KARNAL BUNTthe regional economic effects of a potential incursion

abare e Report 04.4

ABARE report for Plant Health Australia

Lisa Elliston, Alasebu Yainshet and Ray Hinde

January 2004

abare e Report 04.4

ii

© Commonwealth of Australia 2004

This work is copyright. The Copyright Act 1968 permits fair dealing for study, research, news reporting, criticism or review. Selected passages, tables or diagrams may be reproduced for such purposes provided acknowledgment of the source is included. Major extracts or the entire document may not be reproduced by any process without the written permission of the Executive Director, ABARE.

Elliston, L., Yainshet, A. and Hinde, R. 2004, Karnal Bunt: The Regional Economic Effects of a Potential Incursion, ABARE eReport 04.4 Prepared for Plant Health Australia, Canberra, January.

ISSN 1447-817XISBN 1 920925 05 8

Australian Bureau of Agricultural and Resource EconomicsGPO Box 1563 Canberra 2601

Telephone +61 2 6272 2000 Facsimile +61 2 6272 2001Internet www.abareconomics.com

ABARE is a professionally independent government economic research agency.

ABARE project 2875

abare e Report 04.4

iii

foreword

Australia has a valued reputation for supplying high quality agricultural products to export markets with ‘disease free’ status. A potential disease incursion could seriously threaten this reputation and cause harm to the agricultural industry and surrounding regional economies, resulting in considerable losses in trade and incomes.

ABARE was commissioned by Plant Health Australia to develop an incursion management model capable of investigating the regional economic effects of a hypothetical incursion of karnal bunt of wheat in a case study region in south eastern Queensland. Quantifying the losses resulting from a potential karnal bunt incursion involves estimating both the direct losses to the wheat industry and the indirect or ‘spill-over’ losses that would occur across the rest of the regional economy. These include decreased commercial, wholesale and retail sales.

This report describes the model that was developed to simulate different disease incursion scenarios and estimate the resulting regional economy effects across the western Darling Downs. The development of the model enables the exploration and evaluation of the strategic manage-ment options — such as increased surveillance and expanded contain-ment or quarantine activities — available to Plant Health Australia and the Australian Government Department of Agriculture, Fisheries and Forestry in the event of a karnal bunt incursion in Australia.

BRIAN S. FISHER

Executive Director

January 2004

abare e Report 04.4

iv

acknowledgments

The authors wish to thank Pascal Perez, CIRAD representative at the Australian National University, and other model developers at CIRAD (Centre de coopération internationale en recherche agronomique pour le développement) in Montpellier, France, for valuable guidance in developing the exotic incursion management (EIM) model on the CORMAS platform. Simon McKirdy and Ryan Wilson of Plant Health Australia, together with John Brennan and Gordon Murray at the New South Wales Department of Agriculture, provided extensive informa-tion on the scientifi c and epidemiological aspects of Karnal bunt which was used to develop the model. Staff from the Queensland Department of Primary Industries, including John Switala and Peter Whittle, were instrumental in the choice of case study region. Kenton Lawson and Caroline Levantis from ABARE generated many maps for the project, while Stephen Beare, ABARE’s Research Director, provided invaluable oversight throughout the duration of the project.

abare e Report 04.4

v

contents

Summary 1

1 Introduction 5

2 Background 7Karnal bunt 7Case study region 8

3 Methodology 11Integrated bioeconomic modeling 11Multiagent systems 11Exotic Incursion Management model 12

4 Simulation results 17Contractor based incursion 17Contaminated fertiliser based incursion 22

5 Sensitivity analysis 26Identifi cation of bunted grain at silo 26Identifi cation of bunted grain on-farm 27

6 Conclusions 28

AppendixesA The input–output model 30B Regional economywide effect of a $1 a tonne

change in export demand for wheat 34

References 35

abare e Report 04.4

vi

fi guresA Area infested with karnal bunt teliospores 18B Direct economic effects on value of output 19C Economic effect of a contractor based incursion 20D Area planted to wheat for grain export 22E Direct economic effects on value of output 23F Economic effect of a fertiliser based incursion 24G Area infested with teliospores, sensitivity analysis 26H Graphic representation of the I–O table 30

map1 Case study region 8

tables1 Statistical local areas in case study region 92 Number of businesses and employment, by industry 93 Selected agriculture statistics for the case study

region 104 Parameters representing the spread characteristics of

the disease 125 Model yield and price parameters 146 Probability of detection 157 Gross volume and value of production, reference case 178 Changes to production and returns, contractor based

incursion 189 Regional economy effects of a contractor based

incursion 2010 Distribution of adverse fl ow-on effects between

industries in the region 2111 Changes to production and returns, fertiliser based

incursion 2312 Regional economy effects of a fertiliser based

incursion 2513 Distribution of adverse fl ow-on effects between

industries in the region 26

abare e Report 04.4

1

summary

Karnal bunt of wheat is caused by the smut fungus Tilletia indica Mica. Despite only minor yield losses, infected grains emit a strong fi shy odor and are unfi t for human consumption. Air currents, vehicles and farm machinery can spread karnal bunt teliospores. The teliospores have also proven resistant to adverse environmental conditions, remaining viable for two to fi ve years in contaminated soil. The combination of the effect of the teliospores on wheat grain quality and the diffi culty in control-ling an incursion has made karnal bunt the subject of strict quarantine regulations by a number of wheat importing countries.

Australia has a valued reputation for supplying high quality wheat with ‘disease free’ status to export markets. A potential incursion of a disease such as karnal bunt could seriously threaten this reputation and cause harm to the wheat industry and surrounding regional economies, resulting in considerable losses in trade and incomes.

Quantifying the total loss in income from a potential karnal bunt incur-sion is diffi cult because it involves both direct and indirect losses. The direct losses to the wheat industry, such as lost wheat export markets resulting in lost income and lower property values, have been quanti-fi ed by other researchers for various regions under a range of different scenarios. However, the indirect or ‘spillover’ losses that would also occur across the rest of the regional economy, including decreased commercial, wholesale and retail sales, are no less important but signifi cantly more diffi cult to quantify. Providing an assessment of these spillover effects is essential in order to determine optimal or best practice plant surveillance containment, and eradication strategies. This is because optimal surveillance depends critically on a full estimate of the potential losses of income that can result in a region as a result of a disease incursion, and on the costs associated with surveillance.

An agent based simulation model and an input–output model describing the regional economy were used to determine the effects of two hypo-thetical karnal bunt incursions for a case study region in south eastern Queensland.

abare e Report 04.4

2

The region covers agricultural land in the western Darling Downs, south of Roma and west of Dalby, down to the New South Wales border. The region relies heavily on the agricultural sector, with almost 60 per cent of all registered businesses belonging to the agriculture, forestry and fi shing industry. Around 90 per cent of all land in the region consists of agricultural holdings. More than 1.1 million hectares of crops were grown in the region in 2001, and over half of this was sown to wheat. Agricultural land not used for cropping activities in the region was predominantly used in the grazing of beef cattle.

The agent based model integrates the physical process of disease trans- mission with measures to eradicate or contain the incursion. The resulting impact on the farming system is then calculated. Separate modules capture the characteristics of the disease, the farming system, the incursion response and management of an outbreak in the region. A stylised representation of the regional economy is then used to measure the fl ow-on effects of a karnal bunt incursion.

A range of potential transmission vectors are explicitly incorporated into the model. These include farmers and their machinery, contrac-tors and their machinery, the wind, and contaminated farm inputs such as fertiliser. The probability of the disease spreading by each of these vectors varies with the time of year.

In a weekly time stepped model, farmers decide what crops to plant and when to plant them. They decide whether or not to employ contractors to spray for weeds and harvest crops during the year. Based on average yields and prices for agricultural commodities in the region, the total volume and value of production within the region is calculated.

A karnal bunt incursion in the case study region is identifi ed when infected grain is sent to the silo at the end of the year and correctly recognised. When this occurs a quarantine response is triggered to investigate the extent of the incursion and attempt to contain it. Quar-antine offi cers are dispatched to search the farm where the infected grain came from and all neighboring properties. Contractors that have been on infested properties are identifi ed and any other farms they have visited during the year are also investigated for signs of the disease. Farms that have been positively identifi ed as having the disease are quarantined and are unable to grow any wheat for fi ve years. Crops generating lower economic returns are grown on these properties for the duration of the quarantine period. Farms neighboring the positively

abare e Report 04.4

3

identifi ed and quarantined farms form a buffer zone and are only able to grow wheat for feed grain purposes, obtaining a correspondingly lower price for their wheat.

The economy is represented by a set of input–output tables that capture the supply and demand of goods and services in the case study region, as well as the interdependencies among industries and the associated primary factors of production. Changes in the value of wheat and other grain production as a result of a karnal bunt incursion and any subse-quent measures to limit the spread of the disease are then applied to the input–output tables to analyse the effect on other sectors of the regional economy as well as the overall loss of income.

Two hypothetical incursions were analysed to demonstrate the capacity of the model to explore the variations in regional impacts that result from different transmission vectors. In the fi rst scenario, a contractor with an infected piece of machinery introduces the disease to the region. The disease is slow to spread through the region, delaying its identifi cation in many parts of the region. The incursion is unable to be contained within the fi fteen year planning horizon considered. In net present value terms, the overall loss in regional output as a result of an incursion of this nature was estimated as $20 million over the fi fteen year simulation.

In the second scenario, a contaminated load of fertiliser is allowed into Australia and the disease is transported with the fertiliser to a number of farms within the region. Half of one per cent of all agricultural land in the region becomes contaminated with karnal bunt teliospores in the fi rst year of the simulation. The disease spreads signifi cantly faster in this scenario and, despite some limited success with containment part way through the simulation, the disease fails to be contained within the fi fteen year planning horizon considered. In net present value terms, the loss in gross regional product as a result of this much larger incursion was an estimated $126 million.

These scenarios indicate that based on the model parameters it would be diffi cult to eradicate and contain a karnal bunt incursion. The model allows for a range of strategic management options to be explored, such as increased surveillance and expanded containment or quarantine activ-ities. A number of contaminated fertiliser based scenarios were investi-gated where the likelihood of the disease being identifi ed at the silo or by farmers was increased. These results indicated that although the area

abare e Report 04.4

4

infested with teliospores could be reduced, the overall economic effects were unable to be signifi cantly reduced, as the disease failed to be contained. One reason for this result may be the relatively small quar-antine region or buffer zone adopted in the scenario. An increase in the size of the region quarantined around an infested farm may be required before a disease incursion of this nature can be contained successfully; however, the increase in management costs associated with this would need to be taken into consideration.

With additional development, the modeling framework used to analyse the effects of a karnal bunt incursion could be extended to analyse the incursion of different plant diseases and pests in other case study regions.

abare e Report 04.4

5

introduction

In 2001-02, Australia produced almost 25 million tonnes of wheat (ABARE 2003). Almost two-thirds of the wheat produced was exported, and this trade was valued at $4.6 billion. Australia has a valued reputation for supplying high quality wheat with ‘disease free’ status to export markets. A potential incursion of a disease such as karnal bunt could seriously threaten this reputation and cause serious harm to the wheat industry and surrounding regional economies, resulting in considerable losses in trade and incomes.

Quantifying the loss in income from a potential karnal bunt incursion can be diffi cult because it involves both direct and indirect losses. A loss of wheat export markets would lead to a fall in income and property values in the wheat industry. These direct losses have been quantifi ed for various regions under a range of different scenarios (see CIE 2002; Murray 1998; Brennan and Warham 1990). However, indirect or ‘spillover’ losses would also occur in surrounding communities in the event of a karnal bunt incursion and are more complicated to estimate.

A disease outbreak can be especially damaging to a regional economy. A fall in income and asset values in the wheat industry can potentially generate a chain reaction in terms of decreased commercial, wholesale and retail sales throughout the local community and farm region. Providing an assessment of these spillover effects is essential in order to deter-mine optimal or best practice plant surveillance, because the assigned level of surveillance depends critically on a full estimate of the potential losses of income that can result from a disease incursion.

ABARE was commissioned by Plant Health Australia to evaluate the local and regional economy effects of a potential karnal bunt incursion for a case study region in Queensland. This involved the development of an agent based model to simulate the effect of a hypo-thetical karnal bunt incursion and the development of an input–output table to estimate the resulting effect on the rest of the regional economy. The model and subsequent computa-tion of economic effects are documented in chapter 3.

In developing a model that integrates the biophysical aspects of disease incursion with the farming system and the wider regional economy it is possible to extend the analysis to consider the economic implications of incursions with different characteristics. Two hypo-thetical incursions are analysed in chapter 4 and demonstrate the capacity of the model to explore the range of impacts from a disease incursion and the possible strategic responses that may reduce the associated costs.

1

abare e Report 04.4

6

In the fi rst scenario, a contractor with an infected piece of machinery introduces the disease to the region. In the second scenario, a contaminated load of fertiliser is allowed into Australia and the disease is transported with the fertiliser to a number of farms within the region. In chapter 5, some of the model parameters were varied to investigate the sensitivity of the results to different assumptions regarding the likely identifi cation of the disease.

With some additional development, the modeling framework would also be suited to analysing the incursion of different plant diseases and pests in other case study regions. It is anticipated that following peer review, the framework will be extended to model and analyse the spread of both weeds and insects in other case study regions.

abare e Report 04.4

7

background

Karnal buntKarnal bunt of wheat is caused by the smut fungus Tilletia indica Mitra (Bonde, Peterson, Schaad and Smilanick 1997). Teliospores on the soil surface germinate in the presence of water, and are splashed or blown onto developing wheat heads at the time of fl owering. As the wheat kernels mature, large numbers of teliospores are produced and are redeposited on the soil surface at harvest time, continuing the cycle. Long range teliospore dissemination can occur by the transport of infected or contaminated seed. Karnal bunt teliospores can also be spread by air currents and by vehicles and farm machinery (Bonde et al. 1997).

Despite the strong fi shy odor of infected grains, only a portion of an infected kernel is usually replaced with a mass of teliospores, and only a few kernels in some wheat heads are likely to be infected. This makes the disease diffi cult to detect under fi eld conditions (Bonde et al. 1997).

Karnal bunt teliospores have proven resistant to adverse environmental conditions, remaining viable for two to fi ve years in contaminated soil (Bonde et al. 1997). Laboratory tests have shown that the teliospores are also very resistant to both chemical and physical treatment. Teliospores can survive freezing for several months. Seed treatment is not capable of protecting plants from infection if they are planted in infested soil, and fungi-cides applied to the soil have not proven effective. Changes to farm management practices such as delayed sowing, reduced nitrogen fertilisation and reduced planting density have had little effect on the incidence of the disease and may also reduce wheat yields. Similarly, hot water treatment of seed that kill teliospores also reduces seed germination. The only real successful management of the disease has been the development and use of varieties with genetic resistance.

Only minor yield losses are associated with the disease, with studies in India and Mexico showing a decline in yield of less than 1 per cent on average (Bonde et al. 1997). The percentage of bunted seeds in samples of infected wheat has also been consistently low. While karnal bunt infected grain is not toxic, infected grain emits a fi shy odor due to trimethylamine. When more than 3 per cent of grains are infected, wheat products become unfi t for human consumption (Nagarajan et al. 1997).

This quality effect, combined with the diffi culty in controlling a karnal bunt incursion, has made karnal bunt the subject of strict quarantine regulations by a number of wheat importing countries. Currently, more than twenty countries have karnal bunt on their quar-

2

abare e Report 04.4

8

antine list, preventing the import of wheat from any country in which karnal bunt has been confi rmed (American Phytopathological Society 1996).

The disease was fi rst discovered in north west India and is now considered common in the Puinjab region of India. It is largely confi ned to the Middle East, reported in Pakistan, Iraq, Iran, Lebanon, Nepal and Afghanistan (Stansbury, McKirdy, Diggle and Riley 2002). However, in the early 1970s it was reported in Mexico, in the mid-1990s it was found in the southern wheat growing regions of the United States, and in 2000 an incursion was identi-fi ed in South Africa. Karnal bunt has never been identifi ed in Australian wheat.

Case study regionThe analysis of the local and regional economy effects of a karnal bunt incursion in Australia was conducted in a case study region in south eastern Queensland (map 1).

The region consists of twelve ABS statistical local areas, which represent ten shires and two towns (table 1). The region covers agricultural land in the western Darling Downs, south of Roma and west of Dalby down to the New South Wales border.

1 Case study region

abare e Report 04.4

9

At almost 105 000 square kilometres, it represents around 6 per cent of the state of Queensland (OESR 2003). An estimated 50 000 people resided in the case study region in December 2001.

Regional economyAs at September 1998, 3 per cent of busi-nesses in Queensland were located in the case study region. The majority of busi-nesses — almost 60 per cent — belonged to the agriculture, forestry and fi shing industry (table 2). At the time of the 2001 ABS census of population and housing, the agriculture, forestry and fi shing industry employed just over 30 per cent of the workforce in the region. A further 8 per cent of all businesses were in the retail trade sector, employing around 12 per cent of the workforce. The remainder of businesses and employers were in the transport and storage, construction, wholesale trade and services sectors.

2 Number of businesses and employment, by industry

Number of Number of businesses persons employed

Agriculture, forestry and fi shing 3 545 6 942

Mining 38 134

Manufacturing 144 1 248

Electricity, gas and water supply 44 285

Construction 278 1 768

Wholesale trade 218 1 016

Retail trade 487 2 723

Accommodation, cafes and restaurants 177 865

Transport and storage 298 795

Communication services 25 191

Finance and insurance 72 284

Property and business services 187 1 366

Government administration and defence 40 797

Education 116 1 516

Health and community services 189 1 466

Cultural and recreational services 41 149

Personal and other services 216 454

Other (nonclassifi able) 0 582

Total 6 115 22 581

1 Statistical local areas in case study region

Balonne Shire

Bendemere Shire

Chinchilla Shire

Dalby Town

Goondiwindi Town

Inglewood Shire

Millmerran Shire

Murilla Shire

Tara Shire

Waggamba Shire

Wambo Shire

Warroo Shire\

abare e Report 04.4

10

Based on statistics collected by the Austra-lian Taxation Offi ce, the mean taxable income of taxpayers in the case study region for the 1999-2000 fi nancial year was around $30 000, around $3000 less than the mean taxable income across the whole of the state (OESR 2003).

The towns of Dalby and Goondiwindi are both explicitly included in the case study region and are large enough to warrant consideration as a statistical local area in their own right. A number of other smaller towns including Millmerran, Inglewood and St George are located throughout the remaining ten statistical local areas that make up the case study region.

Agricultural sectorThe case study region is dominated by agri-culture. The total area of agricultural hold-ings in 2001 was 9.5 million hectares, repre-senting around 90 per cent of all land in the region (table 3). In 1993 it was estimated that the number of agricultural establishments in the region was around 3000.

The total area of crops in the region was more than 1.1 million hectares — around 12 per cent of agricultural land. Almost half of the area cropped was sown to wheat in 2001. Agri-cultural land not used for cropping activities was predominantly used in the grazing of beef cattle.

The total gross value of agricultural production in the region for the 1998-99 fi nancial year was almost $1.1 billion, representing 18 per cent of the Queensland total (OESR 2003). Despite using only 12 per cent of all agricultural land in the region, cropping activities contributed more than 50 per cent of the total value of agricultural production. This refl ects the generally higher returns to cropping relative to livestock activities, particularly with respect to cotton grown in the region. Livestock disposals contributed around 40 per cent of the value of agricultural production, with the remainder coming from other livestock products.

The total value of crops in the region for the 1998-99 fi nancial year represented 20 per cent of the value of crops produced across Queensland (OESR 2003). Similarly, 15 per cent of the value of Queensland’s livestock disposals were produced in the case study region, and almost 13 per cent of the value of livestock products were produced in the region.

3 Selected agriculture statistics for the case study region

Region totals

– area of holdings ha 9 539 736

– area of crops ha 1 170 025

– beef cattle numbers no. 924 756

Total value

– agriculture $m 1 084.2

– crops $m 540.2

– livestock disposals $m 473.6

– livestock products $m 64.1

Wheat for grain

– area ha 527 489

– production t 489 845

– value $m 113.5

Sorghum for grain

– area ha 185 969

– production t 348 244

– value $m 50.8

abare e Report 04.4

11

methodology

Integrated bioeconomic modelingIntegrated bioeconomic models bring together a representation of biophysical processes with the economic system in which events occur. While biophysical models and economic models in their own right are capable of offering insights into the impact of a disease incursion, they often miss the complex interaction and feedback effects that exist between agricultural production, disease characteristics and economic returns.

In building an integrated bioeconomic model to represent a case study region it is possible to capture both the physical process of disease transmission from a variety of different vectors, and the economic impact of the spreading disease and any resulting management to eradicate or control the incursion. With a more detailed representation of the biophysical system it becomes possible to investigate the economic effects of different assumptions about the characteristics of the disease or the likelihood of detection at various stages.

Multiagent systemsMultiagent systems are increasingly being used to develop computer simulation models where the problem space results from the complex interaction of numerous individual components. It provides a framework for isolating each individual component within a system and programming its behavior as a collection of relatively simple commands. By taking a ‘bottom up’ approach to modeling, the behavior of individual components of a multifaceted system can be developed in detail. The fi nal output from agent based models represents the aggregation of all the micro-level interactions that occur between large numbers of different agents. This approach is often capable of providing insights into complex systems where the multitude of interactions between different components are too detailed to capture with more traditional modeling techniques.

While multiagent systems are somewhat limited as a forecasting tool, they have proven useful as a simulation tool. By changing the behavior of a single agent or set of agents it is possible to investigate the effect of this on the whole of the system. In the case of disease incursion management, the agent based nature of the model enables the simulation of numerous ‘what if?’ scenarios, investigating the sensitivity of both the extent of an incur-sion and the overall economic effect on a region to different assumptions. The results of such simulations can then become valuable input into the design of incursion management and response plans.

3

abare e Report 04.4

12

Exotic incursion management modelThe EIM (exotic incursion management) model developed for this project is an agent based model developed using CORMAS, a spatial natural resource and agent based simulation modeling framework running on the VisualWorks platform. Cellular automata techniques are used to drive the spread of the disease across neighboring paddocks, and a range of potential disease vectors are modeled explicitly. At the same time, numerous agents including farmers, contractors and quarantine offi cers — each with their own specifi c patterns of behavior and movement — are also interacting in the spatial environment.

The model consists of a number of distinct components, which work on a weekly time step. Separate modules attempt to capture the characteristics of the disease, the farming system, the incursion response and management of an outbreak in the region and a stylised representation of the regional economy to measure the fl ow-on effects of a karnal bunt incursion.

Disease characteristicsA number of parameters for the probability of karnal bunt teliospores being spread across the case study region make up a stylised representation of the disease. A number of trans-mission vectors are explicitly incorporated into the model. These include contract workers and their farm machinery, farmers and their farm machinery, contaminated farm inputs and the wind. The parameter values listed in table 4 were provided by Plant Health Australia.

4 Parameters representing the spread characteristics of the disease

Probability of contractor with teliospore infected machinery infesting a wheat paddock while spraying for weeds (v) 0.00001

Probability of a contractor with teliospore infected machinery infesting a wheat paddock while harvesting (v) 0.5 – 0.75

Probability of a contractor’s machinery becoming infected if they spray for weeds in an infested paddock (v) 0.0001

Probability of a contractor’s machinery becoming infected if they harvest an infested paddock (v) 1.0

Probability of a farmer with teliospore infected machinery infesting an uninfested wheat paddock elsewhere on their property during harvest time (s) 0.75

Probability of a farmer with teliospore infected machinery infesting an uninfested wheat paddock elsewhere on their property at any other time of the year (s) 0.0001 – 0.75

Probability of a farmer’s machinery becoming infected if they are in an infested paddock during harvest time (s) 1.0

Probability of a farmer’s machinery becoming infected if they are in an infested paddock at any other time of the year (s) 0.01 – 0.0001

Probability of teliospores spreading from one paddock to a neighboring one (due to wind) at harvest time (s) 0.9

Probability of teliospores/sporidia spreading from one paddock to a neighboring one (due to wind) at any other time of the year (s) 0 – 0.25

All probabilities are expressed per visit (v), or per season (s).

abare e Report 04.4

13

Contract workers, who spray for weeds early in the season and then return at the end of the season, can spread teliospores across the whole of the case study region. After passing through an infested wheat paddock, there is some chance that a contractor’s equipment will have become infected with teliospores. The probability of this happening is greater during harvest time than when spraying for weeds. Once infected, a contractor can then move teliospores to any other farms they subsequently visit. Similar to the probabilities for becoming infected, contractors infest wheat paddocks with different probabilities depending on the time of the year.

Farmers share some of the same behavioral characteristics of contractors. On farms where contractors are not used, farmers spray for weeds early in the season, and harvest at the end of the season. If any of their paddocks contain teliospores there are probabilities associated with the likelihood that their equipment will become infected and that they will spread the infestation to other paddocks on their property. Unlike contractors, however, farmers do not move off their property but randomly move around their property at all other times of the year.

Teliospores can also be spread across neighboring paddocks by wind. The probability of this happening at harvest time is signifi cantly higher than at any other time of the year. Because karnal bunt only affects wheat crops, it is assumed that teliospores are only spread across neighboring wheat paddocks. When another crop is planted in an infested paddock the disease is unable to spread by wind.

To model scenarios where the disease is spread via infected seed, a number of paddocks at the beginning of the fi rst season are randomly infected across the whole of the case study region.

Farm systemThe farm system is modeled within a weekly time-stepped year. Farmers decide what crops to plant and when to plant them within a planting timeframe that starts in early April and fi nishes at the end of June. To focus on the agricultural activities most likely to be affected by a karnal bunt incursion only four activities are included in the model: wheat, feed wheat, sorghum and other agricultural activities.

In the absence of an identifi ed karnal bunt incursion, farmers engage in a range of agri-cultural activities, including the production of wheat and sorghum. While the behavior of farmers is not represented by an optimising process, the limited number of agricultural activities included in the model means that farmers not producing wheat are assumed to produce the commodity generating the next highest economic returns. At the aggregate region level, the area of wheat, sorghum and other agricultural activities are calibrated to data collected as part of the 2001 ABS agricultural census.

After planting, the two major events in the remainder of the year involve spraying for weeds in July and harvesting crops throughout October, November and December. Farmers decide whether they wish to use contractors to spray for weeds and harvest crops, or whether they will do these activities themselves. This decision is made at the beginning of the simulation,

abare e Report 04.4

14

with farmers choosing to hire contractors doing so every year. On average, 65 per cent of all farmers in the case study region choose to hire contractors to spray for weeds and harvest. This was based on ABARE survey data which indicate that an estimated 65 per cent of all broadacre farmers in the case study region used contractors in 2001-02.

Contractors, or farmers, move through all crops during July spraying for weeds. It is possible for teliospores to be spread across the case study region during this period; however, the probability of infestation at this time of the year is low.

At the end of the year, contractors and farmers again move through all the crops on a prop-erty, this time harvesting crops. It is much more likely that the disease will be spread during this period.

Based on average yields reported in the case study region, the total volume of wheat and sorghum produced on each farm is calculated (table 5). The grain produced is sent to the silo and estimates of gross receipts are generated and returned to the farmer, based on average gross returns reported for the various commodities within the case study region.

At all other times of the years, farmers are able to move around their properties. As part of this movement farmers are able to spread teliospores from paddock to paddock.

Incursion response and managementThere are two ways in which an incursion of karnal bunt in the case study region can be identifi ed (table 6). There is a small chance that farmers will identify bunted grains in their infested paddocks during harvest time. In the majority of cases, however, the disease will be identifi ed when harvested grain is sent to the silo at the end of the year. Refl ecting the considerable uncertainty surrounding these estimates, the sensitivity of the results to different parameter values is analysed in chapter 5.

When teliospores are identifi ed at the silo a quarantine response is triggered to investigate the extent of the incursion and attempt to contain it so that it cannot spread further. The farm from which the infected grain came from is immediately quarantined and a quaran-tine offi cer is dispatched to the property. All neighboring properties are placed in a buffer quarantine zone. Quarantine offi cers visit each of the neighboring properties and search for signs of the disease. If teliospores are found on any of these properties, the properties are upgraded to full quarantine status and all properties neighboring this newly identifi ed farm are then searched. As this continues, a quarantine region is established. Where signs of infestation are not found on neighboring properties, those properties remain in the buffer quarantine region and the search of other properties stops.

5 Model yield and price parameters

Yield Price

t/ha $/t

Wheat 0.93 232Feed wheat 0.93 150Sorghum 1.87 146Other agricultural activities 1.00 106

Source: unpublished ABS agricultural data for the case study region.

abare e Report 04.4

15

At the same time, any contractors that have visited infested farms that are now fully quar-antined are identifi ed to trace back the source of the incursion and limit its spread. In the fi rst instance, contractors identifi ed in this process are asked to provide a list of all the farms they have visited during the year. Quarantine offi cers are then dispatched to each of these farms in order to identify the extent of the incursion. Where an infestation is identi-fi ed on a property, that property is fully quarantined and the search through all neighboring properties begins. Any contractors contacted in this trace back process who were carrying teliospores on their machinery are disinfected before the next season begins.

In the event of an identifi ed karnal bunt incursion, farms in the case study region can be classifi ed as: identifi ed as having an infestation and fully quarantined; identifi ed as not having an infestation but in a buffer quarantine zone because neighboring properties have the infestation; not quarantined (either clear of infestation or not yet identifi ed). Because of the limited range of effective eradication measures, any quarantine region must remain in place for fi ve years and farmers within the affected zone face reduced planting options throughout the period.

Fully quarantined farmers are unable to grow wheat of any kind and are assumed to plant sorghum and engage in other agricultural activities such as livestock or other nonwheat crops. Farmers in the buffer quarantine zone are unable to grow wheat for export or human consumption, but are able to grow feed wheat. It is assumed that farmers in this situation plant a combination of feed wheat and sorghum, together with their other nonwheat agri-cultural activities throughout the fi ve year quarantine period.

After fi ve years, the teliospores in infested paddocks are assumed to no longer be viable. The planting restrictions are lifted and farmers are able to resume growing wheat.

The regional economyThe economy of the case study region is represented in the model by a number of local towns and a single representative regional centre. The annual gross values of production for the various commodities at the farm level are aggregated to the regional level. In this manner, changes in the value of wheat production and other grain production as a result of a karnal bunt incursion and any subsequent quarantine measures can be estimated.

6 Probability of detection

Probability that a farmer will identify the presence of teliospores during harvest time 0 – 0.4

Probability that a farmer will identify the presence of teliospores at any other time of the year 0

Probability that teliospores will be identifi ed in wheat at the silo when less than 3 per cent of the grains are bunted 0

Probability that teliospores will be identifi ed in wheat at the silo when at least 3 per cent of the grains are bunted 0.5 – 0.9

Probability that quarantine offi cers will identify teliospores in infested paddocks when sent out into the fi eld following a suspected incursion 1.0

abare e Report 04.4

16

These changes in the value of production within the agricultural sector can then be traced through all other sectors of the regional economy using input–output (I–O) analysis. I–O tables contain the supply and demand of goods and services in an economy over a particular period, along with the interdependencies between the industries and associated primary factors of production. This enables analysis of the economywide effects of an exogenous (determined by factors outside the I–O system) change to an economy. By generating a range of I–O multipliers it is possible to estimate the effect of a karnal bunt incursion on the region’s output, employment, income and imports. (A brief overview of the I–O table, along with a description of the multiplier derivation process is provided in appendix A.)

The I–O analysis provides estimates of both the direct and indirect impacts of a change in agricultural production resulting from a karnal bunt incursion. The direct or initial impact captures the changes in wheat and other grain production and any associated changes in employment and income in the directly affected industries, as well as any changes in imports required by these industries. Subsequent changes in all other industries and the directly affected industries form indirect or fl ow-on impacts.

Estimation of fl ow-on effects involves quantifying the changes that would occur in indus-tries that are directly or indirectly linked to the wheat industry. For example, wheat produc-tion has a direct requirement of inputs from the manufacturing sector. The manufacturing sector, in turn, requires energy inputs to process its products. This creates an indirect depen-dence of the wheat industry on the energy sector. All these direct and indirect relationships within the regional economy need to be aggregated to quantify the overall economy wide fl ow-on effects of a karnal bunt incursion. The overall gross regional product effects — a regional equivalent of gross domestic product (GDP) — are then calculated to capture the income earned by all primary factors of production, labor, capital and management (profi ts) within the region.

The major limitation of I–O analysis is its overestimation of results. Overestimation arises because I–O analysis does not allow for price induced fl exibility between primary factors of production or labor and capital and between different commodities. As a result, substitu-tion between primary factors and between different commodities does not occur and the fl ow-on costs are greater. However, I–O analysis is likely to offer a reasonable approxima-tion for changes in a small region, such as the case study analysed in this paper because the potential price impacts are likely to be small. See appendix A for a more detailed presenta-tion of the underlying assumptions of I–O analysis.

abare e Report 04.4

17

simulation results

All simulations were run over a fi fteen year time horizon to generate an indication of the full extent and costs of a disease incursion. While the disease is not contained within this time horizon for a number of simulations, it is a long enough period to analyse the effect of surveillance and detection measures on the time taken for an incursion to be identifi ed. All fi nancial estimates are reported in constant 2001–02 dollar terms and the total effect over the fi fteen year period investigated is reported in net present value terms.

The probabilistic nature of many of the model parameters means that multiple simula-tions need to be run to generate robust estimates that are not dominated by any one single set of randomly determined parameter values. Therefore, a series of 100 simulations was conducted for each scenario and the results presented in this chapter refl ect the average results of those simulations.

Two scenarios are analysed in the sections below. The fi rst involves the introduction of karnal bunt to the case study region by a single contractor. The incursion begins at a single point and spreads across the region from there by the movement of farmers and contractors, as well as the wind. The second scenario has a diffuse starting point, to refl ect a situation where a load of contaminated fertiliser is unintentionally sold throughout the region and the incursion therefore starts at multiple locations.

A reference case was generated to establish a baseline against which all other incursion scenarios could be compared. The estimates of gross volume and value of wheat and sorghum production in the case study region are reported in table 7. The results are roughly consistent with the ABS data in table 3.

Contractor based incursionIn the scenario where a single contractor has contaminated machinery and introduces karnal bunt into the case study region the

4

7 Gross volume and value of production, reference case

Area sown Gross receipts

Year Wheat Sorghum Wheat Sorghum

’000 ha ’000 ha $m $m

1 503 194 108 532 500 193 108 533 500 193 108 534 502 194 108 535 501 194 108 536 501 196 108 537 497 194 107 538 503 195 108 539 501 193 108 5310 500 194 108 5311 504 193 109 5312 500 193 108 5313 499 193 108 5314 501 194 108 5315 502 194 108 53

All years 1 621 793

abare e Report 04.4

18

disease is slow to spread across the region. Less than 50 000 hectares are infested with teliospores within the fi rst nine years of the simulation (fi gure A). The relatively low levels of infestation on farms delay the identifi cation of the disease in many parts of the region and between the ninth and fi fteenth years of the simulation the area infested more than triples to around 145 000 hectares. The incursion fails to be contained and continues to spread at the end of the fi fteen year planning horizon.

The delay in identifying the presence of the disease in the case study region means that the quarantine effect is correspondingly delayed. The area planted to wheat for grain export

A Area infested with karnal bunt teliospores

21 5 7 8year

9 10 11 12 13 143 4 6 15

’000 ha

500

1000

1500

2000

Fertiliser scenario

Contractor scenario

8 Changes to production and returns, contractor based incursion

Area sown Gross receipts

Wheat Feed wheat Sorghum Wheat a Sorghum

Year ’000 ha ’000 ha ’000 ha $m $m

1 501 0 193 108 532 497 3 194 108 533 494 7 194 107 534 488 12 196 107 545 484 14 195 106 536 485 14 198 107 547 485 13 197 106 548 488 10 197 107 549 485 12 197 106 5410 482 14 198 106 5411 475 18 200 105 5512 471 22 202 104 5513 467 24 202 104 5514 466 25 205 104 5615 463 27 206 103 56

All years 1 587 813

a Sum of feed wheat receipts and wheat for grain export receipts.

abare e Report 04.4

19

purposes is only around 40 000 hectares less than in the reference case where there is no karnal bunt incursion by the fi nal year of the simulation (table 8). This decline in wheat sown for export is offset to some extent by an increase in the area planted to sorghum and feed wheat.

While the differences in gross value of production between the contractor incursion and the reference case are relatively in small in any given year, the total effect in net present value terms over the full fi fteen year planning horizon is not insignifi cant. The model indicates a $34 million (or 2.1 per cent) loss in the gross value of wheat production in the region as a result of a contractor based karnal bunt incursion. However, this is offset somewhat by a $20 million (2.4 per cent) increase in the gross value of sorghum production relative to the reference case.

Regional economy effectsInput–output (I–O) analysis was used to estimate the economic effects over the fi fteen year planning horizon of a single contractor with teliospore contaminated machinery introducing karnal bunt into the case study region. The annual change in the gross value of production of wheat and other grains reported for each of the fi fteen years were converted to equivalent changes in wheat and other grain exports. The estimated change in the value of wheat and other grains production, in 2003 prices, are illustrated in fi gure B.

Figure B shows that the type of incursion considered has relatively modest direct impact on the net value of production as falls in wheat are largely offset by rises in other grains. The impacts of these changes in the value of production of wheat and other grains on main economic aspects of the region are also estimated to be modest. The estimated changes include output, employment and gross regional product (value added) and are presented for each of the fi fteen years in 2003 prices in fi gure C. The values illustrated in fi gure C are cumulative, direct and indirect, effects on all industries in the region.

All three economic indicators continued to trend lower in the second part of the period as the impacts of increasing value of other grains production were not large enough to miti-

B Direct economic effects on value of output

21 5 7 8year

9 10 11 12 13 143 4 6 15

Other grains

Wheat

–5

–4

–3

–2

–1

1

$m

2

3

abare e Report 04.4

20

gate the impacts of falling values of wheat production. The widening gap between changes in production and employment in the later years of the planning period indicates that lower values of regional output resulted in even larger falls in regional employment. The esti-mated disparity in relative changes of employment to output implies that relatively labor intensive industries were more negatively affected by the indirect impact of the karnal bunt outbreak.

The overall effects in output and employment, observed in fi gure C, are split into direct and indirect effects in table 9. The initial effect refl ects the direct impact of the hypothetical

9 Regional economy effects of a contractor based incursion

Output Employment a

Initial Flow-on Initial Flow-on (direct) (indirect) Total (direct) (indirect) Total

Year $m $m $m no. no. no.

1 –0.9 –0.3 –1.2 –40 –25 –652 0.2 0.1 0.3 9 6 153 –0.1 0.0 –0.1 –6 –3 –94 –0.9 –0.3 –1.2 –43 –25 –685 –1.5 –0.6 –2.1 –71 –42 –1146 –0.8 –0.3 –1.1 –40 –23 –637 0.2 0.1 0.2 3 4 78 –1.4 –0.5 –1.9 –66 –39 –1069 –0.7 –0.3 –0.9 –37 –20 –5710 –1.0 –0.4 –1.4 –53 –30 –8311 –2.1 –0.8 –2.9 –106 –61 –16712 –1.0 –0.4 –1.4 –57 –30 –8813 –1.4 –0.6 –2.0 –78 –43 –12114 –1.5 –0.6 –2.0 –81 –44 –12515 –1.5 –0.6 –2.1 –84 –45 –130

Total –14.3 –5.7 –19.9 –751 –421 –1 172

a Employment effects are reported in person years.

C Economic effect of a contractor based incursion

21 5 7 8year

9 10 11 12 13 143 4 6 15

$mno.Gross regional product

Regional value of production

right axis

Regional employmentleft axis

–3.5

–3.0

–2.5

–2.0

–1.5

–1.0

–0.5

0

–180–160–140–120–100–80–60–40–20

abare e Report 04.4

21

karnal bunt incursion on the output and level of employment in the directly affected wheat and, the assumed substitute, other grain industries. The change in employment is inclusive of changes in owner and family labor. Over the fi fteen year planning horizon the direct effect of the hypothetical karnal bunt incursion on the wheat and other grain industries is estimated to be, in 2003 prices, a $14 million decline in output, and a decline in employ-ment of 750 person years (equivalent to around 50 full time jobs for the fi fteen years).

The indirect, or fl ow-on, effects of the incursion on the regional economy are calculated as the combined effect of all subsequent changes in the initially affected as well as in the other industries. They are caused by changes in purchases by the wheat and other grains industries as well as by industries that the wheat and other grains industries would normally purchase from, and from any changes in household demand as a result of changes in wages and salaries fl owing from changes in employment. Over the fi fteen year planning horizon, the indirect effect of the hypothetical incursion on all industries is estimated to be a $6 million (2003 prices) decline in output and a decline in employment of 420 person years.

The total industry and consumption effects, refl ecting the indirect effects along with the initial (direct) effects, capture the overall impact of a hypothetical karnal bunt incursion in the case study region. It is estimated that over the fi fteen year planning horizon, the decline in 2003 prices is $20 million. It is also estimated to result in about 1200 fewer person years (equivalent to around 78 full time jobs across the region).

While the results in table 8 show the cumulative effects across all industries in the region, it is also useful to analyse the relative effects of a hypothetical incursion on individual indus-tries. Table 10 shows how the indirect (fl ow-on) effects are distributed among the industries in the region (excludes positive fl ow-on effects on the other grains industry for ease of reporting). The results indicate that ‘trade’ is the most affected sector in terms of lost output and employment in the event of a karnal bunt incursion in the case study region. The trade sector accounts for 28 per cent of lost output and 39 per cent of lost employment of the region. Other industries that are likely to be signifi cantly affected by the fl ow-on effect include fi nance, business and property, manufacturing, wheat and transport sectors, in that order.

The industries most affected in terms of lost output were also the most affected in lost employment, generally in the same order. However, the magnitude of the effect on output differs from that on employment. The difference arises because of differences in labor intensity across the industries.

10 Distribution of adverse fl ow-on effects between industries in the

region

Output Employment

% %

Wheat 13.1 8.2Livestock 0.2 0.2Other agriculture 3.4 2.6Mining 0.4 0.3Manufacturing 15.3 11.9Public utility 10.2 4.3Construction 2.1 2.8Trade 27.7 39.4Transport and communication 9.1 11.6Finance, business and property 15.3 11.9Public and community services 1.5 3.4Personal and other services 1.8 3.5

Total 100.0 100.0

abare e Report 04.4

22

Contaminated fertiliser based incursion

In the scenario where contaminated fertiliser is inadvertently sold in the region, it is assumed that half of one per cent of all agricultural land becomes infested with karnal bunt teliospores in the fi rst year of the simulation. Compared with the contractor based incur-sion, the number of paddocks infested with teliospores increases sharply within the fi rst few years of the planning horizon (fi gure A). Within the fi rst fi ve years of the period, almost 900 000 hectares is infested with karnal bunt teliospores. Quarantine measures appear to temporarily contain the disease; however, it continues to follow a cyclical pattern and the area infested increases to around 2.2 million hectares by the end of the fi fteen year period.

The greater incidence of the teliospores across the case study region means that the disease spreads further over a shorter period of time and is correspondingly identifi ed much earlier in the simulation compared with the contractor based scenario. The effect of the quarantine restriction is dramatic, with the area planted to wheat falling from around 500 000 hectares to around 200 000 hectares by the fi fth year of the simulation (fi gure D). This corresponds with the decline observed in the area infested with teliospores, but as some of the farms are released from the quarantine restrictions and the area planted to wheat increases, the infestation continues to spread. A second phase of quarantine restrictions reduces the area planted to wheat. By the end of the fi fteen year planning horizon the area planted to wheat declines by more than 60 per cent from the area sown in the fi rst year.

The extent of the incursion is much larger in the fertiliser based scenario and the economic costs associated with the disease are correspondingly larger (table 11). The results indicate that a karnal bunt incursion of this type in the region is associated with a $503 million loss in the gross value of wheat production in net present value terms over the fi fteen year plan-ning horizon. In percentage terms this represents around a 30 per cent decline in the gross value of wheat production in the region. However, this is offset to some extent by a $414 million increase in the gross value of sorghum production relative to the reference case.

The returns to sorghum and other cropping activities may be overstated in this scenario because it is likely that additional costs will be imposed on farmers within the quarantine region marketing their other grain and transporting it to the silo.

D Changes to production and returns, fertiliser based incursion

21 5 7 8year

9 10 11 12 13 143 4 6 15

Fertiliser scenario

Contractor scenario

Reference case

100

200

300

400

500

’000 ha

abare e Report 04.4

23

11 Changes to production and returns, fertiliser based incursion

Area sown Gross receipts

Wheat Feed wheat Sorghum Wheat a Sorghum

Year ’000 ha ’000 ha ’000 ha $m $m

1 501 0 194 107 532 461 32 203 101 553 345 117 234 89 644 260 173 263 79 725 215 201 280 74 766 224 186 282 73 777 301 123 269 80 738 320 106 267 81 739 272 132 291 75 7910 226 154 316 68 8611 187 167 340 62 9312 184 155 354 59 9713 199 133 361 59 9814 189 127 377 56 10315 179 128 391 54 107

All years 1 118 1 207

a Sum of feed wheat receipts and wheat for grain export receipts.

Regional economy effectsThe main economic effects over a fi fteen year planning horizon of a contaminated fertiliser based incursion are estimated. As in the fi rst scenario, farmers who are unable to grow wheat due to quarantine restrictions are assumed to grow other grains instead. The changes in the value of wheat and other grains production in 2003 prices over the fi fteen year period are presented in fi gure E. It is evident from fi gure E that the impact of this scenario on the value of output is much higher than in the previous scenario. However, in this scenario the decline in the value of wheat production is increasingly compensated through time by increasing values of other grains production at a faster rate than in the previous scenario.

E Direct economic effects on value of output

21 5 7 8year

9 10 11 12 13 143 4 6 15

Other grains

Wheat

$m

–60

–40

–20

20

40

abare e Report 04.4

24

The impact of the combined changes in the gross value of production of wheat and other grains on the region’s output, employment and gross regional product (value added) are presented in fi gure F. They illustrate cumulative, direct and indirect, effects on all industries in the region. The much larger impact of this scenario compared with the fi rst scenario is refl ected in the subsequent effects on output, employment and gross regional product.

Sharp falls in the region’s output and gross regional product in the earlier years are reversed in the later years as the impacts of the increasing value of other grains offset the impacts of falling values of wheat production. However, the recovery of employment lagged behind that of output indicating the relatively labor intensive industries were more severely affected by the change.

The overall effects in output and employment are broken down into direct and indirect effects in table 12. The initial, or direct, effect in the wheat and other grains industries was estimated as an $89 million (2003 prices) decline in output over the fi fteen year planning horizon and a loss of 6000 person years (equivalent to around 400 full time jobs). The indirect, or fl ow-on, effects of the incursion in all other industries were estimated as a decline of $38 million (2003 prices) in output and a decline in employment of 3000 person years.

The total effects, capturing the combined direct and indirect effects over the fi fteen year planning horizon, were estimated as a $126 million (2003 prices) decline in output. It was also estimated to result in a loss in employment of 9000 person years (equivalent to around 600 fewer full time jobs across the region).

The fertiliser scenario has a much larger impact on the economy of the affected region because of its larger impact on the value of wheat production. In spite of the larger impact in the fertiliser scenario, lost output and gross regional product were almost restored to the pre-incursion situation and employment regained most of its losses. In the contractor scenario, all the main economic indicators continued to trend lower as the impacts of the increasing values of other grains

F Economic effect of a fertiliser based incursion

21 5 7 8year

9 10 11 12 13 143 4 6 15

$mno.Gross regional product right axis

Regional value of production

right axis

Regional employmentleft axis

–18

–16

–14

–12

–8

–10

–6

–4

–2

–1200

–1000

–800

–600

–400

–200

abare e Report 04.4

25

production were not large enough to mitigate the impacts of the falling values of wheat production.

The relative fl ow-on effects of a contami-nated fertiliser based incursion on individual industries across the case study region are presented in table 12. The results indicate that the trade, wheat, and fi nance, business and property sectors are likely to be the most affected, in that order, in terms of lost output and employment in the event of a karnal bunt incursion of this nature. These three sectors were also identifi ed as being the most likely to lose in the contractor scenario. Unlike in the contractor scenario where the manufacturing sector was among the most likely to lose, it does not appear to be so in the fertiliser scenario. A possible explanation is the signifi cant positive direct and indirect effect of the other grains industry might have mitigated the negative fl ow-on effect on the manufacturing sector in the fertiliser scenario.

12 Regional economy effects of a fertiliser based incursion

Output Employment a

Initial Flow-on Initial Flow-on (direct) (indirect) Total (direct) (indirect) Total

Year $m $m $m no. no. no.

1 –1.7 –0.6 –2.3 –76 –46 –1222 –3.5 –1.4 –4.9 –176 –102 –2783 –7.7 –3.0 –10.7 –409 –228 –6374 –10.5 –4.2 –14.7 –574 –315 –8895 –10.9 –4.4 –15.4 –618 –333 –9516 –11.1 –4.5 –15.5 –624 –337 –9617 –6.3 –2.6 –9.0 –390 –201 –5918 –7.4 –3.0 –10.4 –434 –229 –6639 –6.4 –2.7 –9.1 –424 –211 –63410 –6.1 –2.6 –8.7 –439 –209 –64811 –6.5 –2.8 –9.3 –489 –227 –71612 –4.7 –2.2 –6.9 –426 –183 –60813 –3.2 –1.6 –4.8 –362 –141 –50314 –2.0 –1.2 –3.2 –329 –114 –44415 –0.7 –0.7 –1.4 –283 –80 –362

Total –88.7 –37.6 –126.3 –6 053 –2 954 –9 007

a Employment effects are reported in person years.

13 Distribution of adverse fl ow-on effects between industries in the

region

Output Employment

Wheat 26.2 15.8Livestock 0.1 0.1Other agriculture 2.6 2.0Mining 0.4 0.2Manufacturing 2.9 3.5Public utility 11.3 4.8Construction 2.3 3.1Trade 28.2 40.6Transport and communication 8.0 10.6Finance, business and property 14.5 11.6Public and community services 1.9 4.4Personal and other services 1.6 3.2

Total 100 100

abare e Report 04.4

26

sensitivity analysis

There is considerable uncertainty surrounding the value of a number of parameters included in this model, particularly those relating to the spread and identifi cation of karnal bunt teliospores. By varying the values of key parameters in a systematic way it is possible to gain some insights into the relative sensitivity of specifi c parameter values to the overall model results. In the sensitivity analysis presented below, the probability that bunted grains are identifi ed at the silo or on-farm during harvest time is varied.

Identifi cation of bunted grain at siloThe estimated economic effects of a karnal bunt incursion in the case study region that occurs as a result of contaminated fertiliser being sold into the region were found to be quite high. This was because despite some temporary success with containment measures part way through the simulation the disease was not contained and continued to spread, imposing greater economic costs in terms of forgone wheat production. The scenario assumed that in only 50 per cent of cases, bunted grains were correctly identifi ed as being infected with karnal bunt teliospores, triggering the quarantine response. Two alternative scenarios were therefore considered: one where the probability of identifying the disease at the silo was increased to 70 per cent, and one where the probability of identifying the disease at the silo was increased to 90 per cent. No estimate of the costs associated with the increase in the likelihood of detection was included and all other parameters in the model were held constant.

G Area infested with teliospores, sensitivity analysis

21 5 7 8year

9 10 11 12 13 143 4 6 15

’000 ha

90 per cent

70 per cent

50 per cent

500

1000

1500

2000

5

abare e Report 04.4

27

The results indicate that while an increase in the likelihood of detection is capable of reducing the area infested within the fi rst half of the simulation, it is insuffi cient to contain the incursion over the fi fteen year period considered (fi gure G). In addition, with more farms affected by the quarantine region, particularly those in buffer regions only able to produce wheat for feed grain purposes, the value of wheat production in the region was observed to decline as the likelihood of detection increased.

Not withstanding the decline in the gross value of wheat production as a result of the increase in the likelihood of detection, the results indicate that increasing the likelihood of detection at the silo reduces the time taken to identify the disease in the case study region and to put in place quarantine restrictions. These results provide an improved under-standing of the relationship between the likelihood of detection at the silo and the time elapsed between the incursion and its identifi cation. This is likely to assist in the design of monitoring policies aimed at reducing the potential hazard associated with karnal bunt.

Identifi cation of bunted grain on-farmIn addition to assuming that bunted grains were correctly identifi ed at the silo only 50 per cent of the time, the results presented in the previous chapter also assumed that farmers did not recognise the presence of the disease on their property. A simulation was run where the likelihood of detection at the silo was 90 per cent and the likelihood of a farmer detecting the disease at harvest time was increased to 30 per cent. The results were similar to those presented in the sensitivity analysis in the previous section.

The increase in the likelihood of detection reduced the extent of the infestation. However, the gross value of wheat production was observed to decline because of the increase in the number of farms within the buffer quarantine regions unable to grow wheat for export purposes.

abare e Report 04.4

28

conclusions

The exotic incursion management model developed as part of this project was used to demonstrate the regional economy effects of a potential karnal bunt incursion in south eastern Queensland. By developing an agent based model that integrated the biophysical processes of disease spread with the agricultural production system and regional economy it was possible to generate estimates of two different types of incursions: one beginning from a single point source and one with multiple sources randomly located across the region.

The results indicate that the effects of an incursion with a single point source are much smaller in magnitude than a diffuse incursion as could occur if a load of contaminated fertiliser was inadvertently sold into the region. However, the results also indicated that the slow spread of the disease and the relatively low incidence of bunted grains made it diffi cult to identify and contain, even fi fteen years after the disease had been introduced to the region.

In the scenario where contaminated fertiliser was the source of the incursion, the disease was more widespread and identifi ed earlier. Because of this, containment measures were able to limit the spread of the disease within the fi rst few years, but ultimately failed to fully contain the incursion. The economic effect of such an incursion was estimated as a decline in regional output of $126 million, and a corresponding decline in the gross regional product of the region of almost $60 million. It was also estimated that the incursion resulted in a loss of around 600 full time equivalent jobs in the region over the fi fteen year period.

By increasing the likelihood of the disease being identifi ed at the silo or on-farm during harvest time, the sensitivity of the results to the value of detection parameters was analysed. When the likelihood of detection at the silo was increased there were only minor changes in the overall effects of the disease. While the area infested with teliospores was reduced by the earlier detection, the disease was still unable to be contained and the economic cost associated with the incursion remained high. Similarly, increasing the likelihood of detec-tion on-farm failed to contain the incursion.

With further refi nements, the detail and accuracy with which the disease and case study region are modeled can be improved. This could include the expansion of transmission vectors, or the range of agricultural activities modeled explicitly. Additional sensitivity analysis would prove useful in quantifying the relationships that exist between the like-lihood of detection, the effectiveness of different types of quarantine responses and the economic effects of an incursion. In addition to the sensitivity analysis conducted on the

6

abare e Report 04.4

29

likelihood of detection in this report, other strategic responses such as the incentive for farmers to self report an incursion could be explored. Tactical response issues, including the placement and duration of quarantine boundaries and the resources allocated to containing an incursion, could also be explored with sensitivity analysis. This type of investigation is likely to assist in the further development of policies to both monitor and reduce the risk of a karnal bunt incursion in Australia.

The modeling framework used to analyse the effects of a karnal bunt incursion in this report is capable of being adapted to analyse the incursion of other plant diseases and pests. It is anticipated that following peer review additional development of the framework will occur so that the model can be extended to analyse the spread of both weeds and insect pests in other case study regions.

abare e Report 04.4

30

appendix

the input–output model

I–O tablesThe I–O tables were constructed primarily from adjusted Queensland I–O data, secondary data on the region, ABARE farm survey data and national I–O data.

The core information of the I–O data is contained in the intermediate demand. The inter-mediate demand shows the interrelationship of the industries of the regional economy as a source of inputs or as markets for products. It is this part of the I–O table that is used to calculate I–O multipliers.

The intermediate demand is represented as aij X

j, and the I–O coeffi cient as a

ij=X

ij/Y

j

where aij is intermediate input i used in the production of a unit of output in industry j; X

j

is the total quantity of j output used as intermediate input and Yj is the total production of

industry j.

The intermediate inputs refer to only to materials sourced from within the regional economy. In matrix form a

ij becomes A.

H Graphic representation of the I–O table

Intermediate uses

Prim

ary

uses

Intermediate uses (i x j)*

Household uses(i)

Other final uses(OFD) (i)

Wages and salaries( j)

Gross operating surplus and mixed income (GOS)

( j)

Imports( j)

Final uses

*Figures in brackets denote the dimension of the matrix. i = commodities (13), j = industries (13)

A

abare e Report 04.4

31

Basic description of the regional I–O table

An I–O table records the production and sale of goods and services in an economy for a particular year. The rows show the pattern of sale of the output of each industry to all indus-tries and fi nal users. The columns show the pattern of purchases by each industry from all industries in the economy and from imports. The columns also show industry purchases of labor and other primary inputs.

The I–O matrix contains:

■ Intermediate uses (by commodity and industry).

■ Final uses (by commodity), consisting of:

– household consumption

– export demand (by commodity) includes all sales outside the region

– other fi nal demand (government demand, investment demand and changes in stocks).

■ Primary uses (by industry)

– labor

– gross operating surplus and mixed income (GOS) consisting of capital, land, profi ts, product taxes less subsidies).

■ Imports (by industry), including interregional and international imports. Imports are directly allocated to the industries using them. Therefore, the import data under a certain industry shows the sum of all imports used by that industry. With direct alloca-tion of imports the intermediate inputs show only those inputs that are sourced from within the regional economy. This method, therefore, enables the changes in demand for domestic products in response to changes in fi nal demand to be gauged. However, in indirect allocation where imports are allocated to industries that would have produced them (competing imports) domestic and imported products are not separately identi-fi ed.

The core information of the I–O data is contained in the intermediate demand. The inter-mediate demand shows the interrelationship of the industries of the regional economy as a source of inputs or as markets for products. It is this part of the I–O table that is used to calculate I–O multipliers.

The intermediate demand is represented as: aij X

j

where aij is intermediate input i used in the production of industry j; X

j is the total quantity

of output j used as intermediate input.

The input–output coeffi cient, aij, is the amount of intermediate input required to produce a

unit of output j:a

ij = X

ij/Y

j

The intermediate inputs, as stated above, refer only to materials sourced from within the regional economy. In matrix form a

ij becomes A.

abare e Report 04.4

32

Derivation of I–O multipliers

A and its derivatives and extensions are used to calculate a range of multipliers (see ABS 1994 for details).

The vectors and the corresponding multipliers calculated in this paper are:

Vector Multiplier

Output (Y) Output

Employment (E) Employment

Wages and salaries (H) Household income

Value added (V) Total income

Imports (M) Import

The calculation of multipliers with respect to employment multiplier is provided, as an example, below. All the other multipliers follow the same principle by substituting E (e) with another vector of desired variable.

Direct (initial impact): e = E * 1/Y, the immediate change that occurs in the directly affected industries.

First round impact = e * A, where A is the I–O coeffi cient matrix. [e * A] represents the changes that occur in industries that have direct links with the originally affected industries.

Industrial effect = e * (I–A)’, where (I–A)’ is the inverse of (I–A). [e * (I–A)’] refl ects the changes in industries that are indirectly linked to the directly affected industries.

The production induced effect (the sum of the fi rst round effect and the industrial effect) represents the fl ow-on effects resulting from changes in industrial demand.

The simple multiplier (the sum of the initial effect and the production induced effect) refers to the direct and indirect changes in employment across the region resulting from changes in industrial demand.

A more complete picture of the effects of changes in wheat exports on the regional economy is gained by incorporating the effects of household demand.

The above simple multiplier analysis assumes industries are the only endogenous factors of the I–O system and all fi nal demands are exogenous. Given that households earn their income from within the production process, their demand, through their income, is likely to be affected by what is occurring in the production system. Therefore, treating household demand as endogenous to the system refl ects actual economic interactions more accurately and makes the I–O analysis more complete. This is achieved by ‘closing’ the system with respect to income. The system is ‘closed’ by including the household sector as another sector in the matrix of industries — that is, extending aij, where ij = 13 to cij, where ij = 14. The extended matrix, C, is then used to calculate consumption induced impacts.

abare e Report 04.4

33

The outcome of the extension is that, an increase in fi nal demand not only induces addi-tional output (as in the simple multiplier model) but the associated income generated also induces further increases in household demand which in turn induces further increases in output. These industry and household demand induced impacts are captured by calculating the total multiplier.

Consumption induced effect e * (I–C)’, where (I–C)’ is the inverse of (I–C).

Total fl ow-on effect (4 + a), economywide fl ow-on effects resulting from changes in both industrial and household demand.

Total multiplier (1 + b), direct and indirect effects resulting from changes in industrial and household demand.

The I–O analysis operates under the following assumptions and therefore is subject to the limitations that the assumptions impose.

Constant technology, the relationships between inputs and output remain constant. This means average and marginal relationships are the same, implying no substitution between inputs.

The interdependence of industries is established through purchases and sales of inter-mediate inputs.

Industries do not compete for inputs, implying unlimited resource availability at the given price. Therefore no adjustment mechanism is associated with a given economic change.

There are no macroeconomic constraints such as government budget, price levels, exchange rates etc.

abare e Report 04.4

34

appendixB

B1 Output multipliers with respect to a $1 change in export demand for

wheat

Simple Total multipliers multipliers

Wheat 1.023 1.024Other grains 0.000 0.004Livestock 0.001 0.001Other agriculture 0.005 0.015Mining 0.001 0.002Manufacturing 0.072 0.083Public utility 0.029 0.037Construction 0.005 0.007Trade 0.095 0.104Transport and communication 0.032 0.037Finance, business and property 0.046 0.060Public and community services 0.003 0.005Personal and other services 0.006 0.007

Total effects 1.318 1.385

B2 Employment multipliers with respect to a $1 change in export

demand for wheat

Simple Total multipliers multipliers

Wheat 0.005 0.005Other grains 0.000 0.000Livestock 0.000 0.000Other agriculture 0.000 0.000Mining 0.000 0.000Manufacturing 0.000 0.000Public utility 0.000 0.000Construction 0.000 0.000Trade 0.001 0.001Transport and communication 0.000 0.000Finance, business and property 0.000 0.000Public and community services 0.000 0.000Personal and other services 0.000 0.000

Total effects 0.007 0.007

B3 Household income multipliers with respect to a $1 change in export

demand for wheat

Simple Total multipliers multipliers

Wheat 0.185 0.185Other grains 0.000 0.001Livestock 0.000 0.000Other agriculture 0.001 0.003Mining 0.000 0.000Manufacturing 0.015 0.018Public utility 0.003 0.004Construction 0.002 0.003Trade 0.038 0.042Transport and communication 0.012 0.013Finance, business and property 0.010 0.013Public and community services 0.002 0.003Personal and other services 0.003 0.004

Total effects 0.273 0.290

B4 Total income multipliers with respect to a $1 change in export

demand for wheat

Simple Total multipliers multipliers

Wheat 0.388 0.389Other grains 0.000 0.001Livestock 0.000 0.000Other agriculture 0.002 0.006Mining 0.001 0.001Manufacturing 0.022 0.025Public utility 0.015 0.020Construction 0.003 0.004Trade 0.060 0.066Transport and communication 0.019 0.022Finance, business and property 0.032 0.041Public and community services 0.002 0.003Personal and other services 0.003 0.004

Total effects 0.548 0.582

regional economywide effect of a $1 a tonne change in export demand for wheat

abare e Report 04.4

35

references

ABARE 2003, Australian Commodity Statistics 2002, Canberra.

ABS 1994, Australian National Accounts, Input–Output Multipliers 1989-90, cat. no. 5237.0, Canberra.

American Phytopathological Society 1996, APSnet: intro to karnal bunt, Paper presented to the Karnal Bunt Symposium, 24 June – 18 August (www.apsnet.org/online/karnal/kbspaper/artintro.htm).

Bonde, M.R., Peterson, G.L., Schaad, N.W. and Smilanick, J.L. 1997, ‘Karnal bunt of wheat’, Plant Disease, vol. 81, no. 12, pp. 1370–7.

Brennan, J.P. and Warham, E.J. 1990, Economic losses from Karnal bunt of wheat in Mexico, CIMMYT Economics Working Paper 90/02, International Maize and Wheat Improvement Centre, Mexico.

CIE (Centre for International Economics) 2002, A Tale of Two Models: A More Detailed Look at the Karnal Bunt Case Study, Report prepared for Plant Health Australia, Canberra.

Murray, G.M. 1998, Pest Risk Analysis on Karnal Bunt of Wheat, Risk Analysis Report, NSW Agriculture, Sydney.

Murray, G.M. and Brennan, J.P. 1998, ‘The risk to Australia from Tilletia indica, the cause of karnal bunt of wheat’, Australian Plant Pathology, vol. 27, no. 4, pp. 212–24.

Nagarajan, S., Aujla, S.S., Nanda, G.S., Sharma, I., Goel, L.B., Kumar, J. and Singh, D.V. 1997, ‘Karnal bunt (Tilletia indica) of wheat – a review’, Review of Plant Pathology, vol. 76, no. 12, pp. 1207–14.

OESR (Offi ce of Economic and Statistical Research) 2003, ‘Local Government Area Profi le’, Report generated by the Queensland Government Offi ce of Economic and Statistical Research, Brisbane, September (www.oesr.qld.gov.au).

Stansbury, C.D., McKirdy, S.J., Diggle, A.J. and Riley, I.T. 2002, ‘Modeling the risk of entry, establishment, spread, containment and economic impact of Tilletia indica, the cause of karnal bunt of wheat, using an Australian context’, Phytopathology, vol. 92, no. 3, pp. 321–31.

abare e Report 04.4

36

Asia Pacifi c Economic Cooperation Secretariat

AusAid

Australian Centre for International Agricultural Research

Australian Greenhouse Offi ce

Australian Quarantine and Inspection Service

Australian Wool Innovation Limited

CSIRO (Commonwealth Scientifi c and Industrial Research Organisation)

Dairy Australia

Department of Agriculture, Fisheries and Forestry

Department of Foreign Affairs and Trade

Department of Health and Ageing

Department of Industry, Tourism and Resources

Department of Natural Resources and Mines, Queensland

Department of Primary Industries, Queensland

Fisheries Research and Development Corporation

Fisheries Resources Research Fund

Forest and Wood Products Research and Development Corporation

Grains Research and Development Corporation

Grape and Wine Research and Development Corporation

Horticulture Australia

Institute of National Affairs, PNG

Land and Water Australia

Meat and Livestock Australia

Ministerial Council on Energy

National Landcare Program

National Oceans Offi ce

Natural Heritage Trust

New Zealand Prime Minister and Cabinet

Offi ce of Resource Development, Northern Territory

Organisation for Economic Cooperation and Development

Plant Health Australia

Primary Industries, Victoria

Rio Tinto Energy Marketing Services

Rural Industries Research and Development Corporation

Snowy Mountains Engineering Corporation

Woolmark Company Pty Ltd

Research funding. ABARE relies on fi nancial support from external organ-

isations to complete its research program. As at the date of this publication, the following

organisations had provided fi nancial support for ABARE’s 2003-04 research program. We

gratefully acknowledge this assistance.