Evaluating Agri-Environmental Evaluating Agri-Environmental Schemes – The Marginal Cost of Schemes – The Marginal Cost of

Ecosystem ServicesEcosystem Services

Johannes Sauer and Ada WossinkJohannes Sauer and Ada WossinkEconomicsEconomics, School of Social Sciences, , School of Social Sciences,

University of Manchester University of Manchester

2

What is the issue?

• Growing literature on the societal relevance and valuation of ecosystems services

• This knowledge is important but understanding and modelling the underlying processes leading to service provision is essential for predicting and managing change in ecosystem services (Nicholson et al., 2009).

• We address this issue in the context of agri-environmental agreements

3

Research Question

How to determine the cost of marginal ecosystem changes and the effectiveness of green payments based on a theoretical and empirical analysis of the bio-economic production relationships at the micro (farm) level.

supply curves should be estimated at a low level of aggregation accounting for biophysical and socio-economic variability

relationships between marketed output and non-marketed ES, and assessment of direct and opportunity costs at the margin

4

Approach/Contribution

• New theoretical approach based on generalized joint production model which allows complementary, substitutive and competitive relationships.

• We implement this theoretical framework empirically as a transformation function.

• Third, we include farm/farmer specific impacts and use panel data analysis.

• We apply our approach to UK data on the Environmental Stewardship Scheme (ESS) and the Hill Farm Allowance (HFA).

5

• Change in the supply of marketed output Y simultaneously affects the supply of non-marketed ecosystem service denoted by Z • Inputs contribute to both outputs and Z affects Y • Economies of scale and scope

ES-Dimension of Agricultural Production

Yield

ES positive

Z1ES negative

Z2

PPF1

6

Theoretical Model

• The two outputs are produced simultaneously but since these are multiple outputs a separate production function is used for each output.

• This leads to a generalised joint production model.

• This model allows for joint inputs and the possibility of varying the proportion of agricultural output and ES.

7

Theoretical Model

where Y = agricultural output Z = non-marketed ecosystem service X = input contributing to Y and Z D = site specific biotic and a-biotic environment.

YDZXF );,(

s.t.

cpXCMinX

ZDXG );(constraint on level of ecosystem service to establish marginal cost of trading-off Y for Z

8

F.O.C.: 021 XX GFp

0);( YDXF

0);( ZDXG

*

**1*

2

*

X

X

G

Fp

Z

C

Envelope theorem and FOC:

Theoretical Model

X

Z

Z

F

X

F

X

FFX

dd

Envelope theorem and FOC:

for an optimal solutionwhere 1 and 2 as Lagrange multiplier for technology and ecosystem service constraints

2 as shadow price for Z;techn inter-dependencies and non-allocatable inputs for production of Z and Y

9

Three Cases

• Case 1: Complementary. The sum of the direct yield effect and the indirect yield effect of input use is positive (but decreasing) and the farmer can produce more Z while also increasing his commodity output Y. λ*2 = 0

• Case 2: Substitutive. Either the direct yield effect or the indirect yield effect, is non-positive but the net yield effect of the rearrangement of input X is positive. λ*2 = 0

• Case 3: Competitive. The direct yield effect of reallocating X is nil and there are yield losses caused by the required increase in Z needed to satisfy the constraint on the ecosystem services.

X

Z

Z

F

X

F

X

FFX

dd

10

Implementation

• Two schemes: the Environmental Stewardship Scheme (ESS) and the Hill Farm Allowance (HFA)

Implementation

11

Proportion of land under ESS Agro-env. payment per ha (ESS and other)

Implementation

12

• Panel data collected in England and Wales, 2005-2007

•Theoretical model implemented as a transformation function Y1= G(Y-1,X,T), where Y1 is agricultural output of the farms and Y-1 the vector of other outputs (Z payments and non-agricultural output YNAO).

• Generalized linear functional form, random effects specification.

13

Estimated Model

,euwithutyb

tzb tzbtxbttbtb

y xazxazxa

xxaxayaza za

x2ay2az2aza2ay

itiitititit NAO, NAOT

it5.0

it HFA, HFATit5.0

tiESS,ESSTit5.0

it

K

1k

kTitit Tit T

K

1k

5.0it AO,

5.0it k,kNAO

5.05.0it k,

K

1k

kHFA5.0

it ESS,5.0

it k,

K

1k

kESS

5.0it,

5.0it k,

K

1

k

K

1k

it k,kkit NAO, NAONAOit HFA,HFAHFA it ESS,ESSESS

K

1k

5.0it k,0k

5.0it NAO, 0NAO

5.0it HFA,0HFA

5.0it ESS,0ESS0it AO,

it HFA,

14

Results

Estimated Direct and Indirect Effects

Effect evaluated Mean Std. Dev. Min Max

dYAO/dX 173.978 259.197 -440.066 1591.110

dYAO/dZESS .372 2.887 -8.233 12.288

dYAO/dZHFA -2.529 6.310 -39.071 23.947

(dYAO/dZESS)(dZESS/dX) 0.065 0.032 0.006 0.192

(dYAO/dZHFA)(dZHFA/dX) 0.071 0.058 0.007 0.438

(dYAO/dZESS)(dZESS/dZHFA) -6.61e-04 5.61e-04 -0.004 -7.01e-05

(dYAO/dYNAO)(dYNAO/dZHFA) 9.03e-05 7.74E-05 1.21e-05 5.83e-04

(dYAO/dYNAO)(dYNAO/dZESS) -5.03e-05 3.09E-05 -2.24e-04 5.24e-06

(dYAO/dYNAO)(dYNAO/dX) -0.008 0.005 -0.043 -7.11e-04

15

Results

# Observations per case for Product-Product Relationships

Relation Agric. outputESS

Agric. outputHFA

HFAESS

ESSHFA

Direct effect

dYAO/dX dYAO/dX dYAO/dZHFA dYAO/dZESS

Indirect effect

(dYAO/dZESS)*(dESS/dX)

(dYAO/dZHFA)*(dZHFA/dX)

(dYAO/dZESS)*(dZESS/dZHFA)

(dYAO/dZHFA)*(dZHFA/dZESS)

Case I 314 314 0 0

Case II 0 0 121 202

Case III 79 79 272 191

Total Obs. 393 393 393 393

04/18/23 vTI

Options for Efficient Reorganisation of Production

Y – Zrelationship

Case 1 –complementary

Case 2 –substitutive

Case 3 –competitive

AO - ESS+ agricultural output+ environm output

(314 farms)(79)

AO - HFA + agricultural output+ environm output

(314 )(79)

HFA - ESS (121) + environm output 2 (272)

ESS - HFA

+ environm output 1or+ environm output 2

(202)(191)

HFA - NAO(121)

+ non-agricultural output

(272)

ESS - NAO+ environm output 1or+ non-agricultural output (202) (191)

AO - NAO+ agricultural outputoder+ non-agricultural output (314) (79)

17

Conclusions - Empirics

The majority of farms produce agricultural output and ecosystem services in a complementary relationship (λ*2 = 0 )

Generation of multiple ecosystem services on the same farm showed either a substitutive or competitive relationship.

Changing the composition of the ecosystem services output (HFA, ESS) would have very different implications for individual farms.

18

Current / Further Work

• modelling issues:

• monetary payments vs. “service quantities”• correction for potential selection bias

• look at spatial patterns

• investigate significant characteristics of the farms being part of the classes I-III as estimated in our paper

• more sophisticated modelling approaches (dynamic panel data approach, mixed-effects logistic approach)