You are on page 1of 29

Can rice yield gaps be closed?

Explaining yield gaps


at farm level with a case study for Central Luzon

Joo Vasco Silva


Plant Production Systems
Wageningen University
IRRI CESD/SSD Seminar
24th November 2014

Outline of
the presentation
I.

Introduction and objective

II.

Theoretical framework
A. Concepts of production ecology
B. Decomposing the yield gap

III. Materials and Methods


A. Central Luzon Loop Survey
B. Frontier analysis
C. Crop modelling: ORYZAv3
IV. Preliminary results
V. Summary and final thoughts
VI. Next steps and further research

Introduction:
Rice yield gaps in Southeast Asia

Introduction:
Methodologies for Yg analysis
Yield gap analysis can be used to investigate the relative importance of
different growth factors and to explain actual production levels (van Ittersum
& Rabbinge, 1997)
Crop models have been widely used for yield gap analysis and especially
for assessing the theoretical maximum crop yields from a biophysical
perspective but... they do not take into account the effect of biotic factors
and cannot deal with socio-economic factors such as labour and cash.
Frontier analysis is a highly flexible, versatile and robust benchmarking
method widely applied to farm survey data by (agricultural) economists. The
outcomes tend to be input dependent which makes the comparison across of
studies rather difficult.
Why not combining both approaches in a novel theoretical framework to
take advantage of the opportunities offered by each of them?

Introduction:
Why focusing on the farm level?
Analysis at field level: too fine resolution and does not capture interactions between
different fields/production activities of the same farm which affect operational, tactical
and strategic decisions made by farmers.
Analysis at regional to global levels: too coarse and difficult to capture dynamics and
feedbacks occurring in each particular farming system. Useful for relative comparisons
and identify target regions but no meaningful explanations.

Boling et al. (2010) AgSys

Neumann et al.
(2010) AgSys

Introduction:
Research objectives
The objective of this study is to apply and test a combination of
methods, i.e. frontier analysis and crop modelling, to explain rice yield
gaps in the period 1979 - 2012 in Central Luzon, the Philippines.
In this presentation, I will provide a (very) preliminary analysis for the
periods 1979-1980, 1990-1991 and 2003-2004.

Specific research questions for this presentation are:


1. What is the magnitude of the intermediate yield gaps (i.e. efficiency,
resource and technology) in rice-based farming systems in Central
Luzon?
2. How those intermediate yield gaps have changed with time?

Theoretical framework:
Concepts of production ecology

van Ittersum & Rabbinge (1997)

Production level (y)

Theoretical framework:
Efficiency yield gap

YBF
YTEx
For consistency, important
to control for:

Efficiency
yield gap

YA

Input level (x)

Weather conditions
Soil types
Rice varieties

Analysis in multiple InputOutput setting


It is explained by crop
management referring to
time, space, and form of
application of inputs

Production level (y)

Theoretical framework:
Resource yield gap

YBF = YOpt
Resource
yield gap

YTEx
YOpt
Resource
yield gap

Farmersobjectives given
a set of constraints:

Input level (x)

Maximising profits
Minimising risks
HH food security
Maximising RUE

Theoretical framework:
Technology yield gap
YP
Production level (y)

Technology
yield gap

YBF

Useful to isolate the effect


of growth-reducing and
growth- limiting factors.

Input level (x)

It is explained by current
technologies not being able
to exploit the full climatic
potential.

Theoretical framework:
Methods and integration

See also: Herdt & Mandac


(1981!). Modern Ttechnology
and economic efficiency of
Philippines rice farmers.

Materials and methods:


Central Luzon loop survey (incl. soil survey)
Farm household survey 1966 - 2012
Agro-climatic
characterization
Unbalanced
Panel
Weather station: Munoz, NE
Period: 1983 - 2003

Database of 96 fields (data not


yet available):
Moya
Texture, organic C, T.
P-Olsen,
J. van
Wart et al.
Exch. K, bulk density,
depth
plough pan and bund height

Materials and methods:


Data Envelopment Analysis (DEA)
Years

Output-oriented DEA with


variable returns to scale

1979 (WS) 1980 (DS)


1990 (WS) 1991 (DS)
2003 (WS) 2004 (DS)

Efficiency
Yield Gap
(1 )

Output variables
Rice yield for MV3 varieties (kg ha-1)

Input variables
N application rate (kg N ha-1)
P application rate (kg P ha-1)
K application rate (kg K ha-1)
Seed rate (kg ha-1)
Herbicide (kg a.i. ha-1)
Insecticide (kg a.i. ha-1)
*all fields are irrigated

Data aggregated for an unique


farm x field x year x season
combination.

Kanellopoulos et al. (2014)

Limitations of current analysis!


1.

Biased TE scores: if we have K inputs


and M outputs in DEA, then at least
K+M units will be 100% efficient.

2.

Aggregating data in order to avoid


zero levels of inputs and outputs for
single observation.

Materials and methods:


Estimating YOpt and trade-offs

Resource
Yield Gap

Production perspective:
YBF was quantified as the average of YA values above the 95th percentile.
Economic perspective:
Yopt which corresponds to the maximum revenues, minimum costs or maximum profits can be quantified
using cost production functions, i.e. estimating the production frontier with inputs and outputs expressed in
economic terms.

Environmental perspective:
Yopt which corresponds to the highest resource-use efficiency of a specific input can be quantified by means
of indicators such as (single/total) factor productivity or by taking the first order derivative of the production
frontier. Crop modelling may as well be an option.
Food security perspective:
Approaches such as the land gap (Hengsdijk et al. 2014) can be useful to assess whether or not individual
households match their food security needs by taking into account data on crop yield, farm size and number
of household members.
Risk perspective:
Yopt corresponding to the minimal risk can be quantified using optimisation models based on linear
programming. This needs further thought....

Materials and methods:


Application of ORYZA v3

Technology
Yield Gap

Total Simulations:

Virtual Experiment:
500 N management strategies
500 random N
management strategies
1 production level (N limiting)

1 weather station (Cabanatuan)


Application rate
Application timing
2 soil types (WISE, PhilRice)

1 variety (IR72)

Number of splits

14 years and 36 periods (1999- 2013)

Simulation
Input File Generator
Organizer

= 500.000 simulations!

Preliminary results:
Descriptive statistics, inputs & outputs
Wet Season

Dry Season

P. Moya et al. (in press)

Preliminary results:
Efficiency yield gap (1)

Preliminary results:
Efficiency yield gap (2)
2nd stage frontier analysis
(multiple regression)

Eff
YG

No. land preparation operations


No. crop establishment activities
Seed source
No. fertiliser applications
Timing of fertiliser applications
No. pesticide applications
Timing of pesticide applications
How?

Labour use (ld ha-1) and quality


(hired vs family) per activity?
Importance of off-farm activities
(importance of rice)?
Tenure status of the field (e.g.
investment in infrastructure)?

Preliminary results:
Resource yield gap

Dry Season

Preliminary results:
Technology yield gap (1)

Preliminary results:
Technology yield gap (2)

Sowing Dates?
SSNM?

AWD?

Summary and conclusion:


Decomposing the yield gap in Central Luzon
Wet Season

Dry Season

Data from this presentation


and Yp from Laborte et al. (2012)

Explaining rice yield gaps:


The farm level (1)
Complete and thorough yield gap analysis must be done at the farm level and
hence, requires true multi-disciplinary approaches based on sound agronomy
and social sciences (socio-economic, institutional and political fields).
For the sake of consistency, it is important to have clear and standard
concepts and definitions of the yield gap in order to make explicit the
limitations and the assumptions behind the analysis.
Efficiency, resource and technology yield gaps matter! While efficiency and
technology yield gaps tell us which yield gains can be obtained from
improved crop management (under current and future technologies), resource
yield gaps inform us about possible trade-offs between
economic/environmental and production objectives at the farm level.

Explaining rice yield gaps:


Can they be closed? (2)

Closing rice yield gaps is about on-farm research to disentangle the complexFARM
interactions involved in crop management and to improve all-farmGAP
performance, including off-farm activities and income.

Improved rice crop management is affected by the returns to labour offered


by off-farm activities and by farmers management skills. Closing efficiency
and technology yield gaps might require a more attractive agricultural sector as
well as a greater network for giving local support to farmers.

There are trade-offs at farm level which may not make it interesting for
individual farmers to increase their production and hence close the resource
yield gap. Apart from that, environmental concerns and regulations might
further hinder such yield increases.

Time to break free from the yield gap trap ?? (by: Stephanie Malyon , CIAT)
Yes, rice yield gaps can be closed!!
But at what cost and risk?!

Looking forward...
and further research!
Refine the estimates of YTEx and estimate resource yield gaps: take the advantage
of having a rich panel dataset and use more advanced methodologies (i.e. stochastic
frontier). Make the analysis per soil type.

Finalize the evaluation of ORYZAv3 and refine its application for Central Luzon by
integrating crop management information from the CLLS and the quantitative soil
data collected. Expand frontier to other inputs!
Think further how to truly integrate these two methodologies in order to have as
much explanatory power as possible.
Identify the main causes behind the efficiency, resource and technology yield gaps
either by means of quantitative (e.g. multiple regression) or qualitative approaches
(e.g. Bennedict Kerkvliet: Everyday Politics in a Central Luzon Village).
Test and extend the framework with data from other rice farming systems of SE Asia
(e.g. Mekong Delta, Vietnam)...?

Acknowledgements
Muito obrigado!
@Wageningen:

Prof. Dr. Martin van Ittersum (Plant Production Systems)


Dr. Pytrik Reidsma (Plant Production Systems)
Prof. Dr. Ken Giller (Plant Production Systems)

@IRRI:

Dr. Alice Laborte & team (GIS Lab & MRTeam)


Dr. Piedad Moya & team (SSD)

Dr. Tao Li (Crop Modelling)


Man Marcaida (Crop Modelling)

Ferdinand Corcuera (Crop Modelling)


Dr. James Quilty (Soil Sciences)

Thank you for your attention!


Questions & suggestions?

Joo Vasco Silva


Plant Production Systems
joao.silva@wur.nl

Materials and methods:


Evaluation of ORYZA v3

RMSE = 16.81

Technology
Yield Gap

RMSE = 21.42

You might also like