Landowner Objectives and Nonindustrial

Private Timber Supply

Jarl Kuuluvainen, Heimo Karppinen, and Ville Ovaskainen

ABSTRACT. In this paper, nonindustrialprivate forest owners' objectives are empirically identified,
and the link between ownership objectives and observed harvesting behavior is established by
estimating a theoretically derived timber supply function. Surveydata on 146 Finnishforest owners
and their timber sales in 1987-1991 is used. Prior to estimation, forest owners are classified into four
groups accordingto their ownershipobjectivesby K-means clustering. Dummy variables indicating
cluster membership are included in the supply function. Accordingto the results, "multiobjective
owners"
harvest
significantly
more(m3/ha/yr)thantheotherownergroups
(self-employed
owners,
recreationists, and investors), ceteris padbus. The results further indicate that the multiobjective
owners' harvestingpolicy can be described as present-valuemaximizingwhile the other (single-
objective)groups' harvestingbehaviorseems to reflectthe effect of market imperfections,as assumed
by the theoretical model of the study. FOR.Scl. 42(3):300-309.
Additional key words: Amenities, credit rationing,Tobit model.

fromthenonmarketed amenityservicesof theforest.Assum-

nimportant
conclusion
of
the
empirical
studies
on
the forestmanagement behaviorof nonindustrial
privateforest(NIPF) ownersis thattheirharvesting
decisionsare influencedby a numberof forestandowner
characteristics
(e.g.,Binkley 1981,J'firvel'fiinen
1981, 1988,
Loikkanen et al. 1986, Carltn 1990, Dennis 1988, 1990,
KuuluvainenandSalo 1991). A theoreticalexplanationput
forwardfor thisis thatin the presenceof marketimperfec-
tions,NIPF ownerscannotpursuepresent-value maximizing
harvestingpolicies.Empirically,however,the effectsof
ownercharacteristics tendtobefairlysmallascomparedwith
the effectsof pricesandthetimberstock.
Generally,a significantpart of the variancein NIPF
owners'observed annualharvestratesremainsunexplained.
Thismaybeduetothedifficultyof identifyingtheindividual
restrictionson optimizingbehavior.Anotherexplanation is
theindividualvariationin privatelandowners' management
motivations.Followingthis line of reasoning,we hypoth-
esizethat the diversityof ownershipobjectivesshouldbe
recognized whenmodelingNIPF ownerbehavior.
In the theoreticalmodels,the multiplicityof objectives
hasusuallybeentakeninto accountby assumingthat the
forestownerderivesutility not only from incomebut also
ing that the flow of nontimberservicesis monotonically
increasing in standage(standingstock),thiswill extendthe
optimalrotation(stock)beyondthepresent-value maximiz-
ing one(e.g.,Hartman1976,HybergandHolthausen1989,
Max andLehman 1988, Ovaskainen1992). However, dueto
insufficientempiricaldataon landowners'objectives,har-
vestrates,andstandingstock,thereisnoconclusive
evidence
on whether the formulation is relevant and whether landown-
ers who alsovalue nontimberbenefitsactuallyuse longer
rotationsor harvestless often. The comparisonbetween
industrialandNIPF ownersby NewmanandWear (1993) is
interestingin thissense.
The otherline of researchhasfocuseddirectlyon land-
owner attitudes,beliefs,and objectives(J'firvel'fiinen
1971,
Hahtola 1973, Kurtz andLewis 1981, Ferretti 1984, Gramman
et al. 1985,YoungandReichenbach
1987,Marty et al. 1988,
Blatner and Greene 1989, Bliss and Martin 1989, L0nnstedt
1989,EganandJones1993).While thesestudieshavebeen
ableto identifydifferenttypesof NIPF ownersanddescribe
theirmotivations,therelativeimportance of motivations
vs.
otherexplanatory factorsfor actualbehaviorhasnot been
sufficientlyanalyzed.

The authors are with the Finnish Forest Research Institute, Unioninkatu 40 A, FIN30170 Helsinki, Finland. An earlier version of the paper was
presented at the conference "Forestryand the Environment:EconomicPerspectivesI1," Banff, Alberta, Caneda, October 1994.
Acknowledgments: Helpfulcommentsfromthe anonymousrefereesandthe essociateeditor,as well es ErkkiKoskela,RobertMendelsohn,and
MarkkuOllikainenare gratefullyacknowledged.Responsibility
for remainingerrors is ours alone.
ManuscriptreceivedJune 6, 1994. Accepted September 14, 1995. Copyright 1996 by the Societyof American Foresters

300 ForestScience
42(3)1996
 In this paper,NIPF owners'objectivesare empmcally v2 = vo - h1+ F(vo - hl) - h2 (2b)
identified,and the link betweenownershipobjectivesand
observedharvesting behavioris established
simultaneously
controllingfor otherexplanatoryvariables.We showthat c1 = Plhl + m+ B (2c)
attitudesandownershipobjectivescanactuallyhavea quan-
tffiable role in the variation of timber harvest rates. Thus,
informationontheobjectivesof differentNIPF ownergroups
c2 = p2h2+ m2 - (1+ r)B (2d)
canbeusedto forecastmajortrendsin futuretimbersupply.
Evenmoreimportantmaybetheimplications thatthediver-
sity of motivationshas for the contentand marketingof
forestryextensionservices.Also, differentobjectivesmay B < Bu (2e)
give rise to very differentresponses
to forestpolicy mea-
sures. In Equation(1), ct(0 < ct< 1)is aparameter
thatmeasures
the
We first presenta Fisheriantwo-periodconsumption- relative utility weight betweennontimberamenitiesand
savings modelforaforestownerwhoperceives creditration- consumption
ofgoods
andservices.
Weuse[= (1+ fi)-ito
lngin thecapitalmarketandvaluesthenontimberbenefitsof denotethe discountfactorof futureutility, wherefi is the
basforestproperty.To our knowledge,thesetwo features forestowner'ssubjectiverate of discount.The exogenous
havenot beenconsidered simultaneously
in earliertheoreti- marketinterestrateis denotedby r.
cal modelsof NIPF behavior.We thenpresentour survey Ourinterestisin theshort-term
timbersupply(i.e.,supply
data on 146 NIPF owners in southern Finland and their timber froma givenstockof standingtimber).We thereforefocuson
salesin 1987-1991.Usingprincipalcomponent andcluster thefirst-period
harvesting
decision
only.1Assume
thatthe
analysis,we nextclassifytheforestownersintofour groups borrowingconstraint(2e) is binding,i.e., the forestowner
accordingto the relativeimportanceof threemajortypesof wouldliketo borrowmorethanis allowedbythecreditlimit.
objectives:nontimbervalues,salesincomeand economic In effect,borrowingwill bereplacedby theexogenous credit
security,and self-employment opportunities. Finally, we limit (denotedBU),andwe cansubstituteB =-Bu into(2c) and
estimatea Tobit timbersupplymodelwith dummyvariables (2d). The first-orderconditionsfor utility maximizinghar-
indicatinggroupmembership. Theresultsarereported forthe vestratesareobtainedby substituting (2a)-(2d) into (1) and
entiresampleandfor two subsamples (multiplevs. single differentiating
withrespect
to hI andh2. Rearranging,
the
objectivelandowners). "cuttingrule," or equilibriumconditionfor the first-period
harvest, can be written as follows (an interior solution is
TheoreticalFramework assumed
andprimesareusedto denotederivatives
of func-
tionsof onevariable):
2
Assume that the forest owner behaves as if he maximized
hisdiscounted utilityovertwoperiods(thepresentperiodand
the"future").Utility isanincreasing,
concave,andadditively [1+ F'(v)l(P2/ t5)
(3)
separable functionof theconsumption of goodsandservices = u'(q) / Iu'(c2)-ag'(Vl)/(D- alu'(c2)p
(c1,c2) andthepostharvest stockof standingtimber(v1,v2),
representingnontimberamenities.Modeling nontimber Becauseof thebindingborrowingconstraint,the interest
amenitiesas an increasingfunctionof the standingstock ratedoesnotentertheequilibriumconditionexplicitly,and
aloneis a simplification(cf. Swallowet al. 1990,Swallow themarginalvalueproductof theforestmustbe equatedto
andWear 1993)butit will sufficeto demonstrate the quali- themarginalrateof timepreference.
Thesecondquotienton
tativeimplications. theright-handsideadjuststhemarginalrateof timeprefer-
Timbersales
incomedepends ontheharvestrates(h1,h2) ence to take into account the valuation of the nontimber
andexogenous
timberprices(Pl,P2)duringthetwoperiods.
In addition,the forestownerhas exogenous nonforestry
income(m1, rn2).Capitalmarketimperfections arerecog- 1 When
theterminal
stock
hasvalue,
theentire
stock
willnotbeharvested
duringthe secondperiod.That is, the second-period harvestis also
nizedby assuming thatthe forestownerperceivesa credit endogenous, which is not the casein the two-periodmodelwithout
limit which he cannotexceed(cf. Koskela 1989). Further, nontimberor bequestvalues.For empiricalpurposes, however,we need
timbergrowthis assumed to bea strictlyconcavefunctionof notconsiderthecomparative staticsfor thesecondperiod.As shownin
Ovaskainen(1992), the qualitativeresultsfor the first periodremain
the standingstockafterthe first-periodharvest.The forest unchangedwhenthenumberof periodsis increased.
owner'sproblemcanbe presented asfollows: 2 Thefzrst-order
conditions
fortheoptimization
problem
in(1)-(2e)
canbe
Maximize written as

8U/ 8hI = (1- ct)u'(c

1)pl- ctlg'(vl
) + [g'(v2)[l
+ F'(vI )]} = 0
U =(1-oO[u(cl)+[u(c2)]+ot[g(vl)+[g(v2)
] (1) 3u/ 3h2 = [1[(1
- a)u'(c2)P2- ag'(v2)1 0.
according
towhichthemarginal utilityofconsumption mustbeequatedto
st. the marginalutility of the nontimberservices,i.e., standingstock(cf.
Binkley1981,MaxandLehman1988).Alternatively, thesameresultscan
bederived
byfirststating
(2e)asc1- Plhl- rn1< Buandwriting the
V1 ----V0 --h1 (2a) Lagrangian
functionZ(Cl,hl,h2,,
) = U + X(Bu- c1+Plhl+ rnl).

ForestSctence
42(3)1996 301
amenities.If tz = 0, the nontmberbenefitsare not valued and
this term vanishes.
Note that the modelin (1) through(2e) combinestwo
sources
ofnonseparation
whichhavepreviously
'been
ana-
lyzed separately.Credit rationing(Koskela1989) implies
that the forestownerwill harvestmoreat any given initial
stocklevel,andthattheremainingstockwill be less,thanis
impliedby present-value maximization.Fornontimberben-
efitspositivelyrelatedto the standingstock,the oppositeis
true(Ovaskainen1992).Thus,it is notapriori clearwhether
thefirst-periodharvestandoptimalstockin thepresentcase
will be greateror less than is implied by present-value
maximizationof commercialtimber profits only. This is
becauseonecannotsaywhethertheright-handsideof Equa-
tion (3) is greateror lessthan(1 + r).
The ordinarybehavioraltimber supplyfunctionis ob-
tainedby straightforward comparativestaticsanalysis.The
results can be stated as follows:
hi = h(Pl,P2,r, v0,m,m2,,Bu,
(4)
? ?-+- + +- -
The signsunderthevariablesindicatethedirectionof the
expectedeffects.Becauseof marketimperfections, the sup-
ply decisionisinfluencednotonlyby pricesandtimberstock
but alsoby theowner'sincome,subjectivepreferences with
respecttoconsumption andnontimberbenefits,andthecredit
limit. Prices have both a substitution effect and an income
effect, andtheir total effectscannotbe signeda priori (see
Kuuluvainen1990).Presentandexpected incomehavenega-
tive andpositiveeffects,respectively.Generally,theliquid-
ity effectsof creditrationingdominatethoseof nonmarket
amenities.(Thiscompareswith thecaseof priceuncertainty
under credit rationing(Koskela 1989), where the wealth
effectsdueto uncertainty arereplacedby liquidityeffects.)
For example,theinterestrate,representing therelativeprice
of thefirst-periodconsumption, hasanegativeincomeeffect
only.Thisis consistentwiththegeneralresultconcerning the
priceeffectsof rationedgoods(NearyandRoberts1980).
The effectof thetimberstockis unambiguously positive,yet
not equalto unityasis thecasewith creditrationingalone.
The supplyequation(5) includesthree more variables
which vary over individualsbut whichcannotbe directly
observed: thesubjective rateof discount ,therelativeutility
weightof thenontimberamenities, tz,andthecreditlimit Bu.
The rateof time preferencehasa positiveeffect,while the
effectof thecreditlimit is negative(i.e., improvedavailabil-
ity of creditwill reducetimberharvests).The resultfor the
amenityweight tz providessomepredictionsaboutthe be-
haviorof landowners with differentobjectives.The higher
thevaluationof the amenityservices,the smallerwill be the
harvest
inthepresent
period
and,since
v1= v0- hl, thelarger
the desired timber stock.
Data
We usesurveydatacollectedbyamailinquiryin 1990and
personal
interviewsin 1991froma sampleof NIPF ownersin
302 ForestSctence
42(3)1996
southernFnland.The samphngprocedurewas two-stage
arealclustersamplingwhereafarm'sprobabilityto enterthe
samplewasproportionalto itstotallandarea(seeKarppinen
andHinninen1990).The response ratein the mail inquiry
was78%.Theanalysis of thesampling error(Karppinenetal.
1994)didnotindicateanynonresponse biasthatwouldaffect
the presentresults.In the personalinterviewsthe response
rate was 94%. The number of observations used in the
estimationwas 730 (146 ownersover 5 yr).
Annualmicrodataon the quantifiesandpricesof timber
soldduringthe5 yr precedingtheinterviewwerecollectedin
thepersonalinterviewsin 1991onthebasisof writtensales
contracts.Theannualpricesusedin estimations arequantity-
weightedaveragesof the pricesof differenttimber assort-
mentssold.For the yearswherethe ownerhad not sold,
regionalaveragetimberpricesandownerandforestcharac-
teristicswereusedto forecasttheprice.The forestsof those
interviewed were also inventoried. For each woodlot, the
timber stockat the beginningof eachyear was calculated
using
therecursion
equation
vt= vt_! +f( vt_1) - hrTheaverage
percentage growthwasiteratedassuming thatgrowthf(.) s
linearin the stockandusinginformationon the actualstock
in 1982, 1986,and 1991andthe actualharvestin eachyear
The datasetwas augmentedby annualmicrodataon forest
owners'incomeandwealth.All monetaryunitsaremeasured
in realtermsdeflatedby the wholesalepriceindexwith the
baseyear 1990.
The informationonlandownerobjectives is basedonthe
mail inquiryof thesamesamplein 1990.In themail inquiry,
theforestownerswereaskedto assess theimportance of 21
differentobjectivesof forestownershipusinga three-point
scale(Important,Don't know,Not important).The potential
goalscomprised monetaryobjectives
aswell asrecreational,
emotional, and aestheticconsiderations.
Landowner Objectives
Threeof the variablesin the theoreticalsupplyequation
(4)-- subjectiverateof discount,relativeweightof amenity
services,andthecreditlimit--are directlyunobservable. To
find a proxy variable, we assumethat the unobservable
preferences arereflectedin thelandowner'sobjectives,
which
can be examinedusingsurveytechniques. The problemis
howto utilizethe surveydatain the analysis.
It is, of course,impossibleto includeall 21 different
objectivesdirectlyin theregression equation.Eventhough
the degreesof freedomdo not posea problem,the useof
the original variables is preventedby multicollinearity
Using the most "representative"original objectivess
furtherhandicapped by thefact thatthe objectivescannot
be measuredin a meaningful and comprehensivemanner
with single variables.What we need to find for the statis-
tical analysisis a limited numberof broad categoriesof
objectives.Therefore,we initially usedprincipalcompo-
nent analysisto condensethe informationin the original
variables into a few interpretable combined variables
(e.g., Harman 1970, Mulaik 1972, Lewis-Beck 1994).
Orthogonalby construction, theprincipalcomponents are
a convenientwayto avoidtheproblemof multicollinearity
Theresultingthreedimensionsof objectivesarepresented nthesupplyequauon,theircoefficientslackameamngful
m Table 1. quantitative
interpretation.
Also,it is obviousthatdiffer-
Theinterpretationof theprincipalcomponents isbasedon ent combinations of the main dimensions should be al-
theobjectiveswith thehighestloadingsoneachcomponent lowed.For thesereasons,the principalcomponentscores
(loadingsequalthe correlations betweenthe originalvari- weresubsequently usedasgroupingvariablesin a cluster
ablesand the principalcomponent). As regardsthe first analysis.We usedthe K-meansclusteringmethod(e.g.,
principal component, variables describing various Anderberg1973,Hartigan1975,Engelman1980),which
nonmarketed,recreational,and aestheticaspectshad high is a combinationof hierarchicalstem-to-leafalgorithm
loadings. Consequently,it was taken to representthe anditerative
partitioning.
4
"nontimbervalues"of forestownership.Monetary objec- It turned out that the forest owners could be classified into
tryeswerehighlyloadedonthesecond principalcomponent. four groups:"multiobjective owners"(36% of forestland
This componentwas thereforelabeled"salesincomeand area), "recreationists " (18%), "self-employedowners"
economicsecurity."The third principalcomponentwas (26%), and"investors" (20%).Exceptfor themultiobjective
characterizedby the employmentopportunities and labor owners,the grouplabelsarebasedon theprincipalcompo-
ncomefrom deliverysales,as well as forestwork "for its nentwith the highestmeanscore(Table 2).
own sake." This dimension was therefore called "self-em- Multiobjectiveownersvalueboththemonetaryandame-
ployment
opportunities."
3 nity benefitsof theirforests,asindicatedby thefactthatall
However,theprincipalcomponentsassucharenotvery threeprincipalcomponents have relativelyhigh positive
usefulin explainingtimbersupply.If theprincipalcompo- scores.Recreationists emphasizethe nontimberand non-
nentscoresareincludeddirectlyasexplanatoryvariables monetaryvaluesof forestownership.Self-employed owners
emphasize theemployment opportunities andlaborincome
Table 1. Objectives of forest ownership. Principal component providedby theforestproperty.For theinvestors, theforest
analysis,arimax rotation (loadingsbelow 0.250 are denoted by propertyis anassetanda sourceof regularsalesincomeand
0.000).
economicsecurity.For the interested reader,thegroupsare
characterizedin Table 3 in terms of directly observable
I II III
owner and forest characteristics.
Outdoor recreation 0.735 0.000 0.000 There were few statisticallysignificantdifferencesbe-
Solitude and meditation 0.638 0.000 0.000 tweentheownergroupsatthe5% level.In part,however,this
Residential environment 0.636 0.000 0.386
maybe dueto smallsamplesize.Usingbasicallythe same
Berry-picking 0.608 0.000 0.000
Nature protection 0.587 0.000 0.000 data(N = 245) andlogitmodels,anotherstudyby Karppinen
Aesthetic values 0.580 0.000 0.000 (1995) suggeststhat owner and forest characteristicscan
Inherent value 0.571 0.000 0.000
actuallybe usedto identifyownergroupswith different
Roots in native locality 0.506 0.403 -0.346
Asset motive 0.000 0.655 0.000 objectives.
Funding of investments 0.000 0.602 0.000
Regularsalesincome 0.000 0.578 0.252
Securityagainstold age 0.000 0.505 -0.337 Table 2. Forestowner groups based on the objectivesof forest
Hedging motives 0.000 0.452 0.356 ownership (N= 146).
Speculativemotives 0.266 0.470 0.000
Laborincomeand employment 0.000 0.379 0.651
Forest work 0.000 0.000 0.639
Mean of principal
component scores
Household timber 0.000 0.000 0.611
Securityagainstinflation 0.354 0.322 -0.391 Owner group n I II III
Hunting 0.000 0.322 0.371
Credibility 0.000 0.341 0.000 Multiobjective owners 51 0.649 0.768 0.324
Inheritance 0.280 0.277 0.000
Recreationists 25 0.668 -0.722 -1.254
Self-employedowners 42 -0.262 -0.766 0.675
Eigenvalue 3.430 2.593 2.183 Investors 28 -1.386 0.395 -0.482
Proportion explained 0.18 0.12 0.10 F-ratio 75.843 46.857 44.531
P-value 0.000 0.000 0.000
NOTE; Interpretation of the principal components:
III "ontimber
"Sales income
values"
and economic security" NOTE: Interpretation of the principal components:
III "Sales
"ontimber
income
values"
and economic security"
III "Self-employment opportunities"
III "Self-employment opportunities"

3 Some
seemingly
contradictory
loadings
are
explained
bytheinstitutional
environment in Finland.First,"regularsalesincome"hasa highloading
onprincipalcomponent II but"laborincomeandemployment" oncompo- 4 InK-means
clustering,
each
case
isfirst
assigned
tothecluster
whose
nentI. Theexplanation isthatonlyabout25%oftimbersalesaredelivery center has the smallestEuclidean distancefrom the case. Second,the
saleswherethesellerdoestheloggingandhauling,andevenin thatcase, maximumdistanceof a casefrom the centerof an originalclusteris
stumpage is thedominant partof totalsalesincome.Thus,regularsales minimized.Thus,K-meansclusteringalgorithmiterativelyreallocates
incomeandlaborincomearenotthesamething.Second, huntinghasonly cases into clusters whose center is closest to the case. As the Euclidean
lowloadings because veryfewlandowners ownsufficiently
largeareasto distance
is sensitive
to theunitsof measurementandmulticollinearity
of
enablehuntingontheirownlandalone(hunting rightsaremostlyhiredby thegrouping variables,principalcomponentscores
[standardized
N(0,1)
huntingclubs). andorthogonal]havedesirable properties
asgroupingvariables.

ForestSctence
42(3)1996 303
 Tabla 3. Owner and forest characteristicsby groups of ownarship objectives.

Multiobjective Self-employed
owners Recreationists owners Investors

Ownerbackground,
percentof forestownersor farms
Farmer 58 27 46 61
Nonfarmer 42 73 54 39
Resident owner 75 29 80 43
Absentee owner 25 71 20 57
Small woodlots(5-20 ha) 37 57 40 30

Forestand ownercharacteristics,
groupmeans
Forestland(ha) 37 26 31 40
Standing
stock
(m3/ha) 121 126 132 122
Arable land (ha) 7 3 9 5
Owner's age (yr) 62 60 53 55
Income(1000 FIM/yr) 124 128 112 140
Wealth (1000 FIM) 413 366 360 524

Salesbehavior,groupmeans
Average
sales
(m3/ha/yr) 3.0 3.4 3.0 3.1
Salesinterval(yr) 2.7 3.3 2.7 3.1

Econometric Analysis of Factors Affecting As showntheoretically,the harvestingdecisiondepends

NIPF Timber Supply ontimberpricesand,outof equilibrium,onthetimberstock,
as well as on owner characteristics.
In addition,dummy
EstimableTitnberSupplyFunction variablesrepresentingowner groupswith differentobjec-
A cross-sectional
censoredregressionor Tobit model tives are included.
(Tobin
1958,
Maddala
1983)
isestimated.
Wedenote
byh*ti
j TheordinaryTobitmodelassumes thateachforestowner
thelatentdependent variable--theharvestingintensity.It is in eachyearis anindependent observation, whichmeansthat
assumedthatthe harvestingintensitycanbe positive,nega- all coefficientsareconstant,andthedisturbance capturesthe
tive, or zeroeventhoughtheactualharvesthasto be nonne- differencesover time and individuals(Judgeet al. 1985).
gative,and that the intensityis a linear functionof the Obviously,thisassumption is restrictive.One way to relax
exogenous variables.The Tobit modelusesthe information theassumption, andto allowfortheeffectsof omittedowner-
available
forh*ti
j above
zero.
Therefore, specificfactors,would be to assumeindividual constant
terms(e.g., MaCurdy 1981). Insteadof this, however,we
hti
= h*tj= i'xtq+etij if theright-hand
sideispositive takethevariationdueto the"preference-related" factorsinto
ht=0 otherwise accountby introducingdummy variablesindicatingthe
landowner' smembership in oneof thefourgroupsbasedon
3 ownership objectives.Thisallowsusto test,atthefirststage,
where
hti
j istheactual
harvest
inm/haofowner
i (i = 1,.. whether there are differences in the harvest rates between
, 146) in ForestryBoardDistrictj (j = 1,2,3; relevantfor
forestownerswith differentobjectives.
regional
prices
only)inyeart (t= 1..... 5).Further,
xti
j are Theoreticaljustificationfor the expectedsignsof the
theexogenous
variables,[I is thevectorof theparameters
to
groupdummiesis providedby the resultfor the amenity
beestimated,
andetijisthenormally
distributed
error
term. weight,, in Equation(4). The multiobjectiveowners,who
The independent
variablesand their expectedsignsare as
follows: valuebothmonetaryandnonmonetary benefits,canbe taken
to represent
a "medium"amenityweight(thereferencecase).
Pit = timberpricelevel(netof harvesting (?) In comparisonwith this, self-employedownerscan be ex-
costs),
FIM/m3
pectedto place less weight on nontimberservicesand to
APj(t-1)= firstdifference
ofregional
timber
price (?)
q = 1,2,3) harvestmore, ceterisparibus. The recreationistowners'
vl(t-1)= standing
stockof timberperhectare, (+) emphasison nontimbervaluescanbe interpretedas a high
m3/ha amenityweight,sotheycanbe expectedto harvestless.As
growthi = meanpercentage growthfor the5-yr (?) the investors indicated no marked interest in nontimber
period,% values,they couldharvestintensively.On the otherhand,
rnit = totalannualpretaxincome,1000FIM (-)
theymayprefera relativelyhightimberstockasaninsurance
wit = taxablewealth(permanent income), (+)
1000 FIM againstunexpectedeventsand thereforeharvestconserva-
ageit = forestowner'sage,yr (-) tively. Thus,the total effectremainsa priori ambiguous.
dl= multiobjective
owner (0) Thepresenttimberpricelevelis usedto takeintoaccount
d2 = self-employed
owner (+) the"medium-term"
priceeffect.
5Thepricelevelrefers
tothe
d3 = investor (?) dataon individualpricescollectedduringthe interview,so
d4 = recreationist (-) thereare730 observations
for thisvariablein all. The impact

304 ForestSctence
42(3)1996
(short-term) priceeffectis accountedfor by thefirstdiffer- Tabla 4. Tha astimated coefficiants, t-values and alasticities of
the Tobit modal for Finnish NIPF ownars' annual timber sales
encesof the regionaltimberprices.Usingdifferences is a (m3/ha),1987-1991(N= 730for 146ownamover5 yr).
commonpracticein time-series econometrics (e.g.,Hendry
andEricsson1991).Here,it impliesthehypothesis thateven Coefficient

if the forestownersdo not sell timber every year, they do Independentvariable (t-value) Elasticity
follow short-term pricechanges asindicatedby theregional Constant -8.40
averageprices.Forthepricedifference,we have15observa- (2.67)
uons(3 regionsover 5 yr). Timberpricelevel(Pit) 1.66 0.41
In additionto thevolumeof theperhectaretimberstock (1.49)
at the beginningof eachperiod,we includethe average Pricedifference
(Pjt-Pj(t-1)) 0.06 0.03
(i.55)
percentage growthof thestock.Thelattercaptures
theeffects Standing
stock
(vi(t_l)) 5.06 0.99
of thewoodlot'sagestructure andaveragesitequality.The (4.81)
effectremainsa priori ambiguous, becausehighmeanper- Meanpercentage
growth(growthi) 0.55 0.36
centagegrowthcaneitherbe dueto goodsites(with a high (2.72)
harvest potential) or reflect the dominanceof immature Income
(init) -0.73 -0.13
(1.37)
stands(implyinga low probabilityof harvest).The interest Wealth(wit) 0.31 0.25
rate could not be included,becausewe did not have accessto (2.85)
the actual interest rates that individual forest owners cur- Owner'sage(ageit) -0.06 -0.47
rentlypaidfor theirloansortotheirinterestrateexpectations. (2.03)
The aggregate interestrate,in turn,wasstronglycorrelated Multiobjective owners(dl) Reference
group
with the pricedifference(r = -0.87). Self-employedowners(d2) -1.51 -0.77 a
(1.72)
EstimationResultsfor the Entire Sample
Investors(d3) -1.98 -1.11 a
The estimationresultsfor theTobit modelarereportedin (2.03)
Table4. Column2 givestheestimatedcoefficientsandtheir Recreationists
(d4) -1.90 -1.01 a
t-values(in parentheses).
Theelasticities
for all observations (1.80)
aregivenin column3. Tobit elasticitiesfor all observations
o 8.3
expressthetotaleffectsof theindependent variableson the
(26.53)
expected averageharvest,includingeffectsontheprobabil- Log-likelihood -1,655.3
ity of enteringthe marketandon thequantityharvestedfor PseudoR2 b 0.08
observationsabove the limit (cf. McDonald and Moffit Expectedharvestat means 3.35
1980). Bera-Jarque
Z2 c 1.21
Accordingto the Bera-Jarquechi-squaretest,the sym- a The "elasticities"for the group dummies directly express the average
metricallytrimmedresidualsof themodelarenormal,which difference(m3/ha/yr)in expectedsalescomparedto the reference
group.
suggeststhat the parameterestimatesare unbiased.The 2
b Pseudo
R2= [Z,x/(o
+[Z,x)
where
Z,xiSthesample
covariance
matrixof
reportedR-squaredis almosttwiceashighasit is turnedout the regressors (see e.g. Laitila 1993).
to be withoutthe groupdummies.As is typicalof cross- c The Bera-JarqueZ2teststatistics
for normalityis computedusingthe
symmetrically
trimmedTobit residuals,l(x[[>O)[min(i,2x[[)-xi'[]
sectionalmodels, it is still rather low. where / is the indicator function (cf. Pagan and Vella 1989). The 5%
criticalvalue with two degreesof freedom is 5.99.
Some of the coefficientsfor the group dummiesin
Table 4 are consistentwith a priori expectations.
In part,
however, the results are at variance with conventional Secondly,thereare no significantdifferencesin the an-
presumptions. First,thegroupwiththelargesttimbersales nualper ha harvestratesbetweenthe recreationists, inves-
turns out to be the multiobjectiveowners (the reference tors,andself-employedowners.Thisisinteresting giventhat
group). The elasticity for d3 shows that the investors' on the basisof statedobjectives,the investorsand self-
meanannual
timbersales
perhaare1.1m3/hadyr
smaller employedownersassignthe mostweight to the monetary
than the multiobjectiveowners', ceteris paribus. The benefitsand could be expectedto harvestmore than the
differenceis significantat the5% level. Recreationistand recreationists.
The resultssuggestthatwe couldhaveused
self-employed ownersalsoharvestlessthanmultiobjective only one dummy variable,multiple vs. single objective
owners(at 7.1% and 8.5% levelsof significance,respec- landowners. Whentried,thisdummyproducedan expected
tively). In summary,forestownerswhoassignmoreor less difference
of 1m3/hadyr
(significant
atthe5%level).
How-
equal weights to both the monetary and nonmonetary ever, consideringall four groupsseparatelydemonstrates
aspectsof forestownershipseemalso to be more active thatratherdifferentobjectivescanresultin a very similar
users of their property's harvestingpotential than are harvestingbehavioras regardsat leastthe averageharvest
. level.
"single-objective"owners, who exclusively value one
dimensionof their forestproperty. Of specialinterestis theresultfor therecreationists.
On
thebasisof statedobjectives,thisgroupwouldseemtobethe
5 Intime-serieseconometrics
this isnormally referred
toasthelong-term mostfar removedfromtheinvestorsor self-employed own-
priceeffect,butthemedium-term effectis morerelevanthereas we are
dealingwiththeadjustment
ofthetimbersupply atthegivenleveloftimber ers,who emphasize the monetary aspects.In termsof harvest
stockandsilvicultural
activity. rates,however,they do not differ that much. This should

Forest
Sctence
42(3)1996 305
serveas a warrangthat too far-reachingconclusionsshould
not be madeconcerning the effectsof ownerobjectiveson
timbersupplywithoutsimultaneously considering theobjec-
tives andactualharvestrates.Further,the usualassumption
thatnontimberbenefitsarea monotonically increasing func-
tion of the standingstock might suggestthat recreation-
orientedowners'propertiestend to be dominatedby fully
stockedmaturestands.Ourempiricaldata,however,doesnot
supportthisconjecture.Therewere no statisticallysignifi-
cant differencesbetweendifferent owner groups'per ha
timberstocksat the 5% level. 6
Fortheotherindependent variables,it sufficesto notethat
the effects are rather similar to those obtained in earlier
studies.The dominantdeterminantof timber salesis the per
ha standingstock.The elasticityis closeto unity (cf. Dennis
1988, 1990,KuuluvainenandSalo 1991).The meanpercent-
agegrowthis alsosignificantwith apositiveeffectthatwould
seemto reflectdifferencesin sitequalityratherthanthe age
structureof the woodlot. The medium-termelasticitywith
respectto timberpriceis 0.4, which seemsreasonable. The
fairly high t-valuefor the pricedifferencesuggests that the
ownersdo follow market developments.However, the t-
valuesindicatethatbothpricevariablesareonly significant
at somewhat more than 10% risk.
Multiple vs.Single ObjectiveLandowners
A differencewas foundbetweenthe multiobjectiveand
single-objective owner groupsas regardsharvestlevels.
However,whenusingthegroupdummiesin a singlemodel,
we implicitly assumedthat differentobjectivesgive rise to
differencesin harvestlevelsonly, while the effectsof other
independent variablesarethe samefor all groups.Theoreti-
cally, it is obviousthat we shouldalsotestwhetherowner
groupswith different objectiveshave different parameter
vectorsaswell. The problemis thatoursamplesizedoesnot
supporttheestimationof four separatesupplyequations. We
therefore chose to estimate separate models for two
subsamples: themultiobjectiveownersasdefinedabovevs.
single-objective ownerscomprisingall threeothergroups.
The resultsare reportedin Table 5. The Bera-Jarque
statisticsindicatenormalityof residualsfor both models.
Accordingto the Wald statistics,the parametervectorsfor
thetwosubsamples
dodifferatthe1%levelofsignificance.
7 othertheoreticallyjustified variableswere simultaneously
Someinterestingremarkscanbe made.
For multiobjectiveowners,the only variablethatis statis-
ticallysignificantat the5% levelis thefirstdifferenceof the
regional stumpageprice. The interpretationis that the
multiobjectiveowners'short-term harvestingdecisions actu-
ally follow the present-valuemaximizingharvestingpolicy.
Thatis,theforestownersin thisgroupdonotseemtoperceive
liquidity constraintswhile making their short-termsupply
6 Similar
conclusions
were
obtained
byasking
thelandowners
whether
they
consideredtheirfarm to be usedmainly for agriculturaland/orforestry
production
or for recreation
andresidence. Thefarm'sorientation hadno
impactonharvesting behavior.Further,the averagetimberstockfor the
recreational
farmswas114m3/hawhile129m3/ha
fortheproduction-
orientedfarms. Notably, the statedmain orientationmay not be truly
exogenous. Someanswersin favorof mainlyrecreational usemayreflect
a very smallharvestingpotentialdueto intensivecut earlier.
306 ForestSctence
42(3)1996
decisions. In addmon, their harvest rate and tmber stock
seemto be closeto thelong-termequilibriumlevel,because
eventhestandingstockhasno statisticallysignificanteffect.
(Theinterestrate,whichistheoretically importantforpresent-
valuemaximizingharvestpolicy,couldnotbe included.)
Forthesingle-objective owners,neitherthepricelevelnor
pricechangeare statisticallysignificant.Instead,theeffects
of thetimberstock,meanpercentage growth,presentincome,
andwealth(i.e., expectedpermanentincome)areall signifi-
cant at the 5% level. The resultssuggestthat the harvesting
behaviorof the single-objectivegroupsis more clearly h-
quidity-constrained andtheir lifecycleharvestreflectsper-
ceivedcreditrationing.Unlike the multiobjectiveowners,
they do not seemto follow short-termprice changes.Ths
interpretationis supported by the fact that single-objective
owners are often absenteeowners (46%) for whom timber
salesare less frequentand following the market therefore
more difficult (the sales probabilities for single vs.
multiobjectiveownersare0.46and0.57,respectively).Theo-
retically,however,theresultcanalsobe explainedby substi-
tution and income effects that cancel each other out.
For single-objective
owners,thecoefficients
of ownerand
forestcharacteristics
differfromzeroatthe 1%levelof signifi-
cance,accordingto the Wald statistics.For multiobjective
owners,thesecoefficientsdiffer from zero at the 5% level. Due
tothe"almost"significant
coefficients
of thepresent
incomeand
owner'sage, it remainssomewhatambiguouswhetherthe
multiobjectiveownersfollowtheunconstrained present-value
maximizingharvestpolicy.The positivecoefficientof present
incomesuggests thatthe assumption aboutthe separabilityof
utilityovertimeandbetweenconsumption andamenitiesmay
betoostrong(cf.Binkley1981).Thenegativecoefficient of the
owner'sagerefersto thebequest motive,because
timbersales
werefairlyconstant
in all ageclasses
exceptforownersover65
yearsof age(notreported).
Conclusion
This papersuggests that NIPF owners'timber salesare
connectedwith the objectivesof forestownership.Unlike
earlier studieson landowner attitudes,we consideredthe role
of ownershipobjectivesusinga statisticalmodelin which
controlled.The biggestdifferencein mean annualtimber
saleswas foundbetweenthe multiobjectiveowners,on the
one hand, and investors, on the other. The differences be-
tween the single-objectivegroups(self-employedowners,
recreationists,and investors)were small and statistically
insignificant.On average,the single-objective
groupssold
approximately
I m3lessperhaperyrthanthemultiobjective
owners.
In the earlier discussionon NIPF owners' harvesting
behaviora divisionhasbeenmadebetween"profit max-
The groupsareseparate
subsamples withdifferentvaluesforbothdepen-
dentandindependentvariables,andthe"restricted"modelfor all owners
cannotbededucedfromtheunrestricted
oneby anylinearrestrictiononthe
parameters.Thatis why we usethe(nonnested) Wald testfor structural
change,whichallowsunequalvariances forthetwoownergroups(Greene
1990,p. 215; seealsofootnoteto Table5).
 Table 5. The estimated results for the Tobit tmber supply model for multiple vs. single objective
landowners.

Multiobjective owners Single-objectiveowners

Coefficient Coefficient
Independentvariable (t-value) Elasticity (t-value) Elasticity

Constant -1.37 -12.90

(0.32) (3.00)
Timberpricelevel(Pit) 0.95 0.26 1.91 0.45
(0.69) (1.17)
Pricedifference
(Pjr-Pj(t-1)) O.15 0.08 0.01 0.006
(2.75) (0.21)
Standing
stock
(vi(twl)) 1.37 0.29 6.74 1.28
(0.87) (4.84)
Meanpercentage
growth(growthi) 0.03 0.02 0.72 0.45
(0.11) (2.51)
Income
(rnit) 1.18 0.24 - 1.72 -0.32
(1.59) (2.37)
Wealth(wit) O.13 O.12 0.40 0.32
(0.87) (2.85)
Owner'sage(ageit) -0.06 -0.56 -0.06 -0.46
(1.49) (1.61)

( 6.77 9.17
(17.29) (20.26)
N 255 475
Log-likelihood -620 -1024
PseudoR2 0.10 0.09
Expectedharvestat means 3.47 3.27
Probabilityof sales 0.57 0.46
Bera-Jarque
Z2 2.14 0.71
Wald statistics a 27.7

a TheWaldstatistics
forstructural
change
between
thesubsamples
isobtained
as(1-2)(Vl+V2)-l(1-2)
where Piand
Vi arethe estimatedcoefficient
vectorsandvariances
for thetwo subsamples
(i = 1,2).ThecriticalzZwith8 df at the
1% risk level is 21.7.

mizers,"with monetaryobjectivesonly, and"utility maxi-
mizers" who also value the nonmarket benefits (e.g.,
Hyberg and Holthausen 1989). However, it seems that
NIPF owners' objectives are rather more complicated.
Naturally, landownerattitudesand objectivesdependon
the cultural, institutional, and economic environment in
eachcountryso that the actualgroupingpresentedin this
paper is specific to Finland. However, our general ap-
proach-i.e., attemptingto empiricallyidentifylandowner
objectivesand incorporatingthe informationin a regres-
sion-basedmodel--seemsmore generally applicable.
Contraryto a priori expectations,
themultiobjectiveown-
ers who value both monetaryand nonmonetarybenefits
harvestedsignificantlymore per ha per yr than the other
groups,ceterisparibus. Furthermore,it wasthemultiobjective
ownerswhoseharvesting behaviorseemedtobeclosesttothe
unconstrained, present-valuemaximizingharvestingpolicy.
For the single-objectiveowners,forestandownercharacter-
sticswere importantdeterminantsof timber harvest.Re-
flectingthe effectsof perceivedliquidityconstraints,their
behavior seemed to be consistent with the theoretical model
presentedin thispaper.
Another finding contrary to the expectation that
nontimberconsiderationsimply a larger standingstock
was that the recreationists'per hectare stocks did not
differ significantlyfrom the other groups.In fact, our
theoreticalmodel showedthat the expectationis not even
a priori justified if the forestowner simultaneouslyfaces
constraintssuchasperceivedor actualcreditrationing.On
the otherhand,the assumptionthat nontimberamenities
are monotonicallyincreasingin the standingstock may
not be justified. Empirical findings by Englin and
Mendelsohn(1991) indicate that people eventually be-
comeoversatiatedwith respectto old-growth stands(see
also Binkley 1981, Swallow et al. 1990, Swallow and
Wear 1993).
Theresultsconcerningtherole of ownershipobjectives
can be usedin evaluatinglong-termtrendsin NIPF own-
ers' timber supply and in predicting their responsesto
forest policy. Even more importantly, the knowledgeof
ownershipobjectivesmay help forestryextensionorgani-
zationsin allocatingtheir resourcesand in adjustingthe
supplyandcontentsof extensionservicesto matchNIPF
owners' diverse motivations. For these purposesit is
importantthatownergroupswith differentobjectivescan
be identifiedin termsof moreeasilyobservableownerand
forestcharacteristics.
However,thisis outsidethescopeof
the presentinvestigation.

Forest
$ctence
42(3)1996 307
Literature Cited KARPPINEN, H., AND H. HANNINEN. 1990. Ykstystlojen
hakkuumahdollisuuksien
kiyttOEtelli-Suomessa.
Summary:Actual
ANDERBERG,
M.R. 1973. Clusteranalysisfor applications.Academic and allowablecut in nonindustrialprivate woodlotsin southern
Press, New York. Finland. Folia Forestalia 747.

BINKLEY,
C.S. 1981. Timber supplyfrom private forests.Yale Univ.,
School of For. and Environ. Studies, Bull. No. 92,
BLATNER,K.A., AND J.L. GREENE.1989. Woodland owner attitudes
towardtimberproductionandmanagement.
Resour.Manage.Opt.
6(3):205-223.
BLISS,J.C., ANDJ.A. MARTIN.1989. Identifying NIPF management
motivationswith qualitativemethods.For. Sci. 35(2):601-622.
CARLIgN,
O. 1990. Private nonindustrialforest owners' management
behaviour:An economicanalysisbasedon empiricaldata.Swedish
Univ. of Agric. Sci., Dep. of For. Econ.,Rep. 92.
DENNIS,
D. 1988.An econometricanalysisof harvestbehavior:Integrat-
ing ownershipand forestcharacteristics.
For. Sci. 35:1088-1104.
DENNIS,
D. 1990. A probitanalysisof harvestdecisionsusingpooled
time-series and cross-sectional data. J. Environ. Econ. Manage.
18:176-187.
EGAN, A., ANDS. JONES.1993. Do landownerpractisesreflect beliefs?
J. For. 91(10):39-45.
ENGELMAN,L. 1980. Annotatedcomputeroutput for BMDPKM--K-
meansclustering.BMDP StatisticalSoftware,Univ. of California,
Tech. Rep. 71.
ENGLIN,
J., ANDR. MENDELSOHN. 1991.A hedonictravelcostanalysisfor
valuation of multiple componentsof site quality: The recreation
valueof forestmanagement.J.Environ.Econ.Manage.21:275-290.
KARPPINEN,
H., H. H'fi,
NNINEN,and P. RPATTL1994. Mail inquiry vs
personalinterview:The problemof reliability and non-response
Scand. For. Econ. 35:110-130.
KOSKELA,E. 1989. Forest taxation and timber supply under price
uncertainty:Creditrationingin thecapitalmarkets.For. Sci.35:160-
172.
KURTZ,W.B., ANDB.J. LEWS.1981. Decision-makingframeworkfor
nonindustrial
privateforestowners:An application
in theMissouri
Ozarks. J. For. 79(5):285-288.
KUULUWdNI,J. 1990.Virtualpriceapproach to short-term
timbersupply
undercreditrationing,J. Environ.Econ.Manage.19:109-126.
KUULUVAINEN, J., ANDJ. SALO.1991. Timber supply and life cycle
harvestof nonindustrial
privateforestowners:An empiricalanalyss
of the Finnish case. For. Sci. 35:1011-1029.
LAITILA,T. 1993. A pseudo-R measure for limited and qualitative
dependentvariablemodels.Journalof Econometrics56:341-356
LEWS-BEcK,
M.S. (ED.) 1994. Factor analysisand relatedtechniques
SagePublications,London.
LOIKKANEN,H.A., J. KUULUVANEN,ANDJ. SALO.1986. Timber supplyof
private nonindustrialforest owners.Univ. of Helsinki, Dep. of
Econ., Res. Rep. 50.
LONNSTEDT,
L. 1989. Goalsandcuttingdecisionsof privatesmallforest
owners. Scand. J. For. Res. 4:259-265.

FERRETTi,W.A. 1984. A pilot study of nonindustrialprivate forest MACURDY, E. 1981.An empiricalmodelof laborsupplyin a life-cycle
landownermotivation.Ph.D. diss.,StateUniversityof New York. setting.J. Polit. Econ. 89:1059-1085.
273 p.
MADDALA,
G.S. 1983. Limited-dependent
and qualitativevariablesn
GRAMANN,J.H., T. MARTY,ANDW.B. KURTZ.1985.A logisticanalysisof econometrics.
CambridgeUniversityPress,Cambridge.
theeffectsof beliefsandpastexperienceon management plansfor
non-industrial privateforests.J. Environ.Manage.20:347-356. MARTY, T.D., W.B. KURTZ, AND J.H. GRAMANN.1988. PNIF owner
attitudesin the Midwest: A casestudyin Missouri and Wisconsm
GREENE,
W.H. 1990.Econometric
analysis.MacmillanPublishingCom- North. J. Appl. For. 5(3):194-197.
pany, New York.
MAx, W., ANDD.E. LEHMAN.1988.A behavioralmodelof timbersupply
HAHTOLA,
K. 1973. The rationaleof decision-makingby forestowners. J. Environ. Econ. Manage. 15:71-86.
Acta For. Fenn. 130.
McDONALD,
J.F., and R.A. MONT. 1980. The nsesof Tobit analyses
HARMAN,
H.H. 1970.Modernfactoranalysis.Ed. 2. Univ. of Chicago Rev. Econ. Stat. 62:318-321.
Press,Chicago.
MULAm,S.A. 1972. The foundationsof factor analysis.McGraw-Hall,
HARTMAN,
J.A. 1975. Clusteringalgorithms.Wiley, New York. New York.
HARTMAN,R. 1976.The harvestingdecisionwhena standingforesthas
NEARY,J.P. ANDROBERTS,
K.W.S. 1980. The theory of household
value.Econ.Inquiry 14(1):52-58.
behaviourunderrationing.Eur. Econ.Rev. 13:25-42.
HENDRY,D.F., ANDN.R. ERICSSON.
1991. An econometricanalysisof
U.K. moneydemand.In Monetarytrendsin theUnited Statesandthe NEWMAN, D.H., ANDD.N. WEAR.1993.Productioneconomicsof private
UnitedKingdom.Am. Ecoa.Rev. 81(1):8-38. forestry:A comparisonof industrialand nonindustrialforestown-
ers. Am. J. Agric. Econ. 75:674-684.
HYaERG,B.T., ANDD.M. HOLTHAUSEN.
1989. The behavior of nonindus-
trial privateforestlandowners.Can. J. For. Res. 19:1014-1023. OVASKANEN,V. 1992. Forest taxation, timber supply, and economic
efficiency.Acta ForestaliaFennica233.
J'ARVELINEN,
V-P. 1971.Factorsinfluencing
silviculturalactivity.Comm.
Inst. For. Fenn. 73(2). PaGaN,A.R., ANDVELLA,F.A. 1989. Diagnostictestsfor modelsbased
on individualdata:A survey.J. Appl. Econ. 4:29-59.
J.RVEL.INEN,
V-P. 1981. Hakkuuktyttiiytyminen
yksityismetsiilSilli.
Summary:Cutting behaviourin Finnishprivate woodlots.Folia SWALLOW, S.K., P.J. PARKS,ANDD.N. WEa. 1990. Policy-relevant
Forestalia 499. nonconvexitiesin the productionof multiple forest benefits J
Environ. Econ. Manage. 19:264-280.
JRVEL.INEN,
V-P. 1988. Hakkuumahdollisuuksien kiiytt66nvaikuttavat
tilakohtaisettekijiit maan linsi- ja itiiosissa.Summary:Factors SWALLOW, S.K., ANDD.N. WEa. 1993. Spatialinteractionsin multiple-
affectingtheuseof theallowablecutin westernandeasternpartsof useforestryandsubstitutionandwealtheffectsfor thesinglestand
Finland. Folia Forestalia 707. J. Environ.Econ. Manage.25:103-120.
JUDGE,G.G., R.C. HILL, W.E. GRIFFITHS,H. LUTKEPOHL,
ANDT-C. LEE. TOmN,J. 1958. Estimationof relationshipsfor limited dependentvari-
1985.Theoryandpracticeof econometrics.
Wiley, New York. ables. Econometrica 26:24-36.

KARPPINEN,H. 1995. Forestvaluesand ownershipobjectivesof nonin- YOUNG,R.A., ANDM.R. REICHENBACH. 1987. Factorsinfluencingthe
dustrialprivate forest owners.P. 445 in Abstractsof Invited Papers, timberharvestintentionsof nonindustrialprivateforestowners.For
IUFRO XX World Congress. Sci. 33(2):381-393.

308 Forest
Sctence
42(3)1996
List of Symbols U Utlhty functionfor thewholeplanningperiod

B Borrowing u(ct) Instantaneous

utilityfromconsumption
of goods
and
services
Bu Exogenouscreditlimit
vt Standing
timberstock
ct Consumption
of goodsandservices
wt Realwealth
F(.) Timbergrowthfunction
x Vectorof exogenous
variables
g(vt) Instantaneous
utilityfromthetimberstock
(x Relativeweightof amenities
in theutilityfunction(0 <
ht Harvest
t Time periodsubscript [ Vectorof parameters
to be estimated;
subjective
dis-
i Individualsubscript count
factor(l+fi)-1 inthetwo-period
model
Vector of error terms
j ForestryBoardDistrictsubscript
rnt Exogenous
(nonforest)
income fi Subjective
rateof discount
of futureutility

Pt Net timberprice o Standard

errorestimateof theharvesting
intensityin the
Tobit model
r Interest rate

Forest
Sctence
42(3)1996 309