(Paper prepared for the Victorian Department of Primary Industries, Dairy Program, June 2004)

Abstract

Productivity measurement is useful in some circumstances but not others. Measured productivity is poor for the Australian dairy industry as a whole. This finding is consistent across a range of studies and is confirmed by other information and analysis.

It is useful to explore reasons for this poor performance because some public policy questions are related to overall industry performance. In particular, productivity measurement concentrates attention on industry-based research and extension programs.

Production (and exports) have increased rapidly in the dairy industry but input use has increased faster. The major change has been increased grain feeding. Both increased purchases of grain and a higher proportion of exports exacerbate the financial risks of dairy farming. Recent drought and associated water shortages in irrigated dairying areas have compounded these systematic changes.

By definition, estimates of average productivity in the whole dairy industry have little to say about what is happening on individual farms. Moreover, productivity is measured using annual data on inputs and outputs. While day-to-day technical and management skills are important, many of the crucial economic decisions by farmers are long-term.

Aggregate productivity analysis is a useful first step in analysing industry performance. A next step is disaggregating the data to identify inputs, regions or time periods of particular interest. The time path of prices, policy changes and the weather continue to have most effect on the dairy industry. A conclusion that follows from recent experience is that the change to increased grain feeding has not been well understood in its scientific dimension, nor well executed at the farm level. Furthermore, expected gains from specialisation in manufacturing milk production following deregulation have not been realised for technical reasons, presumably related to poor reproductive performance.

In particular, it appears that farmers have been given poor information on the difference between the marginal costs and marginal benefits of concentrate feeding in different time periods and circumstances. Nor have the financial consequences been properly considered in advice that has been given to farmers.

Production is not the same as productivity. Increased production and exports should not be promoted as such by dairy companies, governments, scientists, industry and research organisations.

Introduction

The productivity of the Australian dairy industry has been closely examined in the last few years by the Australian Bureau of Agricultural and Resource Economics (ABARE), other researchers and industry commentators.

These notes are intended to provide a non-technical account of that discussion, as well as putting productivity analysis in the context of other policy decisions and events affecting the dairy industry.

The notes are mainly based on the following information:

• A textbook on productivity analysis and performance measurement (Coelli, Prasada Rao and Battese 1998). The textbook provides a thorough analysis of traditional total factor productivity (TFP) analysis and its extensions to data envelopment analysis (DEA) using linear programming and econometric estimation of stochastic production frontiers.
• Four reports by ABARE (2001a, 2001b, 2002, 2003).
• Three papers on application of productivity measures to the dairy industry (Fraser and Cordina 2001, Graham and Fraser 2003[1], Graham 2004) and a related paper applying similar methods to the wool industry (Fraser and Hone 2001). These papers are exploratory. A lot more work is required to bring this effort to fruition.

In addition, a paper from Canada (National Farmers Union 2003) that disputes the ethos of productivity improvement that has underpinned mainstream farmer behaviour and many government and industry policies is discussed in an appendix to this paper. The vigorous and dissenting approach of the Canadian paper has attracted local publicity and found some adherents in Victoria (Jackson 2004). Arguably, widespread acceptance of these views would have negative implications for the prosperity of the dairy industry.

An important trigger for the increased interest in the productivity of the dairy industry was that measured productivity in the initial study by ABARE (2001a) was much lower than that expected by the dairy industry. Milk output had grown rapidly in the1990s but TFP was unexceptional, certainly lower than the cereal and beef industries and not that far ahead of the sheep industry that had suffered the collapse of the reserve price scheme for wool.

Significantly, growth in productivity was higher in the 1980s when the rate of growth of output was much lower than in the 1990s. Furthermore, dairy farmers in New South Wales performed better than Victoria.

Nationally, milk production more than doubled from 1980 to 2002 and the number of farms halved (Graham and Fraser 2003). This implies a fourfold increase in milk output for the average farm business. No wonder dairy farmers were puzzled by the ABARE findings.

Closer examination of ABARE reports and other information, however, supports the finding that the productivity performance of the dairy industry has been unspectacular and that there was a productivity slowdown in the 1990s, especially in Victoria.

Many economic aspects of the dairy industry and technical issues in productivity analysis need to be considered in considering the productivity of the dairy industry.

These include:

• State, regional and temporal differences in productivity in relation to policy changes, and world prices.
• The (inverse) influence of weather on output and use of inputs. In years of low rainfall, inputs are up and output down. TFP measures are thus sensitive to weather conditions at the start or end of the period of the analysis.
• Changing technology of dairy farming in the last couple of decades, accompanied by changes in the factor shares of land, labour, capital and materials and services. The main component of materials and services is fodder purchased off-farm.
• The distinction between average industry performance as measured by TFP and the experience of individual farms, and even more so their financial prospects.

With respect to the last point, failure to distinguish between what is true or even relevant for the individual firm (farm), the industry and the economy is the bane of economic policy making and public discussion of agricultural issues. The industry is not a decision-making unit except in the provision of industry goods like research and development (R&D) and generic promotion.

Productivity analysis has attempted to bring individual and industry performance together with the development of DEA and estimation of stochastic production frontiers (Coelli et al. 1998). Coelli et al. are explicit about the difficulties and qualifications that apply to these extensions of productivity studies. 

In essence, the major mistake of many industry observers is to confuse different concepts of productivity. TFP or multi-factor productivity is intended to measure change in output taking into account all inputs.[2] This is totally different in principle from partial productivity measures that report output per unit of individual inputs.

Like other agricultural industries, and small business generally, the distribution of farm size in the dairy industry is skewed. In 1998-99, a quarter of Australian dairy farms ran fewer than 100 cows and about a quarter more than 200 cows (ABARE 2001 a, p.15). The larger farms were concentrated in Victoria.

Performance of individual farms depends on the timing of major investments like land and milking facilities as well as size, not average productivity across the industry. Commodity prices, often expressed as the ‘terms of trade’ – output prices compared with input prices – also need to be considered. It emerges from some of the studies discussed below that the terms of trade of the dairy industry were unusually benign in the 1980s.

Dairying has different financial characteristics to other agricultural industries. In particular, production is less able to be interrupted than cropping. The flow of receipts from dairying is monthly rather than annual as in other livestock industries. Dairying is conducted in higher rainfall areas or under irrigation. Traditionally, irrigation meant less production risk. That assumption has been turned on its head with recent water shortages in irrigation areas.

Interest in productivity is not just about the ‘numbers’ but the interpretation placed on the results. Especially following an unpublished (but widely read) report on R&D by the Centre for International Economics (CIE) for the then Dairy Research and Development Corporation (DRDC).

The CIE included poor productivity performance as part justification for recommending a change in the direction of R&D, broadly favouring more off-farm R&D and down playing expenditure on extension. Others have used poor productivity as an excuse for a ‘big bang’ approach to R&D, seeking one-off improvements in productivity through significant investments in robotic milking and molecular biology.

Whether any of these suggestions is justified has nothing to do with industry productivity and its measurement. The amount of extension, its content and how it is organised needs to be analysed irrespective of industry productivity. Similarly, whether increased investment in robotics and molecular biology is sensible has to be established on its merits.

Productivity measurement is useful in some circumstances but is not a substitute for economic analysis of individual investment decisions for R&D projects, or application of standard farm management analysis to decision-making by individual farms.

The dairy industry is currently experiencing economic difficulties mainly because of the accumulated result of drought but also with product prices closer to longer-term trends. One reaction has been to question the underlying philosophy of investment in productivity enhancement by industry and farmers. As mentioned above, local discussion of these issues has been influenced by material emanating from Canada (National Farmers Union, 2003) although there are many Australian precedents of populism in rural industries, including dairying.

The remainder of these notes is organised as follows. Next follows a summary of productivity concepts and the results of empirical studies. Brief conclusions follow. In an appendix, a recent report of the Canadian National Farmers Union that questions the productivity ethos is discussed setting out its weaknesses, and occasional strength.

Productivity: Concepts, Calculations and Confusions

A Textbook Approach

The following section of this paper provides a summary of the material presented by Coelli et al. (1998), who provide a comprehensive treatment of the why and the how of efficiency measurement and productivity analysis. The ‘how’ is promoted by provision of instructions for practitioners on the use of available software.

The main conclusion to be drawn from the textbook is that productivity and performance measurement has a valid theoretical basis. This is not to say that data limitations – measurement errors, aggregation issues or omitted variables (inputs or outputs) –  do not frustrate application of these techniques.

Productivity analysis based on economic theory should not be confused with unsubtle applications of comparative analysis, often called benchmarking, to identify farm practices or characteristics associated with superior financial performance. Most of these benchmarking and comparative analysis studies are rank empiricism.

Coelli et al. begin with careful definitions and distinctions between key concepts, most importantly technical efficiency and allocative efficiency in the use of inputs. Paraphrasing their thorough and extensive discussion, technical efficiency means attaining the maximum output possible for a given quantity of inputs. Allocative efficiency is selecting the combination of inputs that produces a given quantity of output at minimum cost, for given input prices.

Next follows an excellent introduction to the theory of production economics. Fundamental ideas like diminishing returns; time and length of run; substitution between inputs; returns to scale; cost minimisation and profit maximisation are clearly expressed. Presentations, qualifications and criticisms of mainstream production economics can be found in libraries by the shelf full. Nevertheless, in careful hands, the theory of production economics is useful in understanding the behaviour of firms (farms).

Coelli et al. also present more complex ideas from production economics such as duality that explain the economic properties of TFP measures and allow separation of technical and allocative components of inefficiency estimates based on frontier cost functions.

The next part of the book concerns index numbers. The simplest way of thinking about index numbers is as summary measures of related economic entities. Obvious examples are price indices (the ‘CPI’) and stock market indices (the ‘All Ordinaries’, the ‘Dow Jones’). Index numbers are like an average. TFP measures are an index number based on separate output and input quantity index numbers. As such, the well-established economic theory of index numbers is relevant to productivity analysis. Usually, a logarithmic trend is fitted to time series of TFP so that productivity is expressed as a rate of growth.

TFP indexes reflect aggregate industry performance or the performance of economies as a whole. This is useful information for some purposes but not others, just as a price index is no indication of price changes for single commodities or a stock market index is any help in evaluating a specific share.[3]

The objective of the detailed examination of the role of index numbers is to measure the growth of output that is net of input growth. Output quantity index numbers do not distinguish between the contribution to growth in output due to growth in inputs and the contribution due to technical change or changes in efficiency (Coelli et al., p.122-4).

The authors demonstrate that Tornqvist indexes used in most empirical studies have desirable economic properties. A Tornqvist index is a weighted geometric average, most easily computed as a weighted average in logarithmic form.

TFP indexes – an index of output divided by an index of inputs – and the rate of growth of productivity derived from TFP indexes are straightforward concepts at the industry level. Extension of this analysis to DEA and stochastic frontier analysis to measure inefficiency is less straightforward, in part because of the linear programming and econometric methods used in estimation.

DEA and stochastic frontier analysis is based on the work of Farrell (1957). Efficient firms operate to the limit of technical possibilities at minimum cost. Estimation of the production frontier of fully efficient firms based on a sample of firms in the industry allows decomposition of differences from efficient operation into technical and allocative inefficiency. 

While Farrell presented his analysis from an input orientation – “how much can input quantities be proportionally reduced without changing the output quantities produced”, the problem could also be posed as “how much can output quantities be proportionally expanded without altering the input quantities used?” (Coelli et al., p.137).

DEA uses linear programming methods to construct a frontier over the data. Efficiency measures are then calculated relative to the frontier. The methods can be adapted to identify scale economies or diseconomies. Econometric methods can also be used to estimate frontier functions.

Both DEA and stochastic frontier analysis have been used in some papers analysing productivity in the Australian dairy industry, as discussed below.

ABARE Reports

Initial Work

The initial study of productivity in the dairy industry by ABARE (2001a) covered the period from 1978-79 to 1998-99. This period was split into two decades and also disaggregated by States with further disaggregation into separate regions in Victoria and New South Wales. Thus, the analysis preceded the deregulation of the dairy industry in 2000 when restrictions on interstate trade and the historical separation of the dairy industry into ‘market milk’ and ‘manufacturing’ sectors was ended.

Arguably, the analysis has been overtaken by events. However, the insight that output is not the same as productivity (or efficiency) leads to useful conclusions. In particular, that now that prices have returned closer to historical trends there will be no respite from the pressures for individual farmers and the industry to increase productivity.

Earlier policy changes, and differences in climatic experience, can explain most of the observed differences in productivity over time and between regions. Some variables of interest were very different in the two separate decades.

The following points are worth noting:

• Victoria abolished its quota system for market milk in the late 1970s. The shakeout of small dairy farms close to Melbourne created conditions for Victorian productivity improvements in the 1980s.
• The terms of trade were generally favourable in the 1980s but less so in the second period. There were differences between States in prices received according to the way market milk prices were regulated.
• New South Wales allowed negotiability of milk quotas in the 1990s encouraging larger farms in more favoured areas for dairying. Dairy farmers in coastal NSW were almost in the business of warehousing land. Many took the opportunity to quit when they could transfer quotas. In terms of TFP calculation, this casts doubt on whether ‘output’ in some regions should be restricted to farm output.
• The productivity ‘slowdown’ in the 1990s can be attributed mainly to a faster rate of growth in feed inputs than milk output. Increased grain feeding in the last fifteen years is a fundamental shift in the dairy industry. Prima facie, this suggests that explanations of poor productivity might be found through further experimental investigations in dairy nutrition and related on-farm management issues.
• The weather was less favourable in the 1990s, especially in Victoria.
• The south west of Victoria is a special case. The number of farms increased rather than declined with conversion of sheep and cattle properties to dairying.
• Not just with the benefit of hindsight, the growth rates of output achieved in the dairy industry of around 6 per cent in the second half of the 1990s were always unsustainable for a combination of economic and weather-related reasons.

Follow-up Comments from ABARE

The initial ABARE report was followed by a workshop organised by DRDC. A subsequent report by ABARE (2001b) summarised the key issues to come out of the workshop. The earlier results were updated with TFP measurements for 1999-2000 and recalculation of growth rates. The method of valuing cows slaughtered was also changed to reflect saleyard prices.

Strong productivity growth occurred in the additional year – 8 per cent for Australia as a whole – with increased output and only a small increase in inputs. The revised annual growth rate of productivity was around 0.25 per cent higher. This illustrates the sensitivity of productivity measurement to the period selected, and assumptions about prices.

Another issue explored was the effect of drought on Victorian dairy productivity. Results were as expected. Drought has had a severe effect on productivity in the dairy industry.

However, a more general conclusion can also be drawn. Increased stocking rates, greater dependence on purchased feed and frequent shortages of irrigation water increase exposure and the severity of the financial risks associated with drought. The rules of thumb once used by dairy farmers in financial management are well and truly obsolete.

Without compensating changes in financial management such as farmers holding more liquid assets or controlling debt, the dairy industry is now more vulnerable to weather shocks, and price risks in the grain and dairy markets than in previous decades.

Commodity prices are also subject to exchange rate risks as well as market access and usual supply/demand factors. Increasing dairy production means increased exports and greater exposure to significant fluctuations on the world market. This is a further source of increased financial risks in dairy farming.

ABARE also observe that new dairy technologies require more capital investment whereas some recent advances in broadacre cropping like minimum till cultivation are capital saving. This represents a further increase in risks facing the dairy industry.

ABARE worked through other possible explanations for poor productivity performance by varying assumptions about milk prices and land values. The results do not alter the underlying conclusion that the productivity performance of the dairy industry has been unimpressive. And correspondingly, the dairy industry became increasingly vulnerable to a downturn in prices and/or increased input prices as happened in the intervening period.

ABARE also examined productivity differences according to size of dairy herd. Contrary to the experience of broadacre farms, there was no significant difference in TFP between the two groups. However, larger dairy farms had a superior financial performance. Cash costs per litre were slightly higher but capital value per litre was appreciably lower.

Stochastic Production Frontiers

The third paper by ABARE (2002) uses a technique described in the book of Coelli et al., stochastic production frontiers. As summarised above, the idea is to allow comparisons of absolute productivity levels among farms within and between regions. The techniques used in the paper allow for decomposition of various sources of inefficiency and measurement of technical inefficiency as defined by the ratio of observed output to the corresponding estimated maximum output defined by the frontier production function, given inputs and stochastic variation (ABARE 2002, p.2).

A finding of this study was that the dairy industry exhibits constant returns to scale. Inputs and outputs are proportional. Productivity change will depend on improvements in technology and efficiency, not changes in size. Around 87 per cent of dairy farms in NSW and Victoria were estimated to be operating at maximum potential output – that is, on the production frontier. There were differences within states or regions.

The authors sought to explain differences in efficiency. The major determinants of efficiency differences were the type of dairy shed and the proportion of irrigated area. Walk-through sheds showed up poorly. Less predictably, farms with rotary sheds under-performed compared to herringbone sheds. Rotary sheds do not generate sufficient additional output relative to their capital cost. Some farmers fail to use their costly investments in rotary sheds to their full capacity.

While this is an interesting result that illustrates the financial significance of the timing of key decisions by farmers, and/or the cost some may be prepared to pay for additional convenience, the information is not much use to the individual farmer. The costs of long-lived fixed capital assets like milking sheds are sunk. There is not much that farmers can do if they have made the wrong decision because their timing was astray with respect to climate or prices.

Farm Financial Analysis by ABARE

The final paper discussed in these notes is a recent report by ABARE (2003) discussing the financial performance of dairy farms. Provisional data are presented for 2002-03, a poor year for the dairy industry when there was an 80 per cent fall in average farm cash income, the worst decline in over 25 years. Milk production fell by 8.4 per cent, the largest fall since 1951-52.

Farm cash income is gross receipts less production costs and interest costs. Farm cash income measures the funds available for consumption and farm investment. That is, it excludes depreciation and change in inventories. The latter items are included in the ABARE measure ‘farm business profit.’ Farm business profit fell dramatically in 2002-03 from an average for Australia of around $50,000 to minus $80,000, with increased costs and less cattle carried.

Although not directed at the productivity issue per se, this report is useful background. In common with other ABARE reports and agricultural commentary, this report emphasises the terms of trade of the dairy industry – output prices compared with input prices. While recognition of declining terms of trade puts the emphasis on productivity improvement as a means of maintaining farm incomes, the concept can be over emphasised.

Declining terms of trade is not just a feature of agricultural industries and is common to most goods vis-à-vis services. This is an underlying error in the material from Canada discussed in the Appendix. In part, the declining terms of trade is a statistical artefact or measurement problem for agriculture. The prices of standard products are measured in the numerator – wheat, milk etc – whereas many of the input prices measured in the denominator are subject to continuous quality improvement. In fact, it is the productivity gains implied by improved quality inputs that help offset the profit-reducing effects of rising costs and falling prices.

Other significant information presented for 2002-03 is the evidence of increasing debt. In recent years, debt has been 2-3 times farm cash income but increased to around five times in 2002-03.

Interestingly, this ABARE report in its discussion of earlier productivity studies observes (p.5):

Most of the increase in herd size and milk production in the three years before 2002-03 was achieved through increased grain feeding, but it is likely that farm managers’ skill in managing what is a more complex production system may take longer to develop.

A question that arises in discussion of financial performance in the Australian dairy industry is the reaction of dairy farmers to the generous adjustment package of July 2000 when the dairy market was deregulated. This has been analysed by Harris (2004). Survey results reported by Harris indicate that farmers remaining in the industry have adjusted in two main ways – increasing the number of cows and the amount of supplementary feeding. Some farmers used cash payments to assist exit from the dairy industry. Subsequent events suggest that they may have chosen the better option.

Other Studies

Fraser and Cordina (1999)

This study applied DEA to irrigated dairy farms in Northern Victoria. A major stimulus to the study was an impending shortage of water (subsequently experienced).

The superiority of DEA that accounts for the relationships between all inputs and outputs simultaneously over partial indicators of farm efficiency is asserted, and demonstrated. More controversial is the objective of helping extension efforts by identifying industry best practice through DEA.

This is because industry ‘best practice’ is an invalid concept. Each farm has to judge its management strategy subjects to its distinct resources, goals, skills and preferred risks.

In short, DEA involves estimating a frontier that envelops the input/output data with those farms lying on the frontier described as technically efficient. Conceptually, farms lying below could reduce input use and maintain output or maintain input use and increase output.

Data for the analysis were based on survey data for 50 farms in the central Goulburn Region in 1994/95 and 1995/96. Because the area is relatively homogeneous, the assumption is that the differences in technical efficiency are the result of managerial ability. The data consisted of a single output (milk solids) and six inputs (cow numbers, perennial pasture, irrigation water, supplementary feeding, fertiliser and labour).

A simple test was conducted on the partial productivity measures. Correlation coefficients between partial indicators varied significantly confirming that partial indicators are a poor guide to whole farm performance. DEA results suggested that many farms in the sample are operating at full technical efficiency, however the model is specified. Others are very close to technical efficiency. There is a tail of farmers who could save substantially on input use if they changed the way they operated to resemble more closely these ‘benchmark’ farms. As perhaps might be expected, tests for scale efficiency indicate that while some farms were operating below efficient scale, larger farms tended to exhibit decreasing returns to scale.

The authors do not consider why some farms are judged efficient and others are not. Socioeconomic differences are hypothesised. While this is an interesting paper and a pioneering application of this technique in Australia, perhaps the authors might have considered further the warning of Coelli et al. at page 181 “ standard DEA does not account for multi-period optimisation nor risk in management decision-making.”

This stricture also applies to the stochastic frontier method. Successful extension work needs to take into account financial variables that affect the growth and survival of the farm business in the long-term. Whether such individual financial analysis can be properly provided via group extension is debatable.                                                                                       

Graham and Fraser (2003)/Fraser and Graham (2004)

The two versions of this further DEA study are entitled ‘Scale Efficiency in Australian Dairy Farms’ and ‘Efficiency Measurement of Australian Dairy Farms: National and Regional Performance’ respectively.

The data were based on information on 1800 dairy farmers collected from a telephone survey by a market research company for the DRDC. The data were analysed for Australia as a whole and for the eight regions defined by the regional development programs of the DRDC.[4] Inexplicably, the survey did not collect information on farm labour. Not just for sophisticated techniques like DEA, this must restrict the usefulness of the survey in any application including simple tabulations of the survey data.

Coelli et al. (p.180) also warned that: “ the exclusion of an important input or output can result in biased results.” ABARE (2002) reported that labour was 18 per cent of input costs. In the event, this study set out in part “to illustrate how the choice of variables used in the analysis affects the efficiency estimates derived.” (Fraser and Graham, 2004, p.9).

The variable chosen for these tests was irrigation. As expected, the ranking of regions in technical efficiency changed dramatically. The Western region was the best performer with six inputs but ranked seventh (of eight) without irrigation. Murray jumped from sixth to second because these farms have the same output whether or not irrigation is included in the model. The authors do not draw the logical conclusion that the worth of their other empirical results is limited because labour is excluded from the analysis, perhaps arguing that cow numbers are a proxy for labour use. The authors found that many farms across all regions are operating below the optimal scale of production. With labour left of the analysis, their results should not be taken too literally.

Nevertheless, the paper demonstrates at least in principle that DEA can be applied to regional questions such as the overall performance of R&D and extension efforts in raising productivity and efficiency.

Graham (2004)

This paper is exploratory. It examines ways of accounting for environmental impacts of dairying. This is an appropriate objective given increasing demands on dairy farmers for improved environmental performance, especially limiting effluent run-off to waterways.

Graham identifies a variety of possible approaches. Following from Dutch research, environmental efficiency can be thought of as one aspect of technical efficiency that focuses on the input with detrimental environmental consequences. Alternatively, adverse environmental effects can be thought of as additional undesirable outputs. Work proposed for the Victorian dairy industry will embrace three categories of factors, inputs, desirable output and pollutants n the form of undesirable inputs. Nitrogen will be the main subject of interest.

Fraser and Hone (2001)

This paper investigated the use of farm level efficiency and productivity estimates for benchmarking studies based on panel data for the wool industry in south west Victoria. It was demonstrated that ranks between technically efficient and technically inefficient farms change substantially between years. This reinforces scepticism about benchmarking even with techniques like DEA or stochastic frontier analysis.

Concluding Comments

The economics of agricultural research and productivity measurement emerged as a specialised field of study following the finding by scholars at the University of Chicago in the 1940s that there was a large proportion of the increase in agricultural output that could not be explained by increases in conventional inputs of land, labour and capital. Much of the unexplained increase in output was attributed to agricultural research and its application by farmers. Studying the role of education and the enhancement of human capital in agriculture was another outcome of this research.

Productivity measurement is useful when thinking about an industry as a whole. Changes in industry profitability are the result of changes in productivity after allowing for the net effect of changes in prices and costs. Farmers have some influence over productivity and little over prices, hence the interest in productivity.

Productivity measures describe gains from technical change and improved management and efficiency. But productivity measurement cannot distinguish between the separate contributions of each component.

There is no good reason to dispute the initial ABARE finding that the productivity performance of the Australian dairy industry has been modest compared to the performance of other agricultural industries. But the estimates were an average with some farms doing better and others worse.

In part, the low productivity gains, as distinct from output, of the dairy industry in recent years are a product of its relative prosperity. Productivity increases fastest in recovery periods immediately following slumps. This is not a comforting observation.

There are measurement and conceptual problems in productivity analysis. In principle, the effects of weather could be accounted for by constructing weather indices. Some problems are less tractable. The value of land and capital items such as livestock depend on expected productivity gains as well as expected prices and profitability. This introduces circularity into productivity measurement.

Extensions of productivity analysis to DEA and stochastic production frontiers have had limited application to date. It is worthwhile knowledge to understand the dispersion of technical efficiency between firms in an industry. Information on scale is also useful. By including many important inputs and outputs in the analysis, these techniques attempt to overcome the inherent problems of partial productivity analysis.

However, less tangible or readily measured management inputs to production have to be left out of the analysis.

Furthermore, the search for ‘benchmarks’ that describe the path that individual farms should follow is likely to remain illusory. No approach to farm management analysis is credible that does not include financial and risk analysis of the individual farm business.

The search for benchmarks has at its core the false notion that there are a small number of best systems that all farmers could or should adopt. An alternative view is that there are as many best systems as there are farmers. In any case, the best use of resources changes over time.

Perhaps the most useful outcome of recent investigations of productivity in the Australian dairy industry will be final recognition that output per se should not be an objective in itself, for individual farmers or the industry as a whole.

Appendix

Dairy Industry Populism: A Canadian Diversion or Warning?

The Canadian and Australian agricultural economies have several things in common. Above all, as lands of recent European settlement and major agricultural exporting countries, farmers in both countries have had to cope with the vicissitudes of world trade in agricultural products. Both countries have efficient agricultural sectors in the sense that a high proportion of output is sold on world markets. The important role that export-led agricultural growth once played in economic development has left an enduring political legacy. Farmers and many others still attribute a special status to agriculture, even though the structure of society and the economy has changed radically.

The grain industry is particularly important in Canada. Traditionally, Australia’s comparative advantage was in grazing-based livestock. From time to time, both countries have intervened in agricultural markets to influence prices and control the way that agricultural products are marketed through Government-backed marketing authorities. Occasional periods of low prices and incomes and dissatisfaction with the costs and performance of agricultural marketing were the main stimulus to these interventions.

These beliefs are still embedded in community attitudes and some Government policies.

In recent years, the direction of policy has diverged in Canada and Australia.  Australia has reduced the level and diversity of assistance to agriculture. Marketing boards are no longer significant. This is less so in Canada where some agricultural products are still significantly regulated and protected (for example, eggs and dairy products).

An observation that follows is that if Canadian farmers are as dissatisfied with current conditions as is reflected in the report of the National Farmers Union, Australian farmers must be more tolerant or resilient. For better or worse, Australian farmers have coped with far greater recent changes to the policy environment.

Moreover, while the Canadian climate is harsh, it is less variable than that of Australia in the places where agriculture is practised. Poor seasons in Australia in the last few years have coincided with the period of policy change, compounding the adjustment problems faced by farmers.

The take home message from the Canadian report is that farmers are victims. Not just of economic circumstances, but of Government policy and lopsided relationships with input supplying and marketing firms. In particular, the report rails against the diagnosis that farms need to become bigger to maintain their incomes and sell on world markets.

By definition, larger farms mean fewer farmers. Without doubt, the efficiency message is sometimes oversold. Who survives in farming is of obvious concern to farmers. The impression is often left in naïve versions of the efficiency message that everyone can be a winner in agricultural adjustment. Ultimately, individual decisions to continue or exit farming are a matter of individual choice. If enforced by unanticipated events like commodity downturns or drought, the rate of change may be unacceptable.

Official or industry policies that promote agricultural expansion for all run the risk that many farmers will be disappointed, and disaffected.

There is a pervasive element of conspiracy theory in the NFU report. The following statements appear at page 4.

…when we analyze this prescription and look at the underlying premises, we find that this plan for restructuring agriculture based on competition and efficiency is constructed of myths and false assumptions – some would say “lies.”  This report examines the damaging myths that form the foundation of Canadian agricultural policy and similar policies around the world. This report also investigates who is propagating these myths and who is benefiting by short-circuiting our attempts to understand and remedy the crisis gripping our family farms and rural communities. (Italics in original.)

Although the NFU report discusses nine myths, these myths overlap and need not be discussed one by one. Early material presented in the report could be described as expressions of the ‘marketing margin’ or ‘farmer’s share of the consumer’s dollar fallacy.’

Although consumer prices of food have risen faster than farm prices in developed countries, this is not saying much. Consumers now purchase many more services with food – processing, transport from remote locations, variety, improved quality and so on. The demand for food at the farm-gate is not the same as the demand for food at retail. Taken on its own, the difference between farm and retail prices says nothing about competition and efficiency in agricultural marketing.

Retailing is a labour-intensive industry and costs have increased more in service industries than goods-producing industries. Higher wages reflect higher community living standards. The incentive of higher wages in the urban and service sectors creates the incentive for off-farm migration. Extra costs are incurred in providing additional marketing services.

It is misleading to imply that most of these extra costs are captured by the businesses providing the services. Whether profits are unreasonable ought to be assessed according to the turnover of these businesses, not their absolute size?

The NFU report asserts an inverse relationship between farm size (efficiency) and farm prosperity. There are empirical issues with Myth 4: ‘Economies of scale are the only way to gain efficiency.’ Even though there has been a substantial contraction in the number of farmers, the numbers are still large enough to ensure competition between farmers. The largest firms in agriculture do not represent a substantial share of output.

The extent of competition in the rest of the economy is more contentious. Even in the concentrated off-farm sector, there is effective competition in many capital-intensive industries with numbers of firms in the single digits. The more so when we consider the time dimension of competition. Trade practices law exists as an additional protection.

Cooperatives are common in dairy manufacturing and marketing, with surpluses returned to farmers in farm-gate returns. The profits of cooperatives are not a political issue for farmers. However, bundling of returns affects on-farm decision making. In most other agricultural industries, profits from processing and marketing are earned elsewhere.

The extent of oligopoly (small number of firms) market power is an empirical question. As such, the potential use of market power should not be ruled out of hand. Reference to transnational firms may appeal to Canadian national sentiment given the United States’ base of many transnational firms.

No doubt large enterprises practice discriminatory pricing, charging what the traffic will bear by adding a variable margin to costs according to the capacity of customers to pay. But occasionally that will work to the advantage of farmers. This is especially the case with an international market. There are many examples of fertilisers and other chemicals being supplied from developing countries. This is usually described as ‘dumping’ by transnational competitors.

The NFU report gets itself into a little bit of (acknowledged) trouble with Myth 5: ‘Technology will make farmers more efficient and prosperous’ with the disclaimer at page 12 “the preceding dismissal of technology as a farm-income enhancer is not a Luddite position. Nor is it blanket condemnation of technology.” Neither Canadian farmers, nor their counterparts, want to return to the physical demands of earlier farming methods. Research and development by input-supplying firms has made many modern developments possible. Much biological innovation is the product of public R&D.

This section of the report makes a plea for low-input and organic farming. Organic farming might have application on favoured soils but has limited scope in Australia. Again the claim is made that manufacturers capture all the benefits of technology, a dubious proposition with an international market for farm machinery and inputs.

There is a germ of truth in the discussion of Myth 7: ‘Farmers are producing too little… and too much.’ The authors are reluctant to concede the truth that farm prices decline in the long run. This is because increases in production allowed by the application of technology and improved management run ahead of the growth of the demand for food, which is constrained in Adam Smith’s memorable phrase by the capacity of the human stomach.

However, the NFU report is right to be sceptical about unsubtle exhortations by Government or industry for farmers to increase production, and exports. Decisions by farmers to produce more should be based on their own assessment of their resources and prices and costs, not boosterism. Furthermore, export targets are even more irrelevant to economic policy in a world of flexible exchange rates.

Farmers now make up about 3 per cent of the Canadian workforce, similar to the United States and a little below Australia. In many rich countries, including the neighbouring United States, the response to declining farm incomes has been to increase assistance to farmers. In effect, costs of assistance to a smaller agricultural sector make assistance more feasible because costs can be spread over a larger non-farm population and economy. This is the direction that the NFU and some in Australia would like to proceed.

However, there are dilemmas for countries like Australia and Canada with export-dependent agricultural systems where farming is often conducted in remote locations, many of which are a far cry from an idyllic rural dream. The long-term political constituency for supporting export-dependent agriculture is not the same as in some Western countries. Nor are farmers likely to put up with persistent low incomes in farming. Off-farm migration will remain important for agricultural adjustment if farmers are to maintain their incomes relative to the rest of the economy.

The NFU report provides interesting information on TFP for Canadian agriculture. As is the case in Australia, the Canadian agricultural sector shows out well in comparison with other sectors of the economy. The NFU report concludes with a claim for equity in incomes of farmers with others in the community. But productivity of the sector and profitability of individual farms are not the same thing. Furthermore, in modern mixed economies, income support should be a function for social security not agricultural policy.

Additional statements are made about environmental damage caused by modern farming techniques. Off-site environmental damage should be tackled (or not tackled) irrespective of farm productivity, profitability and efficiency. Especially in European countries, environmental damage has been used as a justification for assistance under the banners of ‘stewardship’ or ‘multifunctionality.’ The argument might have some currency in densely populated Europe, where intensive agriculture results in loss of amenity for the urban population. It is a harder argument to sustain for Canada, or Australia.

References

ABARE (2001a), Productivity in the Australian Dairy Industry, 1978-79 to 1998-99. Report to DRDC. (Prepared by T. Kompas, Nhu Che, S. McMeniman and D. Spencer).

_____ (2001b), Productivity in the Australian Dairy Industry: Supplement to the ABARE Report to DRDC, August. (Prepared by P. Knopke and V. O’Donnell).

_____ (2002), Production, Efficiency and Productivity of Australian Dairy Farms. Report to DRDC. (Prepared by T. Kompas and Nhu Che).

_____ (2003), Australian Dairy Industry: Productivity and Profit.

Coelli, T.J., Prasada Rao, D.S. and Battese, G.E. (1998), An Introduction to Efficiency and Productivity Analysis, Kluwer Academic Publishers, Massachusetts.

Farrell, M.J. (1957), ‘The Measurement of Productive Efficiency’, Journal of the Royal Statistical Society, Series A, CXX, Part 3, 253-290. 

Fraser, I. and Cordina, D. (1999), An Application of Data Envelopment Analysis to irrigated Dairy Farms in Northern Victoria, Australia, Agricultural Systems, 59, 267-282.

Fraser, I. and Graham, M. (2004), Efficiency Measurement of Australian Dairy Farms: National and Regional Performance, mimeo.

_____ and Hone (2001), Farm Level Efficiency and Productivity Measurement Using Panel Data: Wool Production in South West Victoria, Australian Journal of Agricultural and Resource Economics, 45 (2).

Graham, M. and Fraser, I. (2003), ‘Scale Efficiency of Australian Dairy Farms’, paper presented at the 47^th annual conference of the Australian Agricultural and Resource Economics Society Fremantle, February.

Graham, M. (2004), ‘Environmental Efficiency: Meaning and Measurement and Application to Australian Dairy Farms’, paper presented at the 48^th annual conference of the Australian Agricultural and Resource Economics Society, Melbourne, February.

Jackson, M. (2004), ‘Policy Based on Myths’, The Weekly Times, January 7.

Harris, D.N. (2004), ‘Producer Adjustment to Policy Reform: A Case Study on the Australian Dairy Industry’, paper presented at the 48^th annual conference of the Australian Agricultural and Resource Economics Society, Melbourne, February.

National Farmers Union (Canada) (2003), The Farm Crisis, Bigger Farms, and the Myths of “Competition” and ”Efficiency”, November 20, 2003.