Using input-output analysis to estimate the impact of a coal industry expansion on regional and local economies

Impact Assessment and Project Appraisal December 2011 1461-5517/11/040277-12 US$12.00 IAIA 2011 277

Impact Assessment and Project Appraisal, 29(4), December 2011, pages 277–288 DOI: 10.3152/146155111X12959673795840; http://www.ingentaconnect.com/content/beech/iapa

Using input-output analysis to estimate the impact of a coal industry expansion on

regional and local economies

Galina Ivanova and John Rolfe

Since 2004, there has been a major expansion of coal mining in Queensland, Australia. While there have been significant demographic, social and economic impacts at local, regional and state levels, the size and type of impacts are difficult to quantify. This analysis is of the economic impacts of the projected coal mining expansion in the key mining area of Central Highlands regional economy. It is part of the Bowen Basin region. Impacts were also modelled for two smaller communities: the former Duaringa and Bauhinia Shires. The results provide a guide to the size of the distribution of impacts from the coal mining expansion and the potential impacts at the local level. The model had to be adjusted to take account of how a non-resident workforce would transfer impacts from the local or sub-regional area where the project was located to other regions where the population is based. The results demonstrate how ‘fly-in/fly-out’ or ‘drive-in/drive-out’ employment patterns can reduce the level of economic impacts to local economies.

Keywords: impact assessment, input-output, coal mining, Queensland

HE MINING INDUSTRY is a key part of the economy in the State of Queensland, Austral-ia, accounting for 10% of the gross state

product (DME, 2007). It employs about 31 thousand persons (or 1.7% of total employed persons in Queensland) and pays the highest wage levels (three to four times higher than retail and accommodation services) among other industries in Queensland.

At a broad level, the positive impacts of the mining

industry on the state’s economy are substantial because of the mining industry’s expenditure on wag-es, infrastructure and operating costs, as well as main-taining regional employment and population growth. Direct contributions to the state economic activity

flow through to indirect (flow-on) economic and so-cial impacts. This particularly occurs through job cre-ation in industries other than mining to support the

growth in mining industry, with corresponding flows

of income and wealth accumulation in mining and

non-mining communities (Rolfe et al, 2005, 2007). At the regional and local level though, the impacts

are not so easily defined, particularly for a single mine. This is for two key reasons. First, it is difficult to disaggregate the flow-on effects of a single mine separate to the industry that it is in. Second, there is some diversity in the management, use of contrac-tors and supply of labour to mines, making it harder to identify the impacts of a particular operation on economic and social factors (Rolfe et al, 2005). This is particularly difficult when there are non-resident

T

Dr Galina Ivanova is Senior Lecturer in Economics at the Facul-ty of Arts, Business, Informatics and Education, CQUniversity,Rockhampton QLD 4702, Australia; Email: [email protected]; Tel: +61 7 4930 9386 (office), +61 7 4939 1663 (home).Dr John Rolfe is Professor in Regional Economic Developmentat the Centre for Environmental Management, CQUniversity,Rockhampton, QLD 4702, Australia; Email: [email protected].

The research reported in this paper has been supported by theAustralian Coal Association Research Program (ACARP) andthe Queensland Government. We would like to thank ACARPfor their financial support of this research. We also received ad-ditional helpful comments from two anonymous referees. Theviews expressed and the conclusions drawn are the responsibilityof the authors.

Using IO analysis to estimate the impact of a coal industry expansion on regional and local economies

Impact Assessment and Project Appraisal December 2011 278

workforces involved, so that large proportions of primary impacts leak to other regions.

This paper presents the results of an economic impact assessment that was conducted for the Cen-tral Highlands region1 in Central Queensland, which overlays the southern part of the Bowen Basin coal region. The Bowen Basin region produces approxi-mately 85% of the coal in Queensland, and the Cen-tral Highlands area is experiencing major levels of growth. There were two case studies from the Cen-tral Highlands region: the (former) Duaringa Shire including the mining township of Blackwater and the (former) Bauhinia Shire (agricultural shire) in-cluding Springsure and Rolleston townships. Being a part of the Central Highlands region, these shires are experiencing some population increases and substan-tial requirements for accommodation and other resources (PIFU, 2006).

While it is clear that the mining boom is creating major economic impacts in the Bowen Basin region and Central Highlands in particular, it is not so easy to estimate the level of those changes on smaller communities. In this paper, the key focus is to iden-tify the impacts of the mining industry on regional and sub-regional levels and to compare impacts between mining and agricultural sub-regions. A sec-ondary goal is to identify the strengths and challeng-es of using the input-output (IO) method at a local level. For that purpose several input-output tables were constructed from national input-output tables using the Generation of Regional Input-Output Tables (GRIT) technique.2

The rest of the paper is structured as follows. Background information is outlined in the next sec-tion, then an overview of input-output models is provided, explaining what can be measured using

the input-output method and how the tables can be constructed. This is followed by a discussion of some limitations and precautions that need to be taken

when applying input-output techniques to the small mining towns, then a brief description of the former Duaringa Shire (including the township of Black-water) and the former Bauhinia Shire is given. The last two sections report the sensitivity analyses for the two case studies and then sum up the final conclusions.

Background

The changes in the mining industry in Queensland

Since the commodities boom from 2002, the re-sources sector in Australia has experienced strong growth leading to increases in employment and re-gional incomes. The boom in the coal mining indus-try has generated an increase in direct employment in Queensland in the coal sector from about eight thousand employees in 1999/2000 to more than 18.5 thousand employees in 2005/06 (Figure 1), while the traditional sectors of agriculture, accommodation and food services (closely aligned with the tourism industry) have had relatively low rates of growth from 1989.

The resources industry typically has much higher salary levels than many other industry sectors. For example, ABS (2010) data reveals that the mining industry in Australia paid 4.43 times more per em-ployee in average weekly wages and salaries than the accommodation and food services industry, 3.4 times more per employee than the retail trade indus-try, and 1.9 times more per employee than the manu-

Figure 1. Queensland coal production in ‘000 tonnes and the number of employees Data source: NRM (2006)



facturing industry. Higher levels of employment and income provide a base for larger inputs into regional and metropolitan economies.

ABARE (2006) predicted an increase in Australi-an coal exports by 50% between 2011/12 and 2029/30. The Queensland Government has predicted that there will be a growth rate of 7% per annum in the coal industry from 2005 to 2010 (McGrady, 2004). Growth and associated impacts on the re-gional economy are coming from four primary driv-ers: the expansion of existing mines in the Bowen Basin, the development of new mining projects, ex-ploration, and the construction of new infrastructure such as rail, roads, ports and utilities. Increases in production since 2000/01 (Figure 1) have largely come from expansion of existing mines. There has been little change in the number of operating mines in the Bowen Basin from 2000/01 (34 mines) to 2004/05 (37 mines), but there have been substantial increases in mining production (NRM, 2006), as well as major increases in employee numbers (Figure 1).

The increased activity in mining generates im-pacts on local, regional and state economies in a number of ways, including:

The expenditure of mining companies on salaries for mining, development and exploration activities.

The expenditure of mining companies on contrac-tors and suppliers associated with mining, development and exploration activities.

The expenditure of government and industry on infrastructure development.

The flow-on effects of business expenditure back into other sectors of the economy.

The flow-on effects of consumption expenditure back into other sectors of the economy.

Increased dividends to investors in mining firms which are then used for the purchases of goods and services.

Increased royalty payments and tax revenues to government.

There have been changes in employment patterns within the coal industry over the past few years which tend to dilute the economic impact of coal mining at local and regional levels (Zheng et al, 2007). First, there is increased usage of variations on fly-in/fly-out operations, where mining companies no longer assume primary responsibility for housing the workforce. One effect of the changes is that em-ployees now have more choice about where they are located. Many employees and their families now live in the larger centres or coastal cities that are located away from the mining towns and commute back to the workplace for ‘block’ shift work periods. Mining employees stay in company accommodation when they are completing a ‘block’ shift and then drive/fly to the place of their family residence. Another effect is that there has been some turnover of people across mining towns as many mining companies have shed

staff. However, most of the mining towns have maintained population with increased employment by contractors and service industries. As a result there is potentially less spending in the local area than in ‘average’ non-mining towns (Rolfe et al, 2005).

The economic impact assessment in Queensland and input-output modelling

In Queensland, and other states of Australia, major new projects such as mines have to go through a formal approval process. This involves preparation of an Environmental Impact Statement (EIS), with net economic impacts considered as a component. While economic analysis is only one aspect of an in-tegrated EIS, the economic data is attributed signifi-cant weighting in determining the worthiness of a project and its overall direct and indirect ‘benefit’ to the local community as well as state (in particular) and national economic well-being. Some of the measures of economic analysis include employment levels (jobs), value added (or gross regional prod-uct), aggregate wages and salaries, wealth (including property values) and business output (sales volume and spending).

There are a range of tools that can be used for the economic impact analysis such as the use of simple spending multipliers, input-output modelling, or general equilibrium modelling (Jensen and West 2002; Rolfe et al, 2005). To assess the impact of the changes on regional economy crude estimates of ini-tial changes in production, investment or employ-ment in the local or regional economy can be used. The subsequent purchases by firms receiving the ini-tial economic stimulus (direct effect) generate fur-ther spending (indirect effect) by industries in the region (through a series of ‘ripple’ effects). The es-timates of the extent to which the initial changes are multiplied through the economy by secondary ex-penditure (indirect effect) provide a more compre-hensive understanding of the economic impact of a development activity in the region.

Ripple effects vary according to the size of the chosen study area and the level of inputs (including labour) purchased from the regional economy, as well as the level of outputs supplied into the regional econ-omy (compared to being exported out of the region).

Another effect that can be estimated is the con-sumption-induced effect (e.g. the effect of increased salaries to households leading to increased consump-tion). Type I output, income, value added, employ-ment multipliers (closed model) measure the changes from the proposed project without including consumption effects. Type II multipliers (open model) reflect the effect of the induced demand.

Estimates of these effects can be done with so-phisticated economic modelling tools such as input-output (IO) analysis (Jensen and West, 2002) . The scope of IO analysis can be at local, regional, state or national levels. A particular strength of IO



analysis is that it can be used to assess the impact of one influence or several influences on the economy.

Robison (2009) suggested that IO modelling should be an essential part of the region’s toolkit to allow highlighting the areas that are most important to the region in terms of economic development in both descriptive and predictive forms. The descrip-tive use of the IO allows identification of the struc-ture of the economy, and relative importance of industries in terms of output, wages and salaries, value added and employment. The following data and coefficients can also be obtained from IO mod-els: gross regional product, imports and exports, and household consumption. This can provide a starting point for a community discussion of the potential re-gional development options. The predictive use of the IO model allows estimation of the amounts of inputs that will be purchased from the local econo-my and the size and structure of the flow-on effects in the local economy of the proposed activity.

Input-output modelling has been used in econom-ic impact assessments for major projects, helping to identify the direct, indirect and induced impacts that such projects have on regional economies (e.g. Lenzen et al, 2003; Batey et al, 1993). It has also been used for regional economic planning in national and international case studies to identify the impact of mining on an economy (e.g. Cristobal and

Biezma, 2006; IRIS, 2005). ACIL (2002) reported data from Mangan (2001)

to argue that there are significant flow-on effects in expenditure and job creation from the black coal and petroleum sectors in Queensland. Rolfe et al (2004) report the results of an input-output study to predict the economic impacts of a single coal mine on a local and regional area, while input-output studies are often used as part of the economic impact as-sessment performed in the approval process for new mines. However, the results of different input-output models focused on the Queensland coal industry are not always fully consistent (Rolfe et al, 2004). The focus of this paper is to discuss the application of input-output analysis, highlighting the strengths (e.g. ability to identify how different sectors in an econ-omy interact) and challenges (e.g. the data require-ments to construct accurate models) of using such analysis in impact assessment of a mining boom on small regions and local areas.

An overview of the input-output approach

An input-output (IO) model (developed by Leontief, 1956) provides a snapshot of relationships within an economy at a given point in time. IO modelling pro-vides a mechanistic approach to estimate how eco-nomic impacts can ‘ripple’ through an economy. It is typically done by building a model of a regional economy where the transactions between each in-dustry sector, the household sectors, and the econo-my outside the region are summarised in a matrix.

Table 1 provides an illustrative example and shows a transactional table for a three sector econo-my. Each row indicates the flows from one sector to another. From Table 1, sector 1 consumes $25,000 of output internally, sells $20,000 of its output to sector 2 and $15,000 of its output to sector 3. The columns show the purchases of each sector from the other sectors. Sector 3 purchases $15,000 of goods from sector 1, $10,000 of goods from sector 2 and

Table 1. Hypothetical transactional table ($, 000)

Purchasing sectors

Selling sectors

Intermediate sectors processing matrix Final demand Total output

1 2 3 Total Households Others Export

1 25 20 15 60 30 5 5 100

2 14 6 10 30 5 5 10 50

3 20 12 43 75 5 5 15 100

Total 59 38 68 165 40 15 30 250

Households 20 4 5 29 2 3 2 36

Value added 15 4 15 34 1 1 1 37

Imports 6 4 12 22 1 - 1 24

Total inputs 100 50 100 250 44 19 34 347

Source: Mandeville and Jensen (1978) and Ung (1981)

The scope of IO analysis can be at local, regional, state or national levels. A particular strength of IO analysis is that it can be used to assess the impact of one influence or several influences on the economy



$43,000 from firms in sector 3. For labour input, sector 1 pays $20,000 in wages and salaries to the household sector and a further $6,000 for imports of other input factors.

The types of data collection to construct IO mod-els include survey-based methods, non-survey-based methods and hybrid methods. Survey-based methods involve surveys of households and businesses to identify expenditure levels and the proportions of transactions made between different sectors. Where tables are needed at a local or sub-regional level, then the data has to be collected at this level. This method obtains the most accurate3 data representing the local economy but is also the most expensive and time-consuming option (Shaffer et al, 2004). Also, there is a potential reluctance of business owners, especially in small areas, to provide collectors with confidential business information.

Non-survey-based methods involve the scaling-down of an IO model of a higher-level (national/ state) economy, to an IO model representing the re-gional economy. The assumption for such a generali-sation would be that data summarising economic

relationships at the larger level is directly applicable to

the smaller region. This approach would not consider

differences among communities and regional areas;

therefore the risk of inaccurate predictions will be in-creased. Jensen and West (2002) argued a simple re-ductionist approach can miss the essential differences

among communities and regional areas, increasing the

risk of inaccurate prediction and non-adequate plan-ning.4 However, the scaling-down approach can pro-duce fast and inexpensive results, and may be

appropriate in some circumstances (such as where the

collection of any primary data is not possible). Hybrid methods combine the above-mentioned

methods. First, a source IO model is obtained (e.g. a national or state model) and then it is scaled down (e.g. using regional employment data) to suit the re-gion or area of interest. Second, surveys are used to refine the relationships between sectors in that geo-graphic region, so that the model is fine-tuned for the

regional area (Jensen et al, 1977; West, 1981; West et al, 1984). Lahr (1992) has identified several issues

that have to be addressed if the hybrid method is used (such as the level of aggregation, the accuracy of the regionalisation technique, data to be collected from each sector/industry, and the identification of the sectors).

In Australia, the most widely used method of con-structing IO models is the GRIT technique (hybrid approach) developed by Jensen and West (Jensen et al, 1977; West, 1981; West, et al, 1984). The GRIT method is also widely used in the USA and Europe. The GRIT technique allows the researcher to deter-mine the appropriate level of interference in the model construction by inserting available primary or superior data. GRIT relies on a series of mechanical steps to produce regional models from national eco-nomic data. The parent table is adjusted according to the regional employment and/or output data using

various types of location quotients. As well, analysts can balance the table, and insert and delete sectors.

Some limitations in applying input-output techniques to the small mining towns

In Australia, the national IO tables are prepared by the

Australian Bureau of Statistics (ABS). In Queensland, the Office of the Government Statistician derives the

IO tables for state and statistical divisions based on

the national tables. However, it is not appropriate to

use national or state IO tables to model impacts for

sub-regional areas such as shires or population cen-tres. There are several reasons why the usage of those

tables to predict impacts at a single mine or township

level might create inaccurate results. First, the sub-region of interest might have a differ-

ently structured economy compared to the state econ-omy. The adoption of the IO tables from a state or a

statistical division to represent the region of interest assumes that the relationship between industry sectors

in the region under investigation is the same as those

for the region from which the IO tables have been

drawn as a proxy. Leistritz et al (1990) tested the ac-curacy of the IO model on a range of small and large, slow and rapidly growing areas in the USA. They

compared the model’s estimates of income for 1958–1986 to the estimates reported by the United States

Department of Commerce, Bureau of Economic

Analysis, and Census. They found that at the state and

regional levels with relatively stable populations the

IO model provided moderately accurate predictions

of income. However, the projections were less

accurate for the smallest and rapidly growing areas. Second, published IO tables are usually based on

economic data that is several years out of date. The non-frequent nature of compiling (producing) IO tables means that continuous time series are impos-sible to construct. In effect, IO tables provide a snapshot of the economy and its interconnections at one time, and there is an implicit assumption that these relationships remain constant for future analy-sis. There is, however, research estimating IO coef-ficients on a regular basis5 (e.g. Coon and Leistritz, 1985, 2006) allowing tests for the stability of multi-pliers. The accuracy of impact projection was also tested by Coon and Leistritz (1987). They found that on a project-specific basis the assumptions regarding project employment (especially construction em-ployment) and population trends were a larger source of error than variations in IO coefficients.

Third, the level of aggregation of the sectors might not be relevant for a particular study. For ex-ample, the Australian national tables have ‘Mining’ as

an aggregate of several mining activities, but at the regional level there may be only an ‘Underground Coal Mining’ industry. Variations in the economic relationships between the different mining sectors and the rest of the economy reduce the accuracy of the model extrapolation.



When considering whether to use the IO tech-nique to model the economic impact of mining in a region, the following points need to be considered. First, the extent of initial stimulus and direct impacts in a local area should be estimated to determine if impacts are worth modelling. In some cases coal mines may have very slight economic impacts on a local community but have a larger regional or state impact. Second, the structure and diversity of the local economy is important as this provides some indication of the level of economic interaction be-tween sectors. Third, the level of aggregation should reflect the important industries in the area. The accu-racy and cost of constructing IO tables is often directly proportional to the effort, time and money spend on data collection. When high quality data is not available or is out-dated, then the accuracy of results may be diminished.

The analysis of economic impacts using IO meth-ods has other limitations such as not including ‘non-economic’ transactions, for example environmental losses associated with exploration of natural re-sources by coal mines (Jensen and West, 2002; OECD, 1992). There are also a number of technical assumptions made, some of which may not apply very

well to real-world case studies. These assumptions

include:

All local resources are efficiently employed, and no underemployment of resources occurs.

The entries in the IO table do not represent capital goods, or the purchase of capital goods.6

Technical coefficients are fixed; that is, the amount of each input necessary to produce one unit of each

output is constant. The amount of input purchased

by a sector is determined solely on the level of out-put, with no allowance made for price effects, changing technology or economies of scale. The

input-output model involves the assumption that the same relative mix of inputs will be used by an

industry to create output regardless of quantity. There are no constraints on resources or the ex-

pansion of industry sectors. The lack of availabil-ity of local resources can reduce the real impacts compared with the estimated impacts under this assumption.

While an IO technique is very useful in assessing economic impacts, the assumptions and limitations of this method have to be clearly understood before interpreting the results.

The case study communities

The case study areas (the former Duaringa Shire and the former Bauhinia Shire7) are located in the south-ern part of the Bowen Basin region, a region being affected by the recent coal mining boom. The sub-regional area is the Central Highlands (Figure 2), which is part of the Fitzroy Division in Queensland. Until 2008 the Central Highlands comprised five shires (Bauhinia, Duaringa, Emerald, Jericho and Peak Downs) with an estimated population in 2006 of 28,864 people (OESR, 2006). Emerald is the major service town for the region, with an estimated population of 13,000 people.

The two former shires this research is focused on are very different. The Duaringa Shire is a shire ser-vicing an established coal mining industry, and the

Figure 2. The Central Highlands region, Queensland, Australia Source: http://www.centralhighlands.qld.gov.au/



Bauhinia Shire is a predominantly agricultural shire with coal mines being developed (Figures 3 and 4).

The population of the Duaringa Shire was approx-imately 11,000 people in 2004. The Duaringa Shire experienced population growth due to the increasing activities in the mining industry and therefore higher employment in the shire. Blackwater is the dominant town in the Duaringa Shire, and had an estimated population of 7,917 people in 2006 (OESR, 2006). Blackwater is situated 190 kilometres west of Rock-hampton (the major town near the Capricorn coast). This position allows easy commuting between the

coast and mining town. The town of Blackwater has positioned itself as the Coal Capital of Queensland and is the dormitory town for six coal mines.

The former Bauhinia Shire is located further away from the coast and major centres. It is located around 350 kilometres southwest of Rockhampton and had an estimated population of 2,266 people in 2006 (OESR, 2006). The town of Springsure (ap-proximately 780 people) is the population and administrative centre, and the Shire also includes the town of Rolleston (approximately 80 people), locat-ed approximately 70 kilometres east of Springsure.

Figure 3. Distribution of industry inputs, wages, employment and net trade in the former Duaringa Shire

Figure 4. Distribution of industry inputs, wages, employment and net trade in the former Bauhinia Shire



The Bauhinia Shire traditionally has been based on agriculture, but mines have developed in the shire since 2003. The shire contains substantial reserves of coal and several new mines are currently in vary-ing stages of development. The Rolleston and Minerva mines were established in the shire in 2004 and 2005, producing 8 million tonnes of thermal coal and 2.5 million tonnes of medium to highly volatile coal respectively per annum (NRM, 2006).

Both of the case study communities had some employees based in the local area, while others commute to the shires for periods of ‘block’ shifts, and stay in temporary accommodation at work-camps in the towns or at mine sites. Since mining in Queensland is located mainly inland, there is a ten-dency for some workers to locate their families at the coast and to commute to work for their shifts. PIFU (2006) estimated that 10% of the Duaringa Shire population and 6% of the Bauhinia Shire populations were made up of non-residential work-ers who were commuting to the shires for ‘block’ shift periods.

There are some key differences between the case study communities. Blackwater is located within 2 hours’ drive from the major regional centre of Rockhampton, while Springsure is much further away. This means that Blackwater residents have more access to a large centre, and that it is easier for mine workers to base at a coastal or urban centre and commute to the mines for ‘block’ shift periods. A second key difference is that the larger size of Blackwater, close proximity to a number of mines and

a number of existing work-camps in town for mine

and construction workers may mean this community

is likely to have more direct impacts from an expan-sion in mining. The impacts on Springsure could be much smaller because there are not that many min-ing jobs and there are no work-camps in town.

A third key difference between the communities may be variations in the local economies. Blackwa-ter does not appear to retain very much local ex-penditure, despite being larger. Key reasons appear to be:

Proximity of Emerald and Rockhampton as regional centres;

Fluctuating population levels have made it hard for businesses to survive; and

Social and other factors mean many people prefer to access larger centres.

Springsure appears to capture more local expenditure, perhaps because:

Springsure is further inland from large centres such as Emerald and Rockhampton compared with Blackwater;

Springsure and the Bauhinia Shire has a very stable

rural community; There may be strong local support for businesses;

and

There has been little fluctuation in population levels.

Input-output modelling was used in this project to predict the impacts of the increased mining activities at the local and regional level in the Central High-lands region. This involved the construction of a rep-resentative model for each of the economies of interest, where the interrelationships between each business sector, households and the external world (i.e. outside the region of interest) is summarised by a series of multipliers. The GRIT technique was used. Changes in activity in the coal sector were then fed into the model as an input, and the conse-quential (multiplied) changes in each of the other sectors then was read from the model as an output, in terms of changes in employment, expenditure and incomes.

The comparison of the results for the two case studies

The focus of the analysis was to make some general predictions of the impacts of the mining boom on the regional area while recognising a number of data and modelling limitations of the IO application. Some of the modelling limitations have been identi-fied in the section above, where the difficulties of constraining national or state level input-output models to regional or local circumstances have been noted. Data limitations relate to the difficulties in obtaining accurate and timely data on mining activi-ties, and in disaggregating data to the sub-regional level.

The input-output tables for the Australian econo-my in 1997/988 provided the reference point for im-pact analysis. Three separate Input-output tables have been constructed to undertake the analysis: Queensland, Fitzroy and Central Highlands IO tables. Official statistics were used where possible to scale the reference table to reflect the economy of the regions of interest. The adjustment to the Queensland IO table has been made using available data such as gross state product, household expendi-ture, compensation of employees, and gross operat-ing surplus. The tables for the smaller regions (i.e. Fitzroy and the Central Highlands) were derived from the Australian IO table using employment and other regional data for these areas. The tables can only be used as an approximation of the impacts of the projects until more accurate data for these re-gions are obtained through survey or other means.

It was difficult to get actual data from coal mining companies about expenditure and income patterns relating to expansion activities. This is partly be-cause of the confidential nature of the information, and because it might be difficult for mining compa-nies to collate the financial data into the appropriate categories required for an input-output analysis. All estimated impacts therefore reflect the existing



structure of sales and expenditures of the whole9 mining industry in Australia in 1997/98. The table was adjusted to be presented in 2004 dollars. In 1997/98 the coal prices were lower than in 2004/05. CPI (consumer price index) adjusted price per tonne all coals was A$72/tonne in 1997/98, A$57/tonne in 2003/04, A$83/tonne in 2004/05 and A$125/tonne in 2005/06.10 The value of coal was therefore adjusted to reflect higher prices in 2004 onwards to make the predictions of income and expenditure levels more realistic.

An initial assessment was performed assuming that any growth in demand for outputs of various sectors can be supported by local industries and that employment increases will be housed in the sub-regional area. Model results under these assumptions identify the maximum impacts at a regional level (Fitzroy region).

The assessment was modelled under expectations of general growth in the mining industry in the five years to 2010. NRM (2006) estimated that Queens-land coal exports could increase by 48% in the peri-od to 2010, from 145.5 million tonnes in 2004/05 to around 215 million tonnes in 2010. However, the in-creases were not likely to be uniform across the state, with more developments in the northern Bowen

Basin and other parts of Queensland than in the southern Bowen Basin. Preliminary estimates sug-gested that production in the central and southern parts of the Bowen Basin could increase by over 25 million tonnes on a 2005 production level of 106 million tonnes (NRM, 2006). To reflect this, a general increase of 25% in mining activity was modelled for

the sub-regional area by increasing the value of coal exports from the sub-region.

At the price of A$83/tonne a 25% increase in coal mining production would translate into A$2.2 billion additional coal output in the central and southern parts of the Bowen Basin.

Using a 25% expansion scenario, impacts were estimated at the regional level (Fitzroy region), sub-regional (Central Highlands region) and two former shires (Duaringa and Bauhinia Shires). Model results suggest the total (initial, direct, indirect and induced) impacts of a 25% mining expansion in Fitzroy Statistical Division level on industry output, house-hold income and employment in an average year is expected to be $2,074 million, $348 million and 5,775 jobs respectively (Table 2). That translates in-to about an 8% increase in the output from all indus-tries in Fitzroy Statistical Division region, 10% increase in the regional income and 7% increase in regional employment.

At the Central Highlands region level these ef-fects are smaller: $1,219 million, $191 million and about 2,432 jobs11 respectively. That translates into about 16% increase in the output from all industries, 26% increase in regional income and 17% increase in regional employment.

To estimate the level of impacts at the sub-regional and local shire level, the model results had to be calibrated to account for potential ‘leakages’ caused by non-resident employees. PIFU (2006) reported a non-resident worker population of 2,313 employees in the Central Highlands, about 8.5% of the full-time equivalent population. However,

Table 2. Economic impact of the coal mining industry development: the Fitzroy Region, the Central Highlands Region, Bauhinia Shire and Duaringa Shire, 25% increase, 2004

Industry output, $m

Value added, $m

Household income, $m

Employment, persons

The Fitzroy Region Initial 1,161 665 158 1,297 First round (direct) 369 181 67 1,386 Industrial support (indirect) 206 98 41 983 Consumption (induced) 337 163 82 2,110 Total 2,074 1,108 348 5,775

The Central Highlands Region Initial 912 541 128 1,054 First round (direct) 183 96 34 646 Industrial support (indirect) 46 24 9 212 Consumption (induced) 77 39 20 519 Total 1,219 699 191 2,432

The Bauhinia Shire Initial 5.8 3.3 0.8 6.4 First round (direct) 0.9 0.5 0.2 3.7 Industrial support (indirect) 0.2 0.1 0.0 1.1 Consumption (induced) 0.5 0.2 0.1 3.2 Total 7.4 4.1 1.1 14.4

The Duaringa Shire Initial 359 206 49 401 First round (direct) 67 35 11 200 Industrial support (indirect) 13 7 2 52 Consumption (induced) 16 8 4 108 Total 456 256 67 761



housing shortages, demographic changes and em-ployment changes mean that increasing numbers of mining employees are basing in coastal communities and travelling to the area for ‘block’ shift work (Zheng et al, 2007).

To estimate impacts on the Central Highlands, it was also assumed that only 50% of the workforce for the 25% expansion in mining operations would live in the sub-regional area. This reduced the in-come and employment multipliers compared with a 100% workforce location in the Central Highlands by 0.63 and 0.70 respectively. Model results suggest that in this case the total (initial, direct, indirect and induced) impacts of a 25% mining expansion in the Central Highlands region on household income and employment in an average year is expected to be $120 million and about 1,723 jobs respectively, smaller than the situation when all employees are living in the Central Highlands.

The expected impacts on the Duaringa Shire and the Bauhinia Shire will be a subset of the predicted effects. Model results suggest the total (direct, indi-rect and induced) impacts of a 25% mining expan-sion in the Duaringa Shire on industry output, household income and employment in an average year is expected to be $456 million, $67 million and about 761 jobs respectively. In the Bauhinia Shire the effects of a 25% mining expansion of the shire mining industry are more modest due to the agrarian nature of the Bauhinia Shire and modest coal mining expansion: $7.4 million, $1.1 million and 14.4 jobs respectively.

While the former Bauhinia Shire has a capacity to support the growth in mining, larger projects will put a strain on the housing affordability and local busi-nesses. For example, the expected coal mining ex-pansion ($724 million) in the Bauhinia Shire could result in increase of the total output of the shire by $926 million, an increase in households’ income of $143 million and an additional employment of 1,812 persons, given the assumptions listed. These are highly overestimated impacts on the shire, given labour force, housing and local business capacity constraints. In reality, only a small proportion of new employees are likely to live in the Bauhinia Shire, and the capacity of local business to meet the needs of the coal industry may be limited, which will reduce the multiplier effects.

Leistritz et al (1990) noted that it is hard to ex-trapolate models when there are large changes in de-velopments. Therefore the accuracy of the IO models on the sub-regional level using the non-survey approach can be quite low if the region is ex-periencing large developments. The main issue would be to consider whether the industries that support the coal mining industry are available in the region and have enough capacity to meet the new (or growing) demand from the coal mining industry. If

not, then the aggregate impacts will either be spread

over the larger region (e.g. the Central Highlands, Queensland) or the local region will attract new

industries to be located within the local region. Nei-ther the former Duaringa nor Bauhinia Shires has

enough capacity to meet the demand of a growing

coal mining industry. This means that projected im-pacts may be lower than predicted from the unad-justed model.

Nevertheless, the results of the model can be used as a guide to the magnitude of approximate impacts if the following assumptions can be made: (1) that the development will attract the supporting indus-tries in the sub-region, (2) that the relationships among industries will be the same as in the national economy and (3) that the projected numbers of con-struction employees and project employees will live in the area of investigation.

Summary and discussion

IO analysis is an economic modelling technique used

to predict how changes in economic activity impact through various sectors that make up an economy. The technique is generally used to model economic

activity at a national or state level. There is often in-terest in predicting changes in economic activity that might occur at a sub-regional level, particularly when

these relate to a single industry impact such as the de-velopment of a new coal mine. Although the IO tech-nique can be used to model economic activity in a

region of any size, accurate results will depend on

high quality data about the level of economic activity

and interrelationships between economic sectors. However, the smaller the region, the fewer official statistics are available. The best low cost approxima-tion is normally to use the ABS data on employment and adjust the IO table for a region according to local employment figures.

No detailed input-output model was available for the Central Highlands communities and shires. A re-gional IO model was specifically adapted for two case studies using the GRIT technique to approxi-mate the local models. The IO results provided some guide as to the size of distribution of impacts from the current coal mining expansion and the potential impacts at the local level. However, the IO analysis performed was a desktop study that involved limited

Although the IO technique can be used to model economic activity in a region of any size, accurate results will depend on high quality data about the level of economic activity and interrelationships between economic sectors



adjustment to the local economy. The increasing number of workers who locate their families away from mining towns to regional and coastal12 and commute to work for ‘block’ shifts can make the re-sults of the IO model (closed model) unreliable if this effect is not taken into account. These workers and their families tend not to spend their wages in the local employment area but in the region where they live, therefore transferring the economic stimu-lus. Another important issue is the ability of local business to accommodate the increasing demand from mining in terms of services to expanding min-ing operations. A limited local capacity could mean that the impact effect will be more on the wider re-gion but not on the local community. This is particu-larly important in the context of environmental impact assessment because it is providing an insight as to what aspects of the proposed projects can be strengthened.

While more detailed models can be built if better information about local business spending is availa-ble, the preliminary modelling performed in this case study was typical of the economic impact as-sessment reported in EIS studies. An alternative to constructing a table using business and community surveys is to conduct a sensitivity analysis to adjust results to a more realistic framework. The results of unadjusted models should be treated with caution due to limitations of input-output method, data used and assumptions made. The research in this paper has demonstrated how IO modelling can be tailored to smaller case studies where non-resident work forces are involved.

Notes

1. The Central Highlands region comprised Emerald, Bauhinia, Duaringa, Jericho and Peak Downs Shires before amalgama-tion into the Central Highlands Local Government Area (excluding Jericho Shire) in 2008.

2. While there are several software packages that have been used for IO analysis (e.g. GRIT, GRIMP, RIMS II ) the IO modelling can be done using an Excel spreadsheet. The main issue is the data availability. The more sophisticated com-mercial models for interregional analysis such as general equilibrium models (e.g. IMPLAN, MONASH and TERM) de-mand higher requirements on the data inputs, require lengthy and costly training and can also be costly to purchase and update the data.

3. See Jensen (1980) for the concept of accuracy in the IO modelling.

4. For the analysis of accuracy of IO estimates using non-survey methods see Bonfiglio and Chelli (2007).

5. Coon and Leistritz analysed data on an annual basis from 1985 to 2006.

6. Capital goods are treated as final outputs of the economy that enter the capital stock and provide input ‘services’ in subsequent periods (Fraumeni et al, 2004).

7. The shire levels of the study areas were chosen due to data availability. It is also in line with a ‘functional economic region’ notion where the lowest order places such as small town pro-vide goods that are available everywhere and additional goods and services are provided at a higher order places such as cities (Robinson, 2009).

8. The Australian Input-Output table 1998/99 does not reflect the change in the taxes due to the introduction of Goods and Services Tax (GST), but more recent official data were not available at the time of the study.

9. That includes black and brown coal mining, and oil and gas extraction. There are no reasons identified to treat coal indus-try expenditure and income differently to the rest of mining sector.

10. http://www.nrw.qld.gov.au/mines/coal/files/table_14.xls 11. In the Central Highlands region, the labour force is approxi-

mately 61.2% of the population (OESR, 2006). OESR (2006) predicted a population increase of 2,004 people in the sub-region between 1996 and 2007. While the OESR predictions also include the impacts from other industry sectors, the dif-ferences in population growth predictions suggest that the in-put-output model is not calibrated well in explaining the mining boom. The additional workers are likely to be located outside the region.

12. PIFU (2005) estimated that Linvingston Shire (coastal area) has grown by 7.38% or 1,691 people from 2001 to 2004, while Bauhinia Shire’s population has declined by 0.53% (12 people) in the same period of time. The large centre of Rock-hampton has grown by 1.25% (or 1,691 people) during 2001–2004, similar to growth in the Duaringa Shire of 1.23% (or 81 people) in the same time frame.

References

ABARE 2006. Australian Energy. National and State Projections to 2029–30. ABARE research report 06.26. Canberra, Austral-ia: Australian Bureau of Agricultural and Resource Economics.

ABS 2010. ABS Average Weekly Earnings, Australia, May 2010. ABS. Available at <http://www.abs.gov.au/AUSSTATS/abs@ .nsf/DetailsPage/6302.0Aug%202010> last accessed 10 Sep-tember 2010.

ACIL Consulting 2002. Queensland’s mining industries: creating wealth for the community, the sand the nation. Unpublished report for the Queensland Department of Natural Resources and Mines, Brisbane.

Batey, P, M Madden and G Scholefield 1993. Socio-economic impact assessment of large-scale projects using input-output analysis: a case study of an airport. Regional Studies, 27, 179–191.

Bonfiglio, A and F Chelli 2007. Assessing the Behaviour of Non-Survey Methods of Constructing Regional Input-Output Tables through a Monte Carlo Simulation. Quaderno di Ricerca, 29.

Coon, R and L Leistritz 1985. The Contribution of North Dakota’s Lignite Industry to the State Economy: A Statistical Analysis. AE 85016. Fargo: North Dakota State University, Dept of Agr. Econ.

Coon, R and L Leistritz 1987. The North Dakota Economy: Esti-mating Recent Changes and Projecting Future Trends in the Economic Base. Agricultural Economics Statistical Series No. 41. Fargo: North Dakota State University, Department of Agricultural Economics.

Coon, R and L Leistritz 2006. North Dakota Lignite Energy Indus-try’s Contribution to the State Economy for 2005 and Projected for 2006. AAE 06002. Fargo: North Dakota State University, Dept of Agr. Econ. Available at <http://purl.umn.edu-23655> last accessed 18 August 2011.

Cristobal, J and M Biezma 2006. The mining industry in the European Union: analysis of inter-industry linkages using input-output analysis. Resources Policy, 31(1), 1–6.

Department of Mines and Energy, DME 2007. Queensland Mining

Industries. the Economic Significance of Mining and Mineral Pro-cessing to Queensland. Summary report. Brisbane, Australia.

Fraumeni, B, M Harper, S Powrs and R Yuskavage 2004. An in-tegrated BEA/BLS production account: a first step and theoret-ical considerations. CRIW Architecture for the National Accounts Conference, 17 April, Washington, DC.

IRIS 2005. Illawarra Coal: Regional Economic Impact Assess-ment. Measuring the Economic Impact of Illawarra Coal’s Op-erations on the Illawarra and Wollondilly Regions. Prepared by IRIS Research Ltd, The University of Wollongong, Australia.

Jensen, R 1980. The concept of accuracy in input-output. International Regional Science Review, 5, 139–154.

Jensen, R C and G West 2002. Community Economic Analysis. Brisbane: Queensland Government, Department of Primary Industries.

Jensen, R, T Mandeville and N Karunaratne 1977. Generation of Regional Input-Output Tables for Queensland. Report to Co-ordinator General’s Department and Department of Commer-cial and Industrial Development, Department of Economics,



University of Queensland, 249pp. Lahr, M 1992. An investigation into methods for producing hybrid

regional input-output tables. PhD thesis. Dissertations available

from ProQuest. Paper AAI9235166. Leistritz, L, S Murdock and R Coon 1990. Developing economic–

demographic assessment models for substate areas. Impact Assessment Bulletin, 8(4), 49–65.

Lenzen, M, S Murray, B Korte and C Dey 2003. Environmental impact assessment including indirect effects — a case study using input-output analysis. Environmental Impact Assessment Review, 23, 263–282.

Leontief, W 1956. Factor proportions and the structure of Ameri-can trade: further theoretical and empirical analysis. Readings in W Leontief 1966. Input-Output Economics. New York: Oxford University Press.

McGrady, T 2004. Press release by the then Minister responsible for Mining, Tony McGrady, 23 November 2004.

Mandeville, T and R Jensen 1978. The impact of major develop-ment projects on the Gladstone/Calliope, Fitzroy, Queensland, and Australian economies: An application of Input-Output analysis. Report to the Department of Commercial and Indus-trial Development and Comalco Limited, Department of Economics, University of Queensland.

Mangan, J 2001. The significance of the mining industry to the Queensland economy. Unpublished report prepared for the Queensland Department of Natural Resources and Mines, January 2002, Brisbane.

Natural Resources and Mines, NRM 2006. Queensland Mining and

Petroleum 2005: Exploration, Operations and Developments.

Brisbane: Queensland Government. OECD 1992. Structural Change and Industrial Performance. Paris:

OECD Publishing. Office of Economic and Social Research, OESR 2006. Queens-

land Regional Profiles. Brisbane: Office of the Government Statistician.

Planning Information and Forecasting Unit, PIFU 2005. Popula-tion and Housing Fact Sheets. Brisbane: Queensland Gov-ernment Department of Local Government, Planning, Sport and Recreation.

Planning Information and Forecasting Unit, PIFU 2006. Full-Time

Equivalent population estimates for nine Local Government Areas in the Bowen Basin, June 2006. Report prepared in the Queensland Government Department of Local Government, Planning, Sport and Recreation, Brisbane.

Robison, H M 2009. Input-Output Guidebook: A practical guide for regional economic impact analysis. Economic Modeling Spe-cialists, Inc. Available at <http://www.economicmodeling.com/ wp-content/uploads/emsi-io-guide-1.pdf> last accessed 18 August 2010.

Rolfe, J, S Lockie and M Franettovich 2004. Economic and Social Impacts of the Coopabella Mine on the Nebo Shire and the Mackay Region. Centre for Social Science Research Report. Central Queensland University, Rockhampton, 120pp.

Rolfe, J, S Lockie and G Ivanova 2005. Overview of social and economic issues associated with the Bowen Basin coal industry, Research Report 1, Socio-economic impact assessment and community engagement to reduce conflict over mine operations research reports. Rockhampton: Centre for Social Sciences Research, Central Queensland University.

Rolfe, J, B Miles, S Lockie and G Ivanova 2007. Lessons from the social and economic impacts of the mining boom in the Bowen Basin 2004–2006. Australasian Journal of Regional Studies, 13(2), 165–184.

Shaffer, R, S Deller and D Marcouiller 2004. Community Econom-ics: Linking Theory and Practice, 2nd edition. Iowa, USA: Blackwell Publishing.

Ung, B 1981. Assessing Impacts of Mining Development in Cen-tral Queensland within the Context of Strategic Market Plan-ning: An Application of the Input-Output Model. B.Econ, University of Queensland.

West, G 1981. Generation of Regional Input-Output Tables (GRIT): An Introspection. Working paper N29, Department of Economics, University of Queensland.

West, G, J Morison and R Jensen 1984. A method for the estima-tion of hybrid interregional input-output tables. Regional Stud-ies, 18(5), 413–422.

Zheng, C, J Rolfe, L Di Milia and P Bretherton 2007. Strategic people management of coal mining firms in Central Queensland.

Management Research News, 30(9), 689–704.

Using input-output analysis to estimate the impact of a coal industry expansion on regional and local economies

Documents