Agrculural and Rural Fnance Markes n Transon Proceedngs of Regonal Research Commee NC-04 S. Lous, Mssour Ocober 4-5, 007 Dr. Mchael A. Gunderson, Edor January 008 Food and Resource Economcs Unversy of Florda PO Box 040 Ganesvlle, Illnos 36-040
Sae Level Evdence of he Boom/Bus Cycle for Farmland Prepared by Alfonso Flores-Lagunes and Charles Moss Unversy of Florda
Ths paper reexamnes he evdence of he boom-bus hypoheszed by Schmz (995). The boom-bus cycle has dramac mplcaons for he farm secor. As dscussed by Mshra, Moss, and Erckson (007) farmland values have hsorcally accouned for 70 percen of he agrculural balance shee. Thus, fnancal nnovaons ha mpnge farmland values can mply eher sress for he farm secor (as experenced durng he md and lae 980s) or booms (as recenly experenced n several saes as ncreased urban pressures have ncrease farmland prces [Lvans e al. 006]). Mshra, Moss, and Erckson deermned ha hs effec may be amplfed by falures n he agrculural capal marke. The resuls of hs sudy ndcae ha he presen value model proposed by Schmz holds beween saes, bu no whn a specfc sae. Par of hs varaon s explaned by he effec of nflaon on farmland values (conssen wh he fndngs of Moss [997]). However, he panel dmenson of he boom-bus formulaon rases as many quesons as answers. Modelng Changes n Farmland Values Fnancal heory suggess ha a frm should adop a projec f he ne presen value of ha projec s posve NPV I 0 N E CF j ( r ) j () where NPV s he ne presen value of he nvesmen, I 0 s he purchase prce of he nvesmen, E CF s he expeced cash flow n perod j gven nformaon avalable n perod, r j s he approprae dscoun rae n perod j, and N s he economc lfe of he nvesmen. Usng arbrage argumens, he value of he asse n perod, V,, hen becomes
V N E CF j ( r ) j () wh N n he case of land. We derve a model for asse values based on changes n he asse valuaon equaon over me. Specfcally, akng he frs dfference of Equaon yelds V V V E CF E CF. ( ) ( ) 0 r j r j j j 0 (3) Aggregang over lke exponens, we derve V E CF ( r ) E CF E CF ( ) ( ) r r ( r j) j (4) whch can be rewren as V E CF rv ( ) ( ). r r (5) If expecaons are raonal, hen we would expec o be whe nose, or ha no nformaon remans n he error erm. In order o es for boom/bus cycles usng he heorecal resuls n Equaon 5, we assume ha E CF CF or ha expeced cash flow n he nex perod can be proxed by observed cash flow n he nex perod. Wh hs subsuon Equaon 5 can be reformulaed as V CF rv. ( r) ( r) (7) 3
Tesng for boom/bus cycles s hen a wo sep process. Frs, we esmae and usng he sample daa and es for and. If we fal o rejec hs hypohess, he daa are conssen wh he general presen value formulaon and we urn our aenon o he possbly of anomales n whch would pon o boom/bus cycles. Gven ha he general asse valuaon hypohess canno be rejeced, we mpose and and generae he sequence of o es for whe nose. A hs pon we depar from Schmz by usng he Ljung-Box es o examne wheher he resduals are whe nose. Some alernave explanaons for changes n farmland value ousde he presen value framework nclude capal marke mperfecons. Specfcally, Mshra, Moss, and Erckson found ha he farm secor s deb-o-asse poson affecs he prce of farmland. Implcly, hs resul rases quesons abou he equlbrum n he agrculural capal marke. The dsequlbrum n he capal marke generaes several possble asse marke devaons. One as descrbed by Feldsen (980) s dsequlbrum beween farmland markes and sock markes. In essence farmland becomes an nflaonary hedge. We modfy he orgnal Schmz model o allow for hs nflaonary hedge effec as CF r V V V 0 3 3 r r (8) (3) where V farmland value a me mes he nflaon rae a me. Ths formulaon focuses on he effec of nflaon on land apprecaon. From one perspecve, we would ancpae ha changes n he neres rae conan nformaon abou changes n nflaon. Thus, f he opporuny cos of capal explans changes n land values. Daa and Mehods 4
Ths analyss uses he daa used by Mshra, Moss, and Erckson. The daa s he U.S. Deparmen of Agrculure, Economc Research Servce sae-level daa from 960 o 00. These annual daa on land values, neres raes, reurns o farm asses, governmen paymens, and deb servcng raos are derved from a varey of sources such as he Census of Agrculure, varous USDA agences, Federal Depos Insurance Corporaon (FDIC) call repors, and he Farm Cred Sysem. Ths sudy defnes he reurn o farmland as he gross revenues per acre less he expendures on varable npus as descrbed by Erckson, Mshra, and Moss (003). Average real neres rae s he average neres rae on farm busness deb (.e., rao of neres expenses mnus neres expenses assocaed wh operaors dwellng expenses o average farm deb). Our measure of nflaon s he logarhmc change n he Personal Consumpon Expendure componen of he Implc Gross Domesc Produc deflaor. Gven he panel srucure of he daase, we use generalzed leas squares appled o panel daa as descrbed by Hsao (986). Specfcally, we le X be he daa marx for each sae over me, he frs column of hs marx s he observed cash flow for each year dvded by one plus he correspondng neres rae (.e., for ha year/sae combnaon), he second column s he land value for he precedng mes he neres rae dvded by one plus he neres rae, and he fnal column s he land value n he precedng year mes he nflaon rae value of farmland n he precedng year. The dependen varable marx, Y, s he change n farmland values beween year and year. The generalzed leas squares resuls are hen gven by ˆ M M M M X QX x x x x X QY x x y y GLS T T (9) 5
where Q I T ee T and e s a conformable vecor of ones (hs s he sweep marx), s a relave weghng of he varance componens ( u u T ), x denoes he average value of x for a gven ndvdual over me, and x s he average value of x across all ndvduals and me perods. Ths specfcaon allows for he decomposon of he overall esmaor no wo componens: he whn esmaor and he beween esmaor. The whn esmaor ( cv ) s he average regresson coeffcen across all ndvduals ˆ M cv M X QX X QY (5) and he beween esmaor s he regresson relaonshp defned beween he average observaons for each ndvdual ˆ M b M x x x x x x y y (6) Gven hs esmaor, we can hen defne he average nercep value across all ndvduals ( ˆ ) as ˆ y ˆ x. (7) GLS To complee hs formulaon, we esmae u usng he esmaed varance from he uncorreced (.e., homoskedasc) esmaes of ˆ cv and usng a smlar regresson based on he sample means. Resuls The resuls of our esmaon are presened n Table. In general, neher he fxed effecs nor random effecs model conform o our expecaons from Equaon. In boh specfcaons, he 6
effec of cash flow on changes n farmland values s negave whle he opporuny reurn on farmland values s posve. Furher, he effec of nflaon s posve and sascally sgnfcan a any convenonal confdence level. However, he beween esmaor s largely conssen wh Equaon, he effec of cash flow s negave (bu sascally nsgnfcan) whle he effec of he opporuny cos approaches one and s sascally sgnfcan a any convenonal confdence level. The dffculy les n he sascal sgnfcance of he nflaon parameers. Specfcally, hs parameer s posve and sascally sgnfcan. Thus, bous of nflaon ncrease farmland values beyond ha ancpaed by he presen value framework (.e., he effec of nflaon on he neres rae). Ths concluson s conssen wh Feldsen s conjecure ha farmland values could form an nflaonary hedge agans socks. The concluson s also conssen wh Moss who found ha nflaon conaned more nformaon on changes n farmland prces han eher reurns o farmland or neres raes. References Erckson, K., A.K. Mshra, and C.B. Moss. (003) Cash Rens, Impued Reurns, and he Valuaon of Farmland Revsed. In Governmen Polcy and Farmland Markes: The Manenance of Farmer Wealh eded by Charles B. Moss and Andrew Schmz, pp. 3 35. Ames, Iowa: Iowa Sae Unversy Press. Feldsen, M. (980) Inflaon, Porfolo Choce, and he Prces of Land and Capal Sock. Amercan Journal of Agrculural Economcs 6(5): 90 6. Hsao, C. (986) Analyss of Panel Daa New York: Cambrdge Unversy Press. Lvans, G., C.B. Moss, V.E. Breneman, and R.F. Nehrng. (006) Urban Sprawl and Farmland Prces. Amercan Journal of Agrculural Economcs 88(4): 95 9. 7
Mshra, A.K., C.B. Moss, and K.W. Erckson. (007) The Role of Cred Consrans and Governmen Subsdes n Farmland Valuaon n he Uned Saes: An Opon Prcng Model Approach. Emprcal Economcs (Forhcomng). Moss, C.B. (997) Reurns, Ineres Raes, and Inflaon: How They Explan Changes n Farmland Values. Amercan Journal of Agrculural Economcs 79: 3 8. Schmz, A. (995) Boom/Bus Cycles and Rcardan Ren. Amercan Journal of Agrculural Ecomomcs 77(5): 0-5. 8
Table. Panel Esmaes of he Boom/Bus Specfcaon of Farmland Values Random Effecs Fxed Effecs Beween Group Consan 3.7739 * 0.03 *** -5.445 ** (.465) (.678) (.0407) Dscouned Cash Flow 0.36 *** 0.36 *** -0.03 (0.038) (0.0394) (0.0530) Ineres on Land Value -0.3073 *** -0.549 *** 0.9808 *** (0.0899) (0.0974) (0.5) Inflaon 0.75 *** 0.6 *** 0.6047 *** (0.073) (0.0766) (0.70) a Numbers n parenhess denoe sandard errors *** Denoes sascal sgnfcance a he 0.0 level of confdence. ** Denoes sascal sgnfcance a he 0.05 level of confdence. * Denoes sascal sgnfcance a he 0.0 level of confdence. 9