Price Formation on Agricultural Land Markets A Microstructure Analysis

Transcription

1 Prce Formaton on Agrcultural Land Markets A Mcrostructure Analyss Martn Odenng & Slke Hüttel Department of Agrcultural Economcs, Humboldt-Unverstät zu Berln Department of Agrcultural Economcs, Unversty Rostock Paper presented at the 60 th Annual AARES Conference, 2-5 February 206, Canberra

2 Development of Land Prces n Germany (Source: Destats) 2 Prce ( /ha) West East BVVG

3 Relevant lterature 3 Classcal Hedonc land prcng e.g. Palmqust & Danelson (989), Huang et al. (2006) Productvty characterstcs Locaton & regonal characterstcs Envronmental varables Polcy mpact / subsdes Market (mcro)structure Market power Cotteleer & Gardebroek (2008) Role of market mechansm e.g. Lusht (996), Mayer (996), Quan (2002) Forced sales e.g. Allen & Swsher (2000), Campbell et al. (2009)

4 Outlne 4. Motvaton 2. Forced sales and farmland prces 3. Prce formaton n agrcultural land auctons 4. Conclusons

5 Forced Sales: Hypotheses 5 Interference of two, possbly opposed effects. Pressured sale prce 2. Aucton effect prce 3. Net effect foreclosure prce?

6 Emprcal Applcaton 6 Regon: State of Brandenburg/Germany Data Source: Expert panel (Oberer Gutachterausschuss) Indvdual transacton data: 26,786 observatons, % foreclosures Perod: /2000 9/20 Varables: Prce, sol qualty, plot sze, locaton (county), date of sale

7 Econometrc approach 7. Regresson model ln p 0 ATT x 0 n 0 n u p n x : [sol qualty ndex, sze, sze², 8 county dummes, tme dummes] n ˆ 0 x 2. Matchng Drect covarate matchng K-nearest neghbor Mahalanobs dstance ATT n n match p p

8 Results: Effects of Forced Sales (ATT) 8 Year Observed prce forced sales Observed prce nonforced sales Counter factual (regresson) Counter factual (matchng) Prce dfference (regresson) Prce dfference (matchng) () (2) (3) (4) ()-(3) ()-(4) ,246 2,40 2,354 2, ,5 2,390 2,445,985 -, ,692 2,450,958 2, ,368 2,400 2,075 2, ,960 2,340,878 2, ,595 2,450,969 2, ,589 2,470 2,279 2, * -884* ,48 2,620 2,237 2,602,8* 85* ,437 3,070 2,584 2, * ,83 3,470 2,902 3, ,070 3,60 3,295 3, ,25 3,960 3,59 3,625,966* 500* Total 3,075 2,844 2,475 2,78 600* 282*

9 Land Market Auctons n Saxony-Anhalt 9 LGSA (Landgesellschaft Sachsen-Anhalt) Land admnstraton agency. Prvatzaton of formerly state-owned land Frst prce sealed bd aucton wth publc tenders Tenderng results of the LGSA from ; ~760 transactons (arable and grassland) Avalable nformaton: prce, sze, locaton, sol qualty, number of bds, bdder status (resdental farmer, non-agrcultural nvestor)

10 Prces formaton on land auctons: Hypotheses 0 Aucton theory: Optmal bd ncreases wth number of bdders hgher number of bds = hgher prce (prvate value aucton) Dfferent optmal bds among dfferent groups of bdders: non-agrcultural bdders: hgher optmal bds = hgher prce

11 Results: Spatal Lag Model Estmated coeffcent Sol qualty 0.0*** Plot sze (lnear) (ha) 0.03 Plot sze (quadratc) Share arable land 0.44*** Number of bds 0.033*** Share of bds from agrculture -0.27*** Local buyer 0.06* Dummy 2008 (Bass 2003) 0.203*** Dummy 2009 (Bass 2003) 0.286*** Dummy 200 (Bass 2003) 0.370*** ***, ** and * denote statstcal sgnfcance at, 5 and 0 percent, respectvely

12 Conclusons 2 Market structure and characterstcs play a role for the market outcome Prces on auctons appear hgher than on search markets Share of non-agrcultural bdders drves up prces Prces are hgher f wnnng bdder s a resdent Land prces n Germany can be well explaned by market fundamentals (no evdence for land prce bubbles) No need for tghter regulatons of land markets

13 3 Backup Sldes

14 Data: Summary Statstcs of the Farmland Prces and Characterstcs 4 Group Non-forced sales N=26,502 Forced sales N=284 Total N=26,786 Statstc Prce ( /hectare) Sol qualty arable land Sol qualty grassland Area (hectares) Mean 2, Std. devaton, Mn. 58 0,0 Max. 9, Mean 3, Std. devaton 2, Mn Max. 20, Mean 2, Std. devaton, Mn. 58 0,0 Max. 20, Source: Own calculatons based on data from Oberer Gutachterausschuss Brandenburg.

15 EUR/hectare Number of observatons Data: Development of farmland prces forced sales observatons regular observatons forced sales prces regular prces Year 0

16 Results: Hedonc prce model 6 Dependent varable log prce ( /m²) Coeffcent estmates P-value Sol qualty arable land (AZ) *** Area (hectares) *** Area squared (hectares).33e *** D D D D D D D *** D *** D *** D *** D ***

17 Data: Mean prces for countes 7 Regonal average land prces n /ha, Source: Own calculatons; data provded by Oberer Gutachterausschuss Brandenburg and Landesamt für ländlche Entwcklung, Landwrtschaft und Flurneuordnung.

18 Econometrc approach 8 Objectve: Estmaton of the average treatment effect on the treated (ATT) ATT E p Ths can be expanded to p 0 d p d E p d 0 E p p d E p d E p d 0 E observed dfference n prces ATT selecton bas Condtonal ndependence assumpton (CIA) p 0, p d x

19 Explanatory varables: summary statstcs 9 mean s.d. mn max sol qualty (0-02) % arable land lot sze (ha) arable land (ha) grassland (ha) # of bds % farmers' bds prce ( /sq m) acceptance bd n 48% of the cases from tenant N=722

20 Econometrc procedure 20 Spatal lag model ln prce = ρ W j ln prce j + x k β k j k + d l δ l county + d t δ t year + e l t x: explanatory varables; e: error term; W: spatal weghts; ρ: mpact of neghbourng land prces

21 Open questons & further research 2 How do buyer and seller characterstcs affect prce formaton? What are the motves of non-agrcultural nvestors to buy land? What s the effect of farm takeovers on land markets? Do market power and strategc decson makng play a role n (local) land markets? Are (local) land markets spatally ntegrated or separated?