STATE-OF-THE-ART METHODS for Research, Planning, and Determining the Benefits of Outdoor Recreation

STATE-OF-THE-ART METHODS for research, planning, and determining the benefits of outdoor recreation

PACIFIC SOUTHWEST Forest and Ranee Experiment station

FOREST SERVICE U.S.DEPARTMENT OF AGRICULTURE P.O. BOX 245, BERKELEY, CALIFORNIA 94701

USDA FOREST SERVICE GENERAL TECHNICAL REPORT PSW-20 11977 STATE-OF-THE-ART METHODS for research, plan, Â¥qand determining the benefits of oudoor recreation

Gary H. Elsner, Compiler

CONTENTS

Improvement of Demand Studies as Tool for Planning Outdoor Recreation...... 1 H. N. van Lier

Forecasting the Demand-Response to Changes in Recreational Site Characteristics ...... 11 Peter Greiq

On the Use of Home and Site Surveys in Recrea Research . . 23 Mordechai Shechter

Relative Value of Selected Outdoor Recreation Activity Areas . 27 Joseph E. Hoffman, Jr.

A Recreational Visitor Travel Simulation Model as an Aid to Management Planning...... 31 Robert C. Lucas and Mordechai Shechter

A Survey of Wildlife-Related Recreation in the Tennessee ValleyRegion ...... 36 John L. Mechler and E. Lawrence Klein

Mathematical programming in the Context of Planning for Multiple Goals ...... 46 A. B. Rudra

Investigations on Recreational Forested Areas...... 60 Ulrich Amer Elsner, Gary H., compiler. 1977. State-of-the-art methods for research, planning, and determining the benefits of outdoor recreation. USDA Forest Serv. Gen. Tech. Rep. PSW-20, 62 p., illus. Pacific South- west Forest and Range Exp. Stn., Berkeley, Calif. These eight papers were oresented at Working Party S6.01-3, XV I th World Congress of the International Union of Forestry Research Organizations, Oslo, Norway, June 22, 1976. Topics covered include (a) improving studies on demand for outdoor recreation, (b) forecasting changes in number of visitors after a change in recreational qua1 ity at an area, (c) comparing the use of site surveys with hone interviews for recreation planning, (d) measuring the relative value of selected outdoor recreation activity areas, (e) model ing changes in use or area conditions to determine effects on use patterns and encounters between visitor groups, (f) surveying wildlife-related recreation to determine impact on a local economy, (g) applying mathematical programing in the context of planning for multiple goals, and (h) investigating the degree of aforestation preferred for different broad categories of land uses in recreational areas.

Oxford: 907.2 Retrieval Terms: outdoor recreation; forest recreation; recreation area planning.

GARY H. ELSNER is in charge of land management and landscape planning methodology research at the Pacific Southwest Forest and Range Experiment Station, Forest Service, U.S. Department of Agriculture, Berkeley, California. He is chairman, Working Party S6.01-3--Methodologies for Research, Planning and Determination of Benefits of Outdoor Recreation, International Union of Forestry Research Organizations.

The Authors

ULRICH AMMER is professor, School of Forestry, University of Munich, Germany. PETER GREIG is associated with the Department of Forestry, Oxford University, Great Britain, and the Forests Commission, Victoria, Australia. JOSEPH E. HOFFMAN, JR. is assistant professor, College of Forestry, Wildlife, and Range Sciences, University of Idaho, Moscow. E. LAWRENCE KLEIN is staff forester, Division of Forestry, Fisheries, and Wildlife Development, Tennessee Valley Authority, Norris, Tennessee. ROBERT C. LUCAS is research social scientist, Intermountain Forest and Range Experiment Station, Forest Service, U.S. Depart- ment of Agriculture, Missoula, Montana. JOHN L. MECHLER is supervisor of wildlife management, Land Between The Lakes, Golden Pond, Kentucky. A. B. RUDRA is Senior Lecturer, School of Agriculture and Forestry, University of Melbourne, Australia. MORDECHAI SHECHTER is senior lecturer, Faculty of Industrial and Management Engineering, Technion--Israel Institute of Technology, Haifa. H. N. VAN LIER is associate professor, Department of Land and Water Use, Agricultural University, Wageningen, The Netherlands. PREFACE

The challenge of planning for outdoor fects upon use patterns and encounters between recreation is shared by many countries visitor groups. It was developed to help ex- throughout the world. This publication offers plain and predict use patterns and encounters a selection of state-of-the-art papers by au- within U.S. Wilderness areas where solitude is thors from several countries which are active- often a prime objective, and, consequently ly dealing with this challenge. The eight where management would often 1 ike to decrease papers were prepared for the International encounters. However, this same model may be Union of Forestry Research Organizations useful in examining a1 ternat ive management (IUFRO) XVI World Congress, held in Oslo, strategies in areas where management may wish Norway, June 20-July 2, 1976. They are the to increase encounters, e.g., in parks where formal discussion papers for IUFRO Working wildl ife observations are important and where Party, S6.01-3, Methodologies for Research, such an event may be termed an encounter. The Planning, and Determination of Benefits of approach may indeed by appropriate for simula- Outdoor Recreation, which met on June 22. ting a wide range of management alternatives in a wide array of dispersed outdoor recrea- The lead paper by H. N. Van Lier sets the tion systems. stage for the papers that follow by describing past demand mdeling approaches, explaining Determining the relationship between hunt- their strengths and weaknesses relating to ing, fishing, nonconsumptive wildlife use and planning needs, and making several suggestions the local economy is a difficult but worthwhile for improvements including the need to study task for wildl ife-related recreation planning. in depth the separate influence of the three John L. Mechler and E. Lawrence Klein report on basic system components--origin, destination, a major study of wildlife in the southeastern and linkage--and the need to adequately model United States. Their paper is instructive both substitutability. The next paper, by Peter in terms of the area studied and in terms of Greig, investigates in some detail the desti- illustrating how a careful study of gross ex- nation element and the question of substituta- penditures may be useful to the objective plan- bility. The paper explains and illustrates ing of wildl ife-related recreation which my with simple numerical examples for ski areas a have a positive monetary impact on a local model for forecasting the short-term change in economy. number of visitors to an area which result from a change in the destination's quality. Long-range planning for adequate outdoor recreation areas is usually done in a context Mordechai Shechter's paper compares in in which recreation values must be compared close detail two approaches to estimating out- with other uses for the land, such as, timber door recreation use and benefit information. production. A. B. Rudra specifies carefully On the basis of two comprehensive studies of the use of goal programing and explains its ap- the largest national park in Israel, he con- pl ication to an illustrative area containing cludes that site surveys are often more effi- potential for recreation, timber and mu1 t iple cient and cheaper than home interviews. How- uses. Since many of the parameters needed for ever, the paper recommends comprehensive home either conventional linear programing or goal surveys at longer intervals, say every I0 programing are never known with complete cer- years, for the collection of additional data tainty, his paper also includes a brief intro- and for broad area planning. duction to the use of stochastic programing and explains its advantages. Joseph E. Hoffman's paper illustrates the use of two measures of the value/cost ratio in The long-range planning of alternative comparing the perceived value with the devel- uses of landscape units has received intensive opment and management costs for several alter- investigations in Germany. Ulrich Ammer's pa- native outdoor recreation activity areas. His per highlights the results of these studies in data indicate that areas with low development terms of the preferred degree of aforestation. costs may have a higher value/cost ratio than His conclusions apply to forested lands in and normally expected. around populated areas and rural areas. The results are compared with the current propor- The report by Robert C. Lucas and Morde- tions of forested lands in each category and chai Shechter describes an important and prac- implications for changes in planning goals are tical model for simulating changes in manage- described. ment policy or access within dispersed outdoor recreation areas. The model predicts the ef- These papers were prepared for an international conference of forestry objectivity. researchers. But each paper has its own relevance to a specific decisionmaking One goal of IUFb ,s to increase the situation and as such may be useful to communication among forestry researchers recreation managers or planners who are worldwide. This pub1 icat ion was designed searching for a way to gain additional to help achieve that goal.

Improvement of Demand Studies as Tool For Planning Outdoor Recreation H. N. Van Lier

Abstract--Planning for recreation sites in forested areas requires solution to the sequence of problems of determining the type, location, capacity, and layout of facilities. Models have been developed to forecast the demand for specific sites, but they are not necessarily applicable to other sites. A gravity model was developed to overcome this limitation. Alternatives in the form of competing recreation sites are included in the model. But even this kind of model has limitations: the impossibility of clearly separating three basic factors that affect the distribution of trips: origin of travel, destination, and 1 inkage (or reaction to the distance to be traveled). In addition, the meaning of each factor has not yet been thoroughly investigated.

The demand for outdoor recreation is evi- 1 arge recreational areas, special pro-jects, dent. The main problem is determining not only etc. It has often been emphasized that a close what size it will take, but also in what direc- relationship exists between the components1 tion it will change. As Bijkerk (1975) empha- type, allocation, capacity and layout in recre- sized, this demand problem also covers forest ation planning (van Lier and others lq71). areas since lleconomic and social changes in the Nevertheless a sequence-based approach may meet western world increase the need for multiple the first reaui rement. land use, leading to the fact that recreation in forested areas, therefore? is an important Demand studies are a central issue in issue as well from the point of forestry policy these problems. According to Bijkerk (1975), as forestry planning. I' The same author points "adequate planning of the important phenomenon to the fact that ''as a result of changing eco- of recreation, good statistics on participation nomic and social circumstances--the demand rate, distribution over types and distance, created by the happy few now being created by frequency and time of occurrence is vital .'I social groups, the addition of day and weekend recreation to that in vacations, the greater mobility of recreationists and the awakening of Demand studies are needed in the first the urban population to the fact that the abun- place for the determination of type and amount dance in natural resources is dwindling-- of (additional) facilities. Since in many of forests are becoming an increasingly important these demand studies models are used in which feature in outdoor recreation." the distribution-effect of recreationists over the area is taken into account by means of Planning of forests for recreation or distance functions, they also form a basis for planning of recreation in forest areas means the allocation problems as well as for the solving the sequence of problems concerning tbe capacity of projects. Layout means type, size determination of type? location, capacity and and mutual location of different elements in layout of outdoor recreational faci 1 ities. projects or areas. It determines the attract- Facilities in this context mean all kinds of iveness of the total facil ity--a qua1 ity that provisions for outdoor recreation? as for in- affects demand. Demand studies are, therefore, stance small playgroundsy beaches? waters? vital for the planning of outdoor recreation. HOW DID IT ALL START? The second type of research can be called project research. People visiting certain Older studies regarding outdoor recrea- types of outdoor recreational projects are tion can be divided into two types. The first interviewed regarding their origin, the dis- type, in which a sample of a certain popula- tance traveled, the activities performed on tlon has been interviewed at home about their the project, the expenditures, and some back- outdoor recreational behavior, has been car- ground (socio-economic) variables. Based on ried out in several countries in the last 20 these data, very often so-called use-models years. Information is gathered about back- are constructed, of which the general form is: ground variables (as income, family size, age, sex, profession, etc.) on the one hand and number of trips, type of projects, distance traveled, activities performed, etc. on the other hand (Centraal Bureau voor de Statistiek in which the visit to a certain project from a 1966, Outdoor Recreation Resource Review Com- certain origin depends on the population (pi), mission 1962, Rijksdienst voor het Nationale the distance (Oi) and (some) socio-economic Plan 1961). The data are often, but not al- variables (XI ... xn). ways, used for studies regarding the influence of background variables upon behavior For example M.erewi tz (1966) constructed on outdoor recreation. the following model for a lake in the U.S.A.:

in which population (P"), distance (SU) and ed with the gravity model approach of Van population density (PU . Du) are taken into Doren (19671, in wh ch the alternatives are account. Van Lier (1969/70) constructed taken into account n the following way: models for inland beaches in the Netherlands:

in which P = population and D = distance. where the population (Pi), the attraction in- One of the main disadvantages of this dex (AJ), the distance between origin and pro- type of model is the fact that alternative ject (D*~)as also the combined influence of (competing) projects and areas are not expl i- attraction and distance of the competing pro- citly taken into account (one has to bear in jects (zJ=1 AJ D;;), are ~akeninto account. mind'that using visit- umbers means that the influence of competing projects is imp1 icitly SHORTCOMINGS accounted for). This ack in the modeling means that.. . The shortcomings of and problems with the use-models have already been mentioned. 1. It is very hard to transplant a calibrated use-model to other areas, given the fact In recent years, the approaches with that the support-s tuation (types and dis- gravity-models are also criticized by many au- tribution of facil ties) in most cases is thors. Niedercorn and Bechdol t (1969) drew quite different, while it also might be distinctions between an origin factor, a des- that the "demand1' differs. tination factor, and a 1 inkage factor as essential parts of the modeling. On the basis 2. The influence of a to-be-created facility of these distinctions, the following short- or the improvement of existing ones cannot comings can be listed: be calculated explicitly. I. The impossibility of separating and ex- Aside from these limitations, most models tracting these three factors very clearly, also cannot be transplanted in time, since This problem has been emphasized many changes of behavior in time mostly are not times and by many authors. Both statisti- taken into account. This, however, is very cally and conceptually it is impossible to often true for other models also and therefore separate the influence of origin, destina- will be ignored in this paper. t ion and I inkage on visit rates (or numbers) of outdoor recreational faci Iities. For these reasons, new model-types were Statistically it is impossible because .developed in the past 10 years. It all start- the values given to, for instance, the attraction indices of projects or areas (for instance, regarding visits to forest and distance parameters (as part of a spe- areas or projects which are situated in and cific distance function) depend to a cer- closely related with these areas). In this tain degree on the statistical analyses method a two-stage approach is followed. In ~rocedurethat is used (as for instance, the first stage a covariance technique is used covariance techniques, regression analy- to extract systematically origin factors ses). In this resDect the criterion used (called emissiveness) and destination factors as a peasure for the goodness of fit plays (called attractiveness). In the second stage also an important role. Conceptually the an analysis is carried out in order to find 5e??rs:isn is difficult because it assumes the influence of different factors for both :ra: t5e influence of the origin, i.e., the emissiveness and attractiveness. Differ- cPe push-factor, in no way is related with ent techniques can be used for this, as for the suDport situation, while for the sane instance, multivariate analyses, etc. For the -eascn the attraction-index is assumed to emissiveness selected characteristics of popu- be unrelated to the decision of people lation centers can be used. In the same way whether they will have their outing or the attractiveness can be analyzed by using not. project characteristics.

In other words, it is assuved that the decis ionmaki ng process regarding the mak- In the approaches of analyzing trip dis- ing of a trip, yes or no, by an individual tribut ion regarding outdoor recreation, runs as fol lows: (a) first, the Derson Klaassen (1974) distinguishes between projects decides that he definitely wants to go out or areas which are origin-exclusive and those no matter what he can do outdoors; (b) which are destination-exclusive. When a large part (say 70 percent) of the recreationists on second, he makes an inventory of all DOS- sibilities of the projects and the travel a certain project originate from one popula- distances and then chooses which one he tion center, the project has origin-exclusivi- will visit, knowing the properties of the ty. When the majority of all recreationists different projects (attractivity) as also from a certain population center travel to one the barriers (i.e., distance and travel- destination, then that project has destination- tipe costs) to overcome. exclusivity. Studies regarding the problem of planning a large number of small areas as op- There are reasons to be1 ieve that the posed to a small number of large areas are decisionmaking process sometimes Pore cr starting. Klaassen (1974) found that the less runs like this, but in many cases first planning system (a large number of rath- both aspects are interwoven: many Dersons er small areas) might be advantageous. Ac- decide to make a trip because they know a cording to Bijkerk (1975) the same effect very nice place to perform a certain seems to occur in town planning, where "poly- wanted activity. Nevertheless the dis- nucleation seems to k the leading principle." tinction in origin-, destination- and linkage-factors is useful, because it en- It is obvious that future research re- ables one to approach the process system- garding demand for outdoor recreational facil- atically. One has, however, to keep in ities should also focus on these aspects. mind that this distinction is a means, not a purpose in itself. Regard ing Demand -Mode 1 ing 2. The neaning of each of these three factors has up to now not been investigated thor- Many attempts have been made to improve oughly. In other words, what are the the structure of both use-models and gravity- type models. Regarding use-models, Van Lier background variables in the push-factor, (1973) constructed the following one for in- how is a linkage perceived by the recrea- land beach recreation in the Netherlands: tionists, and what properties of the projects determine the attractivity? Re- cently, studies regarding these aspects have started to appear. in which the population (P), the vacationists RECENT DEVELOPMENTS elsewhere (E) , the vacationists in the area the distance the a1 ternatives in- Regarding the Research Itself (B), (D~), side (Acl) and outside (Ac2) the origin, both Cesario (197.5) recently proposed a new weighted according to recreation type and dis- method to analyze outdoor recreation trip data tance are taken into account. The formula can be rewritten as follows:

where g = competitional effect of a certain account. The model simulates trips or project on other projects, c = capacity of nated in a gravitational field. that project and r = reduction coefficient depending on distance. This shows that prop- For the use of wilderness area, models erties of population centers and alternative were constructed by McKillop (1975) of the projects explicitly also are taken into following type:

in which the use level (Yit; for area i in that the modeling itself is something to be year t is described by several variables followed critically. other-simulation proce- (xit). It was found that for U.S. Forest dures for outdoor recreation demand may become Service areas such variables as percentage of operational in the near future. area over 7000 feet in elevation, road construction in adjacent National Forests, travel Regarding the Origin Factor time, precipitation, population within a certain distance, size of wilderness area, number An important aspect of demand-model ing of lakes and number of entry points are for outdoor recreation is the achievement of important. obtaining knowledge on the reasons of people to seek recreation in the outdoors. Many Regarding gravity-type models, many at- ideas have been formulated, less research has tempts are made for improvements and imple- been done, and almost no results have come up. mentation. Freund and Wi lson (1974) give an example of an implementation by estimating a Up to now the research in this field has gravity-model to explain recreational travel been restricted to analyses of the influence and participation. In concentrating on the of SOC io-economic variables upon demand (meas- implementation method and the nature of re- ured mostly as number of trips), although sults, they found that a major task was to other approaches also have been fol lowed. La make physically observed measurements serve as Page and Ragain (1974) found that a large proxies for parameters specified by the grav- change in camping (51 percent of former camp- ity-model. In addition, they found it neces- ers were either camping less or had dropped sary to choose a reasonable set of meaningful out of the camping market) was related to a predictor variables. change in the style of camping itself, and to changes in the family cycle, although the lat- According to Wol fe (I9721, a disadvantage ter gave not in a consistent pattern. These of the gravity-model is the tendency to over- findings point to the problem of the substi- estimate the number of short recreational tutability which has been defined by Hendee trips and to underestimate the number of.the and Burge (1974) as "the interchangeabi Iity of long ones. He therefore constructed a so- recreation activities in satisfying partici- called inertia-model: pants' motives, needs, wishes and desires.'' It is quite obvious that research, especially dealing with this aspect of the demand, needed . How various socio-economic factors have their influence upon outdoor recreation participation is shown by different researchers. in which the same variables (population P, Recently McEvoy (1974) experimentally investi- capacity C, and distance D) are used but the gated the influence of the distribution of the distance function itself (or the description working time. From the research it appeared of the reaction of recreationists on distance) that "substantial increases in the consumption is transformed. Whether this type is more of outdoor recreation will result if the four- adequate to simulate reality still is to be proved for different forms of outdoor recrea- day workweek is adopted by a significant segment of the work force.'' For planning outdoor tion. It would be worthwhile to try it out for forest areas. recreation the future distribution of leisure time will be very important. According to Taking all things together, it is clear Bijkerk (1975) it is important to know whether 'we are moving towards less working hours per analyzed more thoroughly. In his research day, less worki-ng days per week, less working only population and income were accounted for, weeks per year or less working years in a but it is necessary and it should be possible 1ife-time." to enlarge the efforts in this direction. Lintsen (1975) constructed a special demand In the approach of Cesario (1975), the function, in which the distribution effect was demand-part of the trip-distribut ion can be not included, however:

This formulation shows that the demand (ai) nities plays an important role. Is the road a depends on household-categories (hhcat) which factor on its own or is it a part of the site? were based on income and family cycle as also In addition, in many cases the travel itself on the type of house (won), the possession of can be enjoyable. This probably causes the a car (mob) and the level of urbanization (u) inertia of movement (~olfe1972). of the origin. Regarding the Destination Factor Regardinq the Linkage Factor The leading problem in analyzing the des- The problem of the react ion of people to tination is whether it can be analyzed objec- distance (or on factors derived from distance, tively (i .e., based on hard facts such as such as travel-time or costs) has been tackled acreage of parking areas, playing fields, by many investigators. Wolfe (1972) initiated etc.) or whether it has to be analyzed sub- a new approach by using a different reaction- jectively (i.e., using perception of the area to-distance-function based on the so-called by the recreationists) as well. inertia of starting up and on the inertia of movement. The starting up inertia is caused In the approach of Cesar io (1975) on1 y by the fact that "a great many people may not objective variables were used, such as number wish to make a trip of any length, however of acres, number of camping units, length of short," while the inertia of movement is beach, etc. caused by the fact that "among the minority of people who indulge in lengthy trips, a still An inventory of camping-sites in the smaller minority finds travel itself so stimu- Netherlands by Ukelenstam (1974) showed that lating that the farther they go, the farther the preferences of campers with regard to the they want to go." location of the sites are closely related with forests and seacoast (fig. 1). The attractiv- This ha? 'eaman (1974) to analyze the ity (or the attractiviness) in its essence is reaction to as a function of distance. more subjective, however, in his "analysis of On the basis nalysis of five gravity- visits to outdoor recreation sites in the functions, he J that there are cased in vicinity of a large town in the Netherlands which (a) each new mile to be traveled offers (~indoven),I1 Lintsen (1975) found that attrac- more resistance than the last; (b) each new tivity determined by calibration of gravity- mile to be traveled offers less resistance models per socio-economic groups shows large than the last; and (c) each new mile to be differences (table 1). The findings show traveled has a constant resistance. that (a) the ranking of the areas (first, second, etc.) differs from household category These results suggest that the reaction to household category; and (b) the variancy to distance is hard to understand, especially of the attractivity-indices per area among the because it is also related to the distribution household categories is large. of outdoor recreation facilities itself. This last aspect has been stressed by OIRourke From this it can be concluded that the (1974). who says it is "a function of the attractivity of sites depends not only on the structure of opportunities available to the site-properties but also on the differentia- recreat ionists." tion in demand.

From the foregoing it can be concluded This shows that studies regarding the that this part of the demand has to be inves- perception of especially wilderness and for- tigated thoroughly. One has, however, to keep estry areas should be encouraged in the near in mind that it will be very complicated since future. The first steps on this difficult the availability and distribution of opportu- path have been made, however. Figure I--Campers in the Netherlands preferred camping sites in forests or along the seacoast, according to an inventory made by Ijekenstam (1 974).

'able 1 - The attractivi ty-indices of 12 recreation areas for 12 household categories (hhcat) depending on income and family cycle.

hhcat area 4 2 3 4 5 6 7 8 9 10 11 12

1 .070 .HO .053 .069 .074 .036 .049 .I12 .067 -064 -098 -047 2 .026 .060 .028 .033 -042 .069 .049 .067 .041 .027 .049 -055

Source: Lintsen 1975 Tuan (1974) stud ied environmental percep- 4. A separation of origin-, linkage- tion, attitudes, and values, while Peterson and destination-factors in demand-models is (1974) studied the percept ion of wilderness very often used. Both the study of the way among the different groups about some activi- in which these factors should be distin- ties that were approved of (e.g., paddle guished and the analysis of these three canoeing and fishing) and others that were basic factors only have been started, however. disapproved. Differences were found in the More research is needed, especial ly for forest perception by recreationists and managers. areas, given the fact that up to now not Carls (1974) reported that the perception of enough is known about the factors (variables) certain landscapes is negatively correlated determining their attractiveness. Perception with the "area of people and the area of research should be mentioned in this regard. high development," positively correlated with "area of stream, of waterfall and of 5. For the analysis of the origin-factor, lake." studies regarding background variables should be performed. The reaction of recreationists The research on the destination factor on distance to forest areas is a special prob- has been done in two ways: (a) studying the lem since nothing is known about the percep- properties (hard facts) of the area and relat- tion of travel distance as related to the pro- ing these to the attractiveness; and (b) posed visit to forest areas. More and detail- studying the percept ion by recreationists of ed studies on this part of the dernand-model ing the area. are needed.

An approach including percept ion and hara Finally, some remarks about planning in facts is needed. The recently published re- forestry and recreation by Bijkerk (1975) are port of the Committee on Assessment of Demand appropriate: for Outdoor Recreat ion Resources (1975) should be mentioned. In this report attention is 1. A more systematic approach, according paid to the planning and social and economic the sequence a1 locat ion-capacity-layout, to pol icy of outdoor recreation and to methods of determine the requirements of new recreation analyzing demand. Special topics are the facilities in forests, will improve the effec- demand for alternative types, the demand for tiveness of the plans, as well as lower the site-specific outdoor recreation resources, difficulties of acquiring funds for outdoor a socio-psychological definition of recrea- recreation. tion demand, and the estimation and use of models. 2. A classification of forests according to recreational potential wi 11 prove to be of CONCLUSIONS great benefit to planning an increased recreational use. Research on the possibilities of The following general conclusions making such classifications should be encour- regarding the demand for outdoor recvea- aged. tion in forests can be drawn: 3. Data on recreational activities 1 Demand studies are a v ta1 part of should be included in the census, as such data planning outdoor recreation fac lities in are becoming indispensable for adequate plan- general. This also accounts fo the type of ning of facilities for outdoor recreation. facil ities and projects which a e predomi- nantly situated in forest areas or which are formed by the special layout of the forest itself. VERBESSERUNG DER FRAGEUNTERSUCHUNGEN FUR DIE PLANUNG VON FREILUFTERHOLUNG 2. Research regarding the substituta- oility of the demand should be encouraged. Insight in this matter opens the possibility for planners to adjust the plans in a better Zusammenfassung way to the natural sutability of forest areas for outdoor recreation. Die Frage nach Freilufterholung, besonders in Waldgebiete, ist klar. ZukUnftige Aender- 3. Demand studies based on model- ungen dieser Fragen mit Bezug auf Zahl und approaches have been performed in several Richtung, missen untersucht werden wenn es sich ways. Improvements were made in recent urn die Planung von Anlagen hSndelt. Die LUsung years. Nevertheless the type of model to einer Reihe Probleme mit Bezug auf Feststellung simulate trip distribution in particular von Art, Stelle, KapazitSt und Einrichtung von to forest areas should be object of further Freilufterholungsanlagen ist notwendig. Eine studies. Beschreibung 1st qegeben worden von 'Gebrauchs- modelle' die zuerst entwickelt worden sind urn Une description des 'modSles d'util i- die Fragen fUr bestimmte Stellen zu be- sation' qui avaient 6te d6velopp6s initialement schreiben, gleich wie die Nachteile dieser aux fins de decrire la demande pour des sites Model le. Besonders bezUgl ich die Unmffgl ichkeit precises, a 6t6 donnee aussi que des des- diese fUr andere Gegenden zu beniltzen. Um avantages de ces modSles en ce qui concerne diese Nachteile zu Uberwinden 1st das nachste I'impossibilite de Ies utiliser pour d'autres Model!, das 'Schwerpunktmodelll entwickelt reg ions. worden. In diesem Modelle werden Alter- natieven wie konkurrierende Stellen oder Un deuxihe modele, 'du type gravit6', Gegenden planmassig eingebaut. Die Nachteile etait construit pour surmonter ces desavantages dieses Modelles sind zweifaltiq: par 1 ' inclusion expl icite d'al ternatives (d'autres sites ou regions) dans le modgle. - - die Unmoglichkeit drei Grundlaqen dieses Les l imites de ce deuxihe type sont: Modelles, nSmlich Ursprung, Bestimmung und Verbindung, deutlich zu separieren und zu -- L1impossibi1it6 de s6parer clairement les extrahieren. Die Separation der Einfluss trois facteurs de base: origine et destination dieser drei Grundlagen auf Fahrtverteilung ist du voyageur et accessibilit6 du site. A 12 statistisch sowohl wie konzeptisch schwer fois sur le plan de conception et de statis- durchzufuhren tique la destillation de l'influence de chacun de ces trois facteurs sur I'utilisation des - - die Bedeutunq jener drei Grundlagen ist facilites est difficile. nicht einqehend qecruft worden. Einsicht in den Hintergrundvariabelen von 'Push' (Ursprung) -- La signification de chacun des facteurs n'a und 'Pull ' (Bestimmunq) sowohl wie in der oas 4t6 recherchge profond6men t et notre com- Abstandperzept ion (verbindung) fehlen. prehension reste insuffisante.

In Rahmen neuere Entwicklunqen sind Vor- Lors des d4veloppements plus recents des schlage aemacht worden und Untersuchun~en propositions ont ete formul6es et des etudes durchgefuhrt worden un diese Nachteile zu ont 4t6 entames pour faire face ces deux uberwinden. Neue Analysen n-it Sezug auf Fahrt- limites. Des analyses nouveaux relatifs 2 la verteilungen ebenso wie auf Hintergrunde der distribution des voyages ainsi qu'; la destina- Ursprung und Bestimmung sind fertig oder werden tion et 2 llorigine des voyageurs ont 6t6 faits gemacht. Einige Beisoielen sind gegeben ou sont en voie de real isation. Quelques worden. examples ont 4te donnes.

Hinzu sind einige Schlussbemerkungen ge- A la fin il a 4t6 remarque que des etudes macht worden. Frageuntersuchungen sollen in de la demande seront necessa i re dans 1 'aveni r der Zukunft durchqefuhrt werden. Das Problem et que Ie problSme d'une substitution possible der Substitutionsfahigkeit der Fraqe mussten de cette demande dev-a recevoir plus d1atten- bei noch komnende Untersuchungen nehr benach- tion. Ceci est valable aussi bien pour 1es druckt werden. Dies gilt auch fir die Analysen analyses de distribution de voyages, pour la der Fahrtverteilungen, die Separation in Ur- separation des trois facteurs origine, accessi- sprung, Verbindung und Bestimmungsfaktoren und bilite et destination que pour l'analyse de ces die mehr detai 11 ierten Anal ysen dieser groupes de facteurs de facon plus detail lee. Faktoren.

LIAMELIORATION DES ETUDES DE LA DEMANDE LITERATURE C ITED COMME MOYEN DE PLANIFICATION DE LA RECREATION EN PLEIN AIR Beaman, J. 1974. Distance and the 'reaction' to distance as a function of distance. J. Leis. Res. 6(3):220-231 .

Bijkerk, C. La recreation en plein air, plus particu- 1975. Recreation values of forests and 1iSrement sous forgt, est trSs recherchee. Des parks. Phil. Trans. R. Soc. Lond. B. modifications futures de la demande, quant 2 la 271 :179-198. quantite et Iforientation, devront Stre etu- diees quand i1 s'agit du planning des fa- Carls, E. G. cilites. La solution des probl&mes relatifs 2 1974. The effects of people and man-induced la determination proprement dite 2 la location, conditions on preferences for outdoor 5 la capacite et 2 I 'amenagement des projets en recreation landscapes. J. Leis. Res. plein air est d'une importance majeure. 6 (2) : 1 13-1 24. Cesario, F. J. source development. Water Resources Res. 1975. A new method for analyzing outdoor 2 (4) :625-640. recreation trip data. J. Leis. Res. 7(3): 200-21 5.

Niedercorn, J. H., and B. V. Bechdolt, Jr. Centraal Bureau voor de Statistiek 1969. An economic derivation of the 'gravity 1966. Vrijetijdsbesteding in Nederland 1962- law' of spatial interaction. J. Reg. Sci. 1963. Deel 8: Een samenvattend overzicht 9 (2) :273-282. Karakter ist ieke patronen. De Haan, Hilversum. 57 p. O'Rourke, B. 1974. Travel in the recreational experience Centraal Bureau voor de Statistiek --a l iterature review. J. Leis. Res. 1971. De Nederlandse Bosstatistiek, 1964- 6(2): 140-156. 1968. Staatsuitg. Den Haag. 67 p. Outdoor Recreation Resources Review Commission Committee on Assessment of Demand for Outdoor 1962. National recreation survey. Study Recreation Resources report 19. Washington, D.C. 61 p. 1975. Assessing demand for outdoor recreation. Nat. Acad. Sci., Washington, D.C. Peterson, G. L. 124 p. 1974. A comparison of the sentiments and perceptions of wilderness managers and Freund, R. J., and R. R. Wilson canoests in the boundary water canoe area. 974. An example of a gravity model to esti- J. Leis Res. 6(3) :194-206. mate recreation travel. J. Leis. Res. 6(3) :24l-256. Rijksdienst voor het Nationale Plan 1961. Mensen op zondag. Publ. I4 Staats- Hendee, J. C., and R. J. Burdge drukkeri j's Gravenhage. 166 p. 1974. The substitutability concept: Impli- cation for recreation research and manage- Tuan Yi-Fu ment. J. Leis. Res. 6(2):157-162. 1974. Topophilia: a study of environmental perception, attitudes, and values. Pren- Ijkelenstarn, G. F. P. tice Hall, Englewood Cliffs, N.J. 260 p. 1974. Aantal en spreiding van diverse typen recreatieverblijven in Nederland. Recr. Van Doren, C. S. voorz. 6(4): 136-143. 1967. An interaction travel model for pro- jecting attendance of campers at Michigan Klaassen, L. H. State Parks: a study in recreational geo- 1974. Prelude op een recreatieve kostenbaten graphy. Ph.D. Thesis, Michigan State analyse. Recr. Voorz. 3(3) :96-102. Univ. 264p.

La Page, W. F., and D. P. Ragain Van Lier, H. N. 1974. Family camping trends--an eight year 1969/70. Capaciteitsberekening voor nieuw te panel study. J. Leis. Res. 6(2):101-112. stichten strandbaden. Verkeerstechniek 20(12) en 21 (I), bijvoegsel Recr. Voorz. Lintsen, W. 12:186-190 en 1 :2-6. 1975. Een analyse van het bezoek aan open- luchtrecreatiegebieden in de agglomeratie Van Lier, H. N. Eindhoven. T. H. Eindhoven. 144 p. 1973. Determination of planning capacity and layout criteria of outdoor recreation pro- McEvoy, Ill, J. jects. Pudoc Wageningen. Agric. Res. 1974. Hours of work and the demand for out- Reports 795, 156 p. door recreation. J. Leis. Res. 6(2) :125- 139. Lier, H. N., J. G. Bakker, and H. Bergman 971. Onderzoek ten behoeve van openlucht- McKiIlop, W. recreatieve voorzieningen bij de in- 1975. Wilderness use in California: a quan- richting van het platteland. Cult. Techn. titative analysis. J. Leis. Res. 7(3): Tijdschr. 11:97-128. 165-178. fe, R. I. Merewitz, L. 972. The inertia model. J. Leis. Res. 1966. Recreational benefits of water re- ^(I) :73-76. Forecasting the Demand-Response to Changes in Recreational Site Characteristics 1 Peter Greig

Abstract~Anew method is presented for forecasting the short-term change in numbers of visitors at an area, after some change in the recreational quality there. A group of recreation areas may be substitutes for the area of concern. Recreationists decide on trips to particular areas, depending on the relative costs and quality characteristics of an area, and on their particular preferences for those characteristics. Both the characteristics as we11 as the distribution of preferences in the community, can be described mathematically and a model of recreationists' choices developed. Using observed data on costs, characteristics, and numbers of visits at the areas in the group, the model can be used to forecast new choices after some change in the characteristics of an area. Simple numerical examples illustrate application of the model .

In forestry practice, many operations in- The other class is concerned with esti- volve alterations to the forest landscape. mating, by statistical methods, the relation- ' These alterations often affect the recreation- ship between visitor numbers at recreation al quality of the area concerned--even if rec- areas, and the various characteristics of reation is only one of the multiple uses in- those areas. Usually, a causal relationship volved. In evaluating whether or not to make is implied, on the supposition that visitors the a1 teration, an essential piece of informa- choose according to their preferences for tion will be an estimate of the change, if characteristics. Thus, the second class of any, in the number of visitors to the area. methods may be said to depend on revealed preferences. Mostly, the methods have in- This paper describes a new method of volved the use of multiple regression analysis forecasting the change in numbers of visitors (Johnston and Elsner 1972, Lime 1971, Seneca (and their origins) after a specific change in and Cicchetti 1969, Shafer and Thompson 1968, the recreational quality of a forest or any Holman and Bennett 1973, Cheung 1972). More other rural area. recent1y, a method has been developed to take into account the suspicion that characteris- Recreational quality is defined here as tics may not be entirely independent in their comprising the physical characteristics of separate contributions to a recreationist's recreational areas that may influence house- preferences for an area. Thus Cesario (1973) holds in their choices between such areas. used the Automatic Interaction Detector Recreational characteristics have been the (A. I .D.) Analysis which allows for possible subject of much research, which can be divided interact ion between variables. into two classes. The "stated preference" methods a11 place The first class is concerned with estima- the respondent in a hypothetical choice situa- ting, by direct responses from recreationists, tion, so there are serious doubts about wheth- the relative importances of various character- er the results will predict actual choices. istics, the overall "attractiveness" of an The "revealed preference" methods developed so area, or both. This class is characterized by far have modeled the relationships between its dependence on stated preferences (Sinden characteristics and observed choices without 1973, Cordell and James 1972, Shafer and specifying the underlying'causal phenomena. others 1969, Hoinville 1971, Juurand and In that sense, these models are incompletely others 1974). specified, so that predictions from them may be inaccurate.

The author is grateful to Dr. 1. S. The new method described in this paper, Ferguson, Dr. J. A. Sinden, Dr. C. Price, while belonging to the "revealed preference" Prof. J. A. C. Brown, Mr. J. B. Jack, and class, differs from its predecessors by incor- Dr. J. E. Opie for constructive criticism porat ing expl icitly the under1 ying causes of of an earlier draft. observed choices. Specifically, households are considered to make choices that maximize Consider a "representative" household expected satisfaction, subject to budget con- from this population center. A "household" straints. Thus, the method accords with ac- may, of course, comprise an individual, or a cepted utility theory, as expounded, for exam- family--the term is used to denote the unit ple, in Green (l972), with one important dif- that makes the choices. ference. Satisfaction is taken to be a function of recreation area characteristics, rath- The method makes three basic assumptions er than simply of the areas themselves. The about the behavior of a household: method, therefore, follows a development of accepted utility theory attributed most fre- 1. A household is well-informed about quentl y to Lancaster (1966). the characteristics of all the accessible areas, The basic elements of the new method will be presented and illustrated with simple nu- 2. A household allocates a part of its merical examples, with some suggestions for annual income to skiing trips dur-ing empirical e~timation.~The method has three the year, and basic elements: 3. A household chooses the area (or (i) Define the group of recreation areas areas) that provides the best skiing from which the household may choose, quality for the given budget--skiing once it has decided to make a recre- qua1 ity being some function of area ation trip. The group must include characteristics. all areas that are potential substitutes for the area chosen. To illustrate the analytical procedure, consider just one of the household's objec- i i) Isolate and measure the characteris- tives in going to a ski area for a weekend. tics likely to influence the house- Say this objective was to maximize the dis- hold's choices between members of tance covered on the snow, travelling downhill the group. at some given rate of descent. For this objective, several area characteristics would be ii) Estimate the expenditure of house- important, including snow quality, slope qual- holds on recreation trips, and re- ity, lift quality, and others. Taking just late this to the relative prefer- the first two to simplify the illustration, it ences for the characteristics. would be necessary to define the characteristics in terms of some measurable, physical at- The definition of the group of recreation tributes that governed the distance covered by areas is a well-known problem in recreation a skier travelling downhill at a given rate. and econometric studies. Hence, it is left For example, the steeper the slope, the shal- until last in the ensuing description, so that lower the traverses needed for a given rate of attention can be focused on the more innova- descent; so the greater the distance covered tive aspects of the present method. There- in one weekend, ceteris paribus. A similar fore, the description begins on the under- situation exists for "crisper" snow surface standing that, for a recreation area of par- conditions. Of course, allowances would have ticular concern to decisionmakers, the group to be made for other factors, such as icy con- of substitute areas has already been defined. d itions, or dangerous1 y steep slopes, affecting safety, but these need not affect the pre- THE "BEST" AREAS AT A GIVEN COST sent illustpation.

To illustrate the method, consider a hy- The two characteristics might be measured, pothetical problem and setting. Suppose the for each ski area on weekends during a given problem involves alterations to a particular snow season. The results for each area would ski area. For simp1 icity, assume only one be averaged to yield quantities for an average center of population is involved. The group weekend. If, for example, there were just of recreation areas to be considered, there- four ski areas within a weekend's travel1 ing fore, includes all ski areas accessible to time, then the average quantities might be: that population center, assuming for simplici- ty that these are the best substitutes for the area concerned. Area Snow Slope number : quality(~~) quaIity(~,) Empirical work on the model is under way at the Department of Forestry, Oxford University. 25 50 The financial support of the Forests Commis- 8 5 25 sion, Victoria, Australia, is gratefully 25 50 ac know ledged. 75 75 If =weekends were spent at a given because more of both characteristics could be area, then the quantity of each characteristic obtained, for the same budget, at any of the would be twice that shown in the tabulation other three areas, or some combinat ion of because the distance covered on the snow would them. The household was not obliged to choose be doubled. Thus, considering snow qua1 ity on just one of the available areas--it could make its own, the household would have to make a number of trips to more than one area, so three trips to area 1 in order to obtain the long as the budget was not exceeded. In this same quantity as could be obtained on just one case, the maximum quantities of characteris- trip to area 4. But the household must con- tics obtained would fall somewhere along the sider the relative quantities of both charac- straight lines 1-4, or 4-2, or both. Obvious- teristics on a11 the areas. ly, not fl points on these lines are obtainable, because for a single household, frac- In addition, the household has to con- tions of trips cannot be made. But this is sider the relative costs of the four alterna- not a constraint when large numbers of house- tives. Let these costs, including travelling, holds are considered. The 1 ines 1-4 and 4-2 accommodation and skiing, be, on average: represent all efficient combinations of trips to the ski areas; therefore, they constitute what has been cal led the "efficiency frontier" Area number: Average costs (Lancaster 1966). . They represent all the 'best" trips for the given budget. $ (PI The relevant characteristics for the group of areas can thus be identified empirical 1y. If, on adopting some set of characteristics, an area appears to be inefficient i .e., fa1 Is beneath the efficiency frontier), yet receives significant numbers of visitors from the given population center, then the If the household allocated, say $100 for characteristics must have been wrongly chosen, the year's skiing weekends, then the maximum or wrongly measured. Similarly, if an area number of trips to each area would be: appears efficient yet receives no visitors, then the same conclusion would be drawn.

Area number: Maximum Once the efficiency frontier has been de- weekends fined, the actual point on it the household would choose depends on the relative importance the household attaches to the two characteristics. Thus it is necessary to consider the household preferences.

HOUSEHOLD PREFERENCES

Household preferences can be represented by indifference curves which trace out the It follows that the maximum quantities trade-off the household is prepared to make of each characteristic that could be obtained between the characteristics. If, for example, at each area would be: snow quality is poor at a given area, the household might be prepared to select another area with better snow, even if the slope qual- Maximum quantities of ity is a bit worse. This selection can be il- Area lustrated by curves (fig. 2). Each curve number: traces out the combination of quantities of Zi and Z2 between which the household is indifferent. The curves further from the origin represent higher levels of util ity (i .e. "sat- 1 isfaction" or "preference1'); in fact they may 2 be best thought of as contours of a continuous 3 surface. From here on, the term "utility 4 function" will be used to describe this surface. The curves will be inclined more steep- ly toward one of the axes if the household has These results can be illustrated geomet- a stronger relative preference for the charac- rically (fig. 1). The initial choices that teristic identified with that axis than for face the household can be reduced from four to the other characteristic. three. Area 3 would be an inefficient choice In the extreme, if the household was in- An individual household would, therefore, terested in just one of the characteristics, never need to visit more than k areas, where k say Zl, then the utility functions contours is the relevant number of characteristics. would consist of straight lines, thus: But there were other households in the ncreasing population center, and since they may have had Ieve1 s different preferences, they must be considered. of utility z:u+.0 i COMMUNITY PREFERENCES Let the household just discussed by de- If the characteristics come into olay sequen- noted by Hi, and let there be just three other as suggested by Juurand and others households H2, H3, HI,. Consider each in turn. . the contours night appear as: Firstly, imagine H2 has the same annual budget for skiing weekends as HI $100, but it has a stronger preference than Hi for Zl relative to ZT This means that H2's utility function would be inclined more than Hi's towards the Zl axis (fig. 4). As a result of this incl i- nation, the point of tangency on the efficiency frontier is further towards the Zl axis. In fact, it is "tangent" directly upon point It is oossible to suoerimpose figure 2 on 4, implying that H2 devotes its whole budget figure 1, as shown in figure 3. to area 4 exclusively. Thus the optimum combination of trips for H2 is: The household nay seek to maximize the quantity of site characteristics for a oiven Area number: Optimum trips: Hz cost, irrespective of the number of trips. In resoect to the snow quality characteristic, 1 0 for examole, the household could obtain the 2 0 same quantity by taking either one weekend at 3 0 area 4, or three weekends at area 1. The 4 2 household might be indifferent between these two alternatives, ceteris oaribus, if the H3 has exactly the same preferences and budget overall costs of each were equal. The costs, as Hz, so that its optimum combination of of course, would have to include, in addition trips is the same. These can be added to to travel costs, the opportunity costs of yield: time, since one alternative takes up three times as rany weekends as the other. It is oossible to allow for this in empirical work Area number: Optimum trips: Hy + H3 (see, for example, Keith and Workman 1975, Beesley 1965). Under these conditions, the household's choice situation is i 1 lustrated in figure 3.

Under these circumstances, the household would reach its highest level of utility at just one point on the efficiency frontier. H4 has an even stronger preference for Zl This point, T, is where the utility function than for Z2, so that its utility function is is just tangent on the efficiency frontier. T tanqent even further towards the points 4 and falls exactly midway between points 1 and 4, 2. Hence its budget is divided equally be- which means that the household's budget is tween Zl axis: midway between areas 2 and 4. divided equally between areas 1 and 4. Thus, But H4's budget is $300, so that the original letting the household's budget be $100, the efficiency frontier (drawn for a budget of optimum combination of trips is: $100) is expanded by a scalar factor of 3 (f iq. 4). Thus H4's optimum combination of visits is: Area number: Optimum trips: household Hi Area number: Optimum visits: H4

1 0 2 3 3 0 4 3 increasing levals of utility

Efficiency Frontier ($100 budget) \ contours of equal levels of utility

3, ("Snow quality") J. Figure 1--Maximum quantities of two Figure 2--~amilyof indifference characteristics obtainable at four curves (the "utility function") of ski areas iith an annual budget of a household. $100. 300

Figure 3--Optimum choice for household H1 occurs at T, the point of tangency of its utility function on the Efficiency Frontier.

Z, ("Snow quality") The total number of visits for the whole com- rameters such that munity is:

HI H2 + '43 H4 Community - - The relative strength of preference for Zi is 2 0 0 2 indicated by the relative size of ai. -Thus it the example discussed previous1y (fig. 4), Hlls utility function might have been:

Thus, for a given efficiency frontier, a certain distribution of preferences coupled indicating that Hi preferred Z2 more strongly with a certain distribution of expenditures than Zl, in the sense that a 10 percent in- . result in a certain number of visits to the crease in Zl would produce a 2.5 percent in- areas concerned. The same number of visits crease in the utility level, whereas a 10 per- would have resulted if, instead of HI,, there cent increase in Z2 would produce a 7.5 per- had been two households with Hnls preferences, cent increase in the utility level. This forn each with a budget of $150. Similarly, if H3 for the utility function is chosen for mathe- had not existed, but Hz had allocated a budget matical convenience. It seems to imply cardi- of $200, the same results would have been ob- nal utility, though several transformations ta ined . (e.g., a logarithmic transformation) would maintain the same preference ordering, and do This example has illustrated the essen- equally well in the model, yet allow the util- tial components of the model of recreation- ity levels to vary monotonically. istsl choices. In a model of a real situation, there are many characteristics, many It can be seen, therefore, that the util- areas, and many population centers to be con- ity parameters provide a convenient numerical sidered, and the mathematics required uses the index of preferences. In the example with two techniques of non-l inear programing. But be- characteristics, it is necessary to specify fore the model can become useful to the aims only one of the parameters, say al, because set out in this study, it needs to be cali- the other is immediately defined by a2 = 1 - brated with data from actual recreation areas. ai. Thus the preferences of the households (H) discussed previously may be put down as:

CAL I BRATI NG THE MODEL

Of the model components described, all can be measured directly except the distributions of preference and expenditure. Yet Preferences (a ) (114 1 /2 3/41 these d istri but ions--or a comb inat ion of them 1 --are critical to the applications of the model. Therefore, they must be estimated, and The values of these parameters for the the most straightforward approach is to esti- various households are not known, however, and mate a function, F, that relates total expend- cannot be ascertained directly, except by iture (= number of households by average ex- using "stated preference" methods (e.g., pend iture) to preferences, both with respect Sinden 1973), which have a major weakness al- to the community concerned. To do this, some ready discussed. The alternative approach is numerical index of preferences must be speci- to estimate the parameters indirectly, using fied. This can be done by specifying the observed data on actual choices, by adopting utility functions of figure 2 mathematically. the following steps: One simple, convenient form would be: 1. Postulate values of the parameter a: for k discrete util ity functions that Similar calculations for the other utility are though likely to exist in the functions yield the results: community. For example, postulate where u denotes ut ility 1 eve1 , Zi denotes three utility functions, as follows: quantity of characteristic i, and ai are pa- Utility function number k

1 2 3 Such a model has been developed by the au- k thor for use on an ICL 1900 Computer. Parameter (a,) (114 1 /2 3/4) Figure 4~Utilityfunctions for four households, H , H,H, H, and their respective optimum choices.

utility function 1 Zi ("Snow quality")

Figure 5--After characteristic Z is altered at Area 1, the Efficiency Frontier near Point 1 swings outwards to Point 1'. The points of tangency of two of the three utility functions shift towards Point 1'.

utility function

I I I Oo 100 200 ("Snow quality") h he same ut ility functions as used previously t ion of preferences and expenditures. Thus, have been adopted for convenience. ) inserting (I) into (2), derive:

2. Determine the proportions of expendi- -ture, V , allocated to theT6 area by the households having the k'" u- tility function. This information is in which all variables are known except the derived from the points of tangency coefficients, bi, which can be estimated. of the efficiency frontier of the respective utility functions. Conven- To illustrate this estimation procedure, iently, havinq postulated utilty let the observed aggregate number of visits to functions that have already been used the respective areas from the given population in examples, these prooortions of ex- center be (as before): oend i ture (' ) have a1 ready been calculated previously: Area number: Observed visits: community

Area number: Utility function number (x)

Thus there will be three equations of type (3). Inserting values for ?" , and A, these 3. Postulate some 1 i kelv forr for the equations can be solved : function F. For the sinple examle in hand, the obvious choice is a linear function:

F=b0 +b,.\+E (1) k Thus F = kOOal where b are coefficients and E is the random error term. where F indicates the estimated value of F. (ln this example the estimate fits the data perfectly, so K = F precisely,with no random The linear form of F would imply that error.) there is greater expenditure (either more households, or more average f thus estimates the distribution of spending, or both) by people with expenditure as a function of preferences, as stronger preferences for characteris- it exists at the time of estimation. It is, tic 1- than for characteristic ZT therefore, a general statement about the com- h here is no a priori reason why the munity preferences in the population center; form would be 1 inear; it is used here no attempt is made here to associate F with purely for convenience of exposition.) particular households in terms of socio-economic characteristics (though these developments 4. Define the expressions giving the ex- might be undertaken in future work on the pected number of visits to each area: model 1.

Once F has been est imated, the model is fully calibrated and ready to be used for short-term forecasting. where xjt$enotes the number of visits to the j area, denotes the cost pJ SHORT-TERM FORECASTING of one weekend in the jth area, and the other terms are as defined above. When the characteristics of an area are This is simply a formal expression of the changed, the efficiency frontier will be al- statement made earl ier that a certain distri- tered in shape. If the community preferences, bution of preferences and expenditures implies represented by f, remain unchanged (and in the a certain number of visits to the respective short-term, unless there are marked changes areas. Thus, by corollary, a certain observed elsewhere in the economy, there is no reason number of visits implies a certain distribu- to think otherwise), then the community's utility functions will be tangent upon a new set of points in the efficiency frontier. As before, this iniplies a new set of numbers of visits to the areas in the group. The model, once calibrated can be used directly to forecast this new set of visits.

To illustrate this process, imagine that at area I-.,investments are made in slope- grooming and snow-making equipment, so that the average snow-quality rating, El, becomes 30, compared with 25 previously. Thus the maximum amount of this characteristic that can be had at area 1 for a budget of $100 becomes 120, as against 100 before. The new efficiency frontier becomes as illustrated in figure = 3 5, where the point 1' represents the altered site 1. The new points of tangency of the Thus the forecasted set of visits is: three discrete utility functions are also shown. The first two have points of tangency closer to 1' and further from 4 than they did Area number: Predicted visits: originally. The contours of each utility util ity function 2 function are taken to have constant slopes along all rays from the origin (i.e., they are homothetic). Hence, the new points of tangency are readily found, even though higher contours are reached in the process. The face 4-2 is unchanged, so that the point of tangency of the third utility function is not changed either. The new points of tangency Similar calculat ions for the other uti imply a new set of proportions of expenditure, functions yields the results: A', going to the several areas. Let this new set (using the same ordering as before) be: Util ity function Predicted visits: 1 2 3 community

he new matrix, A', has been simplified for This result, compared with the originally ob- illustrative purposes. The actual changes, served set of visits, shows that, as expected, for the data given, would be much smaller than area I is attracting customers away from area shown.) Then, inserting the new values from 4, while there is no change to the visits in \' into equation 3, the new set of visits for area 2. It will be apparent that, if the each postulated utility function can be calcu- change in area 1 had been large enough, area lated directly. For example, taking the sec- 2's customers would have been effected as ond postulated utility function: well. To test the usefulness of the model as a predictor, it will be necessary to obtain dat on variations in area characteristics and vis itor numbers and origins, by time series. These data are not now available in Australia, but in the meantime, it would be helpful to obtain short-term data (say, one season) and (Note again that f is predicting perfectly, in to use the model to identify the relevant a way that could not be expected in a real characteristics. Once identified, these data study.) could then be collected over longer periods for more complete assessment of the model. From equation 3; Finally, there is the problem of defining a separable group of areas, without which the model cannot operate. DEFINING THE GROUP OF RECREATION AREAS Grouping On Basis Of Demand Functions

An alternative method is to use data from A fundamental requirement for the con- a sample of households within the visitor struct ion and appl icat ion of the model is that catchment of an area. As before, it will be the expenditure by the community on the vari- necessary, for purely practical reasons, to ous areas in the group remains (effectively) specify a likely group of n areas, preferably constant throughout the (smal 1 ) changes that on the basis of their characteristics. The might be made in the characteristics of the essential data to be gathered ~il~~includethe area concerned. Essentially, this requirement number of trips xi, made to the i area in entails that when conditions in the group are the group, over a given period. In addition, altered, substitution occurs only within the the costs per trip, Pi would be estimated. qroup. Also, it entails that changes outside Household socio-economic characteristics, that the group have only the effect of possibly might effect demand, may be collected also. changing the budget allocated to the group. Hence, the definition of the grouo is a criti- The relationship between xi and the costs cal element in the modeling process; one that of all n areas can then be examined by multi- in practice rust precede the other work de- ple regression analysis. Thus: scribed so far. Two oossible aoproaches are now given for defining a group of recreation sites.

Grouping On Nature Of Characte-ist ics where f ( ) means "some function of." This function, having been estimated from household Lancaster (1971) has shown that if the data, is called a representative household de- characteristics of the area of particular con- mand function (Burt and Brewer 1971). Recrea- cern are very special and not easily obtain- tion areas with statistically significant co- able except in areas of similar nature, then efficients in this function can be regarded as such areas together are likely to form a grouo substitutes (if the coefficients are positive) susceotible to analysis. This condition need or complements (if negative) for x; . On this not ao~lyto fi characteristics, as lona as basis, the group can be defined. the "non-special" ones are much more readily available outside the group, than in it. On These grouping methods are, at best, only the basis of these criteria, ski areas cleaPly approximate, since households in the community form a 1 ikel y group, for one of the l i kel y are free to substitute whatever they like next characteristics (namely, snow-qua1 ity--a sur- best, when the conditions in a recreation area face suitable for skiing) can be obtained a change. Further, the focus above has been on in ski areas. Of the other likely character- the recreation areas, rather than on the indi- istics (say, slope quality, access, services, vidual households in the community. In other accommodation) most, probably, would be more recreation research into substitution, the fo- readily obtained in non-ski areas. cus has been more on identifying the clusters of household types that have similar substitution patterns (see, for example, Romsa 1973 Such a priori reasoning would need to be and Beaman 1975). supported by evidence, for example, from time series records of expenditure made on the group by households facing the same set of CONCLUSION costs (e.g., in one population center). If this expenditure remains fairly constant, The model can be compared with its re- after allowing for demand shifting factors lated predecessors in terms of three criteria: (such as changes in population size, per capi- (a) data requirements, (b) predictive power, ta real income) despite the normal yearly and (c) usefulness of output. The model re- fluctuations in snow quality at the various quires much the same kind of data used in areas, then there is some ground for believing other "revealed preference" models, but in that the group exists as defined. The evi- smaller amounts--especially area characteris- dence would imply that households allocate a tics. Unnecessary data can be identified be- similar amount each year to skiing, deploying fore collection by having an explicit under- this budget to their best advantage, depending standing of the way in which characteristics on the relative conditions at the various influence choices. This is not so true of the areas at the time of their trips. regression models (e.g., Lime 1971 measured on each campsite more than 70 variables, of which Usually, though, these time series data only 3 were found significant), nor of the are not available for the recreation sector, A.I.D. model, for which at least 2,000 obser- so that an alternative means of identifying vations are required (~esario1973). The the group would be required. 'stated preference" models require in addition to uncontrolled data on characteristics, a great deal of difficult interviewing. On presente 1;-dessus une nouvelle As to predictive power, the present model m6thode pour predire 1es changements 2 court is untried, as yet. But the other "revealed terme dans 1e nombre de visiteurs sur un preference" models, especially the regression certain terrain aprss I'introduction sur ce models, usually are shown to predict only terrain d'un changement dans la qua1 it6 de la within the data set used for their construc- recreation. La methode propose des moyens tion. Whether they will predict as well in pour determiner un groupe de terrains de new situations, when some of the unspecified recreation qui peut remplacer Ie terrain variables (such as those to do with prefer- duquel i l s'agi t. On pretend que "1es ences) are changed, is another matter. The recr6ationistes1' choisissent leurs trajets en ''stated preference" models, when tested as des terrains particuliers 5 cause des frais predictors against the regression models, have relatifs, des particuliarites des terrains et shown 1 ittle consistency (e.g., Lime 1971). de leurs propres preferences pour ces particuliarites. Les caracteristiques et la Finally, the output of predictive models distribution des preferences sont expliqu6es is most useful if there are pol icy imp1 ica- mathematiquement, pour laisser construire sur t ions, which demand that changes and potential modsle 1es choix des "retreat ionistes."' Apres changes in community welfare be estimated. avoir calibre le modsle par I 'emploi des However, often the output is simply given to a donnees observees dans Ie doma ine des decisionmaker in the hope that it will assist finances, de la caracteristique, et du nombre him in some way. The present model, however, des visiteurs aux terrains particulies du is superior in this respect to other models, groupe, on peut I'utiliser pour prevoir des as it can be used to forecast in quantifiable choix nouveaux aprss des changements de terms the social benefits that follow from caracteristique d'un terrain particulier du potential changes in recreat ion-area charac- groupe. La description est augmentee teristics. d'exemples numeriques simplifies.

ZUSAMMENFASSUNG LITERATURE C ITED

Eine neue Methode wird vorgestellt zur Beaman, J. Voraussage von kurzfristigem Wechsel in der 1975. Comments on the paper "The substitut- Anzahl der Besucher eines Gebietes auf Grund abi l i ty concept: imp1 icat ions for recre- einiger Veranderungen, die Qua1 itat der Frei- ation research and management" by Hendee zeitbeschaftigungen dort betreffend. Die and Burdge. J. Leis. Pes. 7(2) :146-152. Methode schlagt verschiedene MOgl ichkeiten vor, eine Gruppe von Freizeitgebieten zu Beesley, M. E. definieren, die als Ersatz fur das in Frage 1965. The value of time spent in travel- gestellte Gebiet dienen kOnnten. Es wird ing: some new evidence. Econornica 32: behauptet, dass Erholungssuchende wahlen, zu 174-185. bestimmten Gebieten innerhalb dieser Gruppe zu fahren, mit Riicksicht auf die relativen Kosten Burt, 0. R., and D. Brewer und die Qualitatseigenschaften dieser Gebiete, 1971. Estimation of net social benefits sowie mit Rucksicht auf ihre eigenen spezi- from outdoor recreat ion. Econometrics ellen Wunsche diese Eigenschaften betreffend. 39(5) :813-827. Beides, die Eigenschaften sowie die Verteilung der Eigenschaften innerhalb der Gemeinde Cesario, F. J. werden mathematisch beschrieben, so dass es 1973. Estimating attractiveness factors in n%gIich ist, ein Model1 der Wiinsche der spatial interaction modeling. Paper pre- Erholungssuchenden aufzustellen. Sobald ein sented to 3rd North East Regional Science Model1 mit Hilfe der Kostenanschlage, Association Meetings, April 27-29, 1973. Qua1 itatseigenschaf ten und Anzahl von Be- suchern in den Gebieten innerhalb einer Gruppe Cheung, H. K. kal ibriert worden ist, kann es dazu verwendet 1972. A day-use park visitation model. J. werden, neue WLinsche vorauszubestimmen, die Leis. Res. 4:139-156. auf Grund einer Veranderung der Eigenschaften eines beliebigen Gebietes innerhalb der Gruppe Cordel 1, H. K., and G. A. James eintreten k6nnen. Die Methode wird an Hand 1972. Visitors' preferences for certain einfacher numerischer Beispiele erlautert. physical characteristics of developed ampsites. USDA Forest Serv. Res. Paper Columbia Studies in Economics 5. 177 p. SE-100, 21 p. Southeastern Forest Exp. Columbia Univ. Press, N.Y. Stn., Asheville, N.C. Lime, D. W. Green, H. A. J. 1971. Factors influencing campground use in 1972. Consumer theory. 344 p. Penguin the Superior National Forest of Minne- Books Inc., Baltimore, MD. sota. USDA Forest Serv. Res. Paper NC-60, 18 p. North Central Forest Exp. Hoinvi Ile, G. Stn., St. Paul, Minn. 1971. Evaluating community preferences. Environ. and Plann. 3:33-50. Romsa, G. H. 1973. A method of deriving outdoor recrea- Holman, Mary A., and J. T. Bennett tional activity packages. J. Leis. Res. 973. Determinants of use of water-based 5 (4) :34-46. recreational facilities. Water Resour. Res. 9(5):1208-1218. Seneca, J. J., and C. J. Cicchetti 1969. User-response in outdoor recreation: Johnston, W. E., and G. H. Elsner a production analysis. J. Leis. Res. 1: 1972. Variability in use among ski areas: 238-245. a statistical study of the California market region. J. Leis. Res. 4(1):43-49. Shafer, Jr., E. L., J. F. Hamilton, and E. A Schmidt Juurand, P., V. Guzel imian, and J. Beaman 1969. Natural landscape preferences: a 974. Perception of quality of wild rivers. predictive model. J. Leis. Res. 1 :1-19 Canada National and Historic Parks Branch C.O.R.D. Study Tech. Note 27, 69 p. Dep. of Indian and Northern Affairs. Ottawa. Shafer, Jr., E. L., and R. C. Thompson 1968. Models that describe use of Adiron- Keith, J. E., and J. P. Workman dack campgrounds. For. Sci. I4 (4) :383- 1975. Opportunity cost of time in demand 391. estimates for non-market resources. J Leis. Res. 7(2):121-127. Sinden, J. A. 1973. Utility analysis in the valuation o extra-market benefits with particular Lancaster, K. J. reference to water-based recreation. 966. Change and innovation in the technol- Water Resources Research Inst. Res. Rep. ogy of consumption. Amer. Econ. Review WRRI-17, 136 p. Oregon State Univ., 56 (2) :14-23. Corval l is. (PB-231 701) Lancaster, K. J. 971. Consumer demand: a new approach. On the Use of Home and Site Surveys in Recreation Research

Mordechai Shechter

Abstract--The pros and cons of home interviews versus on-site user surveys are briefly reviewed. The complementary relationship between them is illustrated by the findings of an extensive study on the demand for outdoor recreation at Mt. Carmel National Park, the largest national park in Israel. The need for expensive home interviews can be greatly reduced when the less-expensive, site surveys are properly constructed and implemented.

Outdoor recreation services are resource- evidence in which for example, a vinyl thread based and as such can be used only at the is used as an indicator of use intensity. If point where the service is "produced" and Dro- placed near an area which draws relatively vided. Market research designed to gage con- large crowds, it of course wears faster than sumer oreferences, consur- tio on patterns. and if placed somewhere else. Mechanical and other demand chayacterist ics should reccgrize electronic devices also serve as use intensity the difference between these services ard indicators. Here we ~.ay1 ist car detectors other consumption goods and services for which which record the nurber of vehicles entering that constraint is less restrictive, or nonex- and leaving an area, ad counts based on istent. Furthermore, unlike many other sen1- aerial photographs. Finally, we should men- ices, includiqg other recreation services, the t ion the method of direct cbservat ion in which consumption unit is usually a ~rou~--~amily, an observer records the behavior of recrea- friends, oeers. The type, duration and inten- tionists. sity of outdoor recreation activities are therefore based on the interplay of desires The other methods for obtaining inforira- and preferences of the individuals making uo tion on consum~tionoatterns and overall de- the group, their intra-group roles, and their -nand for outdoor rec-eaticn are al l based on physical inclinations or limitations. various forms of rarket surveys. There are several means fc- obtaining the information Because of these characteristics, several through such a survey: self-adrr,inistered approaches have been developed to measure the questionnaires, tel eohone interviews, and per- demand for these services, either for the our- sonal interviews. The latter is the most pose of estimating the number of users--present widely used; it assures the highest rate of and future--or in terms of the range of recre- response, and a1 lows a more in-depth analysis. ational activities engaged in at recreation I shall restrict my discussion to this type of sites. Using these approaches, one can dis- survey and concentrate on the two possible tinguish between behavior-oriented methods tactical approaches to the conduct of the which focus on observing recreationists' activ- interview--home versus in-site interview. ities at the site, and the more customary mar- The former is conducted at the intervieweebs ket research methods of interviewing people at abode, while the latter is conducted at the home. recreation site.

This paper seeks to show that out of the The Home Interview more commonly used market survey approach, the less expensive, on-site interview, might be The interview at home allows for a thor- just as efficient in gathering pertinent data ough interview. Since it is usually based on as the more comprehensive home interview. and a representative sample drawn from the entire in many cases the latter can be satisfactorily population of a given geographical area (city, replaced by the former. A brief description region or country), it naturally covers visi- of various data-gathering techniques is in- tors as well as individuals who do not partic- cluded. Two surveys conducted in conjunction ipate in any recreational activity, or seldom with a demand study of a large Israeli recrea- do. It therefore enables the investigator to tion park and a comparison of the two on the analyse factors which affect the level of par- basis of a statistical analysis are described. ticipation in any population group. Its main disadvantage is the need to base responses on SURVEY AND NON-SURVEY TECHNIQUES recall rather than immediate experience. Home interviews are much more expensive than site Of the non-survey group of techniques we interviews for two reasons: (a) since the pop- could briefly mention the method of physical ulation on which the home interview is based is the origin population for an entire system of the study, 1938 households were interviewed, recreation sites, it is generally larger than representing a 0.5 percent sample of the urban the population of park visitors from which the population in Israel. In each household, one on-site sample is drawn. Hence, home survey of the adults 20 years or older was randomly samples should usually be larger for the pur- selected for the interview. In most cases, pose of obtaining estimates with a given con- this was the head of the family or his spouse. fidence interval (assuming the degree of var- The interview was conducted in the interview- iation in both populations is more-or-less ee's home. equal); (b) the amount of interviewer's time per interviewee, including the time required The on-site interview was conducted on to reach the latter's residence, is larger. three representative dates: a mild winter In the site interview, visitors are concentra- weekend, a spring weekend, and a non-re1 igious ted at the site and can be more easily reached. hol iday (Independence Day). Each date was considered a stratum, and the sampl ing rate in The Site Interview each stratum was 2 percent. Altogether, 845 recreationists were interviewed, each inter- The main advantage of the site interview viewee representing a recreation unit, i.e., is the immediacy of recall. It usually en- visitors who arrived to the site as a group ables a more comprehensive coverage of the var- (e.g., a family). The interviewee was select- ious activities in which recreationists engage. ed from among the adult members of the group. Its principal disadvantages are the relatively The interviewers were stationed in the parking short time available for conducting the inter- lots of the major attraction sites in the view, and the often encountered difficulty in park. In addition to the survey, throughout eliciting the cooperation of the recreationist. the day a count was made of all vehicles Obviously, such a survey relates only to that entering the park at its four access points in group in the total population which actually 30-minute intervals. visits the site. Hence, the two surveys com- plement one another to a great extent--if one The two questionnaires contained as far has the means to conduct both. This is rarely as possible ,dentical questions. However, the the case, nor is it really necessary as I now home interview included questions aimed at the wish to demonstrate. Although the aim of the sub-population which had not visited the park study described below was not to compare the during the 12 months prior to the survey, and two methods, I wish to draw on our experience included more detailed questions dealing with in this regard, and arrive at certain conclu- the demographic and soc io-economic background sions which might be of some help in future of the interviewee and the household. The investigations in this field. site questionnaire, on the other hand, was more detailed in respect to quest ions deal ing with recreational patterns within the park THE MT. CARMEL CASE STUDY boundaries. Since the park offers no camping facilities at the moment, all visits started In 1972 the Technion's Center for Urban and ended within a period of about 12 hours. and Regional Studies conducted two extensive surveys aimed at forecasting the demand for RESULTS outdoor recreation activities at Mt. Carmel National Park (~nisand others 1974). Mt. The two surveys were compared on the basis Carmel National Park is the largest national of the estimates yielded by each for the fol- park in Israel, located in the central and lowing key variables: (a) total visitor-days, southern sections of Mt. Carmel, at the north- (b) frequency of visits, (c) range of recrea- ern part of Israel, extending over some 20,000 tional activities, and (d) measures of recrea- acres. The study was also aimed at deter- tion benefits. mining the recreational capacity of the park, and the formulation of a planning pol icy for Visi tor-Days the future development of the area. Both interviews enabled us to derive di- Except in an extensive study by Coppock rect estimates of total visitor-days (or, (1971), we had not come across a study which alternatively, visit rates) broken down by attempted to compare the two interview method- origin, socio-economic characteristics, season ologies on a basis of actual large-scale inves- of the year in which trips are concentrated, t igat ions. Even in the aforementioned study, and so on. In the home interview, however, the analysis was mostly descriptive and the these estimates are calculated on the basis of role of statistical analysis rather limited. the information contained in the sample *,

The home interview was conducted in 1972 Observant Jews do not travel on the Sabbath and concerned recreation activities of the and religious holidays and thus are not repre- population during the preceding 12 months. In sented in surveys conducted on such dates. and--apart from statistical error--they con- Table I--Effect of socio-economic factors on stitute reliable estimates of the relevant var- the frequency of visits, Mt. Carmel iable. Using the sample information on the National Park, Israel number of visits per recreation unit and the size of each unit (usually a household), I estimated the total number of visitor-days Home between 800,000 and 1,070,000 during the Factors 1971-2 season. Origin ++ ++ (place of residence) The site-survey related estimates were based on counts of vehicles entering the site, Country of birth 0 0 and the average number of occupants per car. In addition to the counts on representative Age of youngest chi 1 d 0 dates, it was necessary to allocate the entire recreation period to each of the categories Age of interviewee 0 0 represented by these dates. This was done on the basis of the weather conditions prevailing Car ownership 0 + during the season, as well as factors such as school vacations and religious holidays. Cal- Occupation of head 0 0 culations yielded an estimate of 836,000 of household visitor-days, which falls within the interval of the home-survey estimate, albeit very close to the lower level. The site-survey estimate involves a number of supporting assumptions 0 = there does not exist a significant which are not required in the case of the home relationship between the .ar&bles, survey. In spite of this, however, the site i .e., a > 0.05. survey yielded an independent estimate which would serve satisfactorily in most applications. + = a < 0.05 and V < 0.125.

Frequency of Visits ++ = a < 0.05 and V > 0.125.

The two surveys showed the effect of var- variable (frequency of visits) and the inde- ious socio-economic factors on the number of pendent variable (travel time) is a convenient visits per recreation unit per year (table 1). standardized expression for comparing the two The statistical tests for the significance, as models. If the dependent variable is denoted well as the intensity, of the relationships be- Y, and the independent variable X, exponential tween the variables were x2 and Cramer's V, model becomes respectively. The level of significance for the x2 test was a = 0.05, and the critical value for Cramer's V was set at 0.125. Since V lies between 0 and 1, a lower value of V indicates a weak relationship, and a value closer and linear model becomes to 1 a strong one l la lock 1972).

The results of the two surveys are quite similar qua1 itatively. The only discrepancy exists with respect to car ownership, but this is explained by the fact that the population when Y is evaluated at the mean (7 = 2.0) of of the Mt. Carmel Park visitors is much more the sample. The results are strikingly simi- homogeneous with respect to car ownership than lar, again~recallingthe independence of the the corresponding population in the home two data bases~substantiatingour confidence interview. in the site data.

A regression analysis of the frequency of Range of Recreational Activities visits data yielded similar results. The most significant factor affecting this variable was A high degree of correspondence was found travel time. Since the mathematical forms between the results of the two surveys with underlying these regressions were different--a respect to the popularity of various recrea- linear form in the case of the home survey and tional activities (table 2). an exponential one in the case of the site survey--the coefficients of the travel time Measures of Recreation Benefits factor are not directly comparable. However, the elasticity (denoted by E) of the dependent Various methods for estimating recreation Table ?--Percentage of respondents who ind i- Table 3--Cumulative percentages of respondents cated a specific activity was the who indicated wi 1 1ingness-to-pay for main one in which they engaged during recreational services, by measure of their visit to Mt. Carmel National willingness, Mt. Carmel National Park, Park, Israel. Israel.

In monetary Site Home Activity va luesl survey survey

Picnic - cumulative percentages -

Walking for pleasure

Driving for pleasure

Hi king

Sports, games

Photographing, pa int

Other

In terms of additional driving time (minutes)

60 benefits have been proposed and appl ied (see, for example, Burton and Fulcher 1968). Brief- ly, there are essentially two methods for estimating recreation benefits. One is the Hotelling-Clawson approach in which we derive a demand schedule for the site, based on visit rates from several origins. From this schedule we calculate the consumers' surplus, which is a monetary measure of recreational benefits. in terms of cumulative percentages for the ob- Alternatively, we approach recreat ionists d i- vious reason that a person who is willing to rectly, and elicit from them their own valu- pay a certain fee will certainly be willing to ation of benefits in terms of their will ing- pay a lower fee for the same service (table 3). ness to pay for the recreational services Needless to say that aggregate benefit mea- involved. Theoretically, both methods should sures based on these raw data (of course, yield identical values. Also, both can be after rather elaborate analysis) also fell used in conjunction with either survey. The within the same order of magnitude. results of the wi1 1 ingness-to-pay approach is first described. Of course, the site survey would not yield any data on the willingness to pay of In each survey we queried the interview- those who had not visited the park. If these ees on their willingness to pay for recrea- people are willing in principle to pay some- tional services in terms of an entrance fee thing as an "option value," the total benefit (based on differential admission fees for values for the *would change accordingly. adults and children), and--mainly for the pur- CONCLUSIONS pose of a consistency test--in terms of their willingness to travel an additional distance The results suggest that a decision in to reach the park (because the authorities favor of a site interview can often be an ef- have decided to close an entrance). The re- f icient solution (in terms of the information sponses in both surveys~includingthe time obtained per do1 lar expenditures) and not just and money questions in each survey--were quite a result of limited budget and a "better than close, in spite of the emotional load inherent nothing" alternative. in such quest ions. (present1y, no entrance fees are charged.) The results are reported Much information is obtained through home surveys based on representative samples of the (entrance fees, in IL.) entire population in comparison with the data obtained through on-site surveys of visitors LITERATURE C ITED only. We think, though, that a comprehensive home survey on the consumption of outdoor rec- Blalock, Hubert M. reation services should be conducted periodi- 1972. Social Statistics. 583 p. McGraw- cally, say every 10 years, preferably attached Hill, Kogakusha, Tokyo. to some national survey of households. Such a tool, however, is not recommended for individ- Burton, T. L., and M. N. Fulcher ual sites, existing or planned, where the-much 1968. Measurement of recreational benefits cheaper site interview would yield data of --a survey. J. Econ. Studies 3:35-48. sufficient quality and informational content. These data, combined with that of the house- Coppock, I . T. hold survey, should suffice for forecasts of 1971. Out in the country. Town and Country demand, use patterns, benefit estimates, and Planning 5:253-257. so on. It is, of course, imperative that the questionnaires of both surveys be consistent, Enis, R., M. Shechter, and M. Baron whether with respect to content or background 1974. The demand for outdoor recreation at questions. Mt. Carmel National Park. 266 p. Centet for Urban and Regional Studies, Technion Acknowledgment: I thank Mrs. M. Baron, --Israel Institute of Technology, Haifa, M.Sc., for her help in preparing this paper. Israel. (In ~ebrew).

Relative Value of Selected Outdoor Recreation Activity Areas

Joseph E. Hoffman, Jr

Abstract~Aproblem for the public sector is deciding how their limited budgets should be spent to provide for which outdoor recreation areas. Since the market system is not available to allocate resources efficiently, some other method of determining relative value and allocating efficiency is needed. In this study visitors were asked how much they would be willing to pay to use activity areas, such as campground or swimming beach, within Wil- low River State Park, Wisconsin. A random sample of visitors provided a measure of the relative value of each activity area. Gen- erally, the more developed areas, such as the campground, were considered the more valuable areas. The values were then compared against costs (value/cost) and an efficiency index was developed. This was done for both-annual maintenance-operation costs and total annual costs (maintenance-operat ion costs plus 1/20 of development costs at 5 percent interest). The results show that the most efficient areas are the least developed ones and the least efficient ones are the most costly to provide. Either the agency has over estimated the value visitors receive from the more expensive zones or visitors do not appreciate the cost of providing them.

The publ ic sector is responsible for pro- The purpose of this study was to provide viding many of the outdoor recreation opportu- for more effective and efficient planning and nities available to the publ ic. The problem management of a state park. The specific ob- is decid ing how 1 imi ted budgets should be jectives were first, to develop and use a spent to provide for which outdoor recreation method of measuring the relative value of rec- experiences. Since the market system is not reation activity areas within a park (to visi- available to allocate resources efficiently, tors), and second, to relate these values with some other method of determining relative the costs of providing these areas. The re- value is needed. sulting efficiency indexes, relating value to cost, should indicate which activity areas are newest state parks, having opened for visitors more or less efficient in providing for pub1 ic in 1971. enjoyment. Willow River State Park was separated in- EXPERIMENTAL METHODS to nine activity areas or zones (fig. 1). An activity area is a geographic area wherein The method developed to obtain visitor certain activities are more likely to occur. values of activity areas was willingness to The nine activity areas are: campground (72 pay. Visitors were asked to indicate how much units); swimming beach; picnic area and dam they would be willing to pay to use each ac- near beach; picnic area near boat landing; tivity area. This expression provides a meas- Little Falls Lake and boat landing; Willow ure of relative value of each zone. Only a River below Little Falls Lake (river corridor); few studies have used the willingness-to-pay north side of park (primitive area); south approach and these studies were essentially side of park (primitive area); and any other for an entire park and not areas within a park part of the park (not shown on map; the area (Davis 1963, Glascock and Born 1971, Manning upstream from the lake was accessible from 1973, Randall and others 1974, Romm 1969, public highways and users were not required to S inden 1974) . go through the entrance station).

All v isitors entering the state park on RESULTS 12 random1 y selected days were given questionnaires and asked to return the completed form The most valuable areas are the most de- to the entrance station when they left. The veloped zones, and the least valuable tend to rate of return on completed questionnaires was be the least developed. The campground and 51 .5 percent. swimming beach were the most valuable areas.

The park studied was the 2520-acre Willow In 1974, the development costs for the River State Park in west central Wisconsin, park were approximately $710,000 while the U.S.A. It is oriented around a reservoir maintenance-operation budget was $60,000. To- formed by a dam originally built for hydro- tal annual costs include maintenance-operation electric purposes. It is one of Wisconsin's

WISCONSIN

NORTH SIDE OF PARK

LITTLE FALLS LAKE ZONE H ZONE E

SOUTH SIDE OF PARK

SCALE 0'- 1000' 2000' Figure 1--Willow River State Park in Wisconsin was separated

into nine activity areas or zone for study. Table I--Eff iciency indexes relating wil l'ingness-to-pay values to maintenance and operat ions costs, Willow River State Park, Wisconsin, summer, 1974, by activity area.

Sum of Maintenance- Activity Efficiency do1 lar operat ion Efficiency area indexes rank va Iuesl costs2 1

Campground

Swimming beach 6

Picnic area and (low) dam near beach 9

Picnic area near 8 boat landing

Little Falls Lake 3 and boat landing

Wi1 low River below 1 (high) Little Falls Lake

North side of park

South side of park

Any other part of park

Based on 353 questionnaires and dollar values visitors said they were willing to pay.

Overhead costs allocated back to each area based on the percent of development costs directly attributable to each zone.

Value/cost times 100.

costs plus annual development costs based on a cat ion of the resources at their disposal. 20-year 1 ife and 5 percent interest. The The indexes indicate where resources could campground, swimming beach and the two picnic either be shifted between or withheld from ac- areas are the most costly areas to provide. tivity areas or both to equalize the value/ The river corridor and primitive areas are the cost ratios. For the park manager the main- least costly to provide (tables 1, 2). tenance-operation cost indexes would be more useful, while the total annual cost indexes The two efficiency indexes measured value would be more useful to the park planner. over both maintenance-operations costs and total annual costs (tables 1, 2). The two pic- An example of how management could use nic areas were the least efficient for the efficiency indexes based on value/maintenance- maintenance-operation cost indexes, while the operation costs is illustrated here. The most swimming beach was the least efficient for coefficient zone is Willow River below Little tal annual cost efficiency indexes. The most Falls Lake with a ratio of 6.99. If this zone efficient area on both sets of indexes was the is adopted as the standard for efficiency, Willow River below Little Falls Lake area. then the maintenance-operation costs of the other activity areas should be reduced until DISCUSS ION their efficiency ratio equals 6.99. If this The purpose of developing efficiency in- was (or could) be done, then the maintenance- dexes is to guide decisionmakers in the allo- operations budget required would only be Table 2--Efficiency indexes relating willingness-to-pay values to annual and development costs, Willow Rtver State Park, Wisconsin, summer, 1974, by activity area.

Annua 1 Total Activity Development Efficiency development a nnua l Efficiency area costs1 ndexes4 rank costs2 cos ts3 I

Campground

Swimming beach

Picnic area and dam near beach

Picnic area near boat landing

Little Falls Lake and boat landing

Willow River below Little Fa!Is Lake

North side of park

South side of park

Any other Dart of park

Overhead costs al located back to each zone based on the percent of developmen t costs directly attributable to each zone.

Based on 20-year life span; and capital recovery rru Itiplier = [i (1 + ;)"I/[( I + i)- I], and 5 percent interest.

Annual development costs and maintenance-operations costs.

Value/cost times 100.

$29,899, a saving of $30,101 over the 1974 may be encouraged to start using areas or may budget. derive even higher values. But this change Another approach would be to shift funds may result in a new mix of values being given, from the more efficient areas to less effi- and the original value measurements may no cient areas. The average maintenance-opera- longer be valid. tion efficiency index was 3.48. All areas with ratios higher than this would lose funds CONCLUSIONS to the areas with efficiency indexes lower than the average. This approach would allow The most efficient areas are usually the for improving the lower ranked areas at the least developed ones, especially when total expense of the higher ranked areas. annual costs are used in determining the efficiency ratios. This relationship indicates Decisionmakers must real ize that any bud- that visitors do not closely relate value with get allocation change may result in a differ- cost. Either the agency has overestimated the ent opportunity being provided. Some visitors value visitors receive from the more expensive may become discouraged and either derive lower zones or visitors do not appreciate the cost values or stop using activity areas. Others of providing them. This method, however, is not without its value for the water-based recreation possible 1imitations. The major one is the resources at Lake MacBride State Park in assumptton that the relative willingness-to- the summer of 1970. 161 p. Iowa State pay values expressed by visitors is a val id Water Resources Research Institute, Univ. measure. To verify this assumption would re- of Iowa, Iowa City. quire an experiment in which user fees could be established for each activity area. Then Manning, G. H. use could be measured at the rate of different 1973. Subjective evaluation of recreation user fees. sites. Forest Econ. Res. Inst. Info. Rep. E-X-20, 12 p. Forestry Serv., LITERATURE C ITED Ottawa, Ontario.

Randall, A., B. C. Ives, and C. Eastman Davis, R. K. 1974. Benefits of abating aesthetic envi- 1963. The value of outdoor recreation: an ronmental damage from the Four Corners economic study of the Maine woods. Ph.D. Power Plant, Fruitland, New Mexico. New thesis, Harvard Univ., Cambridge, Mass. Mexico State Univ. Agric. Exp. Stn., Bull. 618, 40 p. Davis, R. K. 1964. Value of big game hunting in a pri- Romm, J . vate forest. p. 393-403. In Trans., 1969. The value of reservoir recreation. 29th North American Wildlife and Natural Cornell University Water Resources and Resources Conf ., Las Vegas, Nevada, March Marine Sciences Center Tech. Rep. 19, 102 9-11, 1964. James B. Trefethen, editor. p. Ithaca, N.Y. Wildlife Management Institute, Washing- ton, D. C. Sinden, J. A. 1974. A utility approach to the valuation Glascock, M., and J. J. Born of recreational and aesthetic experi- 1971. Determining the demand and economic ences. Am. J. Aqric. Econ. 560) :61-72.

A Recreational Visitor Travel Simulation Model as Aid to Management Planning Robert C. Lucas Mordechai shechterl

Abstract~Asimulation model for dispersed recreation areas has been developed that provides a means for experimenting with modifications of use or area conditions to determine effects on use patterns and encounters between visitor groups. The model, the results of a test of it, and potential future applications are discussed.

Numbers of people visiting most kinds of resource damage have been difficult for manag- outdoor recreation areas continue to grow. ers to solve. These problems have been par- This growth often creates problems for manage- ticularly severe on lands establ ished as wil- ment, with the nature of the problems depend- derness. A wilderness, by law, is to be man- ing on the type of area and the management ob- aged to permit natural ecological processes to jectives established for it. In the United operate without alteration by man and also to States, growth in use of dispersed recreation provide visitors with "outstanding opportuni- areas has been rapid (Lloyd and Fischer 1972) ties for so1 itude." and the resulting problems of congestion and Growth in the number of visits to wilderness increased about 15-fold from 1950 to 1975, Dr. Shechter was affiliated with the Re- threatening both natural ecosystems and the sources for the Future, Inc., of Washington, experience of solitude. Poor distribution of D.C., at the time work described in this paper use, both in space and time, is common and ac- was done. centuates the pfoblems of congestion and eco- system damage. Studies of the distribution of structure of modern urban life to the wilder- wilderness use show that very uneven use pat- ness setting intended to offer release from terns are the general rule; use is heavily civilization's pressures. In general, research concentrated on certain portions of each area, has indicated that a desire to escape civili- while larger portions receive little use. zation is a major motivation for wilderness Similarly, a few summer weekends usually expe- visits. Furthermore, most visitors feel as- rience sharp peaks in use. Redistribution of signing itineraries is a highly undesirable some use seems to offer considerable hope for approach to use control (Stankey 1973). reducing the adverse effects of heavy use. If use pressures and encounters resulting Research has shown that visitor satisfac- from any given use level and pattern cannot be tion is influenced substantially by the types predicted, experimentat ion through trial-and- of encounters with other visitors and that error is an apparent alternative. However, visitors report strong preferences for low trial-and-error is not an effective approach. levels of encounters (stankey 1973). There- It is very time consuming; managers would have fore, managers of wildernesses receiving heavy to try a pol icy for a year or more to see how use are beginning to take actions to modify or it worked. Results for any one year could be control use. In the United States, both the heavily influenced by uncontrolled outside National Parks and the National Forests are factors, such as weather. Detailed informa- rationing use of some areas. In some cases, tion on use patterns and encounters would be this is done by limiting the numbers of visi- available only if special studies monitored tors permitted to enter each day at various the area. It would not always be possible to access points. In other areas, managers set create the use pattern the managers desired to nightly capacities for a11 camping areas and test. For example, if managers wanted to know require visitors to establish and adhere to the effect of a doubling in use, there proba- rigid itineraries that will not result in the bly would be no practical way to cause this campsite capacities being exceeded. If this much use in the short run. At least three restriction is impossible, the party is not sorts of high costs could also result from a permitted to visit the area at that time. In trial-and-error approach to use management. a few other areas, managers have attempted to First, serious long lasting or even irreversi- influence visitors to voluntarily shift their ble damage to resources might result from use to other areas or times through educat ion- tests of heavy use. Second, many visitor ben- a1 pamphlets and personal contacts. efits could be sacrificed, either through testinq excessive use levels that seriously However, all the managerial actions, ex- reduced the quality of visitors' experiences cept the establishment of rigid itineraries or through testing low levels of use that re- (which have other problems discussed below), sulted in many oeople being denied entrance. suffer from a major flaw. The manager's ob- Finally, frequent, major changes in use poli- jective is to reduce use at overused locations, cies could lead to controversy and severe pub- and to avoid excessive levels of various types 1 ic relations problems. of encounters (on trails, at campsites, etc.). However, there has been no way to relate chang- Systems that are too complex for analytic es in total use or redistribution of use to the solutions and not suited to real-world experi- number of encounters per party or to the amount mentation are often approached by simulation of use of particular places within a wilder- model ing (~hechter1971). Therefore, a wi l- ness. The complexity of travel routes, which derness travel simulation model was developed characteristically overlap and intertwine, and to provide a better way to formulate and eval- the variability in travel decisions are so uate use management policies. The simulation great that neither intuition nor analytic model provides a practical way to test use solutions are useful predictors of the varia- patterns quickly. Variability in visitor be- bles of interest for a given amount of use. havior is incorporated in the model, but, in just a few minutes, use can be simulated for The rigid itineraries do provide a more an entire season or a number of seasons. The determinate result, at least for use of key model records and displays in appropriate for- locations and encounters between camping par- mats a1 1 the desired information on use and ties, but not for encounters between parties encounters. Because the experimentation takes while traveling on the trails. For many rea- place in the computer instead of the real sons (weather, illness, over-ambitious plan- world, we avoid the high social costs. Even ning, etc.), not a11 parties adhere to their the most extreme patterns can be tested with- itinerary, so results are not as determinate out damage to precious resources. as they seem. More important, such close control of movements seems to detract from the Travel simulation models are common, but visitors' experiences of adventure, explora- the requirements for the wilderness model were tion, and spontaneity, and to transplant the quite different. In particular, the interest in encounters was unique. Therefore, the application and by boat type in another). Ar- United States Department of Agriculture, For- rival timing patterns, travel speed, etc., can est Service, entered into a cooperative re- vary depending on the type of party. search agreement with Resources for the Future Inc., complementing ongoing research at RFF, 3. User-route interactions--Route selec- to develop a general use simulation model for tion can vary between party types, as can wilderness-type areas. Resources for the Fu- travel time in each direction over different ture involved specialists from International trail segments. Business Machines, Inc. in the project.2 The model has been developed, modified and re- 4. User-user interactions--These are the fined, and has been field tested. three types of encounters described above.

This paper describes the model and re- To make the model operational, data are sults of the field tests, and presents conclu- needed on the area and its use. The travel sions about future applications. network must be known, and something about how different types of visitors behave within it-- SIMULATION MODEL their patterns of arrival, various routes followed and relative popularity of each, travel All simulation models are simplified ab- speeds, and so on. This information is sup- stract tons of complex, real-world processes. plied to the model in probabilistic terms. However, the wilderness travel simulator quite real ist ical ly embodies the main characteristics The simulator provides detailed output of wilderness visitor movements and interac- information for each individual simulation of t ions. a particular use situation or "scenario." Since part of the input data is of a probabi- The computer program for the model gener- listic nature, the model has the facility of ates data on visiting parties who arrive at the producing summaries of a series of replica- area at various simulated dates and clock tions of any such "scenario," providing aver- times, enter at particular access points, se- age values of various performance measures, lect routes of travel, and move along these such as the amount, character, distribution, routes. The simulated parties may overtake and and timing of use. For example, the number of pass slower parties moving in the same direc- parties of each type using each trail segment tion (overtaking encounters), pass parties mov- is provided (if desired, even on a daily basis ing in the opposite direction (meeting encoun- in one of the three versions of the model). ters), or pass by parties camped in areas visi- Additional information is available on the ble from trails or other travel routes, such as number of encounters by type of encounter, by rivers (visual encounters). Parties that stay type of party (classified by.rnode of travel or overnight select campsites which they may or by length of stay), and by individual trail may not share with other camping parties (camp segments and campsites (again, in one version, encounters are recorded when they occur). On on a day-to-day bas is). an ensuing day, camping parties leave the campsite and continue on their chosen routes, The model is coded in the IBM-originated and eventually leave the area. language GPSS (General Purpose Simulation Sys- tem), version V. The model to date had been The model consists of four important com- successfully operated on IBM's 360 and 370 se- ponents : ries of computers as well as Control Data Cor- poration's 6600 computer. A version adapted 1 Route network--This consists of entry to the Univac 1108 computer should be com- points, segments of trails or other travel pleted soon. A User's Manual (Shechter 1975) routes, and camping areas. is available.?

2. User characteristics--Simulated par- RESULTS OF FIELD TESTS ties have been differentiated by size and method of travel (hiking or horseback in one The model has been field tested in two areas: the Desolation Wilderness on the Eldo- rado National Forest in California, and Dino- Dr. Kerry Smith of Resources for the Future and David Webster and Norman Heck of IBM constructed and programed the original version of Single copies are available from Robert C. the model. Dr. John V. Krutilla of Resources Lucas, Forestry Sciences Laboratory, Drawer G, for the Future initiated the project and he, Missoula, MT 59801, U.S.A., until the manual Smith, Dr. Charles Cicchetti, and Dr. Anthony and all programs become available from the Fisher a11 contributed to the initial concep- National Technical Information Service, United tualization. See Fisher and Krutilla 1972; States Department of Commerce, 5258 Port Royal Sdth and others 1974; Smith and Krutilla 1974. Road, Springfield, VA 22151, U.S.A. saur National Monument (a National Park Ser- Next, a variety of scenarios were tested. vice area) in Colorado and ~tah.~The Desola- Use was increased and decreased by varying a- tion Wilderness is a high, mountainous, lake- mounts, and uneven distributions were made dotted area of about 26,000 hectares, that is more even by shifting use from popular entries very heavily visited.: to less-used access points, and from heavily used weekends to weekdays. The Green and Yampa Rivers in Dinosaur National Monument are fast-flowing, "white- Some clear relationships, not a11 expect- water" rivers that visitors float in boats, ed, emerged. Changes in total use Call other kayaks, and rafts. Use is much lighter than things remaining the same) produced propor- in the Desolation Wilderness. tionate results. That this would be true for use patterns is probably obvious; if total use Special sample surveys provided the need- doubles, use of nay specific location doubles, ed input information on use and visitor behav- on the average. Encounters, expressed in per- ior. In the Desolation, visitors kept travel party-per-day terms, also would double in this logs, while in Dinosaur, visitors and profes- examole--something not entirely expected. sional boatmen on commercially guided trips kept logs to supplement National Park use This predictable, proportional relation- data. In both studies, information was re- ship provides a convenient base for comparing corded on encounters. results of more complex scenarios in which use is redistributed with an across-the-board In both areas, the first scenario was the chance with the same total use. The use- existing situation, or "base case." One week redistribution scenarios produced lower aver- of peak use was simulated. A 1-week initiali- aqe encounters per-party-per-day than the same zation period achieved a realistic starting total use without redistribution. This was condition. Simulation results were compared especially true for trail encounters (up to with data from the user surveys as one check one-third fewer encounters than comparable on model validity, both for use patterns and across-the-board total use). Camp encounters encounters. Agreement was good, particularly dropped only a 1 ittle below comparable unmodi- in Dinosaur National Monument, where more pre- fied use, oresumably because campsites were cise information on use characteristics and somewhat limited, and most parties camped in travel routes was collected and used in the about the same areas out of necessity even model. In the Desolation, encounters were, thouqh they arrived by different routes or at for a variety of reasons, somewhat higher ac- different times. cording to the model than reported by visitors. Several minor simplifications or departures Average encounters do not tell the whole from reality had compounded effects, but Dro- story, however. The frequency of extreme en- bably the most important reason was that the counter levels (very high levels, especially, I imited number of visitor travel routes used but sometimes also very low and zero levels) in the model (210 different routes) fell short chanqed substantially. A manager probably of reflectinq how variable visitor movements would be more concerned about reducing or e- real ly were. As a result, sl ight over-concen- liminating experiences of unsatisfactory qual- trations of parties increased encounters. On ity than altering averages. In addition, the rivers, where there were fewer entry changes at key trouble spots were even more points and possible variation in routes was pronounced. This would also probably be more less, the problem was less severe. . relevant to a manager's evaluation of the results of a scenario than overal 1 averages.

The application to the Desolation Wilder- POTENTIAL APPLICATIONS ness was a joint venture of USDA Forest Ser- vice Research and National Forest managers and We conclude that the simulator is a use- Resources for the Future, including the pre- ful tool for the manager of a wilderness or sent authors. The Dinosaur National Monument similar area. The model does not, of course, application involved the National Park Service, make decisions for the manager. It does, how- Dr. David W. Lime of Forest Service North Cen- ever, allow him to compare carefully the like- tral Forest Experiment Station, St. Paul, Min- ly results of various possible alternatives nesota, and Professor Stephen F. McCool , Utah before he decides to implement a management State University, Logan, Utah. A paper by plan. This makes it much more likely that the McCool and Lime, "The Wilderness Area Travel plan chosen wil l achieve management objectives Simulator: Applications to River Recreation and that pub1 ic benefits will be maximized. Management," 1976 is available from them. It also appears to us that it provides a A manuscript describing the results of the practical way to achieve desired conditions in Desolation Wilderness application is being terms of the amount of use of key areas and prepared. the quality of visitor experiences in terms of congest ion or sol itude but without requiring experimentieren mit dem Zie1 Auswirkungen auf tight control of visitor itineraries. Verwendungsart und Zusammentreffen von Besuchergruppen bestimmen zu konnen. Das We feel that the simulator should be ap- Modell, die Ergebnisse der Modell-tests unc pl icable to many other sorts of dispersed rec- kLinf t ige Anwendungsm6gl ichkeiten werden reation systems besides U.S. wilderness. In besprochen. fact, we suspect imaginative appl ications to some very different situations might be possible and useful. The elements in the model are perfectly general. For instance, what we have Un modsle de simulation du comportement named "tra il segments" are, in general , "trans- des usagers de sites de loisirs donnant des portat ion l inkages" and could represent any facons d1exp6rimentations avec les modifica- type of movement; for example, traffic on park t ions dues aux diverses util isat ions ou aux roads or bicycle paths. The model provides conditions de terrain permettant de determiner for six types of "transactions" (the general des patterns d'usage et de rencontres entre GPSS term for the entities whose behavior is groupes d'usagers a 6t6 mis au point. Le simulated). We have usual ly named them large, modsle, 1es r6sultats du test affgrant et ses medium, or small hiking or horseback groups of futures applications possibles sont en visitors, but any designation is possible. discussion. Perhaps one type might even represent some kind of wildl ife (say, elephants) if their LITERATURE CITED movement could be described in probabil istic terms, and "encounters" would become "wildl ife Fisher, A., and J. V. Krutilla observations," and the managers' goal might be 1972. Determination of optimal capacity of to increase, rather than decrease, encounters. resource-based recreation fac i 1 ities. Nat. Resour. J. 12(3):417-444. Certainly, the model is clearly appl icable to any type of dispersed recreation area Lloyd, R. Duane, and Virl is L. Fisher where visitor flows are of concern, where 1972. Dispersed versus con entrated recrea- there are capacity constraints, where visitor tion as forest policy. Seventh World encounters are significant, and where it is Forestry Congress, Buenos Ai res 7CFM/C: desired to allow visitors substantial freedom II f/2~(3)196. to move about flexibly. In such situations, the model is particularly well-suited to man- Shechter, Mordecha i agement planning to modify use, that is, to 1971. On the use of computer simulation for alter numbers of visitors entering at differ- research. Simulation & Games 2(l) :73-88. ent places and times. The model can also be used to test effects of alterations within the Snechter, Mordechai area, such as new access points, closure of 197.5. Simulation model of wilderness-area some travel routes, addition of campsites, and use: model-user's manual and program so on. However, to simulate such changes, a documentat ion (revised and expanded ver- basis for specifying how visitors will respond sion). 172 p., app. 312 p. Resources to the new conditions is needed. Observation for the Future, Inc., Washington, D.C. of current behavior cannot directly provide this basis, and other kinds of special infor; Smith, V. K., and J. V. Krutilla mat ion or assumpt ions based on expert judgment 1974. A simulation model for management of would be required. low density recreational areas. J. Environ. Econ. Manage. 1:187-201. The use of computer-based simulation model ing in outdoor recreation management plan- Smith, V. K., D. Webster, and N. Heck ning may arouse fears of depersonalization. 1974. Analyzing the use of wilderness. On the contrary, it may help make it possible Simulation Today 24:93-96. to maintain the traditional values of recreational visitor independence, flexibility, and Stankey, George H. spontaneity as we1 l as to protect resources 1973. Visitor perception of wilderness rec- and preserve the quality of experience in the reat ion carrying capacity. USDA Forest face of growing demands on limited resour--s. Serv. Res. Pap. INT-142, 61 p., illus. Intermountain Forest and Range Exp. Stn., Oqden, Utah. ZUSAMMENFASSUNG

Es wurde ein Simulationsmodell fur verstreut Iiegende Erholungsgebiete entwickelt welches die Moglichkeit bietet mit veriinderten Nutzungen oder Verhaltnissen im Gebiet zu A Survey of Wildlife-Related Recreation in the Tennessee Valley Region John L. Mechler E. Lawrence Klein

Abstract~Thispaper is designed to help planners and administrators better understand expenditures for wildlife-related recreation. Total spending generated by participants in a particular region is considered. In addition, potential benefits that might accrue to the local economy by increasing or enhancing wildlife- related recreation opportunities, or both, based on consumer- related preferences are out1 ined. An analysis of what people are seeking in terms of types of wildlife-recreation and the satisfac- tions gained can lead to planned wildl ife development that will have a positive monetary impact at a local level.

The primary purpose of this paper is to The gross expenditures of selected recre- examine the total spending and the spending ation groups have frequently been used for patterns of wi ldl ife-related recreat ionists in this purpose and not without good reason. The a particular region of the United States. A amount of money spent by these groups can in secondary purpose is to examine the prefer- fact represent a direct increase to the flow ences of these individuals in regard to hypo- of money within an economy (Crutchf ield 1962). thetical situations involving more intensive Examples of wildlife-related studies include management of wildl ife habitat and development Wallace (1956), Utah State Department of Fish of additional facilities. and Game (1957), University of Utah (1960), Kirkpatrick (1965), Thompson and others The data utilized for this paper were (1967), and Nobe and Gilbert (1970) to 1ist a taken from a major study of the monetary val- few. Although such studies have been criti- ues and benefits of wildlife-related recrea- cized by economists as being of little value tion demands in the Southeastern United States in solving actual valuation problems (Wenner- in 1971 (~orvath197'4a, 1974b). This study gren 1967, Kal ter 1971), they have been used was sponsored by member states of the South- by planners and administrators to a great ex- eastern Association of Game and Fish Commis- tent. In short, they have been applied for sioners, the U.S. Forest Service's Region 8, over 20 years with generally good results and the Tennessee Valley Authority. A forth- while economists continue to attack the lack coming pub1 ication by the Environmental Re- of rigor and continue their debate over an search Group at Georgia State University will acceptable method. present these findings and draw conclusions regarding the value of wildlife-related recre- Gross expenditure studies, however, ation. It should be stressed that it is not cannot show the extent of impact that recrea- the intent of this paper--nor is it recom- t ion expend itures have on local or regional mended--that the results be interpreted to economics. Variations in factor endowments, reflect necessarily on the economic evaluation social and political infrastructures, total of wi ldl ife-related recreation. economic composition, cyclical natures of various types of industries, and other vari- The results are strictly intended to con- ables influence the degree of impact that tribute toward a better understanding of wild- gross expenditures have in different locales. life-related recreation, the total spending generated by participants in a particular re- Input-output studies have been shown to gion, and some potential benefits that might be very valuable in examining this degree of accrue to local economies by increasing and/or impact and for planning the utilization of the enhancing their wildlife-related recreation resources of a region and its economic devel- opportunities based on consumer-stated prefer- opment (isard 1960). An example of a wild- ences. For example, planners and administra- life-related input-output study in one county tors at the local level are frequently inter- of Colorado is the one by Rohdy and Lovegrove ested in the potential monetary impacts that (1970). They determined both the primary might result if a greater number of wildlife economic effects of gross expenditures of recreationists were attracted to their area. hunters and fishermen as we11 as the secondary The results of this paper can serve as initial economic effects of these original expendi- input into that decision process. tures within the local economy. Their approach is more useful for detailed planning, description of the sampling methodology and but it is also more expensive; Rohdy (personal data preparation phases. The following is a communication 1975) estimated that a similar summary of his description. study in one county would require approximately $100,000 to complete. Obviously, this is a The procedure used to obtain a probabil- large investment and one that would not be ity sample of households followed closely undertaken without some prior assessment that those procedures used by the United States a potential payoff might exist. Gross expend- Bureau of the Census for the monthly Current iture studies can help in making this prior Population Survey. assessment. Following these Bureau of the Census pro- A particular area covered by this paper cedures permitted the use of the then-current is the Tennessee Valley region, a 41,000- groupings of counties within the Tennessee square-mi le area in the Southeastern United Valley region into essentially the same Pri- States composed of portions of seven states. mary Sampling Units (PSU'S) used by the Bu- reau. However, some modification was required EXPERIMENTAL METHODS in order to ensure that certain of the PSU's as used by the Bureau were restructured so as The primary data were collected by house- not to cross the region's boundary. These hold interviews, mostly in April and May 1972, PSU's were thus combined into several strata by the Environmental Research Group at Georgia within the region in such a way that individ- State University. The sample was composed of ual strata were wholly contained within the 1,076 households selected from the 1,246,168 region. The criteria of heterogeneity within households in the Tennessee Valley region. PSU's and homogeneity within strata of certain Data were obtained on various socio-economic socio-economic characteristics of the popula- characteristics of the members of households tion were the basis of groupings. indicating participation in wildlife-related recreation activities. From each of the strata, one PSU was selected for sampling with probability of selec- Three major wildl ife-related recreation tion proportional to population size. The activities were selected for the study. These total sample of 1,076 households was allocated are hunting, fishing, and nonconsumptive wild- to each stratum proportional to population life use. Each of these was further subdivid- size. The strata samples were allocated to ed into three categories resulting in nine the selected PSU's and that sample proportion- specific wildlife-related recreation activi- ately reallocated among the constituent counties. These are: ties of the PSU's.

I. Saltwater fishing For each county selected to be in the sample, population and household information 2. Warm freshwater fishing was determined for enumeration districts, block groups, and individual city blocks. 3. Cold freshwater fishing The sources of this information were the census summary tapes and county maps from the 4. Small game hunting Geography Division of the Bureau of the 5. Big game hunting Census. A cluster of four households to be interviewed in each block was then marked on 6. Waterfowl hunting the maps based on the "northwest corner clock- wise method" which is utilized in some aqri- 7. Watching or photographing birds cultural surveys. Once these given households were selected to be in the sample, no substi- 8. Watching or photographing animals tution was permitted to replace those households from which responses could not be ob- 9. Watching or photographing aquatic tained, whatever the reason for the nonre- I ife sponse.

Questions were also asked of participants The field interviewing was conducted by regarding the physical characteristics of rec- personnel of a national interviewing firm who reat ion areas which they considered important. received very detailed instructions concerning Detailed data were also sought from partici- interview procedures. A 12-page questionnaire pating households on both capital and variable served as the source of the questions to be expend itures . asked as well as the record of responses. Sampling Methodology and Data Preparation Capital Expenditures

Horvath (l974b) provides a very detailed Respondents were asked for dollar esti- tates for current replacement costs of speci- mated number of miles driven for wild- .'"ic types of property and equipment used pri- l ife-related recreation in fami ly cars marily for wildl ife-related recreation during and recreational vehicles. 1971. Fourteen categories were included in the questionnaire. These were: 2. Commercial transportation costs.

1. Recreation vehicles including jeeps, 3. Lodging. all-terrain vehicles, swamp buggies, fishing cars, etc. 4. Food and refreshments.

2. Motorcycles, motor scooters, bi- 5. Services connected with hunting, such cycles. as guides and equipment rental.

3. Winter equipment including snowmo- 6. Services connected with fishing. biles, snowshoes, etc. 7. Services connected with nonconsump- 4. Horses and riding equipment tive wildlife use.

5. Motorboats, sailboats, houseboats, 8. Supplies for hunting, st ,nmu n i- canoes, motors, etc. tion, clay birds, etc.

6. Recreational land and/or water with- 9. Supplies for fishing, such as hooks, out improvements. line, etc.

7. Cabins, cottages, travel homes, boat 10. User fees for facilities and access. docks, and other improvements. 11. Hunting licenses. 8. Campers and travel campers. 12. Fishing licenses. 9. Camping equipment including stoves, lanterns, tents, etc. The total sample size included 1,076 households. Of these, 917 (85.22 percent) of 10. Hunting equipment, such as guns, the interviews were completed. Of the remain- archery equipment, field glasses, etc. ing interviews, 109 (10.13 percent) were not completed because the residents were not home, 1 . Fishing equipment, such as rods, and 50 (4.65 percent) because of language bar- reels, nets, etc. riers, illness, and refusals.

2. Special leases on land or water for The total number of persons covered by hunting. the sample was 2,288--1,633 adults and 655 children. Of the adults, 1,472 were married 3. Special leases on land or water for and 161 were single. The children in the sam- fishing. ple were 324 sons and 331 daughters.

14. Diving gear used for underwater fishing. RESULTS

The values utilized for capital expendi- The distribution of Tennessee Val ley tures are the annual depreciation costs of households by income categories is shown in these items--not total replacement costs; each table 1. Also included in table 1 is the item was depreciated over a period applicable estimated total Valley income for 1971 based to that particular item (Horvath 1974a), ex- on the median value for each income range, cept categories 6, 12 and 13. Category 6 was except for the first income category where not depreciated or amortized, and, thus, a an arbitrary value of $2,500 was used, and sl ight upward bias was introduced. The annual the last income category where an arbitrary lease fees were utilized for categories 12 and value of $36,000 was used. 13. Income of households may or may not be a Variable Expenditures significant factor helping to determine whether or not to participate in wildlife-related The variable expenditure categories in- recreation, but it would certainly seem to cluded the following: dictate the degree of participation in terms of total money spent, number of trips, and the I. A per-mile cost assigned to the esti- types of activities selected. Table I--Percentage composition of households, by income categories, in the Tennessee Valley Region in 1971 and the estimated total income received

Estimated Percent of total Valley income Income ($1 households ($ thousand)

Under 3,001 22.9 $ 713,431

3,001 - 5,000 17.9 892,256

5,001 - 7,000 16.0 1,196,320

7,001 - 10,000 18.8 1,991,375

10,001 - 15,000 13.4 2,087,330

15,001 - 20,000 6.2 ,352,091

20,001 - 25,000 2.7 757,046

25,001 and over 2. I 942,102

Total 100.0 $9,931,951

The percentage of participation of house- Again, by way of comparison, 34 percent holds, by income categories, in the three of all households in the nation had one or major wildlife-related recreation activities-- more persons who fished in 1970 (U.S. Dep. hunting, fishing, nonconsumptive activities-- Inter., Bur. Sports Fish. 6 Wildl. 19721, but varied according to activity (table 2). for the same reasons as l isted for hunting, no attempt was made to determine if the observed Tennessee Valley households with incomes difference was statistically significant. of $5,000 and less were general ly represented in a11 the three major activities to a smaller Table 5 shows the number and percentage deqree than were households with incomes of households with one or more members report- greater than $5,000. ed to be involved in nonconsumptive wildlife activities. Over 50 percent of these house- Tables 3, 4 and 5 summarize the number of holds reported that birds are the sole wild- households and percentages' of total households life species of interest to them. with one or more members participating in hunting, fishing, and nonconsumptive wildlife Table 6 shows the total expenditures made activities. Also included in each table are by households in 1971 on wildl ife-related rec- the specific activities comprising each major reation activities as well as the percentage activity. breakdown of households participating in wildlife-related activities. As shown, some The U.S. Department of the Interior, households participated in more than one Bureau of Sport Fisheries and Wildlife (1972) activity and also used some equipment for more found 18 percent of all households in the than one activity. Therefore, all combina- United States participated in hunting activ- tions of activities with the percentage of ities in 1970. Thus, the value of nearly 26 participating households and the total expend- percent of households in the Tennessee Valley itures for each are shown. region would seem to represent a greater participation rate. However, no attempt was made to determine if a statistical difference ex- The total expenditures of $211,311 ,000 isted because of differences of when the sur- (from table 6) represent 2.13 percent of the veys were completed and probably differences estimated total Tennessee Valley household in sampling techniques and definitions of income (from table 1) during 1971. A total activity categories. of 47.7 percent of all households in the Table 2--Percentage participation of Tennessee Valley households during 1971 in hunting, fishing, and nonconsumptive wildlife activities, by income categories

Percentage household participation in

Hunt ing Fishing Nonconsumptive Income ($) activities activities activities

Under 3,001 3.3 23.6 5.4

3,001 - 5,000 22.0 35.2 6.9

5,001 - 7,000 33.8 43.7 6.3

7,001 - 10,000 29.9 49.7 11.4

10,001 - 15,000 34.5 53.8 2.6

15,001 - 20,000 32.7 47.3 18.2

20,001 - 25,000 29.2 50.0 16.7

25,001 and over 21 .o 47.4 21.1

A1 I categories 25.7 40.9 10.8

Table 3--Households participating in various wildlife hunting activities in the Tennessee Valley region in 1971

Types of Number of Percent of hunting househo1ds all households

Small game only

Big game only

Waterfowl on1 y

Small and big

Small and waterfowl

Big and waterfowl

AH three

Total hunters

Nonhunters Table 4--Households participating in various wildlife fishing activities in the Tennessee Valley region in 1971

Types of Number of Percent of fishing househol ds a1 1 households

Saltwater only

Warm freshwater only

Cold freshwater only

Saltwater and warm freshwater

Saltwater and cold freshwater

Warm and cold freshwater

Al l three

Total fishermen

Nonfishermen

Table 5--Households participating in various nonconsurn~tivewildlife activities in the Tennessee Valley region in 1971

Types of nonconsumptive Number of Percent of wildlife activities households a1 1 households

Bird only Animal only Fish only Bird and animal Bird and fish Animal and fish All three Total nonconsumptive enthusiasts Nonwildlife enthusiasts Table 6--Total expenditures on all wildlife recreation activities and combinations thereof by households in the Tennessee Valley region in 1971

Percent of Total participating expenditures Activities households ($ thousand) I Fishin? only I 16.8 $ 64,361 Hunting only

Nonconsumptive activities only

Fishing and hunting 17.6 96,572

Fishing and nonconsumptive 2.5 11,919

Hunting and nonconsunptive 0.7 2,231

All three activities

Total

Table 7--Percentage distribution of the total depreciated replacement value for capital items used for wildlife recreation by ~articipatingTennessee Valley households in 1971

Capital items Percentage

Recreational vehicles 14.4

Motorcycles, scooters, and bicycles 4.9

Winter equipment 0. I

Horses and riding gear 2.4

Boats, canoes, etc. 20.8

Cabins and other improvements 20.8

Campers, trai Iers 8.1

Camp i ng equ ipment 4.4

Hunting equipment 13.5

Fishing equipment 9.3

Leases for land/water used for hunting 0.1

Leases for land/water used for fishing 1.1

Diving gear 0.1

Total 100.0 Table 8--Percentage distribution of the total variable expenditures for wildlife recreation by participating Tennessee Valley households in 1971

Expenditure category Percentage

Mileage costs

Commercial transportation

Lodging

Food and refreshments

Services connected with hunting

Services connected with fishing

Services connected with nonconsumptive activities

Supplies for hunti ng

Supplies for fishi ng

User fees for faci ities and access

Hunting licenses

Fishing licenses

Mi see1 Ianeous

Total

Table 9--Characteristics of quality hunting and fishing areas as reported by participatinr Tennessee Valley households in 1971

Weighted percentage importance -- I Characteristic Hunting Fishing

Available overnight facilities 6.7 7.3 Within convenient travel time 35.3 ! 33.8 Abundance of animals or fish Low participant density Presence of trophy animals or fish Appearance of area Total Table 10--Willingness to pay an extra fee for improved quality areas on public and private lands

Type of activity Type of Hunting Fishing Nonconsumptive Iand (percentages) ! I I Pub1 ic Yes 51.3 48.8 49.3 No 38.7 40.3 33.3 No opinion 10.0 10.9 17.4 Frivate Yes 49.3 43.6 41.3 No 37.3 41 .4 36.6 No opinion 13.4 15.0 22.1

Tennessee Valley participated in one or more mill ion in 1971. Although this represented wildlife-related recreation activities during only about 2 percent of the estimated total 1971. household income in 197). it is nonetheless a considerable amount of money which might be The depreciated replacement value for all important in terms of enhancing the economies capital items utilized for wildlife-related of some local areas. recreation activities averaged $166 per participating household. This value included the Nearly one-half of all households in the nonamortized total value for recreation land Tennessee Valley region participated in one or and/or water. When adjusted for this bias by more wildl ife-related activities, ranging from subtractinq the investments in recreational hunting and fishing to nonconsumptive activi- land and/or water, the depreciated replacement ties. Fishing showed the greatest participa- value for all capital items totaled $155 per tion rate at nearly 41 percent; hunting was participating household (table 7). next at approximately 26 percent, followed by nonconsum~tiveactivities at about 11 percent. The average variable cost per household Both the hunting and fishing participation participating in wildlife recreation activi- rates for Tennessee Valley households were ties was $200 (table 8). higher than the national values for the same activities. This difference suggests that Quality Preferences there might we1 1 be greater potential for wildl ife-related development activities in the Characteristics of hunting and fishing Tennessee Val ley than possibly other areas areas which participants regard as important where fewer people participate. should serve as useful guidelines for planners and administrators. Respondents in the par- Even within each of the major activities, ticipating households were asked about what differences in the participation rate existed constituted a quality hunting or fishing area. based on household income categories. It ap- The most important determinant of both quality pears that $5,000 might be considered the hunting and quality fishing areas is that they threshold for hunting and fishing activities are within convenient travel time (table 9). in that participation rates are greatest above Abundance of animals or fish is the next most that figure. The threshold for nonconsumpt ive important characteristic for both activities, activities is $7,000. These figures suggest fol lowed by low density of people. that the evaluation of wildlife-related development potential would definitely include an Most of the respondents expressed a will- examinat ion of the incomes of user groups ingness to pay an extra fee for hunting or which might be attracted. fishing on lands that had improved qua1 ities (table 10). The participation rate varies within each SUMMARY major activity also. For example, almost all the hunting activity was in the small game Wildl ife-related expenditures in the category, and over one-half of the fishing ac- Tennessee Valley region totaled over $200 tivity was in the warm freshwater category. These differences are primarily reflective of hunting and fishing in New Mexico. 94 p the opportunities available for participation. Bur. of Business Research, Univ. of New This was also indicated in the nonconsumptive Mexico, A1 buquerque. area in which birds were the single greatest category. One could not contend that these activities would justify a significant mone- Nobe, K. C., and A. H. Gilbert tary development, since the reason they are 1970. A survey of sportsmen expenditures most participated in is that there is more for hunting and fishing in Colorado, opportunity to do so. 1968. Colo. Div. Wildl. Tech. Publ. 24, 83 p. Colo. Div. Wildl., Denver, Colo. The most important characteristics that (Available in photocopy only.) determine a qual ity hunting and fishing area are convenient travel time and abundance of Rohdy, D. D., and R. E. Lovegrove animals or fish. The least important charac- 1970. Economic impact of hunting and fish- teristics are overnight facil ities and pres- ing expenditures in Grand County, Colora- ence of trophy animals or fish. Thus, in any do, 1968. 36 p. Dep. of Economics, development of a wildlife nature, it must meet Colo. State Univ., Fort Coll ins. the criteria of abundance and convenience to attract the most visitors; but such amenities Thompson, E. F., J. M. Gray, and B. S. as overnight facilities and trophy animals and McG innes fish are not very important. 967. Estimated hunting expenditures in Virginia. Virginia Polytechnic Inst. One factor that is very important to Res. Rep. 116., 8 p. planners and managers is that in a1 l activities, a majority of respondents were willing U.S. Department of the Interior, Bureau of to pay an extra fee for improved qual ity areas Sport Fisheries and Wildl ife on both public and private lands. 1972. National survey of fishing and hunting (1970). Res. Publ. 95, 108 p. LITERATURE CITED University of Utah, Bureau of Economics and Crutchfield, J. A. Business Research 1962. Valuation of fishery resources. Land 960. The economic value of fishing and ECO~.38(2):145-154. hunting in Utah. 36 p. Utah State Dep. Fish & Game,.Salt Lake City.' Horvath, J. C. 19743. Southeastern economic survey of Utah State Department of Fish and Game wild1 ife recreation: detailed analysis. 1957. A study of the econorric value of Georgia State Univ., Atlanta, Ga. 183 p. fishing and hunting in Utah. Utah State Dep. Fish & Game Publ. 7, 73 p. Utah Horvath, J. C. State Dep. Fish & Game, Salt Lake city.' 1974b. Southeastern economic survey of wildlife recreation: executive summary. Wallace, R. F. Georgia State Univ., Atlanta, Ga. 118 p. 1956. An evaluation of wildlife resources in the State of Washington. Bur. Econ. & Isard, W. Bus. Res. Bull. 28. 63 p. Washington 960. Methods of regional analysis: an State College, School of Economics and introduction to regional science. 784 p. Business, Pullman. The M.I.T. Press, Cambridge, Mass. Wennegren, E. B. Kalter, R. J. 1967. Demand estimates and resource values 1971. The economics of water-based outdoor for resident deer hunting in Utah. Utah recreation: a survey and critique of State Univ., Agric. Exp. Stn. Bull. 469, recent developments. Institute for Water 27 P. Resources Report 71, 192 p.

Kirkpatrick, T. 0. ' The Department is now the Utah State Wild- 1965. The economic and social values of 1 ife Resource Division. Mathematical Programming in the Context of Planning for Multiple Goals A. B. Rudra

Abstract--In attacking some forest management problems, multiple goals can be linked through a single performance-function criterion. A variety of such problems have been solved by traditional mathematical programing techniques, such as linear programing. But in another class of problems, such a unified single- dimension criterion is not readily available. Increased concern in environmental quality has emphasized the importance of extra- market benefits. Surrogate measures are often not adequate enough for assessing the totality of such intangible values that result from any activity. Infeasibility of solution is another difficulty encountered in the conventional structuring of multiple confl ict ing goal s through 1 inear programing. The goal programing approach is, therefore, advocated as ascertaining the trade-off in achieving specific goals. The approach is illustrated by a case study.

In recent years, the use of vathematical as forests, national parks, etc., are not ~roqraminqin forestry has undergone a share readily subject to market valuations. This ucsurae. The most comnonly used technique is I imitation has resulted in an increase in re- the linear ~roq'-ar'ina (LP) cortwlation, either search into the evaluation of extra-market in the conventional way, or in one of its benefits. Certain proxy values have been pro- a1 l ied forms. Problem are aeneral ly struc- oosed for the assessment of unpriced goods. tured in the usual oroduct pix allocation and In the field of recreation, for example, the schedulina formats. Studies cover both the use of consumer surplus shows considerable single-staqe and the multistaged formulations. promise as a measure of the society's willing- All such formulation seek to optimize a sin- ness to pay for an amenity. Even the inclu- gle-dimensional objective function criterion, s ion of a surrogate measure to cover the in- while satisfying a set of constraints which clusion of extra-market benefits into the ob- define a number of intermediate goals. jective function does not, however, overcome the problem of using market prices for the This paper seeks to draw attention to the rest of the commodities in an economic ~lan- method of goal proqrami ng (GP) and, hopeful ly, n ing exercise. to nudge the case for stochastic programina-- techniques which do not aopear to have been There are the problems of partial equi- extensively used in forestry so far. librium models and although there is a defi- nite trend towards general equilibrium models In the standard LP format, all activities (~orsund1972, Mul ler 1973) there is still in- are optimized in terms of a common performance adequate econometric information to rely on. function. Thus, if profit maximization is the We must also emphasize that there is some con- goal, financial returns are assessed by d is- t roversy regard ing the evaluation of extra- counted net revenue, present net worth, inter- market benefits in part. Indeed, Schramm nal rate of return, benefit-cost ratio, or (1973) argues that it is impossible to evalu- some such single criterion for all activities. ate all of the extra-market benefits and, Similarly some common measure of costs is used therefore, one should not attempt to assess a in the cost minimization case. However, in fraction of them only. Johnston (1974) dis- planning for multiple-use forestry, particu- cusses some of the difficulties in deciding larly when extra-market benefits are involved, upon a suitable indicator of social economic it is not always possible to select a suitable welfare in environmental qua1 ity control. criterion--for the comprehensive appraisal of the multi-faceted benefits that result from It is important, therefore, to examine the act ivities--for considerat ion in the ob- the trade-off between environmental and con- jective function. ventional materialistic goals--and goal programing provides a method for such analyses. Growing interest in environmental quality A modification of standard l inear programing, has highlighted the need for environmentally it obviates the need for a single dimensional sensitive indicators rather than the usual optimrzation criterion. And by the very na- economic indicators. Natural resources, such ture of its formulation, it removes the in- asibil ity of solution. In conventional LP, and negative deviations from e lack of a feasible solution is often very goals; each of the vectors has m f st rating, as the offending constraints may elements ~t be readily apparent when the initial problem is structured. b = column vector of m elements representing the goal levels Chance-constra ined programing (CCP) is an example of stochastic programing. Constraint c ' = row vector representing the co- equations representing the goals can be re- efficients of the objective placed by a measure of probabil ity or risk of function; these are the weighted not achieving any or all of the constraints. or unweighted priority factors. This removes the necessity that all constraints must be satisfied. One recognizes immediately the similarity between the deviation variables in GP and the GOAL PROGRAM ING slack variables in LP, in changing inequali- ties into equalities. The imposition of a Lee (1972) recounts the history of goal positive and a negative deviation to each con- programing. The concept of GP was developed straint or goal equation, removes the possi- by Charnes and Cooper and they also provided bility of infeasibility. Further the formula- the name of the method (Charnes and Cooper tion ensures that either d? or df will be zero 1961). For a detailed exposition of the tech- The weighted priorities are given by 5. Let u nique, consult papers by Charnes an'd others first consider the effect of minimizing differ (1968~1, b; 1969), Ijiri (1965), Lee (1971, ent sets of the unloaded deviates. Minimizinc l972), and Field (1973). (d- + d+) makes the search for 6 which achieve the goal fi.? = t~,thereby minimizing the abso- The goal programing model sets out to lute value of (A.?- b). Minimizing only (d-) minimize the aggregate sum of the positive and results in minimizing the negative deviations negative deviations from goals which are spec- from the goal. Solution gives 5 which sets ified by constraints. Thus the objective (b - 4.x) to the minimum possible. On the function consists of coefficients for the de- other hand, minimizing only (d+) results in 5 viations of the goals and places no value for which sets (A.?- b) to the minimum possible. the structural variables depicting activities. We may set out the GP model in the famil iar LP Several techniques are available to pre- form as follows: scribe priorities and weights. And it is possible to force an ordering of the attain- Minimize z = c'd- - ment of goals (~ee1972, Field 1973). If no weighted priorities are used, then the elements of 5 will all equal 1. We may ignore some of the deviations in which case these elements in 5 will have zero value. The loading scheme could be an artifact of the data base, or it could be subjectively prescribed. Such ad hoc weighted priority loads may be taken to represent either conventional wisdom or current thinkinq on the relative importance of goals. It would appear (see following An additional group of constraints, not involv- examples) that the relative order of these ing the positive or negative deviations, may loads is more important than actual value of sometimes also be included to deal with rela- these loads. Generally, as long as we give tions between the constraints. relatively large loads to the deviations we seek to avoid (i.e., seek to ensure the attainment of the goals concerned) these Where m = number of constraints or goals deviation variables will not appear in the final basic solution in a minimization scheme n = number of decision variables (compare the case of artificial variables in a 5 = column vector of decision normal LP scheme). variables (n elements) To further clarify the method, we first 6 = the matrix (m x n) of technolog- consider a simple problem. The corresponding ical coefficients LP and GP formulations are tabulated in figures 1 and 2. The details are self explana- I = the identity matrix (m x m) tory. For the GP case (fig. 2) we have first - + considered two unweighted priority schemes, 4 ,c) = column vectors of the positive and the last two schemes use ad hoc loads (03) x 1 X2 RHS Solution (~asicvariables) OBJ i 80 40 (Maximize) X1=24 Constraints I X2=l 6

Slack of (3)=14

OBJ = 2560

Figure I--The l inear programing formulation

Xi X2 di di d3 di dy d3 RHS as Solutionic variables)

For 01 & 02 Objective Functions xi=l 0

x2=30 I 1 1 1 (Minimize) d;=l4 1 1 I I

P3 p3-(weighted priorities) For 03 & 04

1 1 (Minimize) X 1=24

1 1 I I X2=I 6

d3=14 Constraints

Min. aggregate = 40 of deviates

Figure 2--The general programing formulat ion and weighted priorities calculated on the its equivalent deterministic form (Sengupta basis of some algorithm based on the problem 1972). The chance const ra ined programing data. Cases (01) seek to Min (d- + d"), the method (CCP) is one such technique, developed deviations not being loaded. Case (02) is sim- by Charnes and Cooper (1959). i lar (ignores &+ and dcf) to Cases (03) and (04), except that (02) uses no weighted priorities. Many of the parameters of most real-l ife Notice that a variety of solutions are possi- problems are described by random variables ble, depending on the goal ordering scheme. rather than by deterministic quantities. Thus in the customary LP model [maximize Z = c-5, Let us now consider a s imp1 ified version Axs b, 5 2 ?I, the elements of the set (6, b, of a real 1ife problem (fig. 3). A forest has are generally stochastic. It is obvious three zones A, B, and C. For each zone, three that through random variations in 0,where 6 management a1 ternatives are available: denotes the vector with elements (A, b, c), considerable differences could occur in the TP = Area to be worked for timber solution. Partial answers are available, production through standard sensitivity analyses in LP, to study the effect of changes in the problem R = Area to be exclusively used for parameters on the optimal solution. Stochas- recreation tic programing considers these random effects more explicitly in the solution of the model. MU = Area to be managed for multiple use, say recreation and l imited timber In general, CJ, aiJ, and bi are all ran- product ion bees?

The goal programing model provides a Prob [bi ;B~];Ii range of solutions, depending on the pattern of loads attached to the deviation variables. or equivalently These solutions help in deciding on trade-off between various goals. They also help in understanding the extent of sacrifice involved rob [bi < B,]; 1 - B~ in forcing the realization of any sub-set of goals. where the number Bi is the (1 - 81 ) fract ile of marginal probability distribution of bi STOCHAST IC PROGRAM ING . The determination of Bi presents no problem, once the marginal probability distribution of The central idea of a stochastic program bi is known. is to convert the probabi l istic problem into Zone A I Zone B I Zone C I

(I), (2), (3) - Area constraints (in suitable area unite)

(4) - Water qua1 ity constraints (in suitable BOD units)

(5) - Water qua1 ity constraints (in suitable SS [suspended sol ids] units)

(6) - Permissib1e cuts (in suitable volume units)

(7) - Permissible visitor loads (in suitable units: ST-summer tourists)

(8 - Permissib1e visitor loads (in suitable units: WT-winter tourists)

(9) - Budget constraints (C in suitable money units)

(10) - Anticipated income (R in suitable money units)

Figure 3--The basic problem of management alternatives available for three forest zones: timber production (TP), recreation (R), and multiple use (MU) A B C

BOD (bio-chemical ozygen demand) B= R = 120 units MU .009 .009 .008

SS (suspended sol ids) = 120 units

V (volume of outturn) = 2,400 units

ST (summer tourist load) = 2,400 units

WT (winter tourist load) = 10,000 units

C (budget constraints - costs) c= R = 1,700 units MU

R (income constraint) = 3,600 units

Figure 4--Hypothetical data used for the problem in management alternatives. Zone A 1 Zone B Zone C -;dy di CIA d; d< d: d; dk d; dy(, d: d+ d+ I dl(,:+: RHS Remarks TP R MU Objective j 1 x7 X8 XQ functions 1

0 1 1 1 Minimize

05 k k 4 I k = 999999 ad hoc weighted priorities

000 I 1 1 k = 21 603600 weighted priorities calculated as per Priority Factor Algorithm (~ield1973)

Figure 3

Fiqure 4

Fiqure 5--Goal Proqraminq Tableau for the problem in management alternatives Objective !Variable value So1 ut ion Variable function No. (appropriate Remarks number description : used units) 1. 01 XI Area- A - (TP) M'in (d+ d), without weighted priorities X4 Area B - (TP) Activities chosen (i) all A for timber production Xg Area C - (MU) ii) all B for timber production - d3 Unused area of C - (iii) part C for multiple use d4 BOD level by

Area A - (R) Min (d+ d), with positive deviates given equal and very large weights x priority loads Area A - (MU) Activities chosen (i) part of A for recreation and rest Area B - (TP) for multiple use Area C - (TP) i i) a1 1 of B for timber production Area C - (R) i i i) part of C for timber production BOD level

\ 3. 03 Same solution as Solution 2. In this case minimization problem was same as 02, except that positive income deviate was not given large loading factor.

Area A - (MU) Min (d) without weighted priorities Area B - (TP) Activities chosen (i) all A for multiple use Area C - (TP) (ii) a1 1 B for timber production Area C - (R) (iii) part of C for timber production and rest for recreation BOD level by

^igure 6--Solut ions of the model to the problem of management a1 ternatives: timber product ion (TP), recreation (R), and multiple use (MU). Objective Variable value Solution function No. Variable (appropriate Remarks number used description units)

5. Xo Area A - (R) 6000 Min (d+ d) with specified weighted priorities Xo Area C - (R) 2000 Activities chosen (i) all A for recreation i O5 - ii) part of C for recreation 1 d2- Unused B 2000 d3 Unused C 2000 Substantial part of forest left unused. - Water quality constraints satisfactorily below limit. d4 BOD level

6. 06 Same solution as 5. Same minimization problems as 05, except for weight x priority factors.

Figure 6 (~ontd)--Sol ut ions of the model to the problem of management a1 ternat ives: timber production (TP) , recreation (R) , and mu1 tiple use (MU).

Figure 7--Solution for income maximization. All areas (A, B, and C) are chosen for timber production. Opportunity costs for non-basic variables and optimality ranges for RHS constraints and income coefficients are shown.

SUMMARY OF RESULTS

Variable Variable Basic Activity number name Non-Bas ic Ieve1 1 X 1 B 6000.0000000 2 X 2 NB -- 3 x 3 NB -- 4 X4 B 2000.0000000 5 x 5 NB -- 6 x 6 NB - - 7 x7 B 3272.7272727 8 x 8 NB -- 9 x9 NB -- 10 --SLACK NB -- 11 --SLACK NB -- 12 --SLACK B 727.2727273 13 --SLACK B 61 .6363636 I4 --SLACK NB - - 15 --SLACK B 25.4545455 16 --SLACK B 2400.0000000 17 --SLACK B 4490.9090909 18 --SLACK B 21.8181818

Maximum value of the objective function = 3438.181818 igure 7 (Contd)--Solution for income maximization. A11 areas (A, B, and C) are chosen for timber production. Opportunity costs for non-basic variables and optimality ranges for RHS constraints and income coefficients are shown.

RNGRHS .t. .L .1, .it -8. -1- ,\ ,> ,. ,. ,.,\ (OPTIMAL ITY RANGE FOR R I GHT-HAND-S IDE CONSTANTS) NON-SLACK RESOURCES ONLY

Bl XOUT MIN Bl ORIGINAL Bl MAX Bl XOUT ------Z-LOWER Z Z-UPPER

5 7 84.000 20.00 121.40 15 2620.0 3438.2 3470.0

RNGOBJ .!.,..!..$.!.,.,.,..L, .-8. ,. (OPTIMAL ITY RANGE FOR OBJ COEFFICIENTS) BASIC VARIABLES ONLY Solution (~asicvariables)

Variable number Variable description 1 Variable value 1 Remarks

Area A - TP 1442 Activities chosen: parts of Area A get all management alternatives; Area Area A - R 41 I B for timber production; Area C is Area A - MU 4147 partly chosen for timber production Area B - TP 2000 and partly for recreation.

Area C - TP 1432

Area C - R 2568

V 1 eve' - by 851 All constraints bounds are satisfied as in Solution 02 (figure 6). Budget ST level < by 1518 level about same but income level C level by 724 improves . R level < by 1284

Fiqure 8--Problem in fiqure 3 with elements in the BOD qoal, chance-constrained. The solution of the qoal proqraminq model of fiaure 5 L- inq 03 for deviation loads is given below. (compare solution 2 of 'iqure 6.)

Since Bi provides a pleasure of the pro- bab i1 ity that any specific goal can be met or I;=] E(dijj . j + N v ioIated, a re1 iabil ity measure for the system where the standard normal Npi is such couId be found by combining the component *nat the cumulative distribution function of re1 iabilities, i. e. normal distribution,

and We will illustrate the case when a,, is a random variable by a simple example. Assume aii is normally distributed with mean ~(a;,) and variance v(ai,). Consider the ithchance constraint is real ized iff

Now The constraint is no longer linear. There are ways, however, under some relaxing assumptions, when the constraint can be made properly separable. and To study the effect of chance constrain- ing, the BOD constraint elements (in problem 03 of figure 5) on the optimal solution, we adopted the above procedure. The an's of this constraint were assumed to be independently, normally distributed, so that the co- where is the variance-covariance matrix of variance terms in fl were set to zero and an . It follows that the constraint can now BI = .05. The solution only is presented for be written as this study (fig. 8). In some cases it may be appropriate to have manpower planning. Manage. Sci. Rep. 18. one or more joint chance-constraints instead of Grad. School Ind. Adm., Carnegie-Mellon independent constraints. Such formulations can Univ., Pittsburgh, Pa. 12 p. only be solved by techniques of nonlinear programing. Nevertheless the approach of joint Field, D. B. chance-constraints has some attractive features, 1973. Goal programming for forest manage- and deserves attention. ment. For. Sci. 19(2):125-135.

The imp1 ications of chance constraints are Forsund, F. R. that tolerance measures, one for each con- 972. Al locat ion in space and environmental straint, or jointly for chosen goals, may be pollution. Swedish J. Econ. 74(1):19-34. prescribed by the decisionmaker with the implic- it cost of such preassignment. Alternatively, Ijiri, Y. tolerance measures may be optimally solved along 1965. Management goals and accounting for with other decision variables. control. Rand McNally, Chicago. 191 p.

CONCLUSIONS Johnston, R. E. 1974. Quantitative attempts in environmental By its very structure, the goal programing quality control. Appita 28(3):181-186. method removes infeasibility and obviates the necessity for a single dimensional criterion in Lee, S. M. the objective function. Goal programing, there- 1971 . Decision anal ys,i s through goal pro- fore, deserves more attention in the management gramming. Decision Sci. 2(2) :172-180. planning of resources for multiple goals--particularly when no single measure can comprehen- Lee, S. M. sively evaluate all the facets of conventional 1972. Goal programming for decision analysis. material istic as well as intangible values that Auerback Pub1 ., Inc., Philadelphia. 387 p. result from any management alternative. Muller, F. Another form of programing--stochast ic pro- 1973. An operational mathematical program- graming--also merits wider application since ming model for the planning of economic most parameters in real l ife are probabilistic. activities in relation to the environment. Standard sensitivity analysis with LP can pro- Socio-Econ. Plan. Sci. 7:123-138. vide only partial answers. The probabilities of achieving specific goals can be incorporated Schramm, G. in chance-constrained programing formulations, 1973. Accounting for non-economic goals in and some measure of system reliability can be benefit-cost analysis. J. Environ. Manage. assured. 1 (2):129-150.

Sengupta, J. K. LITERATURE CITED 1972. Stochastic programming: methods and appl icat ions. North-Hol land Publ Co., Charnes, A., and W. W. Cooper . Amsterdam. 313 p. 1959. C hance-cons t ra ined programming . Man- age. Sci. 6(I) :73-79. APPEND I X Charnes, A., and W. W. Cooper 1961. Management models and industrial ap- This is a short 1 ist of references showing pl icat ions of 1 inear programming. 2 vols. the use of conventional mathematical programing John Wiley, New York. techniques in forestry. In each case the multiple and intermediate goals are defined by the Charnes, A., W. W. Cooper, J. K. DeVoe, D. B. constraints, and optimization is sought in terms Learner, and W. Reinecke of a common performance criterion in the objec- 1968a. A goal programming model for media tive function. planning. Manage. Sci. 14(8) :B423-430. Amidon, E. L., and G. S. Akin Charnes, A., W. W. Cooper, D. B. Learner, and 1968. Dynamic programming to determine opti- E. F. Snow mum levels of growing stock. For. Sci. 1968b. Note on an application of a goal pro- 14 (3) :287-291 . (uses dynamic programing to gramming model for media planning. Man- determine optimum levels of growing stock.) age. Sci. 14(8) :B431-436. Bare, B. B., and E. L. Norman Charnes, A., W. W. Cooper, R. J. Niehaus, and 1969. An evaluation of integer programming in D. Sholtz forest production schedul ing. Indiana Ag- 1969. An extended goal programming model for ric. Exp. Stn. Res. Bull. 847, 7 p. Purdue Univ., West Lafayette. (Integer programing Wardle, editor. Great Britain For. Comm. for forest product ion schedul ing problems.) Bull. No. 44. H.M.S.O., London. (~e- scribes the optimization of product mix of Boughton, W. C. small size wood through LP.) 1967. Planning the construction of forest roads by 1 inear programming. Australian Hool, J. N. For. 31 (2) : 11 1-1 20. (~nappl icat ion of LP 1966. A dynamic programming-Markov chain ap- to the planning of road construction.) proach to forest production control. 26 p. For. Sci. Monograph No. 12. (provides a Clutter, J. L., J. H. Bamping, J. E. Bethune, dynamic programing-Markov chain approach to J. C. Fortson, L. A. Hargreaves, and L. S. forest production control, to prescribe for Shackel ford any planning interval and at any point 1968. MAX- ILLION, a computerized forest man- within it the optimal control activity for agement planning system, 61 p. Biometrics possible conditions of the system.) Staff, Oper. Res. Sec., School of Forest Resour., Univ. Georgia, Athens. (uses LP Hool, J. N. technique to schedule operative in forest 1966. A dynamic programming-probabil istic ap- management planning.) proach to forest production control. p. 191-193. In Soc. Amer. For. Proc., 1965. Curtis, F. H. (Sets out a dynamic programing-probabil is- 1962. Linear programming the management of a tic approach to forest production control forest property. J. For. 60(9) :611-616. seeking maximization of yield for sets of (LP formulations to aid foresters in devel- production control activities, subject to oping sustained yield cutting schedules.) constraints of the amount of information and re1 iability of data.) Davis, L. S., E. F. Lyons, and H. E. Burkhart 1972. A spatial equilibrium analysis of the Kidd, Jr., W. E., E. F. Thompson, and P. H. Southern Appalachian hardwood lumber-using Hoepner industry. For. Sci. 18 (3) :247-260. (pro- 1966. Forest regulation by linear programming vides a spatial equilibrium analysis for --a case study. J. For. 64(9):611-613. lumber using industry, minimizing costs (LP for forest regulation to derive a through the network. Model solves the schedule of timber harvests over time which transportat ion problem.) maximizes present net worth.)

Kilkki, P., and U. VaisSnen '"ornstad, B. F. 1969. Determination of the optimum cutting 1971. The 1 inear programming planning system policy for the forest stand by means of dy- of the Swedish Forest Service. p. 124-130. namic programming. 23 p. Acta For. Fenn. In Operat ions research and the managerial 102. (uses dynamic programing for the de- economics of forestry. P. A. Wardle, edi- termination of the optimum cutting policy tor. Great Britain For. Comm. Bull. No. for forest stands.) 44. H.M.S.O., London. (~escribesthe Swedish Forest Service LP planning system.) Kostov, P. 1971. Optimisation of the species composition Gangul B. N. i, of a forest area. p. 18-22. In Operat ions 1970. Linear programming as an aid for pulp research and the managerial economics of wood procurement and scheduling. Indian forestry. P. A. Wardle, editor. Great For. 96(10):781-786. (~sesLP for procure- Britain For. Comm. Bull. No. 44. H.M.S.O., ment and scheduling optimization for a hy- London. (optimization of the species com- pothetical case study involving 4 pulp position of a forest area. Maximizes vol- mills and 6 forest areas.) ume production under constraint of growth of the species on each site and market re- Hazard, J. W., and L. C. Promnitz quirement for the particular wood type.) 1974. Design of successive forest invento- ries: optimization by convex mathematical programing. For. Sci. 20(2):117-127. Liittschwager, J. M., and T. H. Tcheng (uses convex mathematical programing to op- 1967. Solution of a large-scale forest sched- timize the allocation of inventory sampling u1 ing problem by linear programming decom- resources under d ifferent sampl ing plans .) posit ion. J. For. 65(9) :644-646. re- sents solution of a large scale forest HOfle, H. H. schedu 1 ing problem by LP decompos ition. ) 1971. Optimisation of the harvest of small- size wood through linear programming. p. Loucks, D. P. -10. In Operations research and the 1964. The development of an optimal program managerial economics of forestry. P. A. for sustained-yield management. J. For. 62(7):485-490. (LP formulations to aid Bull. No. 44. H.M.S.O., London. (~e- foresters in developing sustained yield scribes methods of operat ions research and cutting schedules.) plans selected operat ions in forestry.)

Nautiyal, J. C., and P. H. Pearse Sampson, G. R., and C. A. Fasick 1967. Optimizing the conversion to sustained 1970. Operations research application in lum- yield--a programming solution. For. Sci. ber production. For. Prod. J. 20(5):12-16. 3(2) :l3l-l39. (LP approach to optimizing (LP to optimize efficiency of a saw mill by the conversion to sustained yield.) maximizing net revenue.)

Navon, D. I., and R. J. McConnen Schreuder, G. F. 1967. Evaluating forest management policies 1968. Optimal forest investment decisions by parametric l inear programing. USDA For- through dynamic programming. School of est Serv. Res. Pap. PSW-42. 13 p. Pacific For. Bull. 72. 70 p. Yale Univ., New Southwest Forest and Range Exp. Stn., Ber- Haven, Conn. (A dynamic programing sched- keley, Ca. (~nanalytical and simulation ule to cover the whole production process technique together with parametric LP to from the seedling to final stages of one or explore alternatives of forest management more primary forest industries.) for pol icy decisions.) Schreuder, G. F. Paine, D. W. M. 1971. The simultaneous determination of opti- 1966. Analysis of a forest management situa- mal thinning schedule and rotation for an tion by linear programming. Australian even-aged forest. For. Sci. 17(3) :333-339. For. 30(4) :292-303. (provides an analysis (~imestaged formulation for the simultane- of a forest management situation by LP.) ous determination of optimal thinning schedule and rotation for an even-aged for- Pearse, P. H., and S. Sydneysmith est.) 1966. Method for a1 locating logs among several utilization processes. For. Prod. J. Sitter, R. M. 6(9):87-98. (LP to calculate pattern of 1969. Linear programming. British Columbia allocation of logs and intermediate prod- Lumberman 53(5) :31-34. (LP to provide a ucts.) good source of ideas for creating better prof it levels.) Penick, Jr., E. B. 1968. Application to machine loading in a Stoltenberg, C. H., and G. W. Thomson furniture plant. For. Prod. J. 18(2):29- 1962. Observations on the usefulness of 1 ine- 34. (LP for production planning problems ar programming in farm forestry. J. For. of machine loading in a furniture plant.) 60(10) :724, 728. (LP for farm forestry.)

Ramsina. K. D. Teeguarden, D. E., and Hans-Leopold Von Sperber 1968.' Linear programming for the plywood mix 1968. Scheduling Douglas-Fir reforestation problem. For. Prod. J. 18(4):98-101. (LP investments: a comparison of methods. for plywood mix problem.) For. Sci. 14 (4) :354-367. (Compares LP capital budgeting and rules of thumb for Rornesburg, H. C. schedul ing reforestat ion investments .) 1974. Scheduling models for wilderness recreation. J. Environ. Manage. 2(2):159-177. Thompson, E. F., and R. W. Haynes (Describes schedul ing model s for wi lderness 1971. A 1inear programming-probabil ist ic ap- recreation. Planning of a wilderness rec- proach to decision making under uncerta im- reation trail network that minimizes user ty. For. Sci. 17(2) :224-229. (A LP-proba- encounters and minimizes cost of campground bilistic approach to decisionmaking under maintenance and damage to ecologically sen- uncertainty. Minimizes present value of sitive areas. Problem originally required wood procurement costs for the planning quadratic programing. Simp1 ifying approxi- period .) mation made, to allow LP solution (11 percent less optimal), which can be decomposed Van Buijtenen, J. P., and W. W. Saitta for solving larger problems.) 1972. Linear programming appl ied to the economic analysis of forest tree improvement. Ruprich, J. J. For. 70(3) :164-167. (LP appl ied to eco- 1971. Methods of operational research and nomic analyses of forest tree improvement, planned selected operat ions in forestry. using percentage improvement expected from p. 64-71. In Operat ions research and the breeding program.) managerial economics of forestry. P. A. Wardle, editor. Great Britain For. Comm. Wardle, P. A. harvest operations in an industrial forest. 1965. Linear programming studies. p. 6-18. Maximizes present net worth using con- In Great Britain Forestry Commission, math- straints of annual crop volumes and re- ematical models in forest management: pro- generat ion acreage.) ceedings of the meeting held at the Univer- sity of Edinburgh on 12th and 13th April Yaptenco, R. C., and A. E. Wyl ie 1965. Great Britain For. Comm. Forest Rec. 1970. A quantitative approach to plywood pro- No. 59. H.M. S.O., London. i is cusses duction scheduling. For. Prod. J. 20(3): mathematical models in forest management.) 54-59. (LP for plywood production scheduling to maximize efficiency of machines in Ware, G. O., and J. L. Clutter production with constraints of material 1971. A mathematical programming system for movement, sequential specification of jobs the management of industrial forests. For. and storage facil ities for material -1 Sci. 17 (4) :428-445. (LP for schedul ing

Investigations on Recreational Forested Areas Ulrich Ammer

Abstract--In many Eur-ooean landscaoes, two opposing processes are being practiced concurrently. In urban populated areas forest land is being cleared for development, and in agrarian areas far from any settlement. uncultivated land is often afforested.

While nearly all olanners ayee that the loss of woodlands near pooulated areas is undesirable, they often disapprove of naking forested areas accessible through afforestation programs in agrarian problem areas because It may be harmful to landscape esthetics. The argument is advanced that increasing the amount of forested areas is done at the exoense of landscape amenit ies. Until now, this Judgment has o- iginated from the subjective ideas of several plannina professionals.

In order to present to forest Tanagers and landscape planners ecological, economic, and esthetic decisionmaking aids, invest iga- tions were carried out with the supoort-of the Deutsche Forschungs- gemeinschaft of Bader-Wurte~berg; the Ministry for Nutrition, Agri- cultural Economics, and the Environment; three provincial; and two municipal government organizations. The purpose of the study was to determine minimal, optimal, and maximum densities of representative forest land.

From these investigations, important planning d rectives would be derived which would.. .

1. Intensify the recreational and tourist function in rural areas, taking into consideration socio-ecological points of views, without causing damage to the attractiveness of the restive Iandscape, and

2. Achieve a balance between the quality of life and environmental protection on one side and permit the planning for the least amount of possible forested areas (even in populated sections) on the other.

STUDY AREA of judgment. Moreover, the esthetic value is always tied to associations of education, exper- The esthetics of any object is a matter of ience, and even psychic and physiological con- subjective phenomena, i.e., there exists no ob- ditions resulting even in a measure of personal jective, agreed-upon-abstract, absolute method judgment. This measure of judgment can change itself independently of place, time, social in- The choice of participants in the question- fluences, etc. It is therefore insufficient naire as well as their surroundings and their for landscape-esthetic investigations to call representativeness were chosen so that statis- upon the judgment of a series of "experts1'-- tical ly re1 iable results could be developed. more significant results can be obtained if it is derived from groups of diverse users. The results of the questionnaire confirm first of a1 1 that respondents preferred that a METHODOLOGY landscape not exceed 80 percent forested area. This upper limit supports findings in an earli- Definitions er study (Ammer and Lutz 1972) and is followed in forest management practices. Although a In contrast to the frequent previous in- considerable port ion of those being interview- vestigations in forestry's social aspects in ed felt comfortable with an even higher upper which the area being studied is more or less limit, the majority felt that a landscape con- described in verbal detail, we tried to present taining over 80 percent forested land was to our participants the object of a "forested mirky and oppressive. area" as quantitatively and as concretely as possible. In social science methodology, In terms of landscape esthetics, the lower sketches and pictures have proved eminently limit of preferred forested area among those successful. Consequently, pictorial represen- questioned was 20 percent. If a landscape had tation methods have been used successfully in less than 20 percent forested area, it was con- the conduct of numerous forestry-related social sidered "boring." science questionnaires, because they "concret- ize" the study object and lead the participant to more exact expressions of opinion. In contrast to these general results, no- t iceable d ifferences resulted when respondents Among the various photographic techniques were asked about residential areas in populated which are theoretically possible, (vertical, sections. In this instance, the lower l imit of horizontal , and obl ique), we chose the obl ique 20 percent of forested areas was regarded as aerial photograph. In this manner, a photo- too low, and the preferred proportion of for- graphic perspective was chosen with the part ested to residential areas was at least 30 per- of the forest nearest the study area with high cent. The upper 1imit was still 80 percent, frequency of user contact given more weight than those further away with 1esser frequency even in this instance. A value of 60 percent of user contact. The pictorial excerpt is was considered optimal for the populated sec- therefore del ineated so that the essential tion--this was statistically corroborated. identification feature of the landscape can be grasped. From these oblique photographs, ab- In rural areas, similar studies showed an stract graphical sketches were prepared which optimal value of 50 percent forested land. contain all the identifiable features, such as, This interesting difference in the conceptions site placement, hills and valleys, etc. All between inhabitants of rural and urban regions other factors which could influence the partic- might well be explained to a large degree by pant's response (such as color, shadow and the higher degree of environmental qua1 ity and light effects, clouds, etc.) can be kept con- quietude in rural areas. This explanation is stant. These oblique photo sketches were confirmed by the fact that in populated areas developed into a corresponding montage of the amount of preferred forested areas--from overlays illustrating increasing afforestation the viewpoint of recreation and relaxation-- in 10 percent increments. The degrees of af- was consideredx8 percent and the upper 1 imit forestation were derived in this model with was 90 percent. Interestingly, these atti- regard to economical 1y significant parameters, tudes correspond fairly closely to the wishes such as, site qua1 ity, re1 ief, development, of resort guests (65-66 percent) in invest iga- and ownership patterns. The interrelated tions of the tourist spots of Allgaus, the problems of photographic technique, choice of Black Forest, and the Swabian Alps. pictorial perspectives, and graphic represen- tat ion and other decisions are discussed by The optimal figure for recreation areas Hartweg (I976). was assumed to be 50 percent forested land in Kiemstedt (1967). When you realize that there Other investigators have demonstrated exists 20 percent (in uplands up to a maximum that those being questioned can describe and of 40 percent) of actual forested land, the identify adequately the actual forest cond i- confl ict of goals is apparent among agricul- tions. It was therefore assumed that the in- tural, typical recreational, or optimal recre- terviewees rely upon real representation when ational area landscapes, although in the opin- formulating their responses to questions about ion of the participants, a rational land econ- the experimentally represented forest areas. omy is still possible. Table ]--Recommended lower and upper limits and optimal values of forest density for planning of landscape units, by area and function.

Percent of forest to nonforest land Area and function Lower 1 imit Optimal value Upper limit

Populated areas:

Residential

Nearby recreation site Rural areas:

Residential with industrial parks

Agricultural product ion

Nearby recreation

Vacation and resort activities

From the results, some interesting direc- gefuhrt mit dem Ziel, Aussagen zur minimal, tions for planners are apparent in the recom- optimal und maximal vertretbaren Walddichte zu mendations for planning of landscape units (ta- mac hen. ble 1). The recommendations include lower and upper 1 imits and optimal values for forest den- Aus diesen Untersuchungen sollten sity--the percent of forest to non-forest land. planerisch wichtiqe Hinweise daraufhin erhalten werden: UNTERSUCHUNGEN ZUM FREIZEITORIENTIERTEN WALDANTEIL 1. Wie unter sozio-okologischen Gesicht- spunkten die Waldflache im landlichen Raum und Zusammenfassung in Gebieten mit Erholungs- und Fremdenverkehrs- funktion ansteigen kann, ohne daB Nachteile fur In vielen europaischen Landschaften Iaufe~ die Attraktivitat als Erholungslandschaft ent- 2 gegeneinander gerichtete Prozesse ab: in den stehen und stadtischen Verdichtunqsbereichen werden Wald- flachen fir Siedlung und Verkehr gerodet und in 2. Welche Mindestwaldausstattung andererseits den agrarischen siedlungsfern gelegenen Raumen auch fur Verdichtungsraume aus Grinden der entstehen Brachflachen, die haufig aufgeforstet Lebensqualitat und der Umwe1tvorsorge gefordert werden. und erreicht werden mussen.

WShrend sich nahezu a11e Planer daruber einig sind, daf3 die Waldverluste im siedlungs- LITERATURE C ITED nahen Bereich sehr negativ sind, werden die Waldzugange durch Aufforstung in den agrari- Ammer, V. U., and Werner Lutz schen Problemraumen nicht sel ten a1 s land- 972. Forest pol icy and planning problems schaftsasthetisch unerwunscht abgelehnt. Es in afforesting recreational and tourist wird argumentiert, der zunehmende Waldanteil regions--as shown in the example of the lasse landschaftl iche Reize verlorengehen. Bei Todtnau Gemarkung. Der Forst- und dieser Beurteilung ist man bisher weitgehend Holzwirt 13: 290-296. von subjektiven Vorstellungen einzelner Planungsfachleute ausgegangen. Hartweg, A. 976. Studies on the influence of forests on Urn fur die forstliche Praxis ebenso wie the landscape. 1976 dissertation, fur die Landschaftsplanung neben okologischen Freiburg. und okonomischen auch asthetische Entscheidungs- hilfen anzubieten, wurden mit Unterstutzung der Kiemstedt, H. Deutschen Forschungsgemeinschaft und des 967. h he recreation value of landscapes. 1 Ministeriums fur Ernahrung, Landwirtschaft und ~eitragezur Landespflege, Special Issue Umwel t Baden-Wiirtemberg in drei land1 ichen und 151, Verlag E. Ulmer, Stuttgart. (in 2 stadt ischen Gemeinden Untersuchungen durch- ~erman)