1. Modelling the Spatial Distribution of Agricultural Incomes Cathal O’Donoghue*, Eoin Grealis** *, Niall Farrell*** *Teagasc Rural Economy and Development Programme, Athenry, Co. Galway ** SEMRU, National University of Ireland Galway *** Economic and Social Research Institute, Dublin

2. Context and Background

3. Context  Significant spatial variability in agriculture in Ireland  Better land in the South and East  Poorer land in the North and West  Water Quality is spatially dependent  Significant reliance on off-farm income  Spatial variability in labour markets  Location specific costs – e.g. transport costs  Spatial dimension to Direct Payments  A function of historical production which is location related  Disadvantaged Areas  HNV’s  Localised REPS/AEOS

4. Ideal Data  Ideal Data  Farm Survey Income Data  Spatial Scale  Detail Farm Management Information  However  Not available  Census – Spatial  Teagasc NFS – Income and Management Information  Spatial Microsimulation in Agriculture  Ireland: Hynes et al. (2009b), Hennessy et al., (2007), O’Donoghue (2013)  The Netherlands: Van Leeuwen et al. (2008)  Sweden: Lindgren and Elmquist (2005)  Landscape services from Agriculture (Pfeiffer et al., 2012)  Participation in Rural Environmental Protection Schemes, (Hynes et al., 2008)

5. Methodology  Data Enhancement or Sampling Methodology  Spatial Microsimulation  Currently on third version  Much improved speed and accuracy  Choice of Methodology (Hermes and Poulsen, 2012)  Iterative Proportional Fitting (Deming and Stephen, 1940)  Deterministic Reweighting (Ballas et al., 2005)  Regression Based Reweighting (Tanton et al, 2011)  Combinatorial Optimisation (Williamson et al., 1998, Ballas and Clarke, 2000)  Quota Sampling (Farrell et al. 2013)  Chosen because other methods have issues with  Areas with small sample size  Runtime  Incorporating local stocking rate

6. Methodology  Sample Farms from Teagasc National Farm Survey/Irish FADN (NFS) to generate spatial samples with same structure as Census of Agriculture  By Size, System and Soil Type  NB not the same farms, merely the same characteristics  Adjustment to ensure that stocking rates are compatible  Challenges  NFS is not representative of the smallest farms and  on land with poorest quality  Cell sizes can be challenging  Choice to sample from within region/LFA  Post-sample Adjust for Differences in Income Variability

7. Sampling Choices Scenario101111010001010101001101100110111101111 LFA110000000101 Stocking Rate Adjustment 010011010011 National (1)/Regional (0) Sample 110000111111 Regional Fixed Effect Adjustment 000101001111

8. Validation  Methodology  12 Alternative Choices  Validate against  Census variables  Local stocking rate  Winners and Losers from CAP reform at regional level (using NFS analysis)  Correlations of Best Method  Census Variables ~ 90%  Stocking Rate ~ 70%  Caution on peninsulas, due to limited number of small farms  However good across main farming areas Correlation Range for 12 Choices

9. Validation - System

11. Validation - Soil

12. Validation – Correlation with Census of Agriculture Scenario101111010001010101001101100110111101111 LFA sub-group110000000101 Stocking Rate Adjustment 010011010011 National (1)/Regional(0) Sample 110000111111 Regional Fixed Effect Adjustment 000101001111 Average 0.9370.9400.8840.8890.8900.8910.9370.9400.9370.9390.9440.950 Target Variable LU per ha0.3970.8610.4110.3770.6810.6800.3970.8610.3910.3700.8600.874 1111 – highest correlations Regional sampling – worse Stocking Rate adjustment important from a

13. Structure of Agriculture

14. Tillage Dairy Beef and MixedSheep Tillage – East and South Dairy – South Beef – Midlands and West Sheep – West and North West

15. Farm Incomes Farm Income/ha Market Income/ha Direct Pay per ha Incomes Higher – South and East of Kerry-Dundalk line Direct Payments reflect both SFP and Disadvantaged Areas

16. Viability Viable Income > Min Wage + Return on investment Sustainable Not Viable but with job Vulnerable Not Viable, no job Viability – reflects higher incomes Sustainable in West and Midlands Vulnerable North West, Border, Coast

17. Spatial Distribution Between DistrictWithin District Baseline Direct Payments17.882.2 Family Farm Income14.585.5 Far more variability between farms than between areas But many winners and losers within the same areas

18. Next Steps  Solve the peninsulas/small farms issue  Methodology is currently being used in 3 other countries  Utilises data that is available in all EU members states  Pitch to EU Commission to fund development for all EU members states for use in EU CAP and RDP policy analysis  Gap in capacity at the moment  Model future Less Favoured Area reforms  Improve the consistency with localised environment

19. Thank You www.teagasc.ie www.teagasc.ie