Environmental Cooperation (1997) prepared a detailed report classifying North America in ecological regions based on the similarities in (a) geology, (b) physiography, (c) vegetation, (d) climate, (e) soils, (f) land use, (g) wildlife distributions, and (h) hydrology. The classification system has four levels with each of the levels providing more ecological detail. We applied the first level of ecological attributes, which divides North America into 15 broad ecological regions. This resulted in the states assigned to one of seven ecological regions, with the exception of Hawaii, which we assigned to the Oceania ecological region (Table 1). A majority rule assigned a state to a particular region if most of the state s territory matched the ecological characteristics in that particular region. The proportion of a state s territory in each case matching the characteristics of the targeted ecological region ranged from over 25% to greater than 75%. The bulk of the 50 states (n = 27) were assigned to the eastern temperate forests. Table 5. Assignments of the 50 Park Systems to Level 1 Ecological Regions of the United States Eastern temperate forests (n = 27) Alabama, Arkansas, Connecticut, Delaware, Florida, Georgia, Illinois, Indiana, Kentucky, Louisiana, Main, Maryland, Massachusetts, Michigan, Mississippi, Missouri, New Jersey, New York, North Carolina, Ohio, Pennsylvania, Rhode Island, South Carolina, Tennessee, Virginia, West Virginia, Wisconsin Great plains (n = 9) Iowa, Kansas, Minnesota, Montana, Nebraska, North Dakota, Oklahoma, South Dakota, Texas Mediterranean California (n = 1) California North American deserts (n = 5) Arizona, Nevada, New Mexico, Utah, Wyoming Northern forests (n = 2) New Hampshire, Vermont Northwestern-forested mountains (n = 4) Colorado, Idaho, Oregon, Washington Oceania (n = 1) Hawaii Taiga (n = 1) Alaska Notes. We used a majority rule to assign a state to a particular region if most of the state s territory matched the ecological characteristics in that particular region. The proportion of a state s territory matching the attributes of the targeted ecological region ranged from 25% to over 75%.

19 2012 Outlook and Analyses of U. S. States Park Systems 19 The resulting spatial distributions of the technical measures of performance among ecological regions are in Table 6. Overall, operators in the northern forests (New Hampshire and Vermont), on average, spends the most per visit, whereas the Oceania (Hawaii) operator spends the least. The Taiga (Alaska) and Oceania (Hawaii) operators collect the lowest revenue per visit and the highest are operators in the northern forests (New Hampshire and Vermont). Table 6. Predicted Average Expenditures and Average Revenues by Ecological Regions in the U.S. for FY 1986 to FY 2011 Ecological Region Average Expenditures Average Revenues Capacity Rate Taiga $2.71 $ Northern forests $8.89 $ Northwestern-forested mountains $3.71 $ Eastern temperate forests $5.27 $ Great plains $3.74 $ North American deserts $6.61 $ Mediterranean $4.95 $ Oceania $2.22 $ Mean values $5.03 $ Speculations The implications of the findings from the technical analyses of the average expenditures and average revenues are clear. First, the rules governing the pricing practices for the 50 different state park operations are indeterminate because the determination of user fees depends in part on the distribution of the incomes and the economic conditions, which result in the different user fees. Second, it is apparent that park officials are practicing some form of average cost pricing with respect to the user s ability-to-pay the fees. Third, the states officials may be over-pricing particular services when in actuality user fees are at low prices or park access is free to the public. This situation is evident in the downward sloping average expenditures curve in the Figure 13, where the average expenditures decrease resulting from the increases in attendance to over-capacity conditions. Consequently, the parks systems require continued direct government spending to cover losses above average revenues as shown by the area between the broken and solid lines in Figure 15.

20 2012 Outlook and Analyses of U. S. States Park Systems 20 $14.00 $12.00 $10.00 $8.00 $6.00 $4.00 $2.00 $ Capacity Utilization Rate Figure 15. Predicted average expenditures (bold line) and average revenues (broken line) at varying capacity utilization rates. At capacity (100), the average expenditure is $4.76 and the average revenue is $1.04. We believe that many operators under current economic conditions are attempting to match the annual total expenditures to revenues from all system operations using private sector profitability criteria and their state s land-management orientations in managing the states parks. In this respect, Davis s (2008) measured the state s public-land management orientations on zero to 100 scales. Of particular importance to us are the scores for the preservation (M = 19.4, SD = 13.9) and the recreation (M = 19.1, SD = 17.6) orientations. For FY 2011, the statistically guidance we offer is that systems with recreation orientations have a statistically significant 23% probability (p > z =.018) of higher average revenues, whereas those with preservation orientations have an insignificant negative 0.12% probability of higher average revenues. In addition, past revenue amounts continue to influence annual funding for park systems more so than the states economic conditions. Conceptually, we would like to view the operators as attempting to satisfy capacity by incrementally adjusting the mix of the factors of production until attendances increase (decreases) to breakeven points. In actuality, there are simply no assurances that operators have in the past or in the future will select the least costly combinations of factors for outdoor recreation as efficiencies warrant. As in Figure 15, a parks system incurs a continuous decline in average expenditures, as the amount of under or slack capacity decreases. At capacity, the average expenditure is $4.76 per visit and the average revenue is $1.04 per visit. The charging of user fees appears as entirely reasonable although the collections of fees may be a greater obstacle than the imposition of user fees by operators. However, the installations of modern self-service pay stations and electronic fee collection machines may overcome this obstacle, particularly at remote locations where manned entrance stations are not feasible alternatives. Apparently, the need to break-even financially, although we assume important to the states officials, is not a necessary condition for park operations in the U. S. The consumption of outdoor recreation at the states parks will probably always be everywhere below the average expenditures at any at capacity

21 2012 Outlook and Analyses of U. S. States Park Systems 21 utilization rate. Evidently, the 50 states operators make do with fixed quantities of the production factors for the fiscal year and, in turn, execute the necessary adjustments in the production process, which may include closing state parks, deferring maintenance, reducing visitor services, enforcement, and may even involve marginal increases in user fees. The public however continues to view the quality and quantity of the states parks commensurate with the average state per capita expenditure of $14.75 (FY 2011 to 2010), thereby deterring many market-based suppliers. In terms of equity, the return on $1,000 spent in 2011 deflated dollars on the states parks results in a mean of 324 (SD = 266) visits. We acknowledge that state officials face difficult decisions regarding the choices between the economic efficiency of attempting to maintain minimum total expenditures and revenue self-sufficiency by imposing user fees for outdoor recreation under the current service condition. Our research of the AIX panel data concludes that increases in both average revenues and average expenditures negatively associate with the increases in the capacity utilization rates of states parks. In the appendices to follow, we apply the general linear model to the various analyses of the AIX panel data. For a comprehensive and a readable introduction to the general linear model, Wikens (2004) offers an excellent paper on the subject. Although not cited, the reader may want to access a seminal paper on the economics of congestion, public goods, and welfare authored by Oakland (1972), Journal of Public Economics 1, References Bruno, G. (2005). Approximating the bias of the LSDV estimator for dynamic unbalanced panel data models. Economic Letters, 87, Burns, M., & Walsh, C. (1981). Market provision of price-excludable public goods: A general analysis. Journal of Political Economy, 89(1), Commission for Environmental Cooperation (1997). Ecological regions of North America: toward a common perspective. Retrieved July 18, 2010 from Davis, S. (2008). Preservation, resource extraction, and recreation on public lands: A view from the states. Natural Resources Journal, 48, Gauss, C., & McCullagh, P. (1991). Bias correction in the generalized linear model. Journal of the Royal Statistical Society, Series B (Methodology), 53(3), Hilbe, J. (2008). Negative binomial regression (2 th Edition). New York: Cambridge University Press. NASPD (2012). Statistical reports of State Park Operations, Annual Information Exchange (AIX) (FY 1984-FY 2011). Raleigh, NC: National Association of State Park Directors. Philadelphia Federal Reserve Bank (2012). Retrieved on February 2012: StataCorp (2009). Stata: Release 11. [Statistical software.] College Station, TX: StataCorp LP. Seneca, J., & Cicchetti, C. (1969). User response in outdoor recreation: A production analysis. Journal of Leisure Research, 1(2),

22 2012 Outlook and Analyses of U. S. States Park Systems 22 Siderelis, C., Moore, R., Leung, Y., & Smith, J. (2012). A nationwide production analysis of state park attendance in the United States. Journal of Environmental Management, 99, Shinkle, D. (2011). National Conference of State Legislatures. Retrieved on January 2012: Wikens, T. (2004). The general linear model. Retrieved on March 2012: Wooldridge, J. (2000). Introductory Econometrics: A Modern Approach. U.S.: Southern Western College Publishing. U.S. Department of Commerce, Bureau of the Census, State Government Finances (2012). Retrieved on February 2012: U.S. Department of Commerce, Bureau of the Census, State Government Finances (2012). Retrieved on February 2012: US Department of Labor, Bureau of Labor Statistics (2012). PPI databases. Retrieved on February 2012:

23 2012 Outlook and Analyses of U. S. States Park Systems 23 Appendix A Data Collection The empirically data for the longitudinal and cross-sectional analysis of panel data are from the Annual Information Exchange for the 50 state park systems (NASPD, 2012). The exchange provided for the longitudinal, cross-sectional collection of state-reported annual surveys addressing states park operations. We limited the analysis of the short-panel data (50 states and 27 years) to the annual observations of empirical observable in the normal processes of park operations from FY 1984 to FY The panel is short because the number of states exceeds the number of years. We gave special care to including only the most complete data. We obtain the data for this longitudinal and cross-sectional analysis of panel data from the Annual Information Exchange (AIX), a database for the 50 states park systems (NASPD, 2012). The AIX compiles data from state-reported annual surveys related to states park operations. We limited our analysis of the data to the annual observations from 1984 to 2010 and gave special care to include only the most complete data. In those instances where longitudinal data of interest were missing or seemed to be inaccurately recorded in the database, we interpolated and extrapolated the time series data to fill in the holes (StataCorp, 2009, [D] ipolate). The missing information was probably due to the incomplete self-recording of information by state contact personnel and/or previous data management issues in updating and handling the AIX archive itself. There was, of course, no guarantee of uniform market and labor prices across the 50 state park agencies. In our panel data descriptions of the production and supply variables that follow the number in the parentheses refers to the table and the capital letters to the column in the table (NASPD, 2012). Attendance is the number of day-use and overnight visits that to a state s parks for any lengths of time (3, L). Attendance was a technical measure and should not be confused with the behavioral measure of recreation demand. Labor refers to the number of full-time, part-time, and seasonal employees to maintain, effectively operate, and protect state parks (6, U). Parklands include the amount of acreage of the areas in the states parks system and the water bodies, when the boundaries of the state park unit wholly enclosed the water body (1, AN). The expenditures include the capital expenditures for land acquired by cash or equivalent value (and not by other means), periodic fixed capital expenditures for park construction (5, Q), and operating expenditures for the operation and maintenance of the state park system per se (5, G). Capital expenditures are non-recurring and expected to improve the productive capacity of the states park systems. Operators exclude other related expenditures for such things as grants-in-aid to other entities. We concern ourselves with the direct capital and operating expenditures, and not the foregone opportunity expenditures for the alternative use of park lands because the states parks systems do not ascribe to the multiple-use doctrine of extracting valuable natural resources on parklands, as far as we know. Park-generated revenues are from fees and charges to include such receipts from entrance fees and charges for cabin and cottage rentals, camping, group facilities, concessions, and beaches and so forth in non-fee areas (5, DA). Revenues are recurring year after year. In time-series analysis, we deflate the expenditure and revenue data to FY 2011 to uncover the real growth in the variables. We deflate the operating and capital expenditures with the producer s price index (PPI), which tracks the average change in prices over time of domestically produced and consumed commodities and includes the prices for capital equipment, but excludes prices for services (Bureau of Labor Statistics, 2012). We apportion capital expenditures equally to the year reported and to the preceding year (Bureau of Economic Analysis, 2010). Alternatively, we deflated the park-generated revenues with the consumer s price index (CPI), which measures the average change in prices over time of goods and services purchased for personal consumption and excludes prices for capital equipment.

24 2012 Outlook and Analyses of U. S. States Park Systems 24 Appendix B Time-Trend Analyses We felt that knowledge of the prior values would reduce the uncertainty of future values. Our preliminary time-series analyses reveal the data for the supply variables are absent of lags in the past errors (moving average term) in the time series. Given the knowledge of the current value and our use of a prior or lagged value in the time-series estimator, the expected value of the prediction may be higher or lower than the current value. In our one-step forecast, we applied a finite distributed lag model or simply forecasting equation to the time series of each variable. The trend-following method generates forecasts by extrapolating past sequences of movements in the time series data into the future. The variables were serially dependent over time allowing us to estimate the parameter that describes the consecutive values of the annual series from the specified, lagged (previous) values. In preparing the data for time-series analyses, we transform labor, parklands, revenues, and the expenditures to natural logarithms for statistical analyses. This action meets the statistical assumptions for the normal distribution of the dependent variables and relaxes the assumption that the rate of change in the dependent variables is constant over the 26 years. We begin with unit root tests to determine if the time series for the variables needs differencing (StataCorp, 2009, [XT] xtunitroot). The unit root test (Table B1, column 5) is the augmented Dickey- Fuller test. Wooldridge (2000, p. 804) describes the unit root process as a highly persistent time series where the current value equals last period's value, plus a weakly dependent disturbance. We report the t- value results from the unit root test. All are greater than the 5% level of statistical significance and the time series did not need differencing. In this case, stationary random processes generate the time series where the joint distributions of the series are invariant across time, and the time series data need not be differenced, which is a necessary condition for stable coefficients in autoregressive equations that follow (Wooldridge, p. 805). The four forecasting equations, one for each of the variables include only the lagged variables and are special cases of linear regression models. We apply a dynamic panel-data estimator to address the fixedeffects bias due to the 50 park systems and short time-span (26 years), and to determine the relative magnitudes of the fixed effects on the estimates. We specify the forecasting equation in the following form for each of the variables: 1 The conditional mean of the variable ( for the j = 1 50 parks systems and t = 1 26 years is the natural logarithm of the lagged variable [ ], γ is the parameter to be estimated, u j is the fixedeffect of the park system for which no distributional assumption is made apart from being fixed over time. The remaining term is the unobserved heterogeneity ( ) over the sample and approximates N( 0, ), which is the serially uncorrelated random disturbance. The estimator calculates bias-corrected fixed effects for the standard autoregressive, panel data equations using the bias approximations in Bruno (2005) with the StataCorp (2009) ado command xtlsdv. In turn, we select as a consistent generalized method of moments (GMM) estimator, the standard one-step