Agribusiness Review - Vol. 12 - 2004

Paper 2
ISSN 1442-6951

Productivity in the Australian Dairy Industry *

Tom Kompas and Tuong Nhu Che

Tom Kompas, National Centre for Development Studies, Asia Pacific School of Economics and Government, Australian National University, Canberra, ACT and the Australian Bureau of Agricultural and Resource Economics, Canberra, ACT tom.kompas@abare.gov.au

Tuong Nhu Che, Australian Bureau of Agricultural and Resource Economics, Canberra, ACT nhu.che@abare.gov.au

Abstract

Although the Australian dairy industry has performed well it has also faced considerable pressure over the past twenty years. A decline in the terms of trade and major structural change has provided added incentives for the industry to improve productivity. This paper constructs Tornqvist index values to measure and analyse movements in inputs, outputs, total factor productivity (TFP) and the terms of trade for the dairy industry as a whole and for each state over the years 1979 to 1999. Overall, there is clear evidence of a significant increase in the TFP index in the 1990s relative to the 1980s. However, in terms of fitted annual growth rates, there is also evidence of a productivity 'slow down' in the 1990s, with the principal exception of New South Wales. Average annual growth in dairy total factor productivity in Australia over the entire twenty-year period is 1.5 per cent, but decreases from 1.8 per cent in the first to 0.9 per cent in the second decade. In Victoria, the largest dairy producer, the growth in TFP in the second decade of the study is virtually zero, with poor weather conditions in the second half of the decade partly to blame. Much of the impressive growth in dairy output in the 1990s can thus be simply attributed to a growth in inputs. Index values for the terms of trade, the share of input costs in total costs and potential drives of productivity change are also examined.

1. Introduction

The dairy industry is one of Australia's most important agricultural industries. The farm gate value of production in dairy ranks it as the third largest agricultural industry in Australia (behind wheat and beef), contributing roughly three billion dollars per year (ADC, 2001). In terms of value-added, the dairy industry is ranked among the top four of the largest processed food industries, providing an important source of employment in rural areas. It is also the largest processed food export industry, with export sales of processed milk and manufactured dairy products of $2.3 billion in 1999-2000 (ADC, 2001).

Over the past twenty years there was a substantial amount of restructuring in this industry and especially so in the 1990s with a large increase in milk production and changes in government regulation. A decline in the terms of trade and considerable structural change has provided added incentives for the industry to improve productivity. Productivity growth is one important aspect of farm performance and is a measure of the gains from technological change and better or more efficient farm practices. Changes in productivity can be measured as increases in outputs using the same amount of inputs or by a process that generates the same outputs using fewer inputs. The precise measure of total factor productivity (TFP) used in this study is calculated using a Tornqvist index over heterogeneous groups of inputs and outputs. Dividing the Tornqvist index of total outputs by the Tornqvist index of total inputs generates a TFP index (as a cumulative growth index). Annual growth rates for TFP are derived by fitting a logarithmic trend line with annual indexed data.

Section 2 of the paper provides a short overview of the Australian dairy industry and indicates several key summary statistics such as the number of farms, farm size and milk production by state. Section 3 details the nature of the Australian milk market arrangements and government regulations in each state, factors that directly influence the terms of trade for dairy products. Section 4 describes the methodology used to construct the measures of outputs, inputs, TFP and the terms of trade for each state and Australia as a whole. Section 5 indicates the data sources for estimation and the survey methodology. Section 6 presents the key results for Australia and the two most important dairy states, Victoria and New South Wales. Section 7 provides a comparison of performance across states in terms of the annual growth rate of TFP and the terms of trade. Section 8 indicates the major cost components in dairy and Section 9 concludes. Appendix A and B collect main statistics, detail survey methodology and give variable definitions.

2. Background to the Australian dairy industry

Australia has over two million dairy cows, producing around 10 billion litres of milk each year (ADC, 2001). The advantages of climate and natural resources allow production to be based mainly on year-round pasture grazing, although supplementary feeding with grains is becoming increasingly common, particularly in the last decade. Most dairy farming areas are located in high rainfall zones, where milk production depends on seasonal pastures. However, irrigation is important in northern Victoria, the Riverina in New South Wales and in parts of Western Australia and Tasmania. Australian dairy farmers continue to increase dairy output through improved pasture, feed and herd management techniques. In 1998-99, over 60 per cent of dairy farms were located in Victoria, 14 per cent in New South Wales, 12 per cent in Queensland, 6 per cent in Tasmania, 5 per cent in South Australia and 3 per cent in Western Australia (ADC, 2000).

While seasonal conditions continue to have a large influence on yearly output, Australian milk production has increased dramatically during the 1990s. In line with a pasture-based production system, Australia's milk output follows a strong seasonal pattern, with production peaking in October/November. This seasonal effect is most pronounced in Victoria.  In Victoria and Tasmania milk production depends mainly on pasture conditions with milk output typically lower during the winter months when pasture growth is reduced. In contrast, milk production in New South Wales, Queensland and Western Australia is more uniform throughout the year. The more uniform monthly distribution of milk production in these states reflects less seasonal patterns of pasture growth, differences in production and agronomic conditions and a greater dependence on the fluid milk market.

Australian annual milk production has increased steadily in every state (Appendix A). It is generally thought that Australia has achieved a high growth rate of milk production in the 1990s largely as a result of improved cow yields and in more recent years increasing cow numbers (ADC, 2000). In recent years, less than 20 per cent of Australia's milk production has been used for the domestic fluid milk (drinking milk) market. The remainder has been channelled into the manufacturing milk sector to produce dairy products such as butter, cheese, milk powders and other products. Victoria dominates milk production in Australia, accounting for 63 per cent of the country's total milk production and 72 per cent of manufacturing milk production in 1998-99. However, manufacturing milk production has expanding greatly in all states, with market milk declining as a percentage of total milk production (ABARE, 2001).

There has been considerable structural adjustment in the Australian dairy industry during the period 1979-99. The long-term trend indicates a movement towards larger farms both in terms of area and herd size. [1]

The number of dairy farms has nearly halved between 1978-79 and 1999-2000 with this decline occurring in all states. However, total milk production has increased by approximately 70 per cent (Appendix A). From 1991, milk yields per cow increased at a very fast rate as dairy farmers increased the adoption rate of improved technologies and farm management practices, such as the use of supplementary feeding, improved cattle genetics and better pasture management (ABARE, 1999).

Principal new technologies and dairy farm practices include enhanced feeds, fodder conservation, soil testing, fertiliser and drainage, enhanced herd and herd-health management and new milking sheds and equipment. Such extensive technological change must partly account for the increase in outputs and TFP. On average, during the last twenty years, the output of dairy farms has grown at a rate of 4.2 per cent per year. The growth rate of output has increased even more rapidly in 1990s at a rate of 5.0 per cent per year. Milk yields per cow have also increased strongly at approximately 2.4 per cent per year, and especially so in Western Australia and Queensland where imported genetically enhanced cows have generated high milk yields. Undoubtedly, genetic management and improved breeding practices have contributed significantly to the growth in total factor productivity. [2]

Finally, new technology for milking sheds and equipment, especially sheds and equipment that correspond to the increasing scale of dairy farms, is another important factor contributing to high productivity growth. With this, it is also important to note that much of the gains in TFP in the 1980s may simply be due to the economies of scale associated with the clear tendency toward larger farms in terms of both area and herd size. The measure of the growth of TFP used in this study includes the effect of returns to scale, which in terms of a growth index seems especially important as an explanatory factor in New South Wales, but less so in Victoria. 

Based on varying natural conditions for milk production and the adoption of new technology, farm sizes in Australia have changed considerably among states and over time and average land area per property, or hectares per farm, has generally increased (Appendix A). In the favourable climate regions for dairy production (principally Victoria and Tasmania where most of the manufacturing milk is produced) farm size is relatively smaller and production is much more seasonal.

3. Dairy markets and government regulations

Over the period of this study the Australian dairy market was characterised by a range of regulatory and institutional measures which divided the raw milk market into two separate milk sectors: the market milk and manufacturing milk sector (ABARE, 2001). Separate arrangements applied to the marketing of manufacturing and market milk, despite the fact that milk of only one quality generally left the farm. On the whole, these regulations and policies were instituted for the purpose of affecting the supply and farm gate price of milk according to its end-use. Different regulatory policies in turn affect the terms of trade as well as production and the adoption of new technologies and hence TFP.

The Commonwealth government provided assistance to the farm gate price of manufacturing milk throughout 1979-1999. Prior to 1986, assistance was in the form of a levy on domestic sales of dairy products that was paid to exporters of dairy products to increase their returns and encourage dairy product manufacturers to increase the quantity of their exports. The introduction of the Kerin Plan in 1986 changed the way in which the Australian dairy industry was supported and saw a reduction in the level of support. The previous levy was replaced by a levy collected from farmers on the production of milk paid to exporters of dairy products.

Under the Commonwealth Domestic Support Scheme, introduced in 1992, annual payments were made to dairy farmers based on their production of manufacturing milk.  The scheme did not attempt to regulate the supply of manufacturing milk, although it clearly had an impact on the production of milk in Australia and on resource allocation within the industry. Funds for payments from the scheme were generated via a levy on milk used to produce manufactured dairy products sold on the domestic market and a separate levy on milk used in the market milk sector. 

During the 1980s and 1990s, in most areas of Australia, state governments controlled the pricing and supply of milk for drinking (or 'market milk'). The arrangement segregated raw milk according to end use and guaranteed eligible farmers a fixed price for regulated supplies of market milk. The guaranteed farm-gate price for market milk was substantially higher than the average price paid for non-regulated milk supplies. In quota states (New South Wales, Western Australia and most of Queensland), farmers who held quota received an administered price for all milk accepted by their authorities for use as market milk. All other milk produced was paid at the manufacturing milk price (IC, 1991). Generally, these states were classified as market milk states since the majority of dairy farm revenue was derived from milk directly sold for use as drinking milk. Failure to deliver the designated supply of quota milk would result in a reduction in individual farm supply entitlement. Any surplus of milk produced above entitlement was sold as manufacturing milk. The farm gate market milk price exceeded the price that dairy farmers received for manufacturing milk (non-quota milk), with the manufacturing milk price generally varying in response to movements in the price of dairy products on world markets.

In non-quota states (Victoria, Tasmania and South Australia) farmers received a weighted average price for all milk produced. These states were classed as manufacturing milk states as the majority of dairy farm revenue was derived from milk sold for the manufacture of milk products. The market milk price and the manufacturing milk price were weighted by their respective volumes in each month's production to determine the price received at the farm gate (IC, 1991). Returns from the fresh milk market were pooled and each farmer received payments depending on the percentage of milk used for market milk in each month (Topp et al. , 1989). The manufacturing milk sector was not subject to any government production controls. Dairy farm incomes within these manufacturing milk states thus tended to be relatively more variable since manufacturing milk incomes are derived from the sale of dairy products on world markets.

From July 1, 2000 all state marketing arrangements were removed, resulting in an open market in fluid milk products with no further formal quantitative controls on the supply or price of domestic drinking milk. Currently, over 50 per cent of Australian milk is exported in manufactured forms, with 77 per cent of these sales destined for markets in Asia and the Middle East (ADC, 2000). The steady improvement in international trading conditions, improved Asian demand and efforts by Australian exporters to develop new markets has increased Australia's share of international trade in dairy products to 15 per cent in 1999/2000 (ADC, 2001). 

4. Measuring Total Factor Productivity

Estimates of productivity growth for Australian dairy farms allow one to decompose the growth in dairy farm output over time due to changes in conventional inputs such as labour, capital and land, from the change in the overall growth in productivity as a residual. In general terms, the productivity of a firm or dairy farm can be defined as the ratio of the output(s) a firm produces to the input(s) it uses. When the production process involves a single input and a single output the calculation is straightforward. However, when there is more than one input (or output) in a production process a method for aggregating these inputs into a single index is needed in order to measure productivity. Once obtained, this indexed value of productivity, or total factor productivity (TFP), is thus a measure of the productivity of all inputs or factors of production, in terms of their combined effect on output, and is often accounted for by technological change or more efficient methods of producing output. Alternatively, partial productivity measures the productivity of a change in a specific input alone, such as labour, holding all other inputs and technology constant. While a useful measure of the effect of each input taken separately, partial productivity measures provide no indication of overall productivity.

The most common chain-index method is a Tornqvist index, originating with Tornqvist (1936) and developed by Diewert (1976, 1981) and Caves, Chistensen and Diewert (1982a, 1982b). In basic terms, the concept of a Tornqvist index is straightforward. Since both inputs and outputs are measured in value terms an index is needed to construct real changes in the value of outputs and inputs, relative to a point of comparison or a base year, much like the construction of any price (or quantity) index. More formally, define the value share of the i th commodity (input or output) relative to the value of all commodities as

  (4.1)

in base period s , for n goods, prices p and quantities q . The Tornqvist quantity index ( Q ) in log-change form for periods ( t - 1) to t is

(4.2)

for

(4.3)

The Tornqvist quantity index at t is thus

  (4.4)

Using equations (4.2) and (4.3), a Tornqvist index can be calculated for both inputs and outputs, taken separately, base-normalized to 100 for all variables. The ratio of Tornqvist outputs to inputs is thus the measure of TFP. Comparable Tornqvist indexes can also be obtained for price variables, such as movements in the terms of trade. [3]

Since the chain-indexed method normalizes all states and regions to the same initial starting point, direct level comparisons in TFP across states are not possible. Nevertheless, comparisons among growth rates in outputs, inputs, terms of trade and TFP across states (and within a state or region for levels and growth rates over time) are valid. Given Tornqvist indexes for outputs, inputs, the terms of trade and TFP, estimated annual growth rates can be obtained by OLS estimates as a fitted logarithmic trend line (for time t ). In practice, total factor productivity (unlike the terms of trade) is calculated in this paper using manufacturing milk prices only, as proxies for marginal cost prices. As such, the effects of non-constant returns to scale (if they exist) will also partly account for the changes in TFP (Knopke, 1988).

There are a number of data and conceptual problems associated with this measure of TFP. Basically, the main aim of this study is to measure improvements brought about by changes in technical efficiency and better production methods. One of the major measurement problems relates to the effect of climatic variability on the TFP. For example in the short term, a severe drought will cause the TFP measure to fall, as the result of the use of more inputs (especially purchased fodder) and lower milk yields. Although systematic weather impacts can be expected to decrease the longer the time period involved, longer term trends in measured productivity can still be affected if rainfall over the start or end period are atypical. Another important uncertainty relates to any changes to the quality of the resource base over the measurement period. For example, if there are some resource costs associated with milk production (such as, salinity and soil erosion) that have affected the productive capacity of the land, these costs will not necessarily be reflected in the TFP measure. Obtaining the appropriate prices for outputs and inputs can also present problems. In the case of land, the price variable used is unlikely to be independent of productivity growth, and therefore does not allow for land values being partly influenced by expectations about the future productivity of that land. Other problems relate to the tendency of farmers to defer some input expenditures (such as capital purchases or repairs and maintenance) in low income years; measurement of the amount of capital used in the production process in any given year; and measuring quality changes (such as protein levels) in the milk produced. For this study, however, given the length of the time series data available (22 years) and the sample size for most estimates, the TFP measure is considered to be a reasonable approximation of the gains due to technological advance, enhanced efficiency and potential economies of scale.

5. Data sources for estimation

The two main sources for the database used in this study are estimates from ABARE's annual surveys of the dairy industry, 1978-79 to 1998-99, and ABARE's indexes of prices paid and received. ABARE surveys are designed and samples are selected on the basis of a framework constructed and maintained by the Australian Bureau of Statistics. The Australian dairy industry survey has been conducted annually since 1979. The relevant dairy establishments are defined under the Australian and New Zealand Standard Industrial Classification (ANZSIC) as being engaged mainly in the grazing, farming and the breeding of milk cattle (Australian Farm Surveys Report, ABARE, 1999). Survey methodology and variable definitions for inputs (including land, capital, livestock capital, labour, materials and services) and outputs (including milk and livestock sales) are detailed in Appendix B.

6. Key estimated results for Australia, Victoria and New South Wales

Key estimated results for TFP and the terms of trade for Australia, Victoria and New South Wales are detailed at length in Tables 1 through 6.  Estimates of partial productivity for each input in production (land, capital, plant and structure capital, livestock capital, livestock purchases, labour, material and services) are also reported.

6.1  Estimated results for Australia dairy farms

Taking 1978-79 as a base year for comparison (indexed at 100), there is a significant improvement in productivity in the Australian dairy industry in the 1990s compared to the first decade of this study (see Figure 1). In particular, from the first to the second decade, annual (average) total factor productivity for dairy farms in Australia increased from 97 to 114 (Table 1). Thus, in the second decade, the average index value for TFP is roughly 14 per cent higher relative to the base year.

Figure 1: Outputs, inputs, productivity and terms of trade indexes for Australia

For Australian dairy farms, the Tornqvist output index grew at a rate of 4.2 per cent from 1978/79 to 1998/99. However, the growth in output was much higher (almost double) in the second decade, or 5.0 compared to 2.9 per cent (Table 2). The reason for much of this increase can be attributed to the increase in inputs over the period. The annual increase in the growth of inputs is more than three times larger in the last ten years (or 4.1 per cent) compared with a growth rate of 1.1 per cent in the first ten years. As a result, the annual growth rate of total factor productivity from 1978-79 to 1988-89 is 1.8 per cent.  However, this rate slowed considerably to 0.9 per cent over the years 1989-90 to 1998-99, providing clear evidence of a productivity 'slow down' in the dairy industry. The growth rate in total factor productivity over the entire twenty-year period is 1.5 per cent.

During the last twenty years output prices received by dairy farmers increased at 4.1 per cent per year. However, prices paid for inputs increased at a faster rate or 4.7 per cent per year, causing a decline in the terms of trade faced by dairy farmers at a rate of -0.5 per cent per year. From 1978-79 to 1989-90, both output and input prices increased at a relatively high rate (6.6 and 5.4 per cent per year), causing considerable variance in the terms of trade but with an overall positive growth rate of 1.2 per cent per year. However, in the second decade, output prices increased only slightly (0.8 per cent), whereas input prices increased at a rate 3.0 per cent per year. The result is a substantial decrease in the terms of trade, or -2.2 per cent per year in the last ten years.

Partial productivity measures for land and plant and structures capital, livestock capital, livestock purchases and labour were all positive (Table 2). The results indicate that these inputs grew at a slower rate than output, possibly indicating that these were used more efficiently, or were combined with an embodied technology that is more efficient. The partial productivity measure for materials and services is an exception, with negative rate of growth at -0.7 per cent per year. The growth rate in materials (such as feed) clearly increased faster than the growth in output.

The highest annual growth rate of input use was materials and services (5.0 per cent for the twenty-year period), and especially so in the second decade at a rate of 6.7 per cent per year (Table 2). It is clear that increased feeding (the main part of materials and services) is an important factor contributing to the high growth rate of output in the last twenty years, and especially so in the last ten years. There is also significant positive growth in land capital, plant and structures capital and livestock capital for dairy production.  Nevertheless, the growth rate of livestock purchases was very low compared with the growth rate of output, perhaps indicating a stronger tendency to use artificial insemination and on-farm breeding. Part of the explanation for this tendency may be due to more restrictive quarantine measures, preventing livestock trade between states and regions to reduce the transfer of exotic animal diseases. In the last decade, in particular, quarantine measures have been more extensive and more rigorously enforced throughout Australia.

6.2  Estimated results for Victorian dairy farms

Victoria is the most important dairy state in Australia, accounting for about 60 per cent of total milk production in Australia. During the last twenty years milk production has been increasing over time and almost doubled in the 1994-99 period compared to 1978-84. During the study period the Victorian dairy industry is characterised by a decreasing number of farms, increases in average herd size and land area (Appendix A). 

Dairy herds in Victoria are mainly pasture fed and temperate climatic conditions allow for year-round grazing on permanent pasture. Supplementary feeding of grain is used as an aid to pasture management. Dairying takes place in the higher rainfall areas of the state (>700mm), namely the southwest, northeast and Gippsland regions, and in the irrigation areas of Northern Victoria and Central Gippsland . Production and milk yield per cow have increased substantially since 1985. In 1999-2000 the average milk yield was roughly 4,500 litres per cow. Three main regions produce the major part of dairy output for the state: the southwest areas, where production is mainly pasture based, with temperature climate conditions and rainfall mostly occurring in winter and spring; the Goulburn and Murray Valleys, where production is based almost entirely on irrigated grazing; and the Gippsland area, a relatively temperate and normally high rainfall area with rainfall mainly occurring in the winter and spring and where production is mainly based on grazing, with few farms using irrigation.

Taking the 1978-79 as a base year, the growth indexes of output, inputs, total factor productivity and terms of trade are indicated in Figure 2.  The average annual TFP index increased from 99 in the first decade compared to 112 in the second decade. The average annual index for outputs increased from 109 to 176. Inputs increased from 111 to 157 (Table 3).

Figure 2: Outputs, inputs, total factor productivity and terms of trade in Victorian dairy farms

The annual growth rate of output has increased from the first to the second decade, or 3.8 per cent and 4.6 per cent respectively. However, the annual growth rate of TFP falls from 2.4 per cent in the first decade to virtually zero in the second decade (Table 4). Poor seasonal conditions in the second half of the 1990's may account for some of the poor performance in the second decade. From 1978-79 to 1998-99 output prices increased at 4.4 per cent per year and input prices increased at a rate of 4.8 per cent per year, causing the terms of trade to deteriorate at a rate of -0.4 per cent per year. The terms of trade decreased at a rate of -2.4 per cent in the second decade of the study.

6.3  Estimated results for New South Wales

As the second major dairy state (after Victoria) New South Wales contributes around 13 per cent of total milk production in Australia. Over the twenty-year period average annual milk production increased from 896 to 1184 million litres, although the annual average number of farms fell from 3312 to 1841 (Appendix A). There are two main dairy regions in the state: the coastal areas, the adjacent tablelands, the Hunter and Lachlan Valleys and scattered inland dairy farms, where production is mainly pasture based with some irrigation in the south and drier inland areas; and the Murrumbidgee Irrigation and Murray Valley areas.

Taking the 1978-79 as a base year, the average annual index for TFP increased from 100 in the first decade compared to 116 in the second decade (see Figure 3 and Table 5). The growth in total factor productivity was 1.4 per cent from 1978-79 to 1998-99. In fact TFP increased significantly in this state from 0.9 per cent per year in the first decade to 2.2 per cent per year in the second decade. Output prices increased at 3.8 per cent per year and input prices increased at a rate of 4.5 per cent per year; consequently, the terms of trade deteriorated at a rate of -0.7 per cent per year from 1978-79 to 1998-99 (Table 6).

Figure 3: Outputs, inputs, total factor productivity and terms of trade in New South Wales dairy farms

7. State comparisons for the growth in TFP and the terms of trade

7.1  Annual growth rate of total factor productivity

The growth rates of outputs, inputs and total productivity over the period 1978/89 to 1998/99 for the dairy industry at the national and state levels are summarised in Table 7. The results allow for some rough comparisons among states and regions. It is important to recognise that most measures of interest vary considerably from the first to the second decade of the study.

Table 1: Growth indexes for Australian dairy farms

Table 2: Estimated annual growth rates for Australian dairy farms

Note:  * significant at the 5 per cent level.

Table 3: Growth indexes for Victorian dairy farms

Table 4: Estimated annual growth rates for Victorian dairy farms