USING THE TRAVEL COST METHOD TO ESTIMATE FRESH WATER BASED RECREATION IN NORTH CENTRAL FLORIDA By BRYAN HUY NGUYEN A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2017
2017 Bryan Huy Nguyen
To my niece Sadie
4 ACKNOWLEDGMENTS I would like to thank the University of Department for the opportunity to develop as a graduate student. I will be fo rever indebted. Thank you to the undergraduate professors, who motivated me to attend graduate school. Also I am thankful to have Dr. Xiang Bi, Dr. Kelly Grogan, and Dr. Tatiana Borisova who gave me a chance to conduct research and develop as a graduate student. Again, thank you for the opportunity to see the natural beauty Florida must offer. I am also thankful to have met many wonderful cohorts and would like to give special thanks to Qianyan Wu, Benjamin Avuwadah, Maria Diaz, Jessica Fernandez, Tori Bradley, Luqing Yu, and Mar ia Belen Belencita Medina. Mrs. Charlotte Emerson (Dr. E), I am eternally indebted to you for helping me return to my studies to complete my educational endeavors, and for your guidance (twice) on how to succeed and excel. Lastly to my family for giving me the motivation to finish my degrees. The Nesic family for giving me access to the web and feeding me Serbian cuisine. To all my friends thank you for your support and encouragement to finally cross the finish line.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 6 LIST OF FIGURES ................................ ................................ ................................ .......... 7 ABSTRACT ................................ ................................ ................................ ..................... 8 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 10 2 LITERATURE REVIEW ................................ ................................ .......................... 1 3 3 METHOD S ................................ ................................ ................................ .............. 21 TCM Estimation ................................ ................................ ................................ ...... 21 Empirical Models ................................ ................................ ................................ .... 26 4 RESULTS ................................ ................................ ................................ ............... 32 Survey Results ................................ ................................ ................................ ........ 32 Visitation Characteristics ................................ ................................ .................. 32 Expenditure ................................ ................................ ................................ ...... 33 Spring Activities ................................ ................................ ................................ 33 Travel Dis tance ................................ ................................ ................................ 33 ................................ ............................. 34 Park Pass and Access Fee ................................ ................................ .............. 35 Demographics ................................ ................................ ................................ .. 35 Estimation Resul ts ................................ ................................ ................................ .. 35 5 CONCLUSION ................................ ................................ ................................ ........ 58 APPENDIX A SURVEY ................................ ................................ ................................ ................. 61 B TABLE OF ESTIMATION RESULTS ................................ ................................ ...... 65 REFERENCES ................................ ................................ ................................ .............. 68 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 71
6 LIST OF TABLES Table page 4 1 Showing the total number of observations collected ................................ ........... 42 4 2 After cleaning the data, the total number of observations remaining .................. 43 4 3 Summary statistics of variables from the survey ................................ ................. 44 4 4 List of variables defined ................................ ................................ ...................... 45 4 5 Distribution of recreational activities of respondents visiting the spring park in North Cen tral Florida State Parks ................................ ................................ ....... 46 4 6 Calculated driving distances to four North Central Florida Springs state parks .. 47 4 7 Average level of rating on statements about the four spring park on a scale of one to five. ................................ ................................ ................................ ...... 47 4 8 Demographics of questionnaire respondents versus Florida population ............ 48 4 9 Travel Cost Model output for current visits using the negative binomial model .. 50 4 10 Annual visitation for 2016 and total consumer surplus of Springs; Florida Department of Environmental Protection provided annual visits ........................ 51 4 11 Sensitivity analysis of wage fraction rate using negative binomial model ........... 52 4 12 The adjust total consumer surplus with 1/3, 1/2, 2/3 fraction wage rate ............. 53 4 13 Perception versus physical environmental quality measures using ordered logistic regression model ................................ ................................ .................... 54 4 14 Marginal effects of the independent variables concerning the dependent variable clarity ................................ ................................ ................................ ..... 55 4 15 Predicted Probabilities of Above Average Water Clarity Perception with Nitrogen Mon thly increasing in mg/L ................................ ................................ .. 56
7 LIST OF FIGURES Figure page 2 1 Illustration of a single site model ................................ ................................ ........ 20 3 1 A graphical illustration of the recreational demand function showing consumer surplus and travel costs ................................ ................................ ..... 31 4 1 The rating of site characteristics with one for below average to five with above average ................................ ................................ ................................ .... 46 4 2 Education attainment reported by respondents ................................ .................. 48 4 3 Employment status reported by respondents ................................ ..................... 49 4 4 G raphical illustration of predicted probabilities alo ng the confidence interval ..... 57
8 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science USING THE TRAVEL COST METHOD TO ESTIMATE FRESH WATER BASED RECREATION IN NORTH CENTRAL FLORIDA By Bryan Huy Nguyen December 2017 Chair: Xiang Bi Major: Food and Resource Economics This thesis e xamine s recreational benefits provided by four spring sites located in North Central Florida using the travel cost method (TCM). The first part of the study estimates the travel demand for springs using data collected from on site intercept s urveys from four springs in North and Central Florid a. The second part derives the consumer surplus (CS) that represents the benef its from visiting the springs. A sensitivity analysis of the calculated travel cost was conducted to examine the robustness of the estimated CS Lastly, an ordered logit model is estimated to examine whether the environmental quality measures can explain an individual's perception on water clarity in the springs. The results of the TCM estimation are consistent with previous ly published CS estimates and they are on the high end of published recreational value estimates The CS of the four parks is valued at $144,497,642 with an average trip valued at $177.49 per person per trip, with a 95% conf idence interval (CI) of $141.78 to $ 234.04. T he sensitivity analysis of TCM yiel ds higher projec ted CS values of the springs versus the standard one third wage rate The ordered logit regression also showed a correlation between an lity mea sures using the
9 ordered logit regression estimation output shows there is a link. The individual perception of water clarity is negatively correlated with the concentration of nitro gen reported during the month in which the survey was conducted Thus, restoration efforts to reduce nitrogen concentration in the spring would water clarity of the springs, and increase recreational demand f o r the spring sites The results imply that the antial value on the spr ings in North Central Florida. TCM is one of the several ways to estimate the value of the springs. The results of this study will help inform decision makers regarding policies to protect springs from further degradation While t he study provides estimates of the visitors placed on the springs, decision makers will need to use the information to implement new policies at the s tate and c ounty levels to preser ve the springs for future generations.
10 CHAPTER 1 INTRODUCTION Outdoor recreation has been an important part of the American culture for decades. A ccording to National Survey on Recreation and the Environment (NSRE ) data showed a n increase of 7.5% from 208 million participants to over 224 million participants per year in general outdoor recreation between 2000 and 2009. N ature based recreation 1 increased by 7.1% from 2000 to 2009 from 196 million participants to over 210 million participants Nature base d recreation provided approximately $2.8 billion annually to the State of Florida for the fiscal year 2016 (Friends of Florida State Parks, 2016). The Florida s prings located in North Central Florida are part of a unique system within the riverine ecosyst ems attracting thousands of visitors worldwide. The springs are under significant threats from urban expansion and population growth. Post World War II Flori da experienced tremendous economic growth leading to significant urban dev elopment Due to this g rowth groundwater pumping has increased Over pumping of groundwater and real estate development has resulted in the disappearance of springs in South Florida according to an article by Pittman (2012) The adverse weather patterns further exacerbate the condition, as rainfall has not recharged the water tables due to dr o ught since the 2000s To conserve the unique and fragile system of springs, the State of Florida has invested approximately $15 million sinc e the 2000s to restore protect and educate the 1 Nature based recreation is defined as travelling with the purpose of enjoying natural attractions and engaging in a variety of outdoor activities. Examples consist of fishing, hiking, bird watching and kayaking.
11 public about the importance of these springs. Given the high investment level t here is a need to estimate the value of these sites to assist with future policy development and implications ( Florida Department of Env iron mental P r otection 2011). This study is to provide an economic valuation to assist policymakers a guide to create a plan on preserving or restoring the spring systems across the state of Florida Once an ecosystem has deteriorated beyond its restoration capabili ties the use of restoration techniques may not be able to revive it The economic valuation can provide policymakers with information to conduct a cost benefit analysis for future projects to prevent degradation of the spring s and to improve spring sites to provide value added service (i.e. tourism) The purpose of this study is to estimate a value of freshwater based recreation on spring site s This study use s a non market v aluation technique the travel cost method, to estimate the value of freshwater recreation on four spring site in North Central Florida To achieve the objectives data collected from an on site survey of four spring parks in North Central Florida were used Three of the four sites are state parks, with one site being privately owne d The surveys were collected from May 2016 to August 2016 during the peak season fo r visitation to the springs Additionally, secondary data on observed water quality information was collected from the Suwanee River Water Management District (SRWMD). Th e results of the estimation were found to be in line with previous TCM publication s and the calculated CS to be above the range of other studies The CS per person per trip is valued at $1 77.49. Higher valuation may be due to the uniqueness of
12 the springs to the region with pristine condition s. The n, a sensitivity analysis of the travel cost was conducted; varying the assumed opportunity cost of travel time with implicit wage rate s. The sensitivity analysis shows the CS has a range of $198.38 to $ 218.59 per person per trip when using the wage fraction ratio of one half and two thirds. Finally, an ordered logit estimation was performed to estimate the correlation between perc eived water quality and environmental water quality measures. Since the perceived clarity was rated using a Likert scale from below average to above average an ordered logit model can examine the relationship between water quality measures and specific perceived clarity rating. The ordered logit estimation shows that individ ual perception on water cl arity to be above average has a negative correlation with the nitrogen concentration in the spring.
13 CHAPTER 2 LITERATURE REVIEW The TCM is one of the revealed preference approach es to calculate the economic value of nature based recreation. Using the opportunity cost of travel time as an implicit price for the trip demand functions are estimated relating trip costs and trip frequency (Whitehead, Haab, & Huang, 2000). The TCM can be used to assess the value of a recreation site which is the consumer surplus, measured by the area under the demand function and above the i mplicit price (Freeman, 1993). This section will discuss the theoretical background of recreation demand modeling and the basi s of TCM. The basis for TCM is: as travel cost increases, individuals will make fewer trips to a site com pared to those who reside near the recreation site. The combinations of both travel cost and trips made can be used to creat e a site specific demand function utility function will depend on the number of trips made to a recreation site represented as R and a bundle of other goods, represented by b The price for good R will be referred to the travel cost represented as ; and the price for the bundle d go ods b is represented as The represents the round trip distance j can be represented by: (2 1) W here represents the income level of the individual. The individual will want to maximize their utility function represented by: (2 2)
14 The individual will choose the op timal bundle representing the highest possible level of utility given the income constraint This will lead to the point of tangency between the indifference curve and budget constraint if there is an interior solution. An interior solution or corner solution forms the basis of the Marshallian demand function for recreational demand with the following: (2 3) R represents the numb er of trips undertaken by the individual, represents the total travel cost as a function of trips, income represents the household income, and z represents the vector of demographic variables impacting the frequency of trips. The function also i ncludes access to other sites leading to which represents substitute sites visitors may visit instead of the original intended site. This model is referred to as a single site model to value the site (Parsons, 2003). Including the income, demogra ph ic variables, alternative sites and trip cost can improve estimation of the actual trip frequency in the demand function. I n theory, the demand function (2 3) should show a negative relationship between the quantit ies of trips and price, thus represen ting the travel cost. This implies individuals residing closer to the site will visit the site more frequently I ndividuals living further away will incur higher travel costs leading to fewe r visits. Figure 2 1 illustrates equation (2 3) gr aphically where an individual with a travel cost of will take trips represented as Region A represents the CS for trips taken to the site. Region B is the total trip cost the difference between Region A and Region B is defined as the ind intended site for recreation is closed, Region A cannot be included in the CS estimation.
15 Estimating trips to the alternative site is similar to the estimation of travel cost for the stu dy sites. The choke price ( site will be zero for the model. The CS is represented by the following equation: (2 4) The next discussion will cover TCM estimation with an environmental quality measure to study the quality changes of each site. To reflect the differentiated level of demand due to the quality of the recreati on site, TCM often includes at least one site quality variable The site quality characteristic (e.g., pollution levels) works as a shi fter of the demand curve ; this causes the CS area to widen or shrink. However, the benefits of qual ity improvement have been shown to be hard to measure with the traditional TCM because of problems in identifying the change in recreation al demands from the quality change (Whitehead et al., 2000). The recommendation is to pool data from all the recreation sites with differ ent quality levels and estimate the effect of the quality variation on the number of trips taken (Smith & Desvousges, 1985 ). There are two limitations associated with this approach. First, there may be little variations in the qual ity rating of each rec reational site and the difference observed may only be marginal. For example, if two sites being studied use the same environmental measures t he environmental measures may have similar variances and mean s leading to no differences in the study site. S econd the TCM approach using en vironmental quality measures assumes an are influenced by a scientific collection of envi ronmental measures (total nitroge n, dissolved oxygen, phosphorus and other environmental measures). The use of environmental quality measures may introduce unintended bias to the individual
16 deciding on the environmental measures generally will not vary at the same recreation al site, lead i ng to the difficulty in valuing the environmental quality The concurrent agreement on the specific quality level cannot be explained At each of the recreation site, pollution levels may me et or exceed federal guidelines. S ome visitors may not be able to determine the level of degradation from personal experience. Previous studies have used chemical and physical measures to represent the environmental quality measures studying water quality ( Whitehead et al., 2000 ; Jeon et al., 2005; Smith & Desvousges, 1986; Dumas et al., 2005). The c hemical measures can include examining the concentration of such pollutants within the water source as total nitrogen, phosphorus and metals (Cordy, 2001). In turn, the second aspect of env ironmental measures include s physical measures can include using a S ecchi disk (measures the dep th of clarity within the water), pH leve ls (characterizing the acidity) and turbidity (which measures the cloudiness based on particles in the water) (Cordy, 20 01). One question that continues to puzzle many non market valuation researchers is whether the decision to visit a recreation site is based on the environmental quality measures or the individual perception from previous experiences (Jeon et al., 2005). Phaneuf, Herriges, and Kling (2000) hypothesized a Kuhn Tucker model to analyze the behaviors of anglers visiting the Great Lakes Kuhn Tucker model can track catch rates of several popular fish species, and the average toxin level observed. The researchers found that some toxin levels do impact a user decision to visit the Great Lakes. Specifically,
17 on which site to visit since pollution level affects catch rate and visitors will repeat visits to a site where they are satisfied with the catch rate Egan et al. (2003) researched the demand of visitors recreating the 129 l akes in Iowa. They included 11 environment al quality measures ( total nitrogen, phosphorus, chlorophyll a dissolved oxygen, pH, Secchi disk, etc. ), and included site characteristic (facilities state parks, boating ramp, etc. ). They show ed that environmental quality measures do impact the decision which lake to recreate. However, E determine if there were any links between the environmental quality measures and individual perception. The perception s were not investigated fully due to the prohibitive cost of conducting lengthy surveys (Egan et al. 2003). Neverthel ess, there is one study carried out by Adamowicz et al. (1997) that analyzed individual (objective measures) The study focused on moose hunting measures and used a discrete choice method. The y showed that models usi ng perception variables perform slightly better than models with physical attributes (access fees, environmental quality, etc. ) related to the hunting sites The study found models using the perception variable have lower CS estimates compared to the model using physical quality attributes However, t here are issues with the estimation when calculating the CS with using the perception variable. The weaknesses of the model are (1) determining the base level of users visiting the site versu s the base level set by a regulatory or management agency and (2) measuring the recreation experience of users who are not in the sample of the study (A different than when measuring the objective measures. With objective measures of
18 improving a site quality, the difference between the current quality levels and improved the same cannot be applied to individuals will have different base levels compared to lack of difference in perceptions of the site quality make s it difficu lt to estimate the CS accurately Another study by Jeon et al. (20 05) utilizes the recreation data for lakes in Iowa and the perception data from Egan et al. (2005). The study used a mixed logi t model to determine if: (1) individuals care about the environmental measures and (2) if the perception regarding water quality at the lake influences individual household behavior. They showed that individuals could be influenced to visit a rec reational site based on subjective high quality (subjective) assessment do es no For example, four lakes in Iowa are listed as impair ed by pollution, but received high er rating on their subjective assessments Another important issue in TCM is how to incorporate opportunity cost of lesiure time as a component of travel cost ( Cesario & Knetsch, 1970). Some previous research has demonstrated the CS estimates of recreation al demand models are affected by the cost of time (Bowker et al., 1996). One assumption of the models for recreation al demand is that individuals can freely adjust their hours worked. However, the work hours of the individuals are not flexible but fixed hours per week and this assumption are forced into the model estimation. The typical workweek of an indiv idual consist s of forty hours per week; thus, the fixed schedule is not likely to be optimal (Feather & Shaw, 1999). The total hours
19 an individual will work per year is between 2000 to 2080 hours. Most travel cost estimation uses fixed work hours; howeve r, literature does not provide for why this is so (Blaine et al., 2014; Egan et al., 2013; Parsons 2003 ). This unknown causes individuals to be modeled as under and over employed, leading to an inaccurate wage rate and estimation of the opportunity cost o f time for recreation. However, the method introduced by Cesario (1976) uses the percentage of the wage rate for recreation al demand, and the method is still being used to date because it has been proven to be consistent with many empirical publication s Additionally, the following articles recommend the use of the wage fraction rate to estimate the cost of time (Larson, 1993; McConnell, 1992; Parsons 2003), and conduct robustness checks using different levels of fractions. This thesis contributes to the literature in the following ways. First, i t examines the economic benefits provided by a unique freshwater system of springs in Florida. Secondly are examined to determine which variables are more likely to a ssess spring recreation accurately Given the fact that many visitors have perceived the spring s as a pristine natural system, it is essential to understand how perceived water quality is correlated with the environmental qua lity measures that can be monitored before and after conser vation efforts. Lastly, a s ensitivity analysis is conducted on the travel cost variable, which has not been previously conducted on spring recreation.
20 Figure 2 1 Illustration of a single site model (Parsons, 2003)
21 CHAPTER 3 METHODS TCM Estimation The following illustrates the estimation of the TCM. Given the integer nature of the reported recreation frequency from data collection, the Poisson mod el can be applied to count data estimations, as the estimations are a discrete probability density function and non negative. The Poisson function below will be used to estimate the count data (Parsons, 2003). (3 1 ) Where the parameter represents the value of the mean and variance of the random variable, R implies trip frequency, and all values must b e positive for this function. When the mean and variance a re not equal, there will be an error in the model; thus, correction steps are discussed below. will always be greater than zero leading to an exponential function: (3 2 ) rep resents the vector of variables influencing demand, and represents the parameters used to estimate TCM. The estimation of this single site model results in the following equation: (3 3 ) The coefficients are estimated for the single site model w h ere the number of trips is the dependent variable. The independent variables used to estimate the model are travel cost to the site, travel cost to an alternative s ite income and some social demographic variable. This model uses similar variables discussed in the recreational demand function (1 3).
22 The parameters within this function can be estimated using maximum likelihood. The following e quation is the likelihood function (Haab & McConnell, 2002): (3 4 ) and the log likelihood function is: ( 3 5 ) These equations are concave using the listed parameters, where the maximum likelihood estimation (MLE) will converge. The conditional mean of the Poisson model is using the following equation for expected trips: (3 6 ) The following is the estimated model of the Poisson Model using the previously discussed parameters: ) (3 7 ) The beta coefficients will be the coefficients estimated in the model. The Poisson model does have limitations re garding (1) t conditional mean and variance are equal because of restrictions in the estimation parameters, the variance does not deviate from the mean, and (2) when the variance is more significant than the mean overdispersion is present in the data. Cameron and Trivedi (1998) suggest that the Poisson regression model is usually too restrictive for count data and the presence of overdispersion may be a result of the failure of an assumption of independence of events which is im plicit in the Poisson process. This is due to an unobserved heterogeneity, generated by the process ; the rate parameter is una va ila ble to be specif ied (random variable).
23 The remedial process for overdispersion count data involves relaxing the conditiona l mean and variance constraint, using the negative binomial model (NB). An additional parameter is added to the model for controlling overdispersion The reduction in overdispersion will also result in a narrower confidence interval versus the Poisson re gression. The NB utilizes the same conditional mean from the Poisson model. There are several versions of NB, but for this study, the following will be adapted using the conditional mean of the Poisson model and the unobserved error: (3 8 ) represents the unobserved hete rogeneity. The equation from (3 8 ) will provide the random variation needed across all observation. By substituting the equation from the right hand side of the equation into the probability for the Poisson model, the number of trips based on the condition of (3 9 ) When is the gamma normalized distribution, with the density for is defined as This is the unconditional probability function for trips, by the process of integration of giv es the following: (3 10 ) denotes the gamma integral, the basis to a factorial for an integer argument. The parameter represents overdispersion in the model. Special cases of the NB include the Poisson when which implies there are no occurrences of over dispersion,
24 thus, reducing the equation to the Poisson model. When there is an indication of overdispersion and represents underdispersion. The values of an NB must be non negative similar with the Poisson model stated above for the equation to hold. The correction process will reduce, but do not eliminate issues relating to overdispersion The estimation process of NB model is similar to th e process of the Poisson model when using the equation from (3 7 ) and interpretations of the coefficients. In addition to the issue of overdispersion w hen conducting an on site data collection, endogenous stratification can occur leading to frequent us ers of the site being oversampled, resulting in the bias of the estimates (Shaw, 1988). The method for correcting this occurrence is by removing one trip count, R 1 from the actual reported trips. The estimated travel cost coefficient from the Poisson and NB models can be used to calculate the CS of recreational use rs traveling to the spring parks The following equation is to calculate the CS for per household visit to the spring site: (3 11 ) (3 12 ) represents the total C S for the site, and is the coefficient for travel cost Equation (3 11) is used when travel cost variable is linear in the TCM and equation (3 12) should be used when travel cost variable is in natural log in the TCM. For this study, the estimated CS will be divided by the number of adults within the group d uring the visit. The second equation represents the calculation for CS when the variable is a
25 natural log of the travel cost. The mean of the travel cost is multiplied by the coefficient. To calculate the total C S of the spring park the following equat ion can be used: (3 13 ) TCS represents the total CS of the site and is calculated by multiplying the estimates by the annually reported number of vis h spring park Figure 3 1 represents cost of TCM, and average trips per season graphically Lastly, when calculating the CS, the confidence intervals can be shown (3 14 ) Figure 3 2 represents cost of TCM, and average trips per season graphically Lastly, t he CI (confidence interval) of the travel cost estimated using a bootstrap simulated 5,000 times to obtain the bias corrected CI (Cameron & Trivedi, 2010) will be provided next to estimated value of CS for each spring site. One issue impacting the travel cos t analysis is the use of different accounting approaches for the cost of operating a vehicle. Most papers suggest the use of the American Automobile Association (AAA ), there is a range of costs for operating a vehicle (AAA, 2016). For example, a small sedan has an average operating cost of 43.9 cents per mile based on 15,000 miles driven per year, and a sports utility vehicle (SUV) has a cost of 68.4 cents per mile using the same miles driven per year. Due to constraints of the on site questionnaire, the actual costs were not captured to estimate
26 the CS accurately A skewed CS may result in an underestimation of the actual cost and benefit analysis. Empirical M od els This portion of the thesis will di scuss the empirical models that use the collected data to estimate the TCM whil e maintaining the integrity of the objectives discussed above. represents the trip frequency of the on site intercept including t he previous twelve month visit frequency, and is the dependent variable. The reported number of trips was also corrected for endogenous strat ification using the mentioned (R 1) method. + (3 15 ) Equation (3 15 ) are estimated with three variation s of variables. The first mo del uses the original discrete rating of variables for perceived water clarity, condition of facility, and condition of greenspace; the second consist uses dummy variables of the se perceived site characte ristic in lieu of the discrete rating variables; and the third model adds the environmental quality measures in lieu of the perception dummy variable. The first variable ln travel cost is the natural log of the travel cost variable. Due to the estimated cost being skewed above the mean travel cost, the use of natural log form of the travel cos t to reduce the skewness The travel cost is estimated by using the following equation: (3 16 )
27 represents the average cost of operating a motor ve hicle traveling to the recreation site. This cost variable was chosen from the American Automobile Association (AAA) cost of operating a vehicle at 58 cents per mile based on a yearly mileage rate of 15,000 miles (AAA, 2016). The variable , is th e round trip distance from respondent home to the recreation site j home zip code to the recreation site using a Google Maps application program interface (API) in R The wage fraction rate is represented by with the st andard rate of one third of the annual income (Cesario, 1976; Englin & Shonkwiler, 1995; Parsons, 2003). The wage per year is represented by the response to household income from the on site ques tionnaire. The denominator portion of the equation is divided by the total hours worked per year and varies between 2000 to 2080 per year (Blaine et al., 2014; Parsons, 2003; Bin et al., 2005). For this calculation, 2000 hours per year was used, factorin g in two weeks of vacation time annually. The variable Income is the household income reported by the respondent on the questionnaire. The household income re ported consisted of seven categories and w as reduced to three: 1) lower income ($35,000 and be low), 2) middle income ($35,000 to $89,999), and 3) high income ($90,000 or high er ). The study expects that the effect of this variable on the number of trips to be non linear and converted into a dummy variable A low income respondent may be able to vi sit the springs more often as the opportunity cost of time to work can be easily substituted. The opposite is observed, for a high income respo ndent who can easily afford to visit the springs. As a result of the entrance fee being affordable, the referen ce group will be high income respondents. For respondents who did not disclose their household income during the on site survey,
28 their income was determined by the median reported household income from their home zip code. The characteristic site quest ion asked respondents to rate the respective spring site facilities, and 3) surrounding greenspace area. The rating scale consists of one (1) representing below average to five (5) being above average. The site characteristic variables were estimated with several approaches to determine the robust choice. The average ratings overall for the site characteristic was around 4.37 or higher. As a discrete variable, it was transformed into categorical to determine the effects of the rating. The three categories consist of: below average, average and above average. If the rating of the characteristic was three or less it was classified as below average, a ra ting of four received an average rating, and five received an above average. A summary statistics of the site characteristic variable showed more than 50% of the rating clustering around average and above average rating. Site characteristics can be chan ged into a dummy variable due to the clustering near the upper range of the variable to conserve the degrees of freedom. The dummy variables of the perceived site characteristic are coded as one (1) if the rating of the site received a rating of four (4) or five (5), a rating of three and below is coded as zero (0). The numbers four (4) and five (5) were used as the dummy coding because one (1) is the average response rate for the site characteristics which had a range of 4.37 to 4.71. Also the stud y est imated ratings in discrete form T hus the results do not statistically differ from using the dummy variables for the site characteristics The results were not ideal and the original ratings from the survey was used.
29 The environmenta l quality measur es in the model are nitrogen monthly measured during the season s (May through August) and Secchi disk reading reported from each spring park The SRWMD reports These measures The nitrogen variables are measured in mg/L of water collected. The variable nitrogen monthly is the actual reading of total nitrogen during the survey intercept and collected remotely from each spring. The variable Secchi disk represents the amount of light penetrating the water body with the higher recorded depth implying a significant amount of water clarity. The next model will be used to conduct a sensitivity analysis of TCM by adjusting the fraction of the wage rate. The model is like Equation 3 15 with only the natural log of the travel cost variable adjusted to two levels of a fraction of the wage rate. The two wage rates level being used are one half and two third examining the range of the recreational value of CS. environm ental quality measure variable. The ordered logistic model is used to determine if there is a correlation between perception and environmental quality measure. The following regression model for estimation is: (3 17 ) The dependent variable clarity represents the site perception portion of the questionnaire. As stated above, the variables are kept as discrete rating for regression purpose. The independent variables used for the model consist of an environmental quality measure of nitrogen monthly current trips first visits (first ever trip), alternative
3 0 sites swimming and the social demographic variable income The variable current trips represent the total visits made to the spring site in the past twelve months. The following variable fi rst visit equals 1 if respondents made their first ever trip to the spring site, and alternative sites consist of nearby sites a respondent would visit in the case of closure or congestion. The variable first visit could have an impact on perception as users will have a different level of experience with nature based recreation. The variable swimming represents activities participated at the spring site involving interaction within the water and are listed as dummy variables with one (1) implying the activity is present and zero (0) implying the activity is not. Swimming activities consist of interaction with the water like swimming and tubing. There are t wo hypotheses for this thesis using the TCM valuation. The first null hypo thesis assumes the TCM variable in the model will follow a similar pattern with previously estimated TCM applications (Blaine et al., 2014; Parsons, 2003; Bin et al., 2005) The independent variables ln travel cost shou ld be statistically significant and have a negative coef ficient. The second null hypothesis assumes there is a negative link between the perceived clarity rating and environmental quality measures. There is no literature to support this hypothesis as previous attempts have been difficult to prove. The intuit ion supporting the second null hypothesis is that as nitrogen levels decrease, the perceived clarity rating of each spring site should increase. The level of nitrogen measured can only fall to a certain threshold (not equal to zero ) because there is a nee d for some level of nitrogen to sustain aquatic life in the spring ecosystem.
31 Figure 3 1 A graphical illustration of the recreational demand function showing consumer surplus and travel costs (Sohngen et al., 1999)
32 CHAPTER 4 RESULTS Survey Results A total of 494 surveys were collected from the four spring parks from May 2016 to August 2016. T his period is considered peak recreation season. The survey team randomized data collection between weekdays and weekends (three weekd ays and four weekend days). Table 4 1 shows each spring site and the total amount of surveys collected on site. Blue Springs in Madison County had fewer responses as a result of the du ring Hurricane Hermine. The data was cleane d by removing any discrepancies such as zip code errors, invalid responses and questions left blank. The final sample si ze was 468 respondents. Table 4 2 represents the total number of observation left a fter cleaning the data. Table 4 3 shows the summar y statistics of some of the responses that are used in this study and table 4 4 are the variables defined Visitation Characteristics Among the respondents from the on site survey, 300 out of 468 respondents (64.1%) indicated th e trip was a singl e purpose day trip. The response rate collected on the weekend showed 333 out of 468 respondents (71.1%) versus 28.8% of visits on the weekday. Out of the 468 respondents, 82 respondents (17.5%) stayed overnight, with an average of 3.2 nights at the park site camping area. A total of 419 out of 468 respondents (89.5%) chose the springs and its surrounding area as the primary reason for the trip. The average group size to the spring site was five (5), consisting of three (3) adults and two (2) child ren.
33 Expenditure The mean expenditure reported when visiting the springs was $113.52 per group, and the mean expendit ure per adult was $37.84. Note: was divided by the number of adults reported for the group. The mean was tak en from this calculation. Spring Activities At the four spring parks, when asked about the recreational activities the top responses were swimming (63.8%), tubing (30.5%), and picnicking (26.7%). The other mentioned activities were natur e viewing (9.6%), hiking (8.5%) and kayaking ( 4%). Table 4 5 summarizes the recreational activities during a typical spring site visit. Trip Frequencies Two hundred and ninety four (294) out of 468 respondents (62.8%) stated they sit. These respondents made an average o f four trips to the spring parks in the past twelve months including the reco rded date (day of questionnaire interception). Whe n the respondents were asked on future expected visits to the springs, the responses averaged fo urteen trips to the spring parks in the next twelve months. Travel Distance The travel distances reported b y the respondents reported zip codes had an average of 123.29 miles, with a standard dev iation of 293.83 miles. Table 4 6 When calculating the travel distance, l ocals (defined by zip code) who visited the springs were assumed to live within five miles with a travel time of approximately ten
34 minutes. Given the significant variation in distance traveled approximately 227 out of 468 respondents (48.5%) drove more t han fifty miles to the respective spring site. When the respondents were asked if the original intended site became unavailable, would the y visit a nearby similar site, 365 out of 468 respondents (78%) ded, but calculated in relationship to the alternative site, the average distance calculation was 40.07 miles with a standard deviation of 59.33 miles. Figure 4 1 summarizes the average rating respondents provided with regards to the perception of water quality, conditions of facilit ies and conditions of the green space around the springs. Ratings were based on a scale of one (1) to five (5), where one (1) is below average, and five (5) is above average The averag e response rating was cluster ed between four and five for water clarity facilities and green space Table 4 7 with the following statements, where (1) strongly disagrees, and (5) strongly agrees The questions covered are : 1) If the water in the spring has become clear er; 2) w ildlife in the water has increased; 3) w ildlife in the surrounding green space has increased; 4) w ater flow in th e spring has increased; and 5) w ater level in the spring has increased. Respondents only answered the next question if they had visited the springs before. Table 4 7 represents the summarized the level of agreement to the following statements, where (1) strongly disagrees and (5) strongly agree s The responses for this category received an average response rating of neutral for the level of agreements.
35 Park Pass and Access Fee Three hundred forty three (343) out of 468 respondents (73.2%) reported to not ha ve an annual park pass for the s tate of Florida. 63.4% of respondents stated fewer trips would be made to the spring site if access fees increased. About thirty four percent (34%) respondents reported their trips to the spring parks would remain unchanged if the access fee increased. Demographics Table 4 8 summarizes the demographics of respondents versus the population of Florida based on census data. The respondents in the sample had a similar age and gender composition as the general population but were more likely to be college educated with higher household income levels then what was described by the census data. Figure 4 2 shows the distribution of t he education level, and Figure 4 3 shows the employment status of respondents. Estimation R esults The summary statistics f rom the on site survey showed overdispersion is present in the independent variable s Due to overdispersion the negative binomial model is used instead of the Poisson model. Furthermore, a simple ratio of the me an to variance is 552 Table 4 3 impli es it would not correct ly converge using the Poisson m odel. A concern with the count data model is an excessive number of zeros present in the data for current trips; however, this is not an issue due to the survey being on site. The second concern is the o n site survey may result in endogenous stratification where frequent visitors may have been intercepted on the site which has also corrected Robust standard errors were used due to outliers skewing the data. Also some
36 observations were removed due to incorrect zip codes, respondents who came with large groups, those of 30 individuals or more, and one respondent lived outside the U.S. The travel cost estimation using the Equation 3 19 resulted in a mean travel cost of $203.40 per household per trip. This estimation will be used in conjunction with the estimated beta coefficient from the regression output to estimate the CS for the spring site. The empirical Equation 3 15 was estimated into three different equations. Model 1 uses the or igi nal Equation 3 15 which shows the variable ln travel cost and clarity as statistically significant. Model 2 use Equation 3 15 but with the dummy variable of the perceived site characteristic s, which shows that ln travel cost and clarity as statistically significant. The dummy variable clarity is negative indicating springs with above average water quality received fewer visitations on average than springs with water quality rated less than average after controlling for travel costs and other factors influencing trip frequency. It is likely that respondents may be more familiar with springs with average water quality and visit them more often. The dummy variables for facilities and greenspace are not statistically significant and having above average facilities and greenspace do not statistically increase trip frequency. Model 3 uses Equation 3 15 and adds the environmental quality measure variable nitrogen monthly which is not statistically sig nificant, but the variable s ln travel cost and clarity are still statistically significant. The environmental quality measures variables are not statistically significant for Model 3. The variable nitrogen monthly (level of nitrogen present during the survey month) and secchi disk are not correlated with the number of trips. This is expected, since respondents do not observe physical
37 water quality measures directly. Instead they form their own perception on water quality based on visual cues, and the ir perceived water quality is correlated with the trip frequency, as shown in Model 1 and 2. Model 1 in Table 4 9 is chosen to calculate the CS due to lower Akaike info rmation criterion (AIC). Additionally, mult iple variations of the TCM model were estim ated and are listed in Appendix B and they all have higher AIC than Model 1 thus are not used to calculate the CS The following discussion will cover Model 1 which is used to estimate the model about the study. Results of Model 1 indicate t he beta coeff icient for ln travel cost is negative and is statistically significant at p < 0.01, showing the quantit ies of trips demanded will decrease as travel cost increases This results in a downward demand function slope. The results are consistent with the suggested literature showing downward demand and negative slope (Blaine et al., 2014; Parsons, 2003; Shrestha, Stein & Clark, 2006). The beta coefficient for in the model is not statistically significant. The after controlling for travel costs and other factors influencing trip frequency, it is not statistically significant Previous studies on the effe ct of income yield little results explaining the estimation of TCM (Loomis & Ng, 2010; Nicholson & Snyder, 2012), but a study by Phaneuf & Smith (2005) makes the case for including income in the model as an important requirement. Results of Model 1 shows that t he beta coefficients for , and represent the perceived rating of the site ch aracteristic s while the only statistically significant variable is clarity at p < 0.01 level. Table 4 10 sh ows the estimated non market value of nature based visits to the springs in North Central
38 Florida using Model 1 in Table 4 9 Using the coefficient for ln trav el cost from Table 4 9 and Equation 3 12 to calculate the CS estimate, the following results are derived The values show a mean CS estimate of $177.49 per person per trip with a 95% confidence interval (CI) of $141.78 $234.04 (the total CS per person per trip was divided by the number of adults per group) According to Shrestha and Loomis (2003), who conducted a meta analysis study of recreation sites across thirty years, the national average CS is estimated to be $47.10 (adjustment for inflation $62.81) per day trip visit ing state parks. Comparing the estimated CS to other recreational ac tivities in the state based on previous TCM application, the estimates are relatively high. For example similar recreational activities in the Apalachicola River region have a value of $74.18 ( adjustment for inflation $87.21) per trip (Shrestha et al., 2 007) and Oklawaha River visit is $9.07 ( adjustment for inflation $14.57) per trip (Stratis & Bendle, 1995). Environmental sites untouched by development will tend to be highly valued and considered pristine (Shrestha et al., 2007) The spring parks can b e consid e red a natural location for recreation with no development nearby and (perceived) low pollution. The mean CS estimates from the survey conducted at t he spring parks can be used to estimate the total economic value of the resource. Each spring pa rk has a reported a nnual attendance rate reported in Table 4 11 for 2016. The total economic value of outdoor recreation visiting the springs in North Central Florida is $144.49 mi llion with a 95% CI of $115.61 to $190.83 million The survey had an equal response from each site, but Ichetucknee Springs State Park and Fanning Springs State Park ha d higher visitation rate s
39 The second part of the assessment is to conduct a sensitivity analysis of the different levels of the wage fraction rate. The standa rd wage rate used in the travel cost estimate is one third based on literature ( Cesario, 1976; Englin & Shonkwiler, 1995; T he decision to participate in recreational activities is likely to be as heav ily influenced by time constraints as by money constraints. An employed individual may have fixed work hours of 40 hours a week or 2000 hours per year, or some flexibility in hours. An issue with the actual one third wage rate may be the rate does not a pply to respondents from the on site questionnaire. More than sixty percent of the respondent stated they were employed full time implying a job with fixed work hours. Also most of the trips were completed on the weekend when those fully employed indivi duals may not be giving up opportunity cost of time for work in place of leisure time. A sensitivity analysis of the different levels of the wage f raction rate using Equation 3 15 will be analyzed further. There are three versions of this equation. The f irst version has already been estimated using the standard one third wage fraction level. The second and third version uses one half and two thirds of the wage fraction level. The estimated CS of recreational value of various wage fraction levels can be seen in Table 4 12. The estimated total recrea tional value of the spring parks using one half and two thirds of the wage fraction rate are $161,758,906 and The wage rates used are all statistically significant at p < 0.01. The coefficient s estimated are only slight ly different from the standard one third wage fraction rate. However, the calculated recreational value of the springs is much higher than the original one third wage rate.
40 The last part of the assessment is to determine what factors determine the erceived water clarity with environment al quality measures using the ordered logistic regression model. Following Equation 3 2 0 Model 1 is used to estimate with the dependent variable of Clarity and its corresponding independent variables. The estimated coefficients of the independent variables are summarized Table 4 13. The variable first visit is significant at p with a log coefficient of .520 indicating first visit is positive ly correlated with the resp clarity for the spring. The second variable Swimming which implies respondents who have participated in water related activities only is also positive ly correlated with the perception of water quality. This variable is significant at p with a positive log coefficient of .419 The environmental quality measure of nitrogen monthly is significant at p with a negative log coefficient of 0.206 which implies that as the level of nitrogen in the water decreas es, the perceived clarity rating should increase. Since perceived water quality is positively correlated with the trip frequency, as shown in table 4 9, reducing nitrogen level in the water would be likely to increase the perceived clarity rating, which c an translate into an increased demand for visits to the spring site. None of the marginal effects evaluated at the sample mean are statistically significant at p with one exception (Table4 4 13, Column 2). The marginal effect of nitrogen month ly is statistically significant at p showing that a one unit change in nitrogen level is associated with a 0.31 % increase in the perceived clarity rating. T hese marginal effects are estimated at the sample mean across all categories of clarity rat ings thus are less informative
41 Table 4 14 further separates the marginal effects by clarity rating. Nitrogen monthly has a positive marginal effect on the perceived clarity rating of below average to above average The m arginal effect becomes negative for predicting the probability of above average water clarity, indicating as nitrogen level increases, the probability of the perceived water clarity being above average will reduce. A one unit increase in the nitrogen level will result in a decrease in t he probability by 3.54% of clarity rating of above average and is highly statistically significant at p<0.01 The marginal effect for the variable swimming is negative when the perceived wate r clarity is less than above average It also indicates the activity does have an impact on increasing the probability of the perceived water clarity being rated above average by 7.22% and statistically significant at p <0. 1 0 This also implies swimmers tend to rate water clarity higher T he similar effect is seen for the variable first visit and the probability of an above average perceived water clarity increases by 8.94% and statistically significant at p <0. 1 0 The statistical significance of the variable swimming and first visit show t he impact on the perceived clarity rating is marginal compared to Nitrogen monthly Table 4 15 shows the estimated output of nitrogen monthly as it increases by an increment of one mg/L ; the probability of an above average perceiv ed water clar ity rating is shown. As the level increase from 2 mg/L of nitrogen monthly measured to 8 mg/L the predicted probabilities show a decrease in the probability of an above average perceived clarity ranging from 79.2 % to 52.6 % For example, using the average le vel of nitrogen monthly across the four springs (2.25 mg/L ) the probability of the springs receiving an above average rating is 78.3%. If the level of nitrogen monthly were to increase by 1 mg/L from the average level, the probability would decrease to 74 .7%.
42 The prediction probabilities are statistically significant at p Figure 4 4 shows the graphical illustration of the predicted probabilities of above average perceived c larity rating base d on the incremental increase level of nitrogen monthly in mg/L. In sum, t he null hypotheses statements cannot be rejected based on the estimation results. T he travel cost estimation is statistically significant with a negative coe fficient Additionally, there is a negative correlation between the perceived clarity rating and environmental water quality measures. Table 4 1 Showing the total nu mber of observations collected Spring Site N Weekday Weekend Blue Springs 131 42 89 Fanning Springs State Park 127 40 87 Ichetucknee Springs State Park 126 38 88 Madison Blue Springs 110 19 91 Total 494 139 355
43 Table 4 2 After cleaning the data, the total number of observations remaining Spring Site N Weekday Weekend Blue Springs 119 34 85 Fanning Springs State Park 124 42 82 Ichetucknee Springs State Park 125 38 87 Madison Blue Springs 100 21 79 Total 468 135 333
44 Table 4 3 Summary statistics of variables from the survey VARIABLES N mean sd min max Cost (1/3 wage rate) 468 203.4 335.7 6.300 3,248 Cost (1/2 wage rate) 468 227.2 373.2 6.550 3,564 Cost (2/3 wage rate) 468 250.9 410.9 6.800 3,880 Ln Cost (1/3 wage rate) 468 4.662 1.15 1.840 8.085 Ln Cost (1/2 wage rate) 468 4.772 1.15 1.879 8.178 Ln Cost (2/3 wage rate) 468 4.871 1.15 1.916 8.263 Adults 459 3.510 2.503 1 17 Children 456 2.140 2.428 0 15 Expenditure 468 113.5 245.8 0 4,000 Swimming 468 0.709 0.455 0 1 Tubing 468 0.306 0.461 0 1 First visit 468 0.372 0.484 0 1 Current trips 468 4.002 9.915 1 100 Age 468 40.92 14.51 18 86 Household 467 2.176 0.888 1 8 Alternative visit 468 0.780 0.415 0 1 Secchi Disk 468 3.250 1.302 2.440 5.736 Nitrogen Monthly 468 2.247 1.633 0.645 4.863 Currents trips (corrected) 468 3.748 5.220 0 21 Clarity 468 4.680 0.692 1 5 Facilities 468 4.373 0.922 1 5 Greenspace 468 4.710 0.630 1 5 Household income 468 2.072 0.879 1 3 Male 468 0.449 0.498 0 1 College Educated 468 0.549 0.498 0 1
45 Table 4 4 List of variables defined Variable Definition C ost Travel cost using different wage fraction levels ( 1/3,1/2, 2/3 of wage rate ) in dollars Ln travel cost Log of travel cost using different wage fraction levels (1/3, 1/2, 2/3) Adults Number of a dults visiting the spring parks over the age of 18 Children Number of children Expenditure Reported trip expense from respondents Swimming Activity performed at the springs Tubing Activity performed at the springs First Visit 1 = yes 0 = no Current Trips Dependent variable; number of visits in the past twelve months including today Alternative visits If sites are unavailable, a nearby site will be used 1= yes 0=no Secchi disk A d evice used to measure water clarity assessing the amount of light to water depth ( ft. ) Nitrogen Monthly The amount of total monthly nitrogen in water (mg/L) Current trips1 (corrected) Correction for endogenous stratification due to on site questionnaire Clarity Rating of water clarity on a scale of 1 to 5 ; 1 is below average to 5 for above average Facilities Ratings of facilities offered on site on a scale of 1 to 5; 1 is below average to 5 for above average Greenspace Ratings of the surrounding green space on a scale of 1 to 5; 1 is below average to 5 for above average Male 1 =male 0=female Household Income Household income level (in thousand US dollars) College Educated 1= college educated or higher 0 = less than college
46 Figure 4 1 The rating of site characteristics with one for below average to five with above average Table 4 5 Distribution of recreational activities of respo ndents visiting the spring park s in North Central Florida State Parks (N=468) Activity Observation Standard Deviation Swimming 332 0.45 Tubing 143 0.46 Picnicking 125 0.44 Nature viewing 45 0.29 Hiking 40 0.28 Kayaking 30 0.24 Other 16 0.18 Camping 12 0.16 Scuba/Cave diving 7 0.12 Motorized boating 4 0.09 4.6 4.6 4.8 4.7 4.4 4.8 4.79 4.2 4.5 4.63 4.3 4.7 3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5 Water clarity Facility Green space Site Characteristic Rating Madison Blue Ichetucknee Fanning
47 Table 4 6 Calculated driving distances to four North Central Florida spring parks Springs Number of observations Mean Std. Dev. Min. Max. Fanning Springs 119 113.39 269.52 5 2434.91 Ichetucknee Springs 120 104.77 243.21 5 2419.76 Blue Springs (High Springs) 118 148.23 290.33 5 1853.16 Blue Springs (Madison County) 109 127.48 366.96 5 3711.82 Total 466 123.29 293.83 5 3711.82 Table 4 7 Average level of rating on statem ents about the four spring park s on a scale of one to five. Fanning Ichetucknee Blue (Gilchrist) Blue (Madison) The water in the spring has become clearer 3.3 3.6 3.7 3.5 Wildlife in the water has increased 3.1 3.5 3.0 3.0 Wildlife in the surrounding green space has increased 3.3 3.5 3.0 3.1 Water flow in the spring has increased 3.1 3.7 3.2 3.2 Water level in the spring has increased 2.8 3.5 2.7 3.1
48 Table 4 8 Demographics of questionnaire respondents versus Florida population Demographics Survey Respondents Florida Census Gender Female 55% 51.5% Male 45% 48.5% Household income (median) $60,000 $47,212 Education High school graduate or higher degree 95% 86.5% or higher 54 % 26.8% Household size 2.2 2.62 Percent in labor force (full time) 59% 59.2% Age (mean) 41 40 Figure 4 2 Education attainment reported by respondents 5% 27% 13% 39% 16% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 12th grade or less High school or GED College without degree College degree Grads or above Frequency Education Level Education
49 Figure 4 3 Employment status reported by respondents 59% 8% 6% 7% 11% 9% 0% 10% 20% 30% 40% 50% 60% 70% Full-time Part-time Self-employed Student Retired Unemployed Frequency Employment Employment Status
50 Table 4 9 Travel Cost Model output for current visits using the negative binomial model VARIABLES Model1 Model2 Model3 Ln travel cost 0.381*** 0.382*** 0.377*** (0.046) (0.045) (0.046) Income Below 0.204 0.213 0.188 (0.167) (0.169) (0.171) Income Average 0.026 0.043 0.047 (0.133) (0.135) (0.136) Clarity 0.203* 0.203* (0.112) (0.112) Facilities 0.084 0.084 (0.095) (0.095) Greenspace 0.057 0.057 (0.112) (0.112) Clarity d ummy 0.271* (0.147) Facilities d ummy 0.014 (0.137) Greenspace d ummy 0.078 (0.171) Secchi Disk 0.018 (0.051) Nitrogen monthly 0.033 (0.039) Constant 2.728*** 3.085*** 2.834*** (0.310) (0.287) (0.284) Ln alpha 0.200*** 0.196*** 0.186*** (0.067) (0.068) (0.066) Observations 468 468 468 AIC 2218 2220 222 2 BIC 2252 2253 224 7 Robust standard errors in parentheses *** p<0.01, ** p<0.05, p<0.1
51 Table 4 10 Annual visitation for 2016 and total consumer surplus of Springs; Florida Department of Environmental Protection provided annual visits Recreation Site Annual Visits (FDEP,2016) Recreation value 95% Lower bound CI 95% Upper bound CI Blue Springs 41,000 $7,288,411 $5,879,442 $9,592,133 Fanning Springs State Park 218,963 $38,924,204 $31,399,516 $51,227,369 Ichetucknee Springs State Park 507,238 $90,169,734 $72,738,444 $118,670,588 Madison Blue Springs 48,209 $8,569,927 $6,913,220 $9,592,133 Total 815,410 $144,952,276 $116,930,622 $190,768,799
52 Table 4 11 Sensitivity analysis of wage fraction rate using negative binomial model VARIABLES Model1 Model2 Model3 Ln Travel cost (1/3 wage rate) 0.3814*** (0.0457) Income Below 0.2044 0.1724 0.1464 (0.1674) (0.1691) (0.1705) Income Average 0.0262 0.0073 0.0076 (0.1328) (0.1334) (0.1339) Clarity 0.2030* 0.2034* 0.2037* (0.1125) (0.1127) (0.1129) Facilities 0.0839 0.0847 0.0854 (0.0947) (0.0946) (0.0945) Greenspace 0.0572 0.0568 0.0565 (0.1120) (0.1120) (0.1121) Ln Travel cost (1/2 wage rate) 0.3822*** (0.0460) Ln Travel cost (2/3 wage rate) 0.3826*** (0.0462) Constant 2.7279*** 2.7905*** 2.8438*** (0.3099) (0.3168) (0.3227) Ln alpha 0.1998*** 0.1991*** 0.1985*** (0.0668) (0.0667) (0.0667) Observations 468 468 468 AIC 2218 2219 2219 BIC 2252 2252 2252 Robust standard errors in parentheses *** p<0.01, ** p<0.05, p<0.1
53 Table 4 12 The adjust total consumer surplus with 1/3, 1/2, and 2/3 fraction of wage Recreation Site Annual Visits (FDEP,2 016) Recreational value (1/3) Recreational value (1/2) Recreational value (2/3) Blue Springs 41,000 $7,288,411 $8,133,473 Fanning Springs State Park 218,963 $38,924,204 $43,437,308 Ichetucknee Springs State Park 507,238 $90,169,734 $100,624,550 Madison Blue Springs 48,209 $8,569,927 $9,563,576 Total 815,410 $144,952,276 $161,758,906
54 Table 4 13 Perception versus physical environmental quality measures using ordered logistic regression model VARIABLES Model 1 Marginal Effects Current T rips 0.019 0.000284 (0.022) (0.000341) Income Below 0.139 0.00205 (0.306) (0.00454) Income Average 0.059 (0.256) 0.000909 (0.00394) First Visit 0.520* 0.00775 (0.270) (0.00495) Alternative visit 0.337 0.00502 (0.312) (0.00501) Swimming 0.419* 0.00625 (0.254) (0.00447) Nitrogen monthly 0.206*** (0.067) 0.00307** (0.00151) Constant cut1 4.604*** (0.529) Constant cut2 4.468*** (0.512) Constant cut3 3.372*** (0.424) Constant cut4 1.544*** (0.377) Observations 460 AIC 652.5 BIC 698 Standard errors in parentheses *** p<0.01, ** p<0.05, p<0.1
55 Tab le 4 14 M arginal effects of the independent variables on the probability of the respondent reporting each level of Perception of Clarity (1) Below Average (2) (3) (4) (5) Above Average VARIABLES Current trips 0.000284 3.93e 05 0.000565 0.00239 0.00327 (0.000341) (5.97e 05) (0.000661) (0.00272) (0.00373) Income Below 0.00205 0.000284 0.00408 0.0173 0.0237 (0.00454) (0.000682) (0.00897) (0.0379) (0.0519) Income Average 0.000909 0.000126 0.00180 0.00748 0.0103 (0.00394) (0.000558) (0.00779) (0.0323) (0.0445) First Visit 0.00775 0.00107 0.0154* 0.0652* 0.0894* (0.00495) (0.00121) (0.00883) (0.0335) (0.0460) Alternative visit 0.00502 0.000697 0.0100 0.0422 0.0580 (0.00501) (0.000948) (0.00957) (0.0390) (0.0535) Swimming 0.00625 0.000867 0.0125 0.0526* 0.0722* (0.00447) (0.00101) (0.00814) (0.0315) (0.0434) Nitrogen monthly 0.00307** 0.000425 0.00611** 0.0258*** 0.0354*** (0.00151) (0.000446) (0.00246) (0.00813) (0.0111) Observations 460 460 460 460 460 Standard errors in parentheses *** p<0.01, ** p<0.05, p<0.1
56 Table 4 15 Predicted Probabilities of Above Average Water Clarity Perception with Nitrogen Monthly increasing in mg/L Nitrogen Monthly Predicted Probabilities 2 mg/L 0.792*** (0.0206) 3 mg/L 0.756*** (0.0214) 4 mg/L 0.717*** (0.0288) 5 mg/L 0.673*** (0.0416) 6 mg/L 0.626*** (0.0574) 7 mg/L 0.577*** (0.0747) 8 mg/L 0.526*** (0.0919) Observations 460 Standard errors in parentheses *** p<0.01, ** p<0.05, p<0.1 NOTE: All predictors at their mean value
57 Figure 4 4 Graphical illustration of predicted probabilities along the confidence interval
58 CHAPTER 5 CONCLUSION The TCM survey reveals approximately 68% of spring visitors having experience d the springs before being interviewed The results show visitors make an average of four trips per year and willing to drive long distance s from urban areas to the springs. The results of the survey are encouraging for local and state officials to ensure acce ss to the springs by protectin g and restoring the spring park and increase the public awareness of the importance of having healthy spring systems The purpose of this study was to identify the environmental quality measures in conjunction with the TCM conduct a wage analysis of the different wage ratios in TCM the environmental quality measures can explain water clarity The results of the NB estimation showed the recrea tional value of th e four spring park s to be worth 144.49 million dollars. Using the ratios of the wage fraction level, the recreatio nal value of the spring park has a value of 161.76 to 178.24 million dollars. The results of the recreational value are a conservative lower bounded estimation of the total CS. There is a need for detailed cost analysis of the entire trip expenditure, and the actual wage rate to estimate the exact travel cost estimation. The results from the o rdered logit model indicate that individuals erceived water clarity rating is correlated with the environmental quality measure of nitrogen during the questioning period. The effect of nitrogen level is nonlinear where there are positive coefficients for poor and average quality, but vice versa for above average clarity rating As the amount of nitrogen decreases, the perceived rating of water clarity will increase Swimmers were also found to prefer a higher perceived water clarity rating, but could
59 not differentiate water clarity up to a certain level of nitrogen in the water. The environmental quality measure variable is statistically significant explaining the perceived water quality, but there is a lack of variation in the perceived wat er clarity among visitors a nd nitrogen levels collected at the four springs Thus, f uture research should focus more on investigating a link between perceived clarity rating with the environmental quality measures over more extended periods of time. Additionally, future research should investigate how perceived water quality rating differ by the type of recreational experience using a larger sample over a longer period of time. A small but important group of visitors to the springs are drivers. However, our sample incl udes summer visitors and divers typically visit the springs in the fall and winter. Morever, as springs quality improve, a potential issue would be congestion. Additional data can be collected to investigate how congestion can affect values derived from recreation. Some limitations to the study consist of the quality of data collected and the methods used to estimate the model. The data contained some errors such as invalid zip codes resulting in a small number of observations not having an estimated t ravel cost. The second issue was the trip count skewed above the mean visi t patterns. The third issue was the environmental quality measure data where some sites had high levels of recorded pollution compared to the other spring parks Also, the use of an on site survey may have introduced unintended bias from the surveyor reading the questions to the respondent One overlooked aspect of the survey data collected is the
60 TCM provides a convenient ap proach sh owing dollar values of recreation that can be compared to the potential restoration cost of the spring s T he results can be used for a cos t benefit analysis to determine if allowing for an increase in human interactions such as camping or building a boat ramp is justifiable. Local and state agencies can use this data provided to create poli cies to protect the spring parks in the future. Also this information can be used to raise public awareness regarding the importance of the springs for the communitie s, as well as the importance of the government programs aimed at springs protection. Currently, t he entrance fees charged to visit the springs are relatively low B ased even if there is a marginal price increase. The revenue from the fee increase can be used to offset future protection initiatives to protect the delicate spring system. Furthermore, if the environmenta l quality of the spring site is improved by reducing nitrogen leve ls; perceived water clarity rating could increase which may result in an increase number of visitors to the spring sites. The increase in visitors could generate additional revenue for the state and improve the local economic condition near the spring sit es.
61 APPENDIX A SURVEY First, we have a few questions about your visit to the springs today. 1. Are you you spending one or more nights away from home? O1 Day trip ( Skip to Question 3 ) O2 Staying overnight ( Go to Q2 ) O8 Not sure/Refused 2. How many total nights on this trip will you spend in the area? Number of nights: O88 Not sure/Refused 3. Is outdoor recreation at the spring and surrounding areas the primary reason for your trip to the area today? O1 Yes ( Skip to Question 5 ) O0 No ( Go to Question 4 ) O8 Not sure/Refused 4. What are the other purposes of this trip to the area? [Check all that apply. Do NOT read.] ( 1 or 0) O x0 Visiting family, friends, or relatives O x1 Attending a business related activity or event O x2 Visiting other cities/sites in Florida O x3 Other (describe): in Excel O x4 Not sure O x5 Prefer not to answer 5. How many adults (age 18 or older), including yourself, and how many children ( under age 18 ) are in your party on this trip? Number of adults: O 88 Not sure/Refused Number of children: O 88 Not sure/Refused 6. For your group, how much did you spend or do you plan to spend in total, on your visit to the spring today, including gas, rental equipment, food & beverages, park tickets, and so on? Total $ amount O8888 Not sure/Refused 7. What outdoor recreational activities did your group participate in at the spring and surrounding green space during this trip? [Check all that apply. Do NOT read.] 1 or 0 Ox0 Swimming / snorkeling Ox1 Scuba diving / Cave diving Ox2 Tubing Ox3 Camping Ox4 Non motorized boating (kayaking, canoeing) Ox5 Hiking Ox6 Nature viewing (bird, manatee watching, etc. ) Ox7 Picnicking, barbequing Ox8 Motorized boating (including waterskiing & jet skiing) Ox9 Not sure/Refused
62 Ox1 0 Other (describe): in Excel 8. Using a scale from 1 to 5, where 1 5 please tell me how you would rate the following characteristics of the spring and surrounding area. 1 2 3 4 5 8DK/R A. Clarity (cleanliness) of the water in the spring O O O O O O B. Conditions of facilities at the spring O O O O O O C. Conditions of the green space surrounding the spring O O O O O O 9. Is this your first trip to this spring? O1 Yes ( Skip to Question 12 ) O0 No ( Go to Question 10 ) O8 Not sure/Refused 10. Including today, how many times have you visited this spring in the last 12 months? # Visits O88 Not sure/Refused 11. read you a list of statements about how this spring may have changed over the past year. Using a scale from 1 to 5, where 1 is 1 2 3 4 5 8DK/R A. The water in the spring has become clearer O O O O O O B. Wildlife in the water has increased O O O O O O C. Wildlife in the surrounding green space has increased O O O O O O D. Water flow in the spring has increased O O O O O O E. Water level in the spring has increased O O O O O O 12. How many times do you plan to visit this spring in the next 12 months? # Visits O 88 Not sure/Refused Next, read you some hypothetical descriptions about future trips you might take to visit this spring, or other similar springs. Please tell me what you think would be most likely, if the hypothetical situation were true 13. Suppose this spring became unavailable for recreational use for some reason, would you visit a different location for similar types of outdoor recreation, or would you skip these activities? O1 Ye s What other place would you visit? (name) in Excel O0 No (Would skip activities) O8 Not sure O9 Prefer not to answer 14. Do you O1 Yes ( GO TO Q14A ) O0 No ( GO TO Q16 ) O9 Prefer not to answer 14A. Is that an individual pass or family pass? O1 Individual O2 Family O9 Prefer not to answer
63 15. Suppose the access fee to the spring were increased next year to improve conservation and restoration of the spring. If the access fee were increased by $20 per pass would that make you : O1 Visit the spring fewer times? O2 Not change the number of times you visit the spring? O3 Visit the spring more times? O8 Not sure O9 Prefer not to answer Skip to Question 17 (Demographics) 16. Suppose the access fee to the spring were increased next year to improve conservation and restoration of the spring. If the access fee were increased to $10 per person would that make you : O1 Visit the spring fewer times? O2 Not change the number of times you visit the spring? O3 Visit the spring more times? O8 Not sure O9 Prefer not to answer Demographics Finally, we just have a few demographic questions to be sure talked to all kinds of people who visit this area. 17. What is your home zip code? O Foreign resident 00000 O Refused 99999 18. What is the highest level of education you completed? [Chec k one.] O1 12 th grade or less (no HS degree) O4 College degree ( or ) O2 High school diploma or GED O5 Graduate / Professional degree O3 Some college, no degree O9 Prefer not to answer 19. In what year were you born? O 9999 Prefer not to answer 20. How many adults live in your household? O99 Prefer not to answer 21. What is your employment status? [Check one.] O1 Employed full time O5 Student O2 Employed part time O6 Retired O3 Self employed O7 Other O4 Unemployed O9 Prefer not to answer 22. What was your total household income before taxes in 2015? O1 Below $35,000 O4 $70,000 to $89,999 O2 $35,000 to $49,999 O5 $90,000 or more O3 $50,000 to $69,999 O8 Not sure O9 Prefer not to answer
64 23. Have you donated time or money to any environmental causes in the past 5 years? O1 Yes O0 No O9 Prefer not to answer 24. Respondent gender [ Interviewer: record, do not ask. ] O1 Male O2 Female That completes our survey. Thank you very much for your time and participation
65 APPENDIX B TABLE OF ESTIMATION RESULTS VARIABLES Model1 Model2 Model3 Model4 Model5 Ln travel cost 0.362*** 0.364*** 0.355*** 0.356*** 0.376*** (0.048) (0.048) (0.048) (0.048) (0.046) Income Below 0.212 0.185 0.220 0.193 0.169 (0.168) (0.170) (0.169) (0.172) (0.170) Income Average 0.053 0.036 0.049 0.044 0.044 (0.131) (0.135) (0.131) (0.135) (0.133) Clarity 0.208* 0.202* 0.184 (0.113) (0.116) (0.113) Facilities 0.107 0.064 0.105 (0.095) (0.098) (0.092) Greenspace 0.087 0.081 0.097 (0.114) (0.111) (0.114) swimming 0.363** 0.349** 0.590*** (0.173) (0.171) (0.199) Nitrogen monthly 0.011 0.077 0.005 0.068 (0.061) (0.057) (0.059) (0.056) Swimming Nitrogen 0.010 0.013 (0.065) (0.063) Non Swimming 0.042 0.018 (0.129) (0.127) Nonswimmingnitrogen 0.118** 0.117** (0.054) (0.053) Clarity_dummy 0.324 0.321 (0.266) (0.266) Facilities_dummy 0.349* 0.303 (0.196) (0.201) Greenspace_dummy 0.499** 0.511** (0.248) (0.244) Dissolved_oxygen 0.057 (0.082) Swimming_Dissolved 0.102 (0.079) Constant 2.351*** 2.518*** 2.989*** 3.253*** 2.355*** (0.376) (0.380) (0.465) (0.460) (0.364) Ln Alpha 0.227*** 0.227*** 0.234*** 0.238*** 0.228*** (0.068) (0.070) (0.067) (0.069) (0.067) Observations 467 447 467 447 467 AIC 2210 2113 2206 2109 2208 BIC 2256 2159 2251 2154 2253 Robust standard errors in parentheses *** p<0.01, ** p<0.05, p<0.1
66 VARIABLES Model 6 Model 7 Model 8 Model 9 Ln travel cost 0.378*** 0.362*** 0.357*** 0.363*** (0.049) (0.047) (0.047) (0.047) Income Below 0.187 0.204 0.227 0.207 (0.172) (0.169) (0.167) (0.169) Income Average 0.030 0.066 0.051 0.039 (0.133) (0.131) (0.131) (0.131) Clarity 0.190* 0.188 0.216* (0.114) (0.223) (0.113) Facilities 0.079 0.101 0.102 (0.096) (0.094) (0.096) Greenspace 0.057 0.091 0.070 (0.111) (0.110) (0.113) Secchi 0.018 (0.051) Nitrogen monthly 0.023 (0.041) Swimming 0.462 0.431 (1.116) (0.597) Swimmerclarity 0.014 (0.236) Clarity_dummy 0.288 (0.527) Facilities_dummy 0.356* (0.195) Greenspace_dummy 0.502** (0.243) Swimmerclarity_dummy 0.062 (0.615) Ln nitrogen monthly 0.001 (0.101) Swimming 0.280* (0.161) Constant 2.607*** 2.334*** 2.958*** 2.405*** (0.381) (0.404) (0.650) (0.369) Ln alpha 0.201*** 0.222*** 0.235*** 0.210*** (0.067) (0.068) (0.067) (0.068) Observations 468 460 468 467 AIC 2222 2182 2207 2214 BIC 2263 2223 2249 2255 Robust standard errors in parentheses *** p<0.01, ** p<0.05, p<0.1
67 VARIABLES Model 10 Model 11 Ln travel cost 0.326*** 0.326*** (0.049) (0.049) Income Below 0.256 0.232 (0.172) (0.174) Income Average 0.054 0.044 (0.138) (0.139) Clarity 0.734*** (0.171) Facilities 0.135* 0.131 (0.079) (0.081) Greenspace 0.098 0.085 (0.111) (0.114) S wimming 1.339*** 1.525** (0.348) (0.717) swimmerclarityd ummy 1.004*** (0.378) Non Swimmerclarity 0.242 (0.155) Constant 1.546*** 1.742** (0.580) (0.761) Ln alpha 0.228*** 0.227*** (0.067) (0.069) Observations 421 416 AIC 1997 1977 BIC 2037 2018 Robust standard errors in parentheses *** p<0.01, ** p<0.05, p<0.1
68 REFERENCES A damowicz, W., Swait, J., & Boxall, P. C. (2003). Perceptions versus Objective Measures of Environmental Quality in Combined Revealed and Stated Preference Models of Environmental Valuation. The New Economics of Outdoor Recreation, 165 190. doi:10.4337/9781781950319. 00018 Azevedo, C. D., Egan, K. J., Herriges, J. A., & Kling, C. L. (2003, Augus t 27). Iowa Lakes Valuation Project [Scholarly project]. Retrieved from http://www.card.iastate.edu/lakes/about_project.aspx Bin, O., Landry, C. E., Ellis, C. L., & Vogelsong, H. (2005) Some Consumer Surplus Estimates for North Carolina Beaches. Marine Resource Economics, 20(2), 145 161. doi:10.1086/mre.20.2.42629466 Blaine, T. W., Lichtkoppler, F. R., Bader, T. J., Hartman, T. J., & Lucente, J. E. (2015). An examination of sources of sensitivity of consumer surplus estimates in travel cost models. Journal of Environmental Management, 151, 427 436. doi: 10.1016/j.jenvman .2014.12.033 Bowker, J. M., English, D. B., & Donovan, J. A. (1996). Toward a Value for Guided Rafting on Southern Rivers. Journal of Agricultural and Applied Economics, 28(02), 423 432. doi:10.1017/s1074070800007410 Cameron, A. C., & Trivedi, P. K. (2010). Microeconometrics Using Stata, Revised Edition. College Station, TX: Stata Press. Cesario, F. (1976). Value of Time in Recreation Benefit Studies. Land Economics, 52 (1), 32 41. doi: 10.2307/3144984 Cesario, F. J., and J. L. Knetsch (1970), Time Bias in Recreation Benefit Estimates, Water Resour. Res., 6(3), 700 704, doi:10.1029/WR006i003p00700. Cordy, G. E. (2001). A primer on water quality ( USA U.S. Department of the Interior). Reston, VA: U.S. Dept. of the Interior, U.S. Geological Survey. Downing, J. A., Li, J., Antoniou, G., Kendall, D., Kling, C., Herriges, J. . Boatwright, L. (2005). Iowa Lakes Classification for Restoration (Rep.). Des Moines, IA: Iowa Department of Natural Resources. Dumas, C. F., Schuhmann, P., & Whitehead, J. C. (2005). Measuring the Economic Benefits of Water Quality Improvement with Benefit Transfer: An Introduction for Noneconomists. American Fisheries Society. Retrieved from https://220.127.116.11/cnr ap/UWEXLakes/Documents/p eople/economics/52_waterQualityBenefitTransfer_ dumas_paper.pdf
69 Englin, J., & Shonkwiler, J. S. (1995). Estimating Social Welfare Using Count Data Models: An Application to Long Run Recreation Demand Under Conditions of Endogenous Stratification and Trun cation. The Review of Economics and Statistics, 77(1), 104 120. doi:10.2307/2109996 Feather, P., & Shaw, W. D. (1999). Estimating the Cost of Leisure Time for Recreation Demand Models. Journal of Environmental Economics and Management, 38(1), 49 65. doi:1 0. 2139/ssrn.187411 Florida Springs Initiative Achievements. (2011, January 26). Retrieved from http://www.dep.state.fl.us/springs/initiative.htm Florida Springs. (2014, April 11). Retrieved from http://www.dep.state.fl. us/springs/ Florida State Park Impacts. (2016). Retrieved June, from https://friendsoffloridastateparks.org/FFSP Florida State Park Impacts Jeon, Y., Herriges, J. A., & Kling, C. L. (2004). The Role of Water Quality Perceptions in Modelling Lake Recreation Demand. The International Handbook on Non Market Environmental Valuation. doi:10.4337/9780857931191.00010 Loomis, J., Ng, K., (2012). Comparing economic values of trout anglers and non trout anglers in Colorado's stocked public reservoirs. North Am. J. Fish. Manag. 32 (2), 202e210. McConnel l, K. (1992). On Site Time in the Demand for Recreation. American Journal of Agricultural Economics, 74 (4), 918 925. Retrieved from http://www.jstor.org/stable/1243189 Nicholson, W., Snyder, C.M., (2004). Microeconomic Theory: Basic Principles and Extensions, ninth ed. South Western Cengage Learning, Mason, OH. Pittman, C. (2014, January 23). Florida's vanishing springs. http://www.tampabay.com/news/environment/water/floridas vanishing springs/1262988 Parsons, G. R. (2003). The Travel Cost Model. In A Primer on Nonmarket V aluation (pp. 269 329). Norwell, MA: Kluwer Academ ic. Perlman, U. H. (n.d.). Groundwater depletion. Retrieved from https://water.usgs.gov/edu/gwdepletion.html Phaneuf, D. J., Kling, C. L., & Herriges, J. A. (2000). Estimation and welfare calculations in a generalized corner solution model with an application to recreation demand. Review of Economics and Statistics, 82(1), 83 92.
70 Phaneuf, D. J., & Smith, V. K. (2005). Recreation Demand Models. In Handbook of Environmental Economics (Vol. 2, pp. 672 751). Elsevier B.V. doi:10.1016/S1574 0099(05 )02015 2 Shaw Daigee (1988), On site samples' regression: Problems of non negative integers, truncation, and endogenous stratification, Journal of Econometrics, 37, (2), 211 223 Shrest ha, R. K., & Loomis, J. B. (2003 ). Meta Analytic Benefit Transfer of Outdoor Recreation Economic Values: Testing Out of Sample Convergent Validity. Environmental and Resource Economics, 25(1), 79 100. Retrieved from https: //link.springer.com/article/10.1023/A:1023658501572 Shrestha, R. K., Stein, T. V., & Clark, J. (2007). Valuing nature based recreation in public natural areas of the Apalachicola River region, Florida. Journal of Environmental Management, 85(4), 977 985. Retrieved from http://www.sciencedirect.com/science/ article/pii/S0301479706003689 Sohngen, B., Lichtkoppler, F., Bielen, M., (1999). The Value of Day Trips to Lake Erie Beaches. Technical Bulletin OHSU e TB e 039. Ohio Sea Grant College Program, Columbus, Ohio. Smith, V. & Desvousges, W. (1985). The Generalized Travel Cost Model and Water Quality Benefits: A Reconsiderati on. Southern Economic Journal, 52 (2 ), 371 381. doi:10.2307/1059623 Stratis, N., & Bendle, B. (1995). A Socio economic Study of the Rodman Reservoir (United States, Florida Department of Environmental Protection). FL. Whitehead, J. C., Haab, T. C., & Huang, J. (2000). Measuring recreation benefits of quality improvements with revealed and stated behavior data. Resource and Energy Economics,22 (4), 339 354. doi:10.1016/s0928 7655 (00)00023 3 [Spring Station Water Readings]. (2017, January 17). Unpublished raw data. http://www.mysuwanneeriver.org/portal/springs.htm
71 BIOGRAPHICAL SKETCH Bryan Nguyen is a Florida native looking to learn about global cultures and the paradigms they must offer. Bryan graduated with a Bachelor of Science in f ood and r esource e conomics at the University of Florida and ha d continued with the same department in his graduate school endeavors. With the many experience s his education has deemed him with and a passion for learning Bryan will be moving to New York City to work with an NGO. He will have the opportunity to continue visiting new countries across the globe while focusing on issues in developing countries.