OWNERSHIP AND INVESTMENT BEHAVIOR IN TRANSITION: CASE OF CZECH COLLECTIVE AND CORPORATE FARMS

Transcription

1 OWERSHP AD VESTMET BEHAVOR TRASTO: CASE OF CECH COLLECTVE AD CORPORATE FARMS JARMLA CURTSS* TOMÁŠ RATGER** AD TOMÁŠ MEDOOS** * Lebz sue of Agrculural Developme ad formao Halle (Saale) Germa emal: curss@amo.de ** sue of Agrculural Ecoomcs ad formao Prague Czech republc emals: rager.omas@uze.cz medoos.omas@uze.cz Paper prepared for preseao a he EAAE Cogress Chage ad Ucera Challeges for Agrculure Food ad aural Resources Augus o Sepember ETH urch urch Swzerlad Coprgh b Jarmla Curss Tomáš Rager ad Tomáš Medoos. All rghs reserved. Readers ma make verbam copes of hs docume for o-commercal purposes b a meas provded ha hs coprgh oce appears o all such copes.

2 OWERSHP AD VESTMET BEHAVOR TRASTO: CASE OF CECH COOPERATVE AD CORPORATE FARMS ABSTRACT Cooperave ad corporae farms have reaed a mpora role for agrculural produco ma raso coures of Ceral ad Easer Europe. Despe hs mporace her owershp srucure ad parcularl he owershp's effec o her vesme acv val for effce resrucurg ad he secor's fuure developme are sll o well udersood. Ths paper ams o aalze he owershp-vesme relaoshp usg daa o Czech farms from 997 o 8. We allow for owershp-specfc varabl farm vesme behavor aalzed b meas of error-correco acceleraor model. Emprcal resuls sugges sgfca dffereces he level of vesme acv resposveess o marke sgals vesme lumpess of vesme as well as sesv o facal varables amog farms wh dffere owershp characerscs. Resulg crease farm performace dffereces amog farms ca be expeced o lead o farm resrucurg dreco of lowerg umber of owers ad creasg owershp cocerao. TRODUCTO The dvers of owershp srucures ha have emerged agrculure raso coures offer a uque opporu o sud he owershp effec o farm vesme acv ad hus o suae fuure farm srucural developme. The farm owershp cosellaos mos of he Ceral ad Easer Europea coures are assged b her dual aure.e. b a large umber of small dvduall or faml owed farms ad a small umber of large-scale farms of a cooperave ad corporae form. The large-scale farm despe her relavel low umbers remaed culvag a cosderable share of agrculural lad. For example Bulgara ad Romaa he share of agrculural lad culvaed b cooperave ad corporae farms correspods o earl 5% he Czech Republc o aroud 7% ad Slovaka almos 9% (LERMA e al. 4; Msersvo zemědělsví České Republk 9 Msersvo Podohospodarsva Sloveskej Republk 8). Despe her domace he secor he complex owershp srucures ad goverace of hese farms parcularl her effec o vesme decsos val for he effce resrucurg ad fuure developme of he secor have o receved researchers' suffce aeo. comparso o frms wh corporae goverace maure ecoomes he corporae farms' goverace raso coures was formed codos of weak legal proeco of reewed proper rghs as well as mor shareholder eress. As a drec cosequece of hese codos he proper rghs reforms led o hghl dspersed owershp of he corporae farms wh a grea represeao of exeral as well as sder (emploee ad maageral) owershp. The followg owershp developme has bee furher cramped b ol slowl emerge eral corporae evrome ad uderdeveloped markes for agrculural owershp shares. Ths evrome dcaes hgh probabl of effce proper rghs allocao ad hgh agec coss whch could be parall lowered b effce maageme ceve ad moorg mechasms. The ew owers however dspose of kowledge wh corporae bodg ad corol mechasms lmed ol o her shor rasoal experece. Each of hese aspecs amplfes he agec problems of corporae goverace raso agrculure. The separao of owershp ad corol over he corporae farms ad uderdeveloped corporae mechasms sugges ha he mos dsgushg characerscs of he corporae goverace raso agrculure s he large

3 scope for maageral dscreo. The area whch he scope for maageral dscreo comes o s greaes effec s he geeraed eral fuds dsrbuo ad decsos over producve vesme. Mos emprcal sudes of he owershp-vesme relaoshp were coduced for maure marke ecoomes (see e.g. MORC ET AL. 988; MCCOELL AD SERVAES 99; CHO 998; CHADDAD ET AL. 5; GUGLER 5; DAELSO AD SCOTT 7); ol scarcel were hese sudes doe for raso coures (MUELLER AD PEEV 7; GUGLER AD PEEV 7; DOMADE ET AL. 8; BOUSHEVA e al. 9). The emprcal resuls b MUELLER AD PEEV (7) suppor he exsece of he maageral dscreo effec o vesme more ha seleced Ceral ad Easer Europea coures. The ascrbe he resulg over-vesme o he corporae goverace suos ad weak law eforceme raso coures. DOMADE ET AL. (8) aalzed he effec of he relaoshp bewee maageme ad emploed owers o he Slovea frms vesme. Ther hpohess ha maagers ad emploees barga over he allocao of geeraed eral fuds bewee wages ad vesmes ad hus decrease fuds for fxed vesme less lqud capal markes was o cofrmed b he emprcal daa. her sud of frms vesme behavor 5 raso ecoomes GUGLER AD PEEV (7) foud a decle he vesme sesv o cash flow over he perod 99-. The arbued hs red o he decrease of agec problems as capal markes ad corporae goverace sadards developed. BOUSHEVA e al. (9) foud ha Russa farms wh hgher share of owers amog maagers show a hgher ad posve relaoshp bewee vesme ad cash flow. The erpre hs resul as relaed o hgher margal producv of capal farms wh hgher maageral owershp. LÍAL AD ŠVEJAR () foud a posve relaoshp bewee vesme ad profabl Czech (o-agrculural) cooperaves ad o a lesser exed small prvae frms ad arbue hs observao o hese frms ecouerg facal cosras. he case of cooperaves however hs resul could dcae o ol ecouerg of cred cosras bu also decso-makg cosras relaed o labormaageme ad hus he cooperaves eral goverace. Wh ol a few vesme sudes o raso agrculure he farm owershp-vesme relaoshp raso coures remas vasl uexplored. Ths sud ams o reduce hs research gap usg uque surve daa o owershp srucure of Czech cooperave ad corporae farms. Ths daa combed wh vesme produco ad facal daa from he Farm Accouac Daa ework for he me perod s used a error-correco acceleraor model framework. OWERSHP AD VESTMET - THEORETCAL DSCUSSO Prcpal-age relaoshp's role vesme modelg The moder heor of frm vesme behavor s bul upo s eoclasscal foudao formulaed b MODGLA AD MLLER (958) ad JORGESO (96). The eoclasscal vesme heor cosders a world of perfec capal markes ad opmal accumulao of The reasos for focusg he sud o collecve ad corporae farms ol are mafold. As meoed above cooperave ad corporae farms pla a mpora role ma of he raso coures whle agec problems ha are characersc for corporae or jo owershp wh delegaed maageme s a lkel source of o-opmal vesme decsos ad resrucurg. Oe could argue for he suabl of a comparave aalss bewee dvdual farms ad farms wh jo owershp. However such a comparso would roduce a vesme effec of heerogeeous facal codos (cred cosras) bewee large ad small dvdual farms (see e.g. BEEMER ) whch whou cocree formao o hese codos would dsurb he erpreao of he resuls o agec-vesme relaoshp. To he coveece of hs sud he owershp srucure of corporae ad cooperave farms s so dverse ha hs relaoshp ca be well aalzed wh he sample of hese farms purel.

4 capal. assumes ha he shor-ru vesme of a frm represes a lagged respose o chages uderlg marke codos as well as he ax srucure boh affecg he cos of capal (JORGESO 96). Laer leraure demosraed he breakdow of he eoclasscal vesme model b ackowledgg he vesme role of asmmerc formao bewee maagers of he frm ad he provders of exeral capal (baks) as well as agec coss arsg from he dverge goals of maagers ad owers of he frm (MARESSE ET AL. 999). The ssue of formao asmmeres also called adverse seleco refers o a suao whe maagers ad oher sders kow ha her frm offers aracve vesme opporues whle poeal supplers of exeral capal do o dspose of suffce formao o assess he vesme reurs (e.g. STGLT ad WESS 98; MYERS ad MAJLUF 984). As a cosequece a frm ma deped more o eral resources ha are lmg her amou whch resuls less opmal capal accumulao (uder-vesme). The secod pheomeo ha shaered he eoclasscal fudame of he vesme model apples o corporae (jo) owershp wh separao of owershp ad corol. formao asmmeres bewee owers ad maagers ofe referred o as moral hazard gve maagers a scope for dscreo ha ca be ulzed for pursug goals ad eress ha devae from he goals ad eres of owers ad hus lead o o-opmal decsos from he owers perspecve (JESE ad MECLG 976). Besdes goal dffereces also o havg her persoal wealh a sake represes a reaso for maagers behavg less opmall ha owers (gve same formao ad qualfcao) (JESE 986; AG 99). Also seekg hgher apprasal maager's geerall ed o prese her work ad hus frm performace ad marke sadg posvel whch suggess growh poeal leadg o over-vesme (GRABOWS AD MUELLER 97; MLLER AD PEEV 7). However we ca argue ha he case of raso where capal markes are characerzed b lower lqud (DOMADE ET AL. 8) ad especall he case of agrculure whch s characerzed b sgfcal lower profabl ha oher secors as well as lower aracveess o baks maageral dscreo could be ulzed oward hgher vesme ol f here are alerave accessble vesme sources (such as operaoal leasg) bu wll ulkel lead o over-vesme. Maageral dscreo could raher mafes self hgher depedec o eral fuds ad decreasg relave producv of he frm over me. The dffereces objecves ad ceves ogeher wh he formao asmmeres hus creases he mporace of overall effcec of corporae goverace cludg maageme moorg. everheless boh losses decso opmal or mprovg corporae goverace resul coss agec coss bore b corporae owers. Ther levels ca be assumed o be parcularl hgh raso ecoomes where he legal proeco of shareholders (mal mor shareholders) has bee weak ad he cocep of acve goverace of frms o shareholders was ew. As ouled above he mode whch corporae owershp srucure affecs vesme he mos s s provso of scope for maageral dscreo ad s effec o he ceve srucure ha reflecs maageral echcal ad vesme performace. Oe wa whch boh asmmerc formao bewee he frm ad possble supplers (leders) of exeral capal ad agec problems wh he frm mpac frm vesme s he use of eral versus exeral facal resources. Ths mples a mporace of facal varables (prof or cash flow) he vesme decso. a perfec capal marke or whe o formao asmmeres exs bewee a frm ad a suppler of exeral face a frm should o be lmed s vesme wh hgh reurs b lower eral fuds sce cos of exeral ad eral faces equal ad exeral capal s full accessble (FAAR ET AL. 988). mperfec capal markes wh formao asmmeres or wh dffereal cos bewee eral ad exeral face a frm vesme acv wll be more sesve o geeraed 4

5 prof or cash flow (FAAR ET AL. 988). Evdece of excessve sesv o cash flow has hus bee ofe erpreed as suggesg he exsece of cred cosras. However APLA AD GALES (997) argue ha he sesv of vesme o cash flow s also jusfed b he fac ha exeral fuds are more cosl ha eral fuds for all frms as log as some rasaco coss are volved. Smlarl o POTERBA (988) he also argue ha sce curre vesme depeds o ol o he curre bu also o expeced fuure chages he desred sock of capal s possble ha formao o cash flow helps o forecas fuure profabl ad vesme opporues whch aga resuls hgher vesme sesv o cash flow. everheless mos releval o our sud he vesme sesv o cash flow as a prox for eral fuds ma have a frm eral raoale. As ouled above maageme amg a persoal apprasal s movaed o prese opmsc frm resuls ad follow a sraeg of growh depede of he real reurs o capal ad vesme. hs case maageme s assumed o preferabl face (less effce) vesme projecs from eral resources o avod he projecs' exeral (bak) scru (JOES ET AL. 5 MUELLER AD PEEV 7). Owershp specfc vesme behavor Sce boh formao asmmeres bewee he frm ad supplers of exeral capal as well as frm eral prcpal-age problems should resul a posve vesme sesv o facal varables he dsco bewee hese wo ad possbl oher sources of vesme sesv o eral fuds remas mosl ambguous. Ol a suffcel dealed emprcal daa uderpg he frm owershp varabl vesgaed wh a releva vesme model ad a comprehesve heorecal dscusso ca help o shed more lgh o he deermas of corporae farm vesme behavor ad fler ou he owershp effecs. The varous forms of jo owershp ca be ouled b dffereces he degree of owershp dsperso (sze of owershp shares ad umber of owers) mbalaces he share szes amog shareholders or he dsrbuo of owershp bewee exeral ad eral owers. Each of hese owershp characerscs mpacs dfferel he jo goverace. Dspersed owershp represeed b small owershp shares dsrbued amog a large umber of owers ma provde suffce ceves for a oe vesor o moor ad corol he performace of he frm whereas where here are large doma shareholders he reurs o acve goverace are greaer (MAYER 996). The moorg fucos are he mal delegaed o eral or exeral corollers (e.g. members of he Board of Trusees). For he same reasos above orgazaos wh hgh owershp dsperso lack a srog backcouplg bewee he owers ad corollers. raso ecoomes assged b shareholders' low experece wh corporae goverace ad lackg coeco o releva specalss exeral corollers were ofe proposed b maagg suff mosl from hsorc eworks. Such cosellao of maageme corol ca be assumed less rgorous ha poeal corol performed b full depede ad qualfed moorg ages. Furhermore he drec moorg capaces of small shareholders ca furher be cosdered as hdered b formao asmmeres bewee hem ad maagers as well as her bouded raoal as owershp o agrculural asses was parall acqured b former emploees collecve farms wh echcal qualfcao low educao ad erepreeural kowledge. The erpla of all hese facors relaed o dspersed owershp creae a evrome characerzed b a larger scope for maageral dscreo allowg he maagers Cooperaves are reaed hs sud as farms wh corporae goverace sce eher he oblgao of coecg cooperave membershp o work he cooperave or he oe member-oe voe vog rule are cluded he acual commercal law of he Czech Republc (Law r. 5/99 of he Code of Law Commercal Code). Mos Czech agrculural cooperaves do o chose he rules radoall defg producer cooperaves academc leraure. 5

6 o pursue her ow eress as well as b codos for hgh maageral rasaco coss ad lesser ceve for maageral performace. Regardg he use of eral versus exeral sources for vesme facg he less effce goverace relaed o dspersed owershp ca lead o he avodace of exeral scru of vesme projecs whch s le wh he above preseed argume b MUELLER AD PEEV (7). Lesser exeral moorg (e.g. b baks) he amplfes suffce moorg of maageme frms wh owershp dsperso (DOMADE ET AL. 8) 4. As a resul of hese facors hgh owershp dsperso ca be assumed o lead o hgher vesme sesv o facal varables he hgher owershp cocerao. he codos of (Czech) raso however mos dspersed owershp s characersc for farms where resues' (ew owers') proper clams were (laer raso) rasformed o shares as a form of securg a value of he resued proper (see CURTSS e al. 6). For he agrculural compaes hs was a form of securg her access o capal he form of equ sead of deb ha would expose he compaes o bak scru. Ma of he small shareholders are hus former proper clamas who dd o succeed her effor of moear rerbuo of her proper clams he were orgall amg o. The mposed shareholdg as well as hgher rsk averso he lkel reflec shareholders low eres he farm fuure performace ad hus vesme acv. As he shareholders predomal follow her eres of earl reversal of her "vesme" ad pame of dvdeds he do o rel o maageme o decde abou vesme acv whch ma reduce he effec of eral formao asmmeres ad maageral dscreo o vesme especall he her effec o vesme sesv o eral fuds. Owershp cocerao defes a suao whch vesors decdes o ves o hgher shares ad hece characerzes owers whch lkel have a hgher erepreeural eres ad more rus he performace of he busess. As hese vesors allocae more capal o he frm he could also be assumed o be less rsk averse ha he small shareholders. The fac ha he have more a sake s assumed o smulae hem o develop more effce corporae goverace srucure cludg beer moorg corollg ad ceve mechasm. a smlar ve MAYER (996: -) argues: "Where here s coceraed owershp here ma be a greaer wllgess o dscple poorl performg maageme as well as more ceve o ervee ad exercse 'voce' raher ha 'ex'". f hgher owershp shares also represe hgher relave share he legal capal per shareholders (.e. here are less owers he frm or a umber of doma owers) provdes he shareholders wh hgher decso-makg powers ad hgher reurs o acve dvdual ower's goverace. However as owershp ad corol are separaed hs owershp characerscs sll leads o a hgher vesme acv ha f owershp ad corol are coceraed he same hads. everheless more effce corol of maagers' performace reduces maagers' rasaco coss ad leads o more opmal vesme decso ha dspersed owershp. Due o lower maageral rasaco coss ad more effecve corol of he farm operao farms wh more coceraed owershp are also expeced o acheve hgher performace ha less owershp coceraed frms. More opmal vesme ad hgher ecoomc performace meas hgher reurs o capal ad lesser reaso for avodg bak corol of he vesme projecs. Because of he owers' lower rsk averso ad beer maageme moorg 4 Maageral rasaco coss are coss of free cash flow dsperso replaceme ressace ressace o prof lqudao or merger power sruggles excessve rsk akg excessve dversfcao excessve growh ec. DOMADE ET AL. (8) do o refer hs coex o he maagers choces of avodg bak scru of vesme projecs bu refer o less lqud capal marke ha he see as a vesme cosra raso coures. 6

7 hgher owershp cocerao s expeced o lead o more opmal vesme projecs ad a hgher use of exeral facg hus lower vesme sesv o eral fuds. The eres ad cocer dvergece bewee exeral ad eral owers ca represe a source of vesme behavor dffereces. The mos mpora dfferece bewee he cosderaos of emploed owers ad exeral vesors sems from he emploed owers' lkage of frm performace o emplome secur. O he oe had emplome secur smulaes hgher labor performace provdes hgher ceves o corol over vesme projecs ad moor maageme performace; o he oher had resuls hgher rskaverso owards vesme projecs. DOW () led hs behavor as "face pessmsm". Aoher problem of emploee-owed frms closel alled wh vesme behavor les wha JESE AD MECLG (976) (see also FURUBOT 976) erm "he horzo problem". Ths mples ha whe workers leave he frm afer he ermao of her work corac he lose her share of he value of a capal ha has bee accumulaed b he frm ad hus have suffce ceves projecs wh log paback perods. As a resul emploed owers prefer curre cosumpo o vesme (JOES ET AL. 5) or vesme projecs wh shor paback perod. Ths vesme coservasm of emploee owershp ca be expeced o resul lower ad less ha opmal vesme level. Also he hgher rsk averso of emploed owers could lead o preferrg exeral facg due o baks revsg he qual of he vesme projec whch would lead o lower vesme sesv o geeraed cash flows ha frms wh hgher share of exeral owers. Hgher exeral owershp o he oher had ca be assumed o provde he farms wh a lesser maageme scru ha allows for maageral dscreo ad hus maagers o carr easer hrough her vesme eress. METHOD OF VESTMET AALYSS vesme model Followg MARESSE's e al. (999) delberao o he developme of vesme models ad BOUSHEVA's e al. (9) dscusso o vesme models' suabl for he case of modelg vesme behavor raso agrculure we chose o appl he error-correco acceleraor model. Ths mples ha we are o amg o look for he "correc" vesme model bu we selec a model based o s heorecal specfcao ad performace of he alerave models prevous applcaos. The advaage of he error-correco specfcao of he acceleraor model s ha allows o separae he log ru vesme deermas from he shor ru vesme adjusmes ad s qual of reag formao he levels of oupu ad capal sock (o ol formao frs dffereces). f daa allow hs specfcao he hs characerzao of vesme behavor makes hs model superor o oher vesme models appled o raso agrculure such as he basc acceleraor model adjusme cos or Euler equaos. A alerave Tob q model s less releva for raso agrculure sce he q (marke) value of he corporae farms does o exs. Aoher advaage of he specfcao of he error-correco acceleraor model s ha does o requre a specfcao of adjusme cos. Due o spaal cosras o hs paper we descrbe he org of error-correco acceleraor model ol brefl. See e.g. MARESSE e al. (999) for a more dealed descrpo of hs model. The error correco ecoomerc approach was roduced o vesme modelg b BEA (98). The error-correco specfcao of he vesme acceleraor model ess he demad for capal equao k a j (JORGESO 96) wh he damc (acceleraor) vesme equao wh a auoregressve-dsrbued lag of legh wo (ADL () fuco) he equao for he frms desred capal sock k deoes he (aural) logarhm of he desred capal sock for frm perod deoes he logarhm of oupu 7

8 8 (or sales) ad j deoes he log of he real user cos of capal. he error-correco acceleraor model damc adjusme capal k s approxmaed b / where represes vesme ad he capal sock for frm a he ed of perod. also assumes ha he varao he user cos of capal j ca be corolled for b cludg ear-specfc ad frm-specfc effecs. The error-correco model ca be wre as follows: v d k. () The error correco coeffce capures he log ru vesme adjusme o he desred level of capal d s a me dumm η s a uobserved frm-specfc effec ad v s a error erm (rasor shock). The remag parameers capure he shor ru damcs. The varable - s added o he error correco acceleraor model o allow for a es of he assumpo of cosa reurs o scale ha s ecessar for he mposed log-ru proporoal he model. Sce he commol used acceleraor model was developed for secors oher ha agrculure we eed o cosder some of he specfcs of agrculure for he eded applcao. Characerscs such as lower reurs o capal hgh suk-coss of capal ad seasoal of produco sugges possble vesme coservasm ad delas adjusmes of he desred sock of capal ad slower resposveess o marke sgals. Therefore we also cosder he error-correco acceleraor model o es damc vesme equao wh a auoregressve-dsrbued lag of legh hree (ADL () fuco). The resulg errorcorreco model has he he followg form: v d k. () Smlarl o umerous vesme sudes we furher add curre ad lagged cash flow scaled b he prevous perod s value of fxed capal o he rgh-had sde of he vesme equao o es he vesme effec of hese facal varables. Ths exeso s aalog for boh equaos (4) ad (6) herefore ol he laer s preseed whch elds followg specfcao: v d k. () hs error-correco specfcao of he acceleraor model we ca es wheher he cash flow (prof rae) plas he role of a log ru deerma of vesme or wheher s ol a shor-ru varable whch ca be erpreed as reflecg he rasor avalabl of fuds for vesme purposes. everheless he erpreao of he sgfcace of vesme effec of he cash flow varable s e ambguous. ca reflec he presece of facal cosras o vesme due o asmmerc formao bewee vesors ad he frm (e.g. FAAR HUBBARD ad PETERSE 988) however he presece of adjusme coss he level of cash flow o capal could corbue o he formao o fuure prof or oupu expecaos

9 9 (CELL 978) or vesme opporues ha were o oherwse accoued b such hgs as sales growh (SAMUEL 996). Also as dscussed he heorecal seco he presece of agec coss arsg from he dverge goals bewee maagers ad owers he γ parameers could capure maagers' sraeg owards he use of avalable eral fuds for vesme projecs. Therefore aalzg he vesme behavor he coex of he frm specfc owershp srucure ad relaed facal codos wll allow o shed more lgh o he sources of he vesme sesv o cash flow. The heorecal predcos of frm ad parcularl owershp-specfc adjusmes capal sock ad dffereces resposes o varous shocks cludg avalabl of eral fuds drecs he ex exeso of he vesme model. Log-ru pael models wh heerogeeous damcs were esmaed prevous sudes. For example Pesara e al. (999) specfed a co-egrag log-ru developmes for varous ecoomes (coures) bu allowed varg u-specfc shor-ru damcs. he coex of vesme behavor modelg Bokusheva e al. (9) allowed boh shor-ru as well as log-ru damcs o var across observaos cocreel farms. We wll follow hs laer approach ad wll allow vesme behavor o var across four owershp varables ( =... ; = ) - owershp cocerao ( ) exeral owershp ( ) ad owers' umber ( ). Ths elds followg model exeso: v d k k (4) A more dealed descrpo of he varables ad her daa sources follows he ex seco. Daa ad varables Daa o farm vesme produco capal ad facal varables orgae from he offcal balace shees come saemes ad supplemear forms of he Farm Accouac Daa ework of he Czech Republc (FAD C) surve for he ears Daa o farm owershp srucure comes from a srucured daa colleco he Czech Republc 4. Ths daa surve was orgazed ad fuded b he sue for Agrculural Developme Ceral ad Easer Europe (AMO) Halle Germa jol wh he Research sue for Agrculural Ecoomcs (VUE) Prague Czech Republc. The sample coas 7 agrculural compaes wh combed crop ad amal produco of a legal e saus (cooperaves JSC ad LLC) for a mmum of 7 ears of cosecuve aual daa bewee 997-8; from hese daa o 4 farms are avalable for he ere perod of ears. The emprcal model varables are all expressed real values ad coa followg formao: - farm sock of capal; cludes all log-erm agble agble ad facal asses; k deoes aural logarhm of ;

10 - value of gross vesme bewee sequeal perods calculaed as he chage capal sock (represeg e vesme) plus deprecao ad amorzao; - logarhm of farm oal sales; s he chage bewee wo followg perods; - value of he farm s cash flow ha s avalable a he ed of a gve perod for purchasg ew capal sock a he begg of a followg perod; he cash flow dcaor s uavalable he double-er accoug herefore s calculaed as reaed eargs (prof or loss) plus deprecao ad amorzao; - -vecor of owershp varables ( = ) ha are me-vara as daa o owershp srucure are avalable ol for he ear. These varables are defed as dumm varables akg he value of for values larger or equal o meda oherwse. Frs owershp varable deoes a average (per ower) share farm's equ shorl owershp cocerao. represes exeral owershp.e. a share of exeral vesors o he oal umber of owers. deoes he umber of owers umber. 5 TD - farm rasformao debedess owards elgble persos o asses from resuos ad asse rasformao ( he case of former collecve farms). Ths dumm varable wll be used purel as a corol varable.e. varable corollg for possbl relaed varabl asse valuao (degree of capal deprecao) ad cred cosras. s corporaed he model he same wa -varables are. d - me dumm varables ha are cluded o accou for me-specfc shocks commo o all farms. Esmao mehod The above specfed model characerzes a damc process whch he depede varable curre vesme o capal rao s flueced b s pas levels. Besdes he auoregressvedsrbued lag he vesme model cludes explaaor varables ha cao be cosdered srcl exogeous. The lagged vesme o capal rao ca be assumed o be correlaed wh frm-specfc effecs. Also growh oupu (sale) ma be correlaed wh hese effecs ad he curre chage oupu (sale) s lkel o be correlaed wh shocks o vesme va he produco fuco (Bod e al. ). hs case a pure occurrece of frm-specfc (uobserved) effecs correlaed or ucorrelaed wh oher varables o he rgh-had sde of esmaed equao requres more ha radoal daa wh-frm rasformao or frs dffereao whch ca be appled he case of smpler pael model specfcaos. The reaso for hs s ha esmaes o such rasformed daa are o cosse o shor me seres (e.g. Maresse a kol. 999; Roodma 9a). Mos advaced mehod of solvg hs ecoomerc ssue s he full effce Geeralzed Mehod of Momes (GMM). For he esmao of emprcal models wh auocorrelao ad oher possble edogeees explaaor varables Arellao ad Bover (995) ad Bludell ad Bod (998) sugges a use of ssem GMM whch allows a combao of wo equaos ad wo samples of srumeal varables o rasformed ad urasformed daa whch ca be more effce 5 Owershp varables are geerall assumed o be edogeeous o he performace ad he marke value of he frm whch aga should smulae vesmes ad herefore hs edogee should be corolled for he model. These heores are geerall appled o frms raded o he sock exchage or frms maure ecoomes wh full proper rghs legal eforceme develop capal marke ad vesors culure. The corporae owershp srucure he Czech agrculure s mal a resul of he rasformao ad prvazao process domaed b former maageme ad ew owers' resrucurg objecves ad sraeges. Due o he sll uderdeveloped capal marke ad assumed maageral dscreo lle (parcularl effcec-drve) damcs he owershp srucure ad hus dsmssble edogee he owershp varables s assumed.

11 ha frs-dfferece GMM 6. Because of he relavel large umber of srumes compared o he umber of observao we esmae oe-sep ssem GMM. We appl he programmg package provded b Roodma (9a) desged for sascal sofware STATA. RESULTS Table preses parameer esmaes of he error-correco acceleraor model whou owershp varables derved from a damc vesme equao wh a auoregressvedsrbued lag of legh wo (ADL ()) ad hree (ADL ()) (see equao ). A comparso of he esmaes of hese wo models shows ha accoug for ADL () damcs mproves he model sgfcal. The es of he presece of he lag hree effec (es of he jo sgfcace of ρ θ ad γ ) reveals ha hese parameers are jol sgfca a 5 % sgfcace level.e. he corbue hghl sgfca formao o he model. Therefore we furher erpre Model ol. Table : GMM esmaes of he error-correco acceleraor vesme model for Czech cooperave ad corporae farms Depede varable / - Model - Basc AR( ) error correco model (s = ) Model - AR( ) error correco model (s = ) dep. var. Coef. Coef. esmae P-value Coef. esmae P-value Cosa α / - ρ / - ρ θ θ θ k -s - -s ϕ s φ / - γ / - γ / - γ # of obs Overall f (F-es).5... AR() es Hase es oe: Coeffces for me dummes are o cluded he able. ) srumes used: a) for frs dffereces equao - lags o ( o ) of - / - ad / - Model ( Model ); b) for level equao - all remag explaaor varables (urasformed) cluded he equao cludg me dummes ad frs dffereces of lags of - / - ad / -. The esmaes of lagged depede varables are foud he rage bewee her OLS esmaes ad wh group (fxed effec) esmaes whch s a dcao of good esmaes (see BOD ). The frs wo parameers followg he cosa refer o he shor ru effec of he pas growh capal sock o he curre vesme acv. Smlarl o he sud b BOUSHEVA e al. (9) for Russa farms hs effec s foud sgfca egave however he rage bewee - ad whch mples a cclcal developme of vesme acv oscllag aroud ad approxmag o zero over me. The cclc vesme acv ca be well 6 he leraure he "frs-dffereced" GMM proposed b Arrelao ad Bod (99) s also called Dfferece GMM. Boh les are refer o he esmao procedure usg frs-dffereces of he daa order o elmae he fxed effecs. Ssem GMM augmes Dfferece GMM b esmag smulaeousl dffereces ad levels; each equao beg dscl srumeed (Roodma 9b). Advaage of applg ssem GMM also s ha allows o clude me-vara regressors our case owershp varables ha would dsappear frs dfferece GMM (Roodma 6: ).

12 explaed b capal marke cosras raso agrculure ad relave depedec o erall geeraed fuds ha eed o be accumulaed over few ears for larger vesmes. The decreasg edec he lumpess of vesme over me could refer o mprovg cred marke codos or beerg performace of he farms due o srucural adjusmes. The followg hree parameers capure he rasor vesme effecs of he pas growh oupu (sales) - -. The sroges posve vesme respose o chages sales (o creasg or dmshg marke opporues) s he respose o he mos ermedae chages; slghl weakes wh he me dsace of he sale chages. The ex varable s he error correco erm roducg he log-ru vesme adjusme o he opmal capal level. Uder he assumpo of opmal vesme behavor he coeffce s expeced o be egave sce he acual capal level lower ha s desred level should be followed b hgher fuure vesme ad coversel (see e.g. BOD e al. 997: 5). Ths behavor s cofrmed b he hghl sgfca egave coeffce esmae. Farm vesme acv s hus cogruece wh he log-ru effce adjusme o he "desred" fuure level of capal however he sze of adjusme s uexpecedl low. The parameer sugges 7% approxmao rae capal sock o log ru capal opmum over he aalzed perod 7. Such low capal adjusme rae was also foud b BOUSHEVA e al. (9) for Russa farms. Ths could mpl hgh capal adjusme coss raso agrculure or raso-specfc capal valuao errors embedded he daa (e.g. dscrepaces bewee capal deprecao ad use produco ha var across farms). The hree γ parameers embod vesme sesv o he level of geeraed eral fuds or sgals o fuure profabl. The mpl relavel hgh vesme sesv o cash flow o capal rao however he sesv weakes wh he lag of he facal varable. The es of her jo sgfcace also mpl ha he do o capure ol he rasor effec of facal cosras o farm vesme. The cash flow level could raslae o a expecao of fuure profs ad he sgfca parameer he mpl a vesme reaco o hs expecao. he coex of raso agrculure s sll reasoable o expec ha he γ parameers capure a leas o a degree persse facal cosras bu possbl also farms' cauous behavor owards bak creds due o useled proper rghs o agrculural asses ad due o a fear of bakrupc he case of vesme projec falure volale marke codos. The las parameer o be dscussed s he parameer wh respec o he scale facor -. s value o sgfcal dffere zero mples ha he log ru elasc of capal o sales s u. The produco fuco s hus characerzed b cosa reurs o scale whch s a produco fuco characersc ha s cosse wh he heorecal assumpos of he vesme acceleraor model. Before preseg ad erpreg he owershp varable effec he framework of he errorcorreco model Table provdes mea sascs of vesme owershp ad oher varables for farm groups buld based o he hree owershp characerscs of our eres. P- values from he es dcae whch varables he farm groups of dsc owershp characerscs dffer sgfcal. vesme acv level gve b he rao / dffers sgfcal ol bewee farms wh lower ad hgher umber of owers. Farms wh lower umber of owers dspla a sgfcal hgher vesme acv ha farms wh hgher umber of owers. The are also assged b hgher sales o capal rao ad hgher cash flow o capal rao. Ths could mpl ha he hgher vesme acv could be relaed o he umber of owers drecl bu could also mpl ha farms wh lower umber of owers 7 For dusres maure marke ecoomes Maresse e al. (999) fd he capal sock error correco o be of a value bewee o 5%.

13 perform beer ad he hgher reurs o capal smulae hgher vesme rae or boh. le wh our heorecal dscusso he lower vesme acv of farms wh hgher umber of owers could sugges her lower eres he farm fuure performace sce for ma of hese owers becomg shareholders represeed he ol alerave o losg owershp clams (or her sgfca value). The wo groups of farms wh dffere share of exeral owershp var sgfcal sales o capal ad cash flow o capal raos. The hgher performace dcaors relaed o lower share of exeral owers he oal umber of owers does o seem o have a mpac o vesme acv. The hgher performace could hs case also relae o he sgfca dffereces bewee hese wo groups he umber of owers. Table : Mea sascs comparso ad wo-sample -es for farm owershp groups Capal owershp cocerao ) Exeral owershp ) Owers umber < meda meda p-value* < meda meda p-value* < meda meda p-value* / S / /S / S / Owers r Ex. ow. ) Trasf. deb ) Cap. co. ) Cap. co. 4) oe: * P-value for a wo-sample -es; ) per owers share equ ( housads C); ) share of he umber of exeral owers/vesors oal umber of owers ( housads C); ) debedess rae from owershp rasformao (debs oward elgble persos from rasformao value of oal asses); 4) per ower share legal capal. The comparso of farm groups based o capal cocerao reveals a mpora capal srucure characersc ha mgh have a cosderable mpac o he vesme modelg resuls. Based o he heorecal predcos s expeced ha hgher owershp cocerao would be relaed o more opmal vesme acv. raso agrculure ha s geerall assged b udervesme ad cred cosras hgher vesme acv s expeced o be more opmal. The dfferece he vesme o capal rao s however sgfca. Moreover he sales o capal rao s hgher for he group of farms wh lower per ower share equ. The vesme o sales ad cash flow o capal raos o he oher had cofrm he predced effec of owershp cocerao o vesme acv. These resuls are possble f here are ssemc dffereces capal value (capal deprecao) bewee farms wh hgher ad lower capal cocerao.e. lower capal value group of farms wh lower capal cocerao creases he vesme o capal rao. The capal deprecao dffereces bewee farms are geerall assumed o be depced b he uobserved frm-specfc effec erm η he vesme model. our case however hs wll be capured also he capal cocerao varable ad all s cross erms. The mos sgfca effec ca be expeced he parameer of he capal cocerao specfc errorcorreco erm. Ths s due o he fac ha farms wh hgher owershp cocerao could be foud wh less opmal capal adjusme o log ru capal sock opmum possbl

14 eve overvesme due o her hgher vesme acv bu sgfcal lower sales o capal rao whch deermes he log ru opmum. The advaage of he vesme model aalss ha follows s he smulaeous cosderao of all hree owershp varables whch allows o depc he vesme effec of each of he owershp varables whle corollg for he effecs of he remag varables. Table preses parameer esmaes of he error-correco models wh owershp-specfc varabl vesme behavor. Model he able s he mos parsmoous verso8 of he model as defed equao 4. Model 4 we addo corol for he vesme effec of rasformao debedess whch s foud o mprove he overall f of he model. Therefore we furher erpre he esmaes of hs model. The esmaes of he capal cocerao-specfc varabl he vesme model parameers dsclose ha hgher owershp cocerao creases he cclcal vesme acv respose o pas growh capal sock. Farms wh hgher per ower shares equ (capal cocerao) respod less o chages sales ad her vesme acv s less sesve o erall geeraed fuds he ear of vesme. O he oher had vesme acv of farms wh hgher capal cocerao s more sesve o pas cash flow o capal rao whch could mpl her hgher resposveess o fuure prof sgals. O he whole he hgher vesme lumpess lower vesme sesv o curre cash flow ad sales sgals as well as hgher vesme sesv o pas cash flow po ou o hgher owershp cocerao beg relaed o hgher cred facg of vesme projecs 9 ha s codoed o pas profabl lowers depedec of vesme of curre profs ad ca cause spkes vesme. These resuls sugges ha maagers of farms wh hgher owershp cocerao perform beer ad have lesser reasos o avod exeral scru o ga access o facg. Hgher capal owershp cocerao ca hus be cosdered as a eleme of a more effce goverace srucure. The las sgfca parameer hs group of farms mples capal cocerao-specfc log ru capal sock adjusme o capal opmum. Ths parameer suggess ha capal cocerao creases he value of hs parameer o a exe ha observed vesme hs group of farms could be cosdered o exceed he log ru opmum capal level. As dscussed above however here are sgfca dffereces bewee he groups of farms wh lower ad hgher capal cocerao acve capal valuao (deprecao) whch resuls relave overvaluao of capal he group of farms wh hgher capal cocerao ad lower oupu o capal rao (see dscusso of Table ). Therefore hs parameer eeds o be erpreed wh cauo; s cluso he model however s mpora as flers ou formao o hs sgfca owershp-specfc varabl ad hus mproves he esmaes of he remag parameers. The group of farms wh hgher exeral owershp shows sgfcal lower vesme respose o pas chages sales ad sgfcal hgher vesme sesv o cash flow o capal rao. Maageme of farms wh hgher share of exeral owers ca have mulple reasos o use eral fuds sead of seekg cred. Beg less producve (smaller sales o capal rao) ad less profable (smaller cash flow o capal rao) could lm her applcao success. A rejeco of cred applcao afer dsclosg performace dcaors o exeral scru could provde sgals o owers ha would furher dscourage hem from 8 9 Due o he large umber of parameers he complee model we amed for he mos parsmoous model. We sep-wse elmaed all varables wh p-value hgher ha.. Our daa cofrm ha group of farms wh hgher owershp cocerao has sgfcal hgher cred debedess (.4% compared o 6.9% he group of farms wh lower owershp cocerao). The gap he cred debedess bewee he groups wh hgher ad lower capal cocerao has creased over he ears. 4

15 usg eral fuds for vesme. Also gve ha exeral owershp was a form of selg rasformao debedess afer owers usuccessful aemp o whdraw her owershp clams from he compa or cooperave beg able o face vesme projecs from geeraed profs sead of dsrbug dvdeds s uacpaed. Ths se of argumes suggess less effce goverace srucure ad maagers hgher scope of dscreo farms wh hgher share of exeral owers. Table : GMM esmaes of he error-correco acceleraor vesme model for Czech agrculural eerprses Depede varable / - Model - AR( ) error correco model wh owershpspecfc damcs Model 4 - AR( ) error correco model wh owershpspecfc damcs (cl. corol varable TD) dep. var. Coef. Coef. esmae p-value Coef. esmae p-value Cosa α / - ρ θ θ k ϕ / - γ (cap. owersh. coc.) α * - / - ρ * θ * - θ * - θ * (k ) ϕ....6 * / - γ * - / - γ (exeral owershp) α * θ * - θ * / - γ (umber of owers) α * - / - ρ * - / - ρ * θ * - φ TD (rasf. debedess) α TD * θ TD * - φ # of obs Wald es (F-es) 4.5 ()..7 (). Wald es of jo sgfcace*. ().6.4 (7).5 AR() es Hase es oe: Coeffces for me dummes are o cluded he able. srumes used: a) for frs dffereces equao - lags o of - / - lags of - ad - / - boh models; b) for level equao - cosa - k (plus TD) me dummes ad frs dffereces of lags of - / - - ad - / - Model (Model 4). * Wald es of jo sgfcace of owershp-specfc vesme varabl. The parameers o he vesme effec of he hrd owershp varable mpl ha creasg umber of owers reduces cclcal vesme red. Ths could relae o he larger sze 5

16 capal sock of hese farms ad lower vesme o capal rao (see Table ). Larger umber of owers also reduces resposveess o chages sales. Due o he hgh correlao wh he sze of capal sock we also fd he scale parameer φ o var wh he umber of owers. Wh he group of farms wh hgher umber of owers he reurs o scale decrease wh creasg sze. Lower profabl ad reurs o capal dcaors (Table ) sugges ha farms wh hghl dspersed owershp srucure do o perform as well as farms wh smaller umber of owers. COCLUSOS Emprcal resuls from error-correco acceleraor model esmaed o Czech cooperave ad corporae farms from show ha owershp srucure sgfcal affecs farm vesme behavor. Farms wh lower umber of owers dspla a sgfcal hgher vesme acv ha farms wh hgher umber of owers. Togeher wh dcao of hgher producv ad profabl lower umber of owers could be cosdered o corbue o effcec of cooperave ad corporae farms' goverace. Group of farms wh hgher exeral owershp shows sgfcal hgher vesme sesv o cash flow o capal rao whle beg a he same me less producve ad less profable. Owers of farms wh hgher share of exeral owershp hus seem o be o a hgher degree cosraed her corol over maageme. The emprcal resuls hus are le wh heorecal expecao ha exeral owershp provdes wh more scope for maageral dscreo ha ca lead o less opmal vesme decsos ad overall performace ad hus hgher agec coss. Resuls o he vesme effec of owershp cocerao o he oher had sugges s sgfca corbuo o vesme performace. Relag o a hgher degree o cred facg of vesme projecs farms wh hgher per ower shares farm equ dspla far hghes profabl amog cosdered groups of farms. Ths observao suppors he heorecal expecao ha hgher owershp cocerao provdes ceves o more effecve jo owershp goverace. Despe her cogruece wh heorecal predcos he emprcal resuls could be affeced b raso specfcs of he farm srucures ad characerscs of equ shareholders. Small shareholder farms wh hgh umber of owers are lkel shareholders who despe her effors dd o succeed facal seleme of her rasformao clams ad for whom becomg shareholders represeed he ol alerave o losg owershp clams or her sgfca value. Ths could resul her lower eres he farm fuure performace reducg farms' vesme acv. Oe could also argue ha he whole process of farm rasformao was subjeced o former maagers' dscreo ad he resulg owershp srucure hece reflecs her maageral ables possbl her deologes degree of socal resposbl or prefereces. everheless f ohg more he emprcal resuls delver grea sdes ha ca sugges fuure developmes of he Czech farm srucure uder compeve pressure. Depedg o he sregh of he compeve pressure farms wh hghl dspersed owershp amog large umber of owers wll requre marked resrucurg ha mgh be possble afer bakrupc or gradual small shareholders bu-ous leadg o hgher owershp cocerao. Exeral owers wll be requred o mpleme ools of more effce corporae goverace whch s lkel o be realzed ol wh ceves from hgher shares a sake. All resuls hus verge oward creasg fuure owershp cocerao of oda's cooperave ad corporae farms. Developg suos suppor of he agrculural share marke ha would arac vesors from ousde he curre farms could corbue o he speed ad effecveess of he farm owershp resrucurg. 6