Application of system dynamics in the analysis of economic impacts of Rift Valley fever in Kenya

Karl M. Rich and Francis Wanyoike International Livestock Research Institute

Presented at a Dynamic Drivers of Disease in Africa Consortium (DDDAC) project workshop Naivasha, Kenya, 24 – 26 June 2014

Background

• In 2007, ILRI carried out a study of the socio-economic impacts of the 2006 Rift Valley fever (RVF) outbreak in Kenya

• Results shed light on how various value chain actors were affected by the outbreak including also the macroeconomic impacts

• The analysis we carried out was however static

• We now wish to conduct a more dynamic kind of analysis of the disease impacts and also simulate how these effects would change in the light of various interventions (policy or otherwise) that could be instituted

Use of system dynamics in disease impact analysis

• This method has been used by ILRI in Botswana to analyze the impacts of FMD outbreaks on trade and competitiveness (Hamza et al. 2014) and Cambodia (Rich and Roland-Holst 2013)

• What is system dynamics (SD)?

o SD is a computer-aided approach to policy analysis and design

o It applies to dynamic problems in complex social, managerial, economic, or ecological systems — such dynamic systems are characterized by interdependence, mutual interaction, information feedback, and circular causality.”

o A methodology for studying complex dynamic systems that include nonlinearities, delays, and feedback loops

• When used in analysis of disease impacts, SD provides a way to integrate epidemiological and economic and other impacts in one platform

Conceptual framework

Financial

Performance

Policy

Risk

Production

Motivation

• An integrated epidemiological-economic model captures 2

types of disease related impacts:

i. Disease outbreaks that influences productivity or are associated with

significant mortality, and also the control measures taken to control

them will influence stocks of animals held by farmers

ii. Animal diseases can have demand impacts, either by reducing

domestic demand due to perceived food safety concerns or

international demand through trade bans, or both

• Note that in both cases, feedback effects matter

o When you have an outbreak, this influences behavior of actors

(Distress selling by farmers) and this in turn influences the spread of

the disease

The Epidemiology module

The diseases spread is based on the RVF transmission

model of Gianni:

Model captures dynamics of mosquito populations (Aedes

and Culex) and their interactions with livestock

The idea is to program the existing model from R into

STELLA (a system dynamics software) to allow integration

with other modules.

Model has been parameterized based on R model, though a

few aspects remain problematic (e.g., endogenizing sizes of

ponds for mosquitos)

The Epidemiology module: Culex dynamics

The Epidemiology module: Aedes dynamics

The Epidemiology module: Disease spread dynamics

The economics/value chain component also has 3

modules /sectors

The economic/value chain module

Market dynamics module: Based on the supply

and demand model of Whelan and Msefer (1996).

Financial costs module

Herd dynamics module: based on the DynMod

model developed by Lesnoff et al. (2008)

Herd Dynamics

Market dynamics module

Financial costs module

Results

The model is still under construction so the results

shown below are just illustrative of how we aspire

to use it

The graphs simulate what happens to cattle

population in Ijara district in the event of an RVF

out break with and with out vaccination

Similar kinds of analysis will be done for value of

livestock, costs, sales, prices etc.

Preliminary results

5:03 AM Wed, Jun 25, 2014

Disease ev olution in liv estock

Page 1

1.00 15.75 30.50 45.25 60.00

Day s

1:

1:

1:

2:

2:

2:

3:

3:

3:

4:

4:

4:

0

45

90

0

20

40

0

150

300

0

200

400

1: Exposed liv estock 2: Inf ected liv estock 3: Recov ered liv estock 4: Suspectible liv estock

11

1 12 2 2 23

3 3 3

4

4 4 4

Disease evolution in the event of an outbreak with no vaccination

4:57 AM Wed, Jun 25, 2014

Disease ev olution in liv estock

Page 1

1.00 15.75 30.50 45.25 60.00

Day s

1:

1:

1:

2:

2:

2:

3:

3:

3:

4:

4:

4:

0

3

6

0

2

3

280

310

340

0

40

80

1: Exposed liv estock 2: Inf ected liv estock 3: Recov ered liv estock 4: Suspectible liv estock

1

1

1

12

2

2

23

3

3 34

4

4 4

Disease evolution in the event outbreak but 80% of animals

had been vaccinated

Total number of animals with and without

vaccination

4:57 AM Wed, Jun 25, 2014

Untitled

Page 1

1.00 15.75 30.50 45.25 60.00

Day s

1:

1:

1:

270

315

360

Total number of liv estock: 1 - 2 -

1

1 1 1

22

2 2

Data needed • Livestock demographics & population dynamics

• Animal movements data

• Elasticities (demand, supply, income)

• Value chain process variables

– Period of time taken between farm sales and market arrivals

– Period of time taken between sales from farms and slaughter (weeks)

– Inventories of meat (weeks)

• RVF epidemiological data

• Market prices

• RVF control costs

• Draught labour parameters

Thank you