Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Performance, Profit and Risk in Pasture-based Dairy Feeding Systems Findings from the TasMilk60 study Produced by Dairy Australia’s Grains2Milk program Performance, Profit and Risk in Pasture-based Dairying Systems – Findings from the TasMilk60 study © Copyright Dairy Australia 2011. All rights reserved. ISBN: 978-0-9871000-5-4 Publisher: Dairy Australia Acknowledgements: Dairy Australia’s Grains2Milk program gratefully acknowledges the many contributors to the TasMilk60 study: • The participating Tasmanian dairy farmers for the goodwill and support offered to the project team over three very different and sometimes trying years. • Gordon Cleary (Agrilink FarmStats), whose meticulous management of the data collection provided the high-quality dataset essential for the study. He provided much of the source material for this booklet. • John Morton (Jemora), whose statistical and analytical input was critical to extracting full value from the dataset. He also provided much of the source material for this booklet. • Pam Hartin (Agrilink FarmStats) for her diligent efforts in on-farm data collection. • Mark Fergusson and his team (Tasmanian Institute of Agricultural Research) for their efforts in on-farm data collection from farmers associated with the DairyTas Benchmarking project, Dairy Business of the Year (DBOY). • Courtney Gronow (Agrilink FarmStats), whose work with the Monte Carlo simulations contributed significantly to the risk assessment studies described in Section 4. • Those whose technical support strengthened the study outcomes: Brendan Cullen (The University of Melbourne), Karen Christie and Richard Rawnsley (TIAR), Don Thomson (Landscape & Social Research) and Warwick Waters (Waters Consulting). • Those who provided data and/or relevant commentary to the project team, including: Andrew Box and staff (Fonterra), Darren Smart (Cadbury), National Foods, Keith Davis (Impact Fertilisers), Brett McGlone (Incitec Pivot), Trevor Macleod (Tasmanian Stockfeed Services), Hugh McMullan (Animal Mineral Solutions), Andrew Radford (John Lawrence), Aaron Robertson (WHK Pinnacle), Peter McBain, Dan Huggins (Maxi Cow Consulting), Andrew Angelino (Andrew Angelino Consulting ), Peter Harrisson (Farm Mapping Services) and Laurie Hooper (Agritech). This project was supported by funding from Dairy Australia and the Dairy Service Levy. Enquiries: Steve Little Grains2Milk program leader for Dairy Australia Locked Bag 104, Flinders Lane, Victoria 8009 e-mail: [email protected] web: www.dairyaustralia.com.au Copyright permission: If you wish to reproduce information contained in this booklet, contact Grains2Milk program leader, Steve Little, or Dairy Australia’s Corporate Communications team. This booklet is published for your information only. It is published with due care and attention to accuracy, but Dairy Australia makes no warranty with regard to the accuracy and reliability of the information provided, and accepts no responsibility for loss arising in any way from or in connection with errors or omissions in any information or advice or use of the information. The information is a guide only and independent professional advice should be sought regarding your specific circumstances. Table of Contents Foreword .............................................................. 3 How to use this booklet ....................................... 4 Quick Quiz – True or False? ...................... fold out Executive summary .............................................. 5 4. Risk – Insights from TasMilk60....................... 39 Risk and uncertainty................................................... 39 Risk management ...................................................... 39 Risk measures ........................................................... 40 Monte Carlo simulation study using TasMilk60 data..... 40 Where to from here? .................................................. 45 1. The TasMilk60 study......................................... 7 5. Other uses of TasMilk60 data to date............ 47 How TasMilk60 was done ............................................ 7 DairyMod pasture study.............................................. 47 The TasMilk60 farms................................................... 14 Greenhouse gas emissions study................................ 49 2. Farm performance – Findings from TasMilk60.....17 6. Conclusions ................................................... 51 Pasture utilised per hectare ........................................ 17 Total feed intake per cow............................................ 17 Milk yield and composition.......................................... 18 Feed conversion efficiency.......................................... 19 7. Appendices .................................................... 53 Tips and traps when collecting farm data (Gordon Cleary) .......................................................... 53 Operating costs.......................................................... 22 Six principles for identifying determinants of profitability by analysing data from groups of dairy farms (John Morton) .......................................... 54 Change in physical and financial aspects year to year................................................................ 24 8. References ..................................................... 57 Income and milk price................................................. 21 3. Farm profitability – Findings from TasMilk60.... 25 Associations between profitability measures................ 25 Farm profitability for each study year........................... 26 Determinants of profit within each study year............... 27 Change in profit due to management changes............ 30 Effect of enterprise size on farm profit ........................ 33 Degree of consistency of profit outcomes achieved year to year................................................................ 34 Determinants of consistent relative profit performance over the three study years........................................... 36 Correlation between farm profitability and ‘farming style’............................................................. 37 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 1 2 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Foreword Soon after Dairy Australia’s Grains2Milk program commenced in 2007 a need was identified for a comprehensive, multi-year dataset of farm performance across a spectrum of farms using different feeding strategies within, as best was possible, the same climatic and market context. The farm dataset was intended mainly for internal use by Dairy Australia and its research, development and extension partners to enhance the industry’s understanding of the interaction between grain/ concentrate input, feeding system performance, profit and risk in pasture-based dairy feeding systems. The results of this Grains2Milk farm monitoring study needed to be relevant to all dairying regions. However it was necessary to select one region from which to collect the dataset to ensure the same climatic and market context. Tasmania was selected for the study because it offered two important features. First, its dairy industry is predominantly a pasture-based one and has been less affected by drought than mainland dairying regions in recent years. Second, the Tasmanian dairy industry has a wide range of feeding approaches. While many farmers choose to operate a low input/low production system with little or no grain/concentrates, in recent years others have adopted the moderate to high grain / concentrate feeding levels typically seen in mainland dairying regions. This divergence made it possible for Grains2Milk to collect physical and financial performance data across a wide range of feeding approaches, extending to the more extreme ends of the performance spectrum than would be available in any other Australian dairying region. This booklet summarises the findings from the TasMilk60 study. It explains the methods we used to analyse the dataset, the key findings and what these mean for industry. While confirming many widely held views, these findings also dispelled several myths, and provide new insights into farm performance, profitability and risk, and how they inter-relate. The booklet will help farmers and their advisers make feeding system decisions relevant to their own situations with less imperfect knowledge. We hope you find the booklet valuable. Steve Little Grains2Milk program leader for Dairy Australia Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 3 How to use this booklet You will get the most out of this booklet if you firstly take 10 minutes to complete the Quick Quiz. To see how many statements you correctly answered True or False, open the fold-out panel. Follow the signpost to the specific page in the body of the booklet that provides the supporting evidence from the TasMilk60 study for each answer. See Section 1 for information about how the TasMilk60 study was done and a summary of the physical features and financial characteristics of the farms that were studied. This booklet can be used as a personal reference. However, it may also be a useful resource to support group discussions among farmers and advisers on aspects of farm performance, profitability and risk. The Table of Contents will also guide you to particular aspects of farm performance, profitability and risk you are interested in which are discussed in Sections 2, 3 and 4. 4 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Quick Quiz – True or False? Read each statement below and circle whether you think it is true or false. When you have completed the quiz, open the fold-out panel to see how many statements you correctly answered. Follow the signposts to the supporting evidence from the TasMilk60 study in Sections 2, 3 and 4 of this booklet. Statement True or False? (Circle below) 1. Farm performance Answer Go to: 1. Farm performance a. Farms that feed higher levels of concentrates per cow per year achieve poorer pasture utilisation (tonnes / hectare / year). True False False Despite fears of substitution, there is no simple relationship between the level of concentrates fed per cow and pasture utilised per hectare. Good and poor pasture utilisation occurs at all levels of concentrate feeding. b. Farms that feed lower levels of concentrates per cow get their cows to eat more pasture (tonnes / cow / year). True False True However, there is a larger variation in pasture intake per cow between farms feeding the same level of concentrate than there is between farms feeding different levels of concentrate, so herd and grazing management are critical. c. The milk protein to fat ratio does not vary between farms feeding different levels of concentrates per cow. True False False Milk protein to fat ratio is higher on farms where higher levels of concentrates are fed. This is largely due to the reduced fat concentrations found in milk with higher daily milk volumes per cow. p18 d. Feed conversion efficiency (FCE) is generally better on farms where higher levels of concentrates per cow are fed. True False True This is largely due to the higher total feed intakes on high concentrate-feeding farms, and hence, the greater proportion of nutrients that are used for milk production and the lesser proportion used for maintenance. pp19-20 e. Most pasture-based dairy farms feeding one to two tonnes of concentrates per cow in the bail achieve about the same feed conversion efficiency (FCE). True False False Feed conversion efficiency (FCE) is highly variable at all levels of concentrate feeding. pp19-20 f. Changing to a higher level of concentrates per cow, if well managed, should guarantee higher milk production and improved feed conversion efficiency (FCE) within one year. True False False When a dairy farm changes its production / feeding system, adjustments are needed (cows, people, farm infrastructure) to realise the full benefits. It may take several years before the benefits are fully realised. p20 g. Increasing feed conversion efficiency (FCE) within a farm’s chosen feeding system can be a powerful lever for increasing farm profit. True False True This is particularly so in higher milk price years, when every 0.1 increase in energy-corrected litres of milk / kilogram feed DM may be worth $200+ extra milk profit per cow per year. p20 h. Dairy farmers have little control over the price they are paid for their milk. True False False Prices paid to farmers (even those supplying the same manufacturer) vary widely due to on-farm factors including milk volumes, milk fat and protein concentrations and seasonality of milk supply pattern. p21 i. Herd, shed and overhead costs are the most important operating costs to control in a dairy farm business. True False False Feed costs, and labour and management costs are generally the highest proportions of operating costs and vary widely between farms. Conversely, herd, shed and overhead costs are relatively small components of operating costs and are less variable between farms. 2. Farm profitability p17 pp17-18 pp22-23 2. Farm profitability a. The best measure of farm profitability to use when comparing farms with different land and herd resources is ‘Milk EBITD per hectare’. True False False In most farm populations, milk EBITD per cow, hectare, litre, kilogram milk solids, and return on capital (based on milk EBITD) are closely correlated with each other, so are likely to tell the same story about farm profitability. p25 b. There is no ‘best’ concentrate feeding level or production / feeding system. True False True Any concentrate feeding level or production / feeding system can be profitable in any year, given an appropriate mix of management, milk price and input costs. p26 c. There are large variations in profitability between farms at each level of concentrate feeding (low / moderate / high), and from year to year. True False True The differences in average or median profits achieved between farms using low / moderate / high concentrate feeding levels are small compared with the variability in profit within each concentrate feeding level. p26 d. With feed being the greatest operating input cost on any dairy farm, low feed costs per cow are essential to achieve high farm profitability. True False False Dairy farms are complex systems, and there are many ways to make a profit (or a loss). Determinants of profit should be assessed collectively, not separately when analysing farm performance. Income from milk sales is an important component of profit as are costs – higher feed costs may be justified if they generate extra profit by lifting milk income. pp27-29 e. In high milk price years, most dairy farms actively pursue higher profit by implementing major management changes. True False False Most farms take a ‘business as usual’ approach in high milk price years and don’t take opportunities to substantially increase farm profit above what the milk price alone delivers. Few farms make large systemic management changes year to year. pp30-32 f. Milk EBITD per cow and per hectare increase markedly with increased enterprise size (more cows, hectares or total milk production). True False False Any beneficial effects of ‘dillution’ of costs per cow with more cows are not large. Increases in total farm profitability with increased enterprise size are likely to be due to scale, not efficiency. p33 g. The most profitable dairy farms are consistently profitable year after year. True False False Relative farm profitability (high or low) is not very repeatable from year to year, so results in a single year may not reflect profit performance over the longer term. h. Farms that have consistently higher profits usually have relatively higher milk price and milk yield per cow, lower pasture and fodder costs, and lower labour and management costs. True False True However, they tend not to be outstanding performers for any of these particular determinants of profit – they tend to be consistent all-rounders whose efforts for these profit determinants collectively are superior. p36 i. Farmers with a wide range of management styles and sets of attitudes and beliefs run successful dairy farm businesses. True False True Farmers don’t require a certain management style or set of attitudes and beliefs to be highly profitable. p37 3. Risk pp34-36 3. Risk a. Something that is risky is best avoided. True False False Attempting to avoid all risk is often futile or counter-productive. Instead, risk exposure can be reduced by spreading, selling, or shifting risk or by risk averaging. p39 b. For the dairy farmer, there is always a trade-off between risk and reward. True False True Farmers with a low risk tolerance will seek options where little risk is involved and require a very high reward for the risk involved. Those with a higher tolerance for risk will be willing to accept risk without such a big potential reward. p39 c. If you want to minimise risk, aim to maximise the amount of pasture in the diet and feed a low level of concentrate. True False False For all production / feeding systems, there is a mix of risk, performance and management principles whose understanding can improve the chances of successfully balancing risk and reward. p39 d. A farm’s unit cost of milk production and operating profit margin are important risk measures for a dairy farm business. True False False These are economic measures, not risk measures. They do not provide information about the probability of favourable and unfavourable outcomes. p40 e. Pasture utilisation, pasture quality and core costs per cow are key profit drivers in all pasture-based production / feeding systems, regardless of the level of concentrates fed. True False True These three on-farm profit drivers can have a greater impact on profit variability than the off-farm profit drivers of milk price, concentrate price and purchased fodder price over which farmers have less control. pp40-44 Executive summary The TasMilk60 study was an observational study done to better understand the interaction between grain/ concentrate input, risk, management skills and profit in pasture-based dairying systems. It involved the collection and analysis of a comprehensive dataset of farm physical and financial performance over three years (2006/07, 2007/08, and 2008/09) across a spectrum of farms using different grain/concentrate feeding rates in the same climatic and market context. We chose to do the study in Tasmania for two main reasons. First, it has been less affected by drought than mainland dairying regions in recent years. Second, Tasmania enabled Grains2Milk to collect physical and financial performance data across a wide range of pasture-based feeding approaches, extending to more extreme ends of the performance spectrum than would be available in any other dairying region of Australia. The study enrolled: • 21 farms that were low concentrate feeders (<1t/cow/year) • 27 farms that were moderate concentrate feeders (1 to <2t/cow/year) • 21 farms that were high concentrate feeders (≥2t/cow/year) The study did not aim to enrol farms in proportion to the distributions of Tasmanian or Australian farms by concentrate feeding category. Similar numbers of farms in each of the three categories were enrolled to allow much more precise statistical analyses to be performed. Data were collected from managers of selected farms by either Agrilink FarmStats P/L (AGFS) or the Tasmanian Institute of Agricultural Research (TIAR), and from relevant milk supply companies, accountants, stockfeed and fertiliser suppliers with the manager’s permission. Data collection was managed by Agrilink FarmStats. All statistical analyses were performed by epidemiologist John Morton (Jemora P/L). The three years of the TasMilk60 study – 2006/07, 2007/08 and 2008/09 – were one of the most volatile periods in terms of trade seen for many, many years, and in times of less favourable climatic conditions. As such, the findings of how well farmers using different feeding approaches responded and reacted to the challenges thrown at them, particularly over the full three-year timeframe rather than just a single year, provide a telling testament to what worked and what did not, and lessons relevant for dairy farmers and advisers managing pasture-based feeding systems across Australia. The findings from Grains2Milk’s TasMilk60 study support the following conclusions: 1. Farm performance Despite fears of substitution, there is no simple relationship between the amount of concentrates fed per cow and pasture utilised per hectare. Both good and poor pasture utilisation is seen at all levels of concentrate feeding. Farms that feed lower levels of concentrate per cow get their cows to eat more pasture. However, there is a larger variation in pasture intake per cow between farms feeding the same level of concentrate than there is between farms feeding different levels of concentrate, so herd and grazing management are critical. Milk protein to fat ratio is generally higher on farms where higher rates of concentrates are fed. This is largely due to the reduced fat concentrations found in milk with higher daily milk volumes per cow. Feed conversion efficiency (FCE) is highly variable at all concentrate feeding levels. It is generally better on farms where higher amounts of concentrates per cow are fed. This is largely due to the higher total intakes on high concentrate feeding farms and, hence, the greater proportion of nutrients that are used for milk production and the lesser proportion used for maintenance. Total feed intake per cow explains more than half the variability in FCE. However, there are also several other important factors known to help optimise FCE, including maintaining high feed quality and good rumen function, and minimising feed gaps, feed wastage and energy losses. Increasing FCE within a farm’s chosen feeding system can be a powerful lever for increasing farm profit, particularly in higher milk price years. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 5 When a dairy farm changes its production / feeding system, adjustments are needed (cows, people, farm infrastructure) to realise the full benefits. It may take several years before the benefits are fully realised. Milk prices paid to farmers (even those supplying the same manufacturer) vary widely due to on-farm factors under their control, including milk volumes, milk fat and protein concentrations, and the seasonality of milk supply pattern. Farmers aiming to increase profitability by reducing costs should review and monitor feed costs, and labour and management costs, as these generally constitute the highest proportions of operating costs and vary widely between farms. Conversely, herd, shed and overhead costs are relatively small components of operating costs and are less variable between farms. 2. Farm profitability In farm populations such as that examined in the TasMilk60 study, milk EBITD per cow, hectare, litre and kilogram milk solids, and return on capital (based on milk EBITD) are closely correlated with each other, so are likely to tell the same story about farm profitability. Most farms maintain a ‘business as usual’ approach in high milk price years and do not take opportunities to substantially increase farm profit above what the milk price alone delivers. Few farms make large systemic management changes year to year. Beneficial effects on farm profitability of increasing enterprise size are generally relatively small and are likely to be due to scale, not efficiency. Relative farm profitability is not very repeatable when milk prices fluctuate widely year to year, so results in a single year may not reflect profit performance over the longer term. Farmers do not require a certain management style or set of attitudes and beliefs to be highly profitable. 3. Risk Risk and uncertainty are central to all dairy farming decisions. Attempting to avoid all risk is often futile or counter-productive. Instead, risk exposure can be reduced by spreading, selling or shifting risk, or by risk averaging. There is no ‘best’ concentrate feeding level or production / feeding system. Any concentrate feeding level or production / feeding system can be profitable in any year, given an appropriate mix of management, milk price and input costs. For the dairy farmer, there is always a trade-off between risk and reward. Those with a low risk tolerance will seek options where little risk is involved and require a very high reward for the risk involved. Those with a higher tolerance for risk will be willing to accept risk without such a big potential reward. The differences in average and median profits achieved between farms using low, moderate and high concentrate feeding levels are small compared with the variability between farms within each concentrate feeding level. For all production/feeding systems, there is a mix of risk, performance and management principles whose understanding can improve the chances of successfully balancing risk and reward. Dairy farms are complex systems and there are many ways to make a profit (or a loss). Determinants of profit should be assessed collectively, not separately, when analysing farm performance. Farms that have consistently higher profits usually have a relatively higher milk price, higher milk yield per cow, lower fodder costs, and lower labour and management costs. However, they tend not to be outstanding performers for any of these particular determinants of profit – they tend to be consistent all-rounders whose efforts for these profit determinants collectively are superior. Income from milk sales is as important a component of profit as are costs – higher feed costs may be justified if they generate extra profit by lifting milk income. Risk measures should provide the probability of unfavourable and favourable outcomes occuring. Unfortunately, many so-called risk measures currently used in the dairy industry do not. These erroneous risk measures include pasture as a percentage of total feed consumed, unit cost of milk production and operating profit margin. Pasture utilisation, pasture quality and core costs per cow are key profit drivers in all pasture-based dairy feeding systems, regardless of the level of concentrates fed. These three on-farm profit drivers can have a greater impact on profit variability than the off-farm profit drivers of milk price, concentrate price and purchased fodder price, over which farmers have less control. Steve Little Grains2Milk program leader for Dairy Australia 6 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Section 1 The TasMilk60 study This section explains how the TasMilk60 study was done and describes physical features and financial characteristics of the farms that were studied. TasMilk60 was purely an observational study. No attempt was made to influence the physical and financial performance of the study farms. Findings were based entirely on natural variations in these, between farms and farm years. Three remarkable years to study farm performance, profit and risk The three years of the TasMilk60 study – 2006/07, 2007/08, and 2008/09 – were remarkable in many ways. Seasonal conditions Seasonal conditions were generally drier than is usual for most of northern Tasmania during these three years (Figure 1). Only Smithton, in 2008/09, exceeded the decade average rainfall, with Deloraine and Scottsdale receiving generally below average rainfall in all three years. Milk prices Market conditions were highly volatile over the threeyear study period, as reflected in the milk price table below (Table 1). After being paid an average milk price in 2006/07, farmers enjoyed the highest milk price ever in 2007/08. However, 2008/09 saw the first milk price step-down in more than 30 years, due to the flow-on of the global financial crisis to the export-orientated milk companies. Announced in December and effective from February, this created substantial financial pressures for many farmers, including debt pressures from new working capital requirements. 04/05 05/06 06/07 07/08 08/09 25.9 27.2 30.9 33.6 36.5 50.2 41.3 $/kg/MS 3.43 3.54 4.05 4.39 4.79 6.63 5.40 Grain prices As shown in Figure 2, the three years of the TasMilk60 study coincided with an extended period of relatively high cereal grain prices. Like milk prices, grain prices peaked in 2007/08 with the global financial crisis. Cumulative monthly rainfall TasMilk60 was a retrospective/prospective longitudinal study. Annualised physical and financial data were collected from 69 convenience sampled Tasmanian dairy farms for between one and three consecutive financial years from 2006/07 to 2008/09. 02/03 03/04 ¢/litre Cumulative monthly rainfall Study overview Table 1: Milk price history – Tasmania (Source: DA). Year Cumulative monthly rainfall How TasMilk60 was done 1200 1000 Seasonal rainfall – Smithton TasMilk60 study 800 600 400 200 0 1400 1200 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Seasonal rainfall – Deloraine TasMilk60 study 1000 800 600 400 200 0 1200 1000 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Seasonal rainfall – Scottsdale TasMilk60 study 800 600 400 200 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Figure 1: Cumulative monthly rainfall (Source: Bureau of Meteorology). Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 7 Other farm inputs Many other input costs, particularly fertiliser and fuel, escalated substantially in 2007/08, the year of the highest milk price seen, as shown in the index of farm input costs below (Figure 3). As a result, for many farmers the increased income arising from the higher milk price was offset by the increased overall cost of providing feed for their cows. The milk price downturn in 2008/09 coincided with an improvement in fertiliser costs, but trading conditions again clearly put the different feeding approaches in use among TasMilk60 farms to the test. Summary In summary, the TasMilk60 study was conducted during one of the most volatile trade periods seen for many, many years, and in times of less-than-favourable climatic conditions. As such, the findings of how well farmers using different feeding approaches responded and reacted to the challenges thrown at them, particularly over the full threeyear timeframe rather than just a single year, provide a telling testament to what worked and what did not. Farm selection Data were collected from the managers of selected farms by either Agrilink FarmStats (AGFS) (46 farms) or personnel from the Tasmanian Institute of Agricultural Research (TIAR) (23 farms). AGFS farms were selected from those using consulting nutritionists, Andrew Angelino (Andrew Angelino Consulting) and Daniel Huggins (Maxi Cow Consulting), and TIAR farms were selected from those participating in the Dairy Business of the Year benchmarking event in Tasmania. $500 Barley Triticale Wheat This farm selection process was chosen so that approximately similar numbers of farms where <1t dry matter of grain/concentrates, 1t to <2t and ≥2t were fed per cow per year would be enrolled. This was desirable to maximise precision when analysing data within each of these three concentrate feeding categories. The study did not aim to enrol farms in proportion to the distributions of Tasmanian or Australian farms by concentrate feeding category. Data collection Data collection was managed by Agrilink FarmStats (AGFS). Farm data were collected by AGFS or TIAR personnel from managers of selected farms using a joint data collection instrument developed by AGFS specifically for the project in conjunction with Mark Fergusson of TIAR. More detailed data were required than are routinely collected with benchmarking programs such as AGFS’s MilkMark®, or Red Sky,® as used in the Dairy Business of the Year (DBOY) event. For most farm years, data were collected within 6-12 months of the end of each financial year, although for a small number of farm years the data were collected more than 12 months after the end of the relevant financial year. While a great deal of the data was obtained from the participating farm managers, some were collected from the relevant milk supply company, accountant, stockfeed or fertiliser supplier or other sources with the manager’s permission. Physical and financial data were recorded and analysed separately by financial year for each study farm. Customised spreadsheets were developed for handling field data on-farm. Data collectors had the choice of using a hard-copy version of the survey instrument or working through the electronic version on-screen. 140% TasMilk 60 study 120% $400 Price index vs. 2000 $300 $200 $100 100% 80% TasMilk 60 study Electricity Feed, grazing, cultivation and harvesting Fertiliser, lime and seeds Fuel Wages and salaries Overall farm expenses 60% 40% 20% 9 Ja 6 n 9 Ja 7 n 9 Ja 8 n 9 Ja 9 n 00 Ja n 0 Ja 1 n 02 Ja n 0 Ja 3 n 0 Ja 4 n 0 Ja 5 n 0 Ja 6 n 0 Ja 7 n 0 Ja 8 n 0 Ja 9 n 1 Ja 0 n 11 n Ja Ja n 95 $0 Figure 2: Cereal grain prices, ex-Melbourne (Source: AWF). 8 0% 2001 2002 2003 2004 2005 2006 2007 2008 Figure 3: Index of farm input prices, base = year 2000 (Source: DA). Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study On-farm sessions inevitably raised data issues that required further clarification and follow-up. Given that the TasMilk60 study required a high-quality dataset, extraordinary efforts were made by AGFS to validate data with help from participating farmers and many others (see acknowledgements on the inside front cover of this booklet). Farm physical data Milk outputs are easy to find, as they are recorded by the receiving milk factory. Various source documents help record what has happened to physical inputs once these arrived on-farm. These include grain, hay or silage invoices of purchase, contractors’ invoices for fodder conservation and fertiliser delivery dockets. However, it is tired farmers who are often the only source of hard data about these. Given its focus on feeding approaches, the TasMilk60 study needed accurate information about cow feeding, so that accurate pasture back-calculations could be carried out. As a result, farmers were asked questions about their feeding approach from different angles to help verify what was fed to which livestock. For example, invoices for grain might show 765 tonnes purchased. The farmer’s daily feeding rate might be nominated at 6kg/day/cow for a 400-cow herd averaging 300 day lactations. Thus, 720 tonnes has purportedly been fed to the milking herd, leaving 45 tonnes to be accounted for, usually through the feeding of dairy replacements. When fertiliser was applied to areas used for potatoes, fodder crops, poppies or vegetables, both the quantity and cost was isolated from the dairy enterprise. Per cow inputs and outputs are central to exercises such as TasMilk60. The effective herd size number to use as the denominator is often subject to negotiation, as issues such as attrition and culling, peak cows and carried-over cows. These were discussed and resolved with each farmer in the TasMilk60 study. The other key denominator, hectares of milking area, is usually a firmer number to define, although seasonal conditions can change how some land is used over the season. That non-milking area on the hillside was defined as a milking area if a proportion of the milking herd spent time on the hill. Farmers were asked to estimate the average body weight of the herd when in early lactation and average changes in body condition from the start to the end of each year. However, these data were not used in statistical analysis of determinants of profitability due to concerns about estimation errors and lack of standardisation between assessors. Fodder reserves carried over between financial years represented both an inventory issue in respect of pasture utilisation in the year if such reserves were fed out and also a cost issue if the cost was effectively carried over into the next year. Special-purpose validation tools were developed by AGFS for fodder inventory. Stock inventory changes also required careful consideration. We included stock sales income in dairy income (i.e. milk plus stock sales plus other dairy income), but did not include costs of purchased cattle as operating costs. This methodology could affect comparisons across farm years where some farms sell unusually high numbers of cattle, and/or receive breeding cattle rather than chopper prices, and where some farms purchase breeding cattle. As explained in Section 3, in TasMilk60 milk and dairy EBITD were closely correlated (Figure 25), demonstrating that these effects were minimal for the study farms. However, to minimise these effects, we focused on milk EBITD rather than dairy EBITD throughout. Fodder wastage was not measured. A default wastage figure of 20% of quantity feed (offered) was assumed for both homegrown and purchased hay and silage across all farms and used in all energy computations. The physical data collected needed to include those inputs necessary for pasture consumption to be estimated by the Department of Primary Industries Victoria (DPIV) Pasture Consumption & Feed Conversion Efficiency Calculator (2010) and those inputs necessary for the estimation of pasture growth using DairyMod (as described on pages 47-48). Examples of the type of measurements required were: • For pasture growth: – GPS identification of the dairy farm; – climate data from the Bureau of Meterology (BOM) and/or SILO database; – soil types; – N fertiliser applications; – pasture types; and – usual irrigation protocol. • For pasture consumption: – amount and estimated ME of supplementary feeds fed during the period; – milking area; – cow details, including estimated body weight and condition; – proportion of first calvers; Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 9 – numbers of productive and non-productive animals on the farm over time; and – daily distance walked by cows and type of terrain. Farm financial data Farm financial data are generally easier to acquire than physical data, with manual or computerised farm accounting systems and many source documents available, including cheque books, bank statements, invoices, receipts, tax returns, wages book, etc. However, financial data collection still presents challenges. Of principal concern in identifying revenue and expenses within a given financial year is whether accounts have been prepared on a cash or accrual basis. Cash is often the visible tip of the iceberg, with accruals the unseen bulk below. TasMilk60 was conducted on an accrual, not cash, basis. • does not include income from stock sales and other dairy income; and • includes non-cash costs (imputed labour and management). Farm EBITD is a ‘whole of farm’ measure of profitability, but is inappropriate for comparisons between farms of different sizes, as it can increase markedly with farm size, as reflected by number of cows, hectares and farm milk production. So, to account for farm size, milk EBITDs per cow and hectare were used extensively in this study; milk EBITDs per litre and kilogram of milk solids were also used. Hard copies of the relevant financial statements for the most recent financial year were sought by the AGFS or TIAR personnel charged with data collection. By choice, the process placed emphasis on the farmer’s own management accounts, rather than his accountant’s annual tax summary, because the former was likely to be more in tune with practical farm management (and the needs of the study) than the accountant’s taxbased chart of accounts. Analyses of study data indicates that any bias due to costs not increasing proportionately with farm size is probably quite small, and certainly far less than the effect of farm size on farm EBITD, EBITD per cow, hectare, litre and kilogram milk solids. Accordingly, these measures were superior to farm EBITDs when making profitability comparisons between farms of different sizes. For example, tax returns frequently show between four to six separately described Repairs and Maintenance cost accounts – Repairs (tracks), Repairs (fences), Repairs (Irrigation), Repairs (Dairy), Repairs (Structures), etc. It is not unusual to see $60,000 worth of expenditure spread across these detailed accounts. However, this flight to detail can be neutralised entirely if $400,000 worth of grain, $100,000 worth of hay and $80,000 worth of silage-making are bundled together into a single ‘Fodder’ cost account. Disaggregation is a time-consuming activity. • The approach taken to stock trading was to include stock sales as dairy income, stock purchases as capital expenditure (rather than as operating costs), and replacement costs as operating costs, given they are necessary for ongoing operation of the farm enterprise. These decisions were irrelevant to most of this study as most analyses are based on milk EBITD rather than dairy EBITD. Financial categories and calculations The income and operating cost categories used in TasMilk60 are shown in Figure 4, and categories of non-operating costs, and assets and liabilities in Figure 5. Methods of calculating various measures of farm profitability are shown in Figure 6. Since the principal focus of the TasMilk60 study was to gather information about the financial performance of different feeding approaches, it was decided to focus on Milk EBITD as the key profit parameter, rather than Dairy EBITD, which includes income from stock sales and other dairy income. 10 Milk EBITD stands for Earnings Before Interest, Tax and Depreciation. It is calculated by subtracting dairy operating costs (herd + shed + feed + labour and management + overheads) from milk income. As such, Milk EBITD: Note: • Changes in fodder inventory were accounted for by including fodder costs only for fodder consumed in the particular farm year. Average unit costs were used. • Changes in grain inventory were disregarded as most farms had little scope for storing substantial quantities of grain. • Similarly, changes in fertiliser inventory were disregarded, as few farms would have purchased and stored fertiliser on-farm for the following financial year. • Advance payments for grain and fertiliser were attributed to the year in which these inputs were used, rather than to the year of purchase. Statistical analysis All statistical analyses described in this report were performed by epidemiologist John Morton (Jemora P/L), using data supplied by AGFS for annual periods (2006/07, 2007/08 and 2008/09). The farm year was the unit of analysis for almost all analyses. A ‘farm year’ consisted of data from one farm for a single year. A farm enrolled in all three years of the study contributed three farm years of observations. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Figure 4: Income and operating cost categories. OPERATING INCOME Milk income Dairy Income Stock sales Gross milk income less levies & volume charge Other dairy income OPERATING COSTS Herd costs Mature stock costs Animal health, identification, herd recording, dogs, sundries Replacement costs Artificial breeding, agistment, animal health, calf rearing supplies, feeds & fodder, identification Dairy supplies & rubberware, chemicals & sanitation, electricity, garbage disposal Shed costs Feed costs Labour & management Overhead costs Milker supplements Concentrates (protein grain/meals, other grains, minerals, additives), by-products, purchased hay & silage Pastures (inc forage crops) Fertiliser & spreading, land lease, soil testing, effluent ground-working, sprays (weeds & pest), seed, diesel, fuel & oil Fodder & irrigation Milking area fodder, agistment for mature stock, outpaddock, irrigation power & fuel, water licence & rates, temporary water purchase, water monitoring Labour costs Gross wages (incl PAYG), allowances, Workcover, superannuation, training courses, protective clothing, staff amenities Imputed management & labour Repairs & maintenance Dairy, drains & lanes, fences & gates, irrigation, machinery, stock water, structures, vehicles, workshop & tools Other overheads General freight & cartage, donations, equipment hire, farm vehicles, farm insurance, memberships & subscriptions, shire rates, travel expenses Administration Accountant, consultants, legal fees, office supplies, telephone/fax & Internet Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 11 Figure 5: Categories of non-operating costs, and assets and liabilities. NON OPERATING COSTS Overdraft interest, loan interest, lease charges, HP charges, bank charges, cow leases Finance costs Personal costs Drawings Principal Co-op shares Capital costs Land & water Livestock Machinery ASSETS & ASSETS & LIABILITIES LIABILITIES Land & improvements Dairy assets improvements Livestock Machinery Dairy liabilities 12 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Figure 6: Methods of calculating various measures of farm profitability (EBITD = earnings before interest, taxation and debt repayment; *capital costs are restricted to those for the dairy enterprise). Milk income Operating costs Milk EBITD Dairy assets Milk return on capital Dairy assets less dairy liabilities Milk return on equity Dairy assets Dairy return on capital Dairy assets less dairy liabilities Dairy return on equity Stock sales and other dairy income Dairy EBITD Non-cash items (Imputed labour & management) Dairy cash surplus Finance, personal & capital costs * Change in overdraft & cash reserves Change in asset value Change in net dairy worth Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 13 Physical features The TasMilk60 farms Predominant breed Number of farms Sixty-nine farms were studied for between one and three years. Fifty-six farms were studied for all three years. The predominant breed was Holstein-Friesian on almost 60% of the study farms and Holstein-Friesian X Jersey cross on a further 26% of farms. Herds where cows were fed more concentrates were more likely to be Holstein-Friesian and less likely to be cross-bred. Cow bodyweight Grain / concentrate feeding categories Farms were categorised by the amount of grain/ concentrate (‘concentrates’) fed to the cows in each year. The categories used were less than 1t DM fed per cow per year, 1t to less than 2t, and 2t or more (throughout this report, concentrate feed amounts are described as tonne DM, rather than on an ‘as fed’ basis, unless otherwise stated). Study farms were distributed approximately evenly across all three categories (Table 2) and most remained in the same category for all years when studied. The estimated weights of cows were higher on farms where ≥2t of concentrates were fed across all farms and among farms with predominantly Holstein-Friesian herds. Calving system Over three-quarters of the herds on the study farms were seasonal calving, with 68% (47/69) calving in spring and a further 9% (6/69) calving in autumn. Herds on the remaining farms were split calving or yearround calving. Calving systems were similar across concentrate feedings categories. Herd size Table 2: Distribution of study farm years by concentrate feeding category and year. Category Year Pooled 2006/07 2007/08 2008/09 <1t 21 18 17 56 1t to <2t 27 26 21 73 ≥2t 21 22 19 62 Pooled 69 66 57 191 Based on Grains2Milk’s feeding system classification (Types 1 to 5) (Little, 2010), about two-thirds of farms used System 2 (moderate-high bail), with most of the remaining farms using System 1 (low bail). Only nine of the 191 farm years studied were in years when farms were using System 3 (PMR). Concentrate feeding experience The concentrate feeding experience of the managers of the study farms at the start of 2006/07 varied widely. The AGFS-recruited participants generally had more experience than the TIAR-recruited participants. While some managers with no experience fed more than 1t concentrates in 2006/07, all managers of farms where ≥2t were fed had at least one previous year of experience feeding concentrates. 14 Across the 191 farm years, the median herd size was 381 cows. Herd sizes ranged from 106 to 1,400 cows and most herds were between 200 and 800 cows. Herd sizes were similar across concentrate feeding categories, but increased as the study progressed. Milking area Across the 191 farm years, the median milking area was 153 hectares. The median milking area was 177ha on farms where cows were fed less than 1t concentrates, compared to 140ha and 148ha for farms where cows were fed 1t to <2t and ≥2t concentrates, respectively. Stocking rate Stocking rates were mostly between 1.5 and four cows per milking area hectare. Across the 191 farm years, the median stocking rate was 2.5 cows per hectare. The median stocking rates were similar in all three concentrate feeding categories. Stocking rates were higher in 2007/08 and 2008/09. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Use of irrigation Equity and debt Irrigation was commonly used. Part or all of the milking area was irrigated in 84% (160/191) of farm years (86%, 79% and 87% in years where cows were fed less than 1t, 1t to <2t and ≥2t concentrates, respectively). Across farms using any irrigation, the median percentage of the milking area that was irrigated was 53%; the median was 41% on farms where cows were fed less than 1t concentrates, compared to 64% and 56% on farms where cows were fed 1t to <2t and ≥2t concentrates, respectively. Median equity in dairy assets in each farm’s first year in the study was 70% (range 12% to 100%; equity on most farms was initially between 40% and 100%). Financial characteristics Assets The median value of dairy assets in each herd’s first year in the study was $3.7m (range $1.1m to $9.3m; most farms were between $2m and $7m). This equated to medians of $9,329 per cow (range $5,355 to $20,657) and $22,348 per hectare (range $7,629 to $61,970; most farms were between $15,000 and $45,000). Assets per cow and hectare did not vary significantly with concentrate feeding category. On all the farms, the highest proportion of assets by value were land and improvements; this proportion was a little lower on farms where more concentrates were fed due to the greater contribution of livestock on those farms. Machinery constituted a relatively small proportion of dairy asset value on all farms. Median debt in each farm’s first year in the study was $1m. This equated to medians of $2,811 per cow (range $0 to $8,195) and $6,049 per hectare (range $0 to $22,483). Income Milk income contributed between 80% and 96% (averaged within farm over all study years) of total dairy income (median 94%). Income from stock sales contributed between 2% and 14% (median 5%). Other income contributed between 0% and 6% (median 1%). Most farms supplied Fonterra, but the milk supply company was National Foods and Cadbury in about 20% and 10% of farm years, respectively. Relative to farms supplying Fonterra, those supplying National Foods were more likely to feed ≥2t of concentrates, and to either calve seasonally in autumn or use split-calving. Breed distributions did not differ markedly between farms supplying National Foods and those supplying Fonterra. For details of milk income received per kilogram milk solids, per cow and per hectare on farms across the three concentrate feeding categories, see page 21. Note: The next two sections of this booklet represent data from the TasMilk60 study in a series of box plots, otherwise known as ‘box and whiskers’ graphs. These are based on percentiles. 1,500 How to read a ‘box and whiskers’ graph ▼ ▼ ▼ • Each value outside of the whiskers is considered an ‘outlier’ and shown as a dot ▼ −500 • The shaded box includes the middle half of the farms. The lower border of the shaded box is the 25th percentile, the upper border is the 75th percentile and the horizontal line within the box is the middle farm, i.e. the 50th percentile. • The lower limit of the whiskers is the value for the lowest farm that is on or within 1.5 interquartile ranges* below the 25th percentile 0 500 1,000 • The upper limit of the whiskers is the value for the highest farm that is on or within 1.5 interquartile ranges* above the 75th percentile * Interquartile range is the difference between the 25th and 75th percentiles. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 15 16 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Section 2 Farm performance – Findings from TasMilk60 This section describes findings from the TasMilk60 study about key physical and financial aspects of farm performance, and explains what these mean for industry. Pasture utilised per hectare Total feed intake per cow What we did What we did Pasture utilised per hectare of milking area was estimated for each farm year using DPI Victoria’s Pasture Consumption & Feed Conversion Efficiency Calculator (2010). Total feed intake per cow was calculated by adding the estimated quantity of pasture consumed per cow (pasture utilised per milking hectare divided by stocking rate) to the quantities of homegrown and purchased fodder, grain/concentrates and any other feeds fed to the milking herd throughout the year. What we found What we found Total feed intake per cow was not always higher on farms where more concentrates were fed (Figure 8). Pasture intake per cow in situ was generally between 2.5t and 4t DM per cow per year (Figure 9). Mean intakes were 0.2t and 0.3t DM less on farms where 1t to <2t and ≥2t of concentrates were fed, compared to where <1t concentrates were fed. These differences were small compared to the large variation in pasture intake per cow in situ between farms. 08/09 Figure 7: Distribution of pasture utilised per hectare from the milking area by concentrate feeding category and year. What this means for industry the amount There is no simple relationship between ure utilised of concentrates fed per cow and the past agement per hectare. Good and poor pasture man ing. feed te entra occurs at all levels of conc 6 4 <1 1 to <2 >=2 07/08 2 <1 1 to <2 >=2 06/07 0 <1 1 to <2 >=2 Total feed intake per cow (t DM) 8 5 10 Pasture intake per cow in situ as a percentage of diet was generally highest on farms where <1t concentrates were fed (typically 60-90%) and much lower (typically 40-55%) 0 Pasture utilised per hectare (t DM) 15 The study farms generally utilised between 5t and 15t DM of pasture from the milking area (both grazed and removed as fodder; Figure 7). The median amounts of pasture used were similar on farms in all three concentrate feeding categories across all study years, showing that there was no simple relationship between the amount of concentrate fed per cow and pasture utilised per hectare. However, the amount of pasture utilised varied markedly between herds. <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 06/07 07/08 08/09 Figure8: Distribution of total feed intake per cow (t DM) by concentrate feeding category and year. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 17 5 4 What we did 2 3 A farm’s average milk solids yield per cow was calculated as the sum of farm fat and protein yields divided by the peak number of cows milked for at least 60 days. <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 06/07 07/08 08/09 Figure 9: Distribution of pasture intake per cow in situ (t DM) by concentrate feeding category and year. on farms where ≥2t of concentrates were fed. However, there was considerable overlap in this variable between concentrate feeding categories in each study year. Farms feeding moderate-high levels of concentrates generally also fed cows more conserved fodder. However, percentages of diets that were fodder overlapped considerably between these groups of farms (Figure 10). Concentrates generally constituted <20%, 20-30% and 28-45% of total feed intake on farms feeding <1t, 1t to <2t, and ≥2t concentrates, respectively. What this means fo r industry There was an important variation in fat and protein concentrations in milk between farms in the same concentrate feeding category. • Fat concentration was lower on farms where more concentrates were fed, but protein concentrations did not differ significantly between concentrate feeding categories. • Protein to fat ratios were higher on farms where more concentrates were fed, largely due to the reduced fat concentrations in milk from those farms. Among the farms with predominantly Holstein-Friesian herds, the fat percentage was generally lower, while the protein percentage was slightly higher on farms where more concentrates were fed (Figure 12). Higher protein to fat ratios on these farms were largely due to lower fat percentages. 700 600 500 400 300 Milk solids per cow (kg) 100 0 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 06/07 07/08 08/09 06/07 07/08 08/09 Figure 10: Distribution of fodder intake per cow (t DM) by concentrate feeding category and year. 18 200 2 1 0 Fodder (home−grown and purchased) intake per cow (t DM) 3 Total feed intakes pe r cow are not alway s higher on farms where more concentrates are fed . Farms feeding mode rate-high levels of co ncentrates generally also feed more conserved fod der. It is difficult to get a cow to consume mu ch more than 3.5t DM of pasture per year, regardless of the amount of concentrates and fod der fed. Milk yield per cow varied substantially between herds within the concentrate feeding category (Figure 11). Generally, higher milk yields per cow on farms where more concentrates were fed translated into higher milk yields per hectare in this group of farms. 800 1 What we found 0 Pasture intake per cow in situ ie while grazing (t DM) Milk yield and composition Figure 11: Distribution of milk solids per cow by concentrate feeding category and year. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 1.1 1 Determinants of feed conversion efficiency were assessed across all concentrate feeding categories. .9 Annual milker feed conversion efficiency (FCE) was calculated using the DPIV Pasture Consumption & Feed Conversion Efficiency Calculator as litres of energycorrected milk (corrected to 3.14 MJ NE per litre) per cow divided by total annual feed intake (expressed as kilogram of DM) (Beever and Doyle, 2007). .8 Protein to fat ratio .7 What we did .6 Feed conversion efficiency (FCE) <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 06/07 07/08 08/09 Figure 12: Distribution of protein to fat ratio by concentrate feeding category and year for farms with predominantly Holstein-Friesian cows. What this means for industry litre can vary The average kilogram milk solids and from farm to farm. significantly in their protein to fat ratio er on farms where Protein to fat ratio is generally high more concentrates are fed. Marginal versus average feed conversion efficiencies Annual feed conversion efficiencies can be either marginal or average. Marginal annual feed conversion efficiency describes the expected increase in the farm’s annual energycorrected litres per cow following feeding of an extra kilogram DM of annual feed intake per cow. Marginal annual feed conversion efficiencies were not estimated in this study. The feed conversion efficiencies reported above are average annual feed conversion efficiencies. They should not be interpreted as the expected annual milk yield response to extra annual feed intake for the farm. For any particular farm year, they may be either less or more than the marginal annual feed conversion efficiency. These annual feed conversion efficiencies should also not be confused with short-term milk yield responses to additional feed intakes. The feed conversion efficiency achieved by the TasMilk60 farms feeding 1t to <2t concentrate per cow over all three study years were examined more closely using data on years of experience feeding at this level of concentrate to see if there was any correlation. A modeling exercise was done with the TasMilk60 dataset to estimate the financial value of an incremental change in feed conversion efficiency. What we found The FCE was generally higher on farms where higher amounts of concentrates were fed, but there were large variations between farms in all concentrate feeding categories (Figure 13). Across all farm years, 41% of farms feeding <1t concentrate per cow achieved Grains2Milk’s recommended target of 1.0L/kg DM for System 1 farms, while 16% of farms feeding 1t to <2t and 31% of farms feeding ≥2t concentrate per cow achieved Grains2Milk’s recommended target of 1.2L/kg DM for System 2 farms. Key drivers of FCE Total feed intake per cow was positively associated with and explained 55% of the variance in feed conversion efficiency (Figure 14). This occured because, with increasing feed intake, a larger proportion of the diet was available for milk production, i.e. maintenance requirements were ‘diluted’ by increasing total feed intakes. In addition to optimising total daily feed intake and providing a nutritionally balanced diet that meets all the cows’ requirements, other things known to help optimise FCE are to: • maintain high feed quality; • maintain good rumen function; • minimise feed gaps throughout the year; • minimise feed wastage; and • minimise energy losses. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 19 Table 3: FCE by number of years concentrate feeding (farms feeding 1t to <2t conc. / cow over all three years). 1.4 1.2 <1 1 to <2 >=2 2007/08 2008/09 Figure 13: Distribution of feed conversion efficiency by concentrate feeding category and year for farms with predominantly Holstein-Friesian cows. 1.4 50% 56% 14% 11% 11% Increasing feed conversion efficiency within your chosen feeding system (Types 1 to 5) can be a powerful lever for increasing profit, particularly in higher milk price years such as 2007/08 (Figure 15). For that year, milk EBITD increased by between $200 and $500 per cow for every 0.1 improvement in FCE, depending on the grain feeding category. This contrasted with negligible to modest benefits of $40 to $120 / cow in the low milk price year of 2006/07. 3 4 5 6 7 8 Total feed intake per cow (t DM) Figure 14: Association between total feed intake per cow and feed conversion efficiency. −1000 Milk EBITD per cow ($ ) 1500 Beneficial effects of FCE were due to combined effects of increased total feed intakes, and/or increased FCE at a given feed intake. ributor to high Total feed intake per cow is a major cont r management feed conversion efficiency. However, othe factors are also important. ding system, If a dairy farm changes its production/fee le, infrastructure) adjustments are needed (cows, feed, peop some years. to realise the full benefits. This may take feed costs for There is significant potential to reduce feed conversion the same milk yield through managing feed nutrient costs). efficiency (and also through controlling in your chosen Increasing feed conversion efficiency with erful lever for pow a be feeding system (Types 1 to 5) can price years. milk er increasing farm profit, particularly in high 1.2 41% 13% 1 19% ≥4 years .8 2-3 yrs .6 33% .4 39% .2 45% 0 69% 1000 0-1 yrs 500 Top quartile (top 25%) 0 3rd quartile −500 2nd quartile $ value of an incremental change in FCE 20 <1 1 to <2 >=2 2006/07 Feed conversion efficiency (energy−corrected lites per kg DM) Bottom quartile (bottom 25%) What this means for industry 1 <1 1 to <2 >=2 Feed conversion efficiency quartiles No. of years concentrate feeding .8 Two-thirds of farms in the top quartile for feed conversion efficiency had at least 2-3 years concentrate feeding under their belt, whereas more than two-thirds of farms in the bottom quartile for feed conversion efficiency had 0-1 years feeding concentrates (Table 3). The major reason for this was because farmers with more prior concentrate feeding experience fed higher total feed intakes per cow, a strong determinant of feed conversion efficiency. .6 Years of experience feeding concentrates versus FCE Feed conversion efficiency (l of energy−corrected milk per kg DM) Visit www.dairyaustralia.com.au for more detailed Grains2Milk information about strategies for optimising feed conversion efficiency in each of the five main feeding systems. .6 .8 1 1.2 1.4 Feed conversion efficiency (energy−corrected litres per kg DM) Figure 15: Relationship between FCE and milk EBITD per cow in year two of the study (2007/08). Green represents <1t concentrate per cow, blue 1 to <2t and gold ≥2t. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Income and milk price Milk income per cow and hectare What we did Milk income per cow (Figure 17) and per hectare (Figure 18) were generally higher on farms where more concentrates were fed, but varied substantially within each concentrate feeding category. Milk prices received were analysed by year and manufacturer supplied, and then compared across concentrate feeding categories. dustry What this means for in 5,000 4,000 <1 1 to <2 >=2 <1 1 to <2 >=2 06/07 07/08 08/09 15,000 10,000 5,000 Milk income per hectare ($ ) 7 6 5 0 4 Milk income per kg milk solids ($ ) <1 1 to <2 >=2 Figure 17: Distribution of milk income per cow by concentrate feeding category and year. 8 Milk price was lower on farms with seasonal spring calving (the most common calving system among the study farms) in two of the three study years, but was higher in 2007/08. After adjusting for milk solids concentration and calving system, milk protein to fat ratio was positively associated with milk price in 2007/08, but not in the other two years. This occurred despite the milk protein to fat ratio being higher in all three study years on farms where more concentrates were fed (Figure 12). 3,000 However, there was marked variability in milk prices between herds within years, even among farms that supplied the same manufacturer, Fonterra (Figure 16). This was potentially due, in part, to farm-level factors under management control, including milk volumes, milk fat and protein concentrations and the seasonality of the supply pattern. 2,000 The median milk prices received by the study farms varied markedly between years, from 33¢ per litre in 2006/07 to 50¢ in 2007/08 to 39¢ in 2008/09; median milk incomes per kilogram milk solids were $4.35, $6.57 and $5.11 per kilogram, respectively. ce than re control over milk pri Dairy farmers have mo luding gh on-farm factors inc they may realise throu tions and and protein concentra milk volumes, milk fat ly pattern. seasonality of milk supp 1,000 Milk income per kilogram milk solids Milk income per cow ($ ) What we found <1 1 to <2 >=2 06/07 <1 1 to <2 >=2 07/08 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 08/09 06/07 07/08 08/09 Figure 16: Distribution of milk income per kilogram milk solids by concentrate feeding category and year for farms supplying Fonterra. Figure 18: Distribution of milk income per hectare by concentrate feeding category and year. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 21 Operating costs Table 4: Percentages of operating costs by category for 191 farm years. What we did Cost category: The cost categories described on page 11 were analysed by year and then compared across concentrate feeding categories. What we found Operating costs per cow, hectare and kilogram milk solids Operating costs per cow (and per hectare) were generally higher on farms where more concentrates were fed, but varied substantially within each concentrate feeding category (Figure 19). Median operating costs per cow were approximately $1,400 and $3,000 on farms feeding less than 1t and 2t or more of concentrates, respectively. Operating costs per litre and per kilogram milk solids were generally higher on farms where more concentrates were fed, but differences between these groups of farms were much less marked (Figure 20). All of these operating cost measures varied substantially between farms in the same concentrate feeding category, indicating that there is considerable potential for reducing costs at similar levels of concentrate feeding. Operating cost components Range Feed 56% 19-73% Labour and management 24% 7-54% Herd 8% 2-20% Shed 3% 1-11% Overhead 9% 4-24% Feed costs generally constituted higher proportions of operating costs on farms where more concentrates were fed (Figure 21), and labour and management costs generally constituted lower proportions (Figure 22). Herd, shed and overhead costs were generally relatively small components of operating costs and proportions of operating costs that were herd, shed and overhead costs varied less between farm years. As would be expected, feed costs were predominantly supplement costs (concentrates, purchased and non-milker area fodder, and other purchased feeds) on farms where 1t or more of concentrates were fed, whereas pasture costs were generally major feed costs on farms where less than 1t of concentrates were fed. The cost for milking area fodder and irrigation generally constituted less than 15% of feed costs. Costs of concentrates (including protein and other grains and meals, and minerals) were mostly between $300 and $600 per tonne as fed; they were lower for feeds fed in 2006/07 (median price per tonne $433) than 2007/08 ($550) and 2008/09 ($545). They were generally higher on farms where more concentrates were fed as these farms were more likely to include protein supplements, minerals and other additives with grain to provide a more nutritionally balanced diet 8 6 4 0 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 06/07 07/08 08/09 06/07 07/08 08/09 Figure 19: Distribution of operating costs per cow by concentrate feeding category and year. 22 2 Operating costs per kg milk solids ($ ) 4,000 3,000 2,000 1,000 0 Operating costs per cow ($ ) 5,000 Feed costs generally constituted the highest proportion of operating costs, followed by labour and management costs. Herd, shed and overhead costs were generally much smaller proportions of operating costs (Table 4). Median Figure 20: Distribution of operating costs per kg milk solids by concentrate feeding category and year. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 400 300 200 0 100 Pasture, fodder and irrigation cost s per tonne DM utilised from milking area ($ ) 70 60 50 40 30 20 10 Percentage of operating costs that were feed cost s 0 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 06/07 07/08 08/09 Figure 21: Distribution of percentages of operating costs that were feed costs by concentrate feeding category and year. to their cows. However costs of concentrates varied substantially within concentrate feeding categories in each year. These differences are likely to be due to variation in both types of feeds and unit prices. Interestingly, farms where more concentrates were fed also generally spent more on pasture, fodder and irrigation costs per milking hectare. This was due, in part, to higher fertiliser expenditure on these farms. It was not explained by greater use of pasture renovation. <1 1 to <2 >=2 <1 1 to <2 >=2 06/07 07/08 08/09 Figure 23: Distribution of pasture, fodder and irrigation costs expressed per tonne DM pasture utilised from the milking area by concentrate feeding category and year. What this means for industry operating There is considerable potential for reducing ng. costs at any given level of concentrate feedi ing Farmers aiming to increase profitability by reduc and , costs costs should review and monitor feed labour and management costs. Large increases gh in profitability are unlikely to be achieved throu they as costs ead overh reductions in herd, shed and costs ting are relatively small components of opera and are less variable between farms. rtunities Further research is required to identify oppo income. for reducing feed costs without reducing milk Marginal versus average costs per tonne of pasture 20 30 40 50 The costs per tonne of pasture calculated in this study can be considered as average annual costs of pasture. They should not be interpreted as marginal costs. (Marginal annual costs are calculated as the extra pasture utilised divided by the extra annual input costs for a given strategy. The marginal cost per tonne of pasture can be either more or less than the average cost of pasture at the same input costs). 10 Note, too, that only pasture, fodder and irrigation costs were included in these calculations; other operating and non-operating costs were not included. 0 Percentage of operating costs that wer e labour & management cost s Average pasture, fodder and irrigation costs per tonne of pasture were calculated as pasture, fodder and irrigation costs divided by tonnes DM of pasture utilised from the milking area. These costs were higher on farms where more concentrates were fed (Figure 23) because pasture, fodder and irrigation costs per hectare were generally higher, but pasture utilised per hectare was no higher on these farms. <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 06/07 07/08 08/09 Figure 22: Distribution of percentages of operating costs that were labour and management costs by concentrate feeding category and year. These annual costs of pasture should also not be confused with the costs of short-term pasture responses to additional inputs to grow more pasture. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 23 Change in physical and financial aspects year to year What we did Changes in physical features and financial characteristics over successive years were described for 65 Tasmanian dairy farms from 2006/07 to 2007/08, and for 57 farms from 2007/08 to 2008/09. These changes were of particular interest as milk prices varied substantially between these years, with the majority of farms receiving typical long-term prices in 2006/07, very high prices in 2007/08, and high opening prices that were reduced markedly mid-year in 2008/09. Two types of change were considered: • Systematic changes across the group of study farms in which the group generally (i.e. collectively) shifted (up or down). • Changes within farms between years. It is possible to have no important systematic change but substantial changes within farms. This would occur if changes within farms were evenly balanced with increases on some farms and decreases on others. Changes within farms between years were assessed using absolute differences, and measures of agreement and correlation. What we found Small systematic changes occurred for some variables. Herd size generally increased across years, consistent with a long-term trend in the Australian dairy industry. Pasture utilised per hectare generally increased during the study, as did stocking rates. Milk yields also generally increased from 2006/07 to 2007/08, but increases in medians were relatively small. None of the pasture, fodder or concentrate intakes changed systematically by important amounts over this period, despite the large differences in milk price between these years. 24 Despite large differences in milk price from 2006/07 to 2007/08, there were no large systematic changes to milk yield, pasture, fodder or concentrate intakes between those years, even though there were large changes in some variables in some herds. Most farms took a ‘business as usual’ approach; only a few farms took a ‘make hay while the sun shines’ approach and made significant changes to optimise their profit in the high milk price year (2007/08). This may have been because farmers were uncertain as to the optimal changes for their circumstances and farm goals. Alternatively, rapid changes in response to the high milk price may have been impractical or difficult. It is possible that these farmers generally aimed to satisfy rather than maximise income, and so made few management changes to maximise income in the high milk price year. In an increasingly volatile operating environment (climatic and markets), dairy farmers do need to ‘make hay while the sun shines’ in the good years to increase equity and/or farm performance so they cope better during more difficult years. We need to better understand farmer decision making in the face of sudden, substantial increases in milk price, and to identify practical management strategies that can be rapidly implemented to capitalise on this situation. What this mean s for industry For many reason s, most dairy farm s do not make large systemic m anagement chan ge s from year to year in response to changing mar ket conditions. It is critical that fa rms have a stable and confident platform from wh ich to respond or ch ange. Change due to instability is often eroded or unprofitable. Farmers need m ore support in id entifying practical management stra tegies that can be rapidly implemented to ta ke greater advant age of high milk price years when they occur. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Section 3 Farm profitability – Findings from TasMilk60 This section describes findings from the TasMilk60 study about farm profitability, and explains what these mean for industry. Milk EBITD per cow, hectare, litre, kilogram milk solids and return on capital (milk EBITD) were closely correlated with each other (Figure 24). Least close correlations were with return on equity. This was presumably because return on equity is determined not only by farm management decisions and external factors, but is also sensitive to equity, and equity was not closely correlated with management decisions and external factors. The close correlation between milk EBITD per cow and per hectare occurred because, in the study population, neither milk EBITD per cow nor per hectare were closely related to stocking rate. Milk EBITD per cow Milk EBITD per ha 1500 1000 −500 0 What we found 500 Twelve measures of farm profitability that were comparable across farms of different sizes were assessed. Dairy EBITD per cow ($ ) What we did 2000 Associations between profitability measures −500 0 500 1000 1500 2000 Milk EBITD per cow ($ ) Figure 25: Associations between milk and dairy EBITD per cow for all farm years combined. The same relationships were observed for the corresponding dairy EBITD measures. In addition, the corresponding milk and dairy EBITD measures (e.g. milk EBITD per cow and dairy EBITD per cow) were all closely correlated (Figure 25). That was because dairy EBITD differed from milk EBITD only in that it included stock sales and other dairy income, both of which were relatively minor compared to income from milk sales. What this means for in dustry Milk EBITD per litre Milk EBITD per kg milk solids Return on capital (milk EBITD) In farm populations su ch as that examined in the TasMilk60 study, milk EBITD per cow, hecta re, litre, kilogram milk solids an d return on capital (milk EBITD) are closely correlated with each other, so are like ly to tell the same story abou t farm profitability. Return on equity (milk EBITD) Figure 24: Associations between profitability measures based on milk EBITD for all farm years combined. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 25 Farm profitability for each study year What we did Given the close correlations between most measures of profitability, we focused on two measures: milk EBITD per cow and milk EBITD per hectare. What we found With these two measures, there were large variations in profitability between farms in various concentrate feeding categories and years (Figures 26 and 27). In the lowest milk price year (2006/07), milk EBITD per cow was most variable on farms where 1t to <2t concentrates were fed. The reasons for this are unclear, but this result may reflect greater management difficulty in optimising this system in low milk price years or less experience feeding concentrates among managers of these farms at the time of the study. <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 06/07 07/08 08/09 Figure 26: Distribution of milk EBITD per cow by concentrate feeding category and year. 4,000 2,000 0 −2,000 Milk EBITD per hectare ($ ) 1,000 500 0 −500 Milk EBITD per cow ($ ) 1,500 6,000 In the year with highest milk price (2007/08), the median milk EBITD per cow was higher on farms where 1t or more of concentrates were fed. In the other two years, milk prices were lower and the median milk EBITD per cow was higher on farms where <1t of concentrates were fed relative to farms where 1t to <2t concentrates were fed. However, these differences in medians between concentrate feeding categories were small compared with the variability within each category, and the highest values were similar for all three categories. <1 1 to <2 >=2 <1 1 to <2 >=2 <1 1 to <2 >=2 06/07 07/08 08/09 Figure 27: Distribution of milk EBITD per hectare by concentrate feeding category and year. s for industry What this mean ing level for concentrate feed There is no ‘best’ entrate feeding system. Any conc ing ed fe n/ tio uc prod n be profitable eding system ca fe / n tio uc od pr ement, level or riate mix of manag op pr ap an n ve gi in any year ut costs. milk price and inp profits average or median The differences in / moderate / high n farms using low achieved betwee all compared with ing levels are sm ed fe te tra en nc co ncentrate ofit within each co the variability in pr feeding level. 26 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Determinants of profit within each study year It is unlikely that highly profitable farms have every indirect and direct determinant at values for maximum (indirect or direct) effects on profit. To explore how farmers managed selected profit determinants collectively on individual commercial farms, the five most and five least profitable farm years were identified within each concentrate feeding category and values for selected indirect and direct profit determinants assessed. These profit determinants were identified from the final path models as those most likely to have important indirect or direct effects on milk EBITD per cow. What we did Direct and indirect determinants of profitability were modelled using path analysis. Indirect determinants were those whose effects were assumed to be mediated through an intervening variable. For example, milk protein percentage was assumed to influence milk EBITD per cow through its effect on milk price per litre. Direct determinants of profitability were assumed to directly determine profitability, rather than via an intervening variable. What we found Some of the results of these analyses are shown in Table 5 below. These show the values for selected profit determinants of milk EBITD per cow for the five most profitable farm years where 1t to <2t of concentrates were fed per cow. Figure 28 (see following page) shows the null path model of milk EBITD per cow. Separate final path models were developed for each of the three concentrate feeding categories. Importance of direct determinants was assessed based on the estimated changes following increases in each determinant from the 25th to 75th percentile values observed on the study farms. Thus, importance was determined by the combination of the strength of association of the relationship and the amount of variation in the determinant in the study farms. The best way to summarise all these data is using Table 6. This shows, for each profit determinant, the proportions of farm years with values more than 10% better and more than 10% worse than the median. Within each concentrate feeding category, for all profit determinants other than milk price, some of the five highest profitability farm years had values worse than the median and/or some of the five lowest profitability farm years had values better than median. Many of the same determinants were retained in the final models for all three concentrate feeding categories, indicating that some determinants affect profitability regardless of concentrate feeding category. Table 5: Values for selected indirect and direct profit determinants of milk EBITD per cow for the five most profitable and five least profitable farm years where 1t to <2t of concentrates were fed per cow. Median* Farm number Year Milk EBITD per cow Milk price (¢/litre) Milk yield (litres per cow) Feed costs per cow Labour and management costs per cow Herd costs per cow Shed costs per cow Overhead costs per cow Total costs per cow Highest milk EBITD per cow** 58 32 18 2007/08 2007/08 2007/08 1,243 1,141 1,091 33 2007/08 1,082 56 2008/09 -534 Lowest milk EBITD per cow** 64 65 66 2006/07 2006/07 2008/09 -555 -588 -631 247 1 2007/08 1,657 68 2006/07 -661 38 6,870 48 9,362 48 7,858 45 8,423 51 7,262 58 5,836 38 7,363 30 7,789 31 7,158 37 6,366 28 5,567 1,365 1,647 1,475 1,559 1,834 1,301 1,825 1,443 1,132 1,704 1,072 565 516 565 626 300 446 806 906 999 739 791 177 194 159 214 158 153 255 237 413 196 154 59 135 87 72 100 82 72 28 76 79 51 224 302 259 138 221 301 346 287 162 291 154 2,454 2,794 2,545 2,609 2,614 2,282 3,304 2,902 2,782 3,009 2,222 *Median for all farm-years where 1t to <2t of concentrates were fed per cow. **Values above median that result in indirect or direct increases in milk EBITD per cow if other paths to milk EBITD per cow are unchanged are in blue font; others are in black font. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 27 28 Pasture, milking area fodder and irrig. costs per ha Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Pasture and milking area fodder cost per t DM % diet as pasture and milking area fodder Purchased fodder costs per t DM % diet as purchased fodder Other feed costs per t DM Fat% Milking area fodder intake per cow % diet as other feed % diet as fodder Pasture intake in situ per hectare Calving system Year Protein% Stocking rate (cows per ha) Pasture utilised (t per ha) % diet as pasture % milking area irrigated % milking area regrassed Fertiliser costs per ha Irrigation volume per irrig. ha Winter grazing rotation Autumn grazing rotation Summer grazing rotation Spring grazing rotation Key approach to pasture management Land and impr. value per ha Manager’s years of experience feeding concentrates Concentrate costs per t DM Feed costs per cow Cows per FTE38 % of milker area irrigated Annual salary/wages plus on-costs per FTE38 Labour and mgt. costs per cow Herd costs per cow Shed costs per cow Overhead costs per cow Milk EBITD per cow Amount fodder conserved Mature cow animal on farm per cow health costs per cow Cows milked per Total non-pasture operator-hour feed intake per cow Replacement feed costs per cow Proportion of FTE38s unpaid Feed costs per t DM % diet as concentrates Diet ME content Feed conversion efficiency Milk yield (l/cow) Concentrate intake per cow Total feed intake per cow Predominant breed Pasture intake in situ per cow Other feed intake per cow Purch. fodder intake per cow Milk price (c/l) Manufacturer Figure 28: Null path model of milk EBITD per cow. Table 6: Proportions of 15 farm years with highest and 15 farm years with lowest milk EBITD per cow* that had values for each profit determinant more than 10% better and more than 10% worse than median. Profit determinant Highest milk EBITD per cow farm years (n=15) Milk price (¢/litre) Lowest milk EBITD per cow farm years (n=15) >10% better than median >10% worse than median >10% better than median >10% worse than median 100% 0% 0% 60% Milk yield (litres per cow) 60% 13% 20% 40% Feed costs per cow 13% 53% 47% 27% Total feed intake per cow 13% 7% 7% 33% Feed conversion efficiency 33% 0% 7% 20% Feed conversion efficiency above maintenance 33% 7% 13% 13% Dietary ME 0% 0% 0% 0% Feed costs per t DM 7% 27% 33% 27% Concentrate costs per t DM 7% 40% 40% 7% Pasture and milking area fodder cost per t DM 20% 47% 20% 67% Pasture, milking area fodder and irrig. costs per hectare 20% 53% 47% 47% Pasture utilised per hectare 40% 13% 27% 53% Stocking rate 27% 13% 40% 53% Purchased fodder costs per t DM 60% 13% 20% 73% Labour and management costs per cow 33% 20% 7% 80% Cows per FTE38 40% 20% 7% 80% Herd costs per cow 47% 40% 27% 53% Shed costs per cow 13% 67% 40% 60% Overhead costs per cow 33% 40% 27% 53% Total costs per cow 7% 40% 7% 60% *The five farm years with highest and five farm years with lowest milk EBITD per cow were selected within each concentrate feeding category. Across all 30 farm years, the following profit determinants generally deviated from median by 10% or more in expected directions for both high and low profitability farm years: milk price, milk yield (litres per cow), pasture utilised per hectare, purchased fodder costs per tonne DM, labour and management costs per cow, and cows per 38-hour full-time equivalent (FTE38). However, exceptions to these patterns were common. In addition, pasture and milking area fodder cost per tonne DM deviated from median by 10% or more in expected directions for low profitability farm years but not for high profitability farm years. These findings emphasise the need to assess all direct determinants of profitability collectively when analysing farm performance, rather than assessing performance against achievable or target values separately for individual indices. What this means for indus try Dairy farms are complex sys tems and there are many ways to make a profit (or a loss). Determinants of profit should be assessed collectively, not separately. Farms that have consistent ly higher profits usually have a relatively higher milk price, higher milk yield per cow, lower fodder cos ts, and lower labour and management costs. Howe ver, they tend not to be outstanding performers for any of these particular determinants of profit – the y tend to be consistent allrounders whose efforts for these profit determinants collectively are superior. See Appendix on page 54 for John Morton’s six principles for identifying the determinants of profitability by analysing data from groups of dairy farms. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 29 Change in profit due to management changes What we found Changes in milk EBITD per cow What we did Changes in milk EBITD per cow from 2006/07 to 2007/08 were explored using 65 farms enrolled in both 2006/07 and 2007/08. What we found The changes are shown in Figure 29. As milk prices increased substantially, milk EBITD per cow increases were expected and this occurred on most farms. On 61 of these 65 farms (94%), milk EBITD per cow increased by $100 or more. Farms feeding moderate-high levels of concentrates tended to achieve higher increases in profit than the low concentrate feeding category. Causes of changes in milk EBITD per cow What we did 1,500 1,000 In summary, these results indicate that the increases in milk EBITD per cow in 2007/08 were largely due to increased milk price on most farms, partly negated by substantially increased operating costs (most commonly due to increased feed costs per cow) and on some farms, by reduced litres per cow. 51 12 2 20 39 24 31 19 45 29 42 22 30 10 28 43 21 14 67 9 500 0 1,000 2,000 0 Change in milk EBITD per cow ($ ) <1 1 to <2 >=2 Figure 29: Distribution of changes in milk EBITD per cow from 2006/07 to 2007/08 by concentrate feeding category. 30 Feed costs per cow increased substantially on most farms, typically by larger amounts on farms where 1t or more of concentrates wee fed. Costs per cow for other cost categories were typically small, although these decreased substantially on some farms and increased substantially on others. −1,000 −500 Change in milk EBITD per cow ($ ) 2,000 Changes in milk EBITD per cow from 2006/07 to 2007/08 were explored using novel but related modelling methods based on a concept developed by Agrilink FarmStats. This allowed components of changes in milk EBITD per cow to be identified. These were analysed separately for farms in each of three concentrate feeding categories. See Figures 30, 31 and 32, and Table 7. On most farms, increases in milk price made large contributions to the changes in milk EBITD per cow. These were typically larger in farms where 1t or more of concentrates were fed. Changes in litres per cow generally made only small contributions to changes in milk EBITD per cow. Total operating costs per cow increased on most farms, commonly by substantial amounts. These increases were typically larger in farms where 1t or more of concentrates were fed. Figure 30: Changes in milk EBITD per cow from 2006/07 to 2007/08 due to increased milk price (dark blue), changed litres per cow at 2007/08 milk price (light blue) and changed total operating costs per cow (orange) in farms where <1 t concentrates were fed in 2006/07. Farms are listed in descending order of actual increase in milk EBITD per cow. Farm number is shown. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 58 64 59 52 68 53 33 5 1 36 65 41 18 25 55 56 23 8 6 61 16 54 34 15 62 40 35 47 38 63 32 7 66 49 48 46 44 27 17 3 4 37 57 50 13 −1,000 0 1,000 2,000 −1,000 Change in milk EBITD per cow ($ ) 0 1,000 2,000 Change in milk EBITD per cow ($ ) Figure 31: Changes in milk EBITD per cow from 2006/07 to 2007/08 due to increased milk price (dark blue), changed litres per cow at 2007/08 milk price (light blue) and changed total operating costs per cow (orange) in farms where 1t to <2t concentrates were fed in 2006/07. Farms are listed in descending order of actual increase in milk EBITD per cow. Farm number is shown. Figure 32: Changes in milk EBITD per cow from 2006/07 to 2007/08 due to increased milk price (dark blue), changed litres per cow at 2007/08 milk price (light blue) and changed total operating costs per cow (orange) in farms where ≥2t concentrates were fed in 2006/07. Farms are listed in descending order of actual increase in milk EBITD per cow. Farm number is shown. Expected changes in milk EBITD per cow had no management changes been made To explore impacts of any management changes, expected changes in milk EBITD per cow from 2006/07 2007/08 were calculated had there been no changes in feed or labour inputs, or in yields per cow. Expected changes were based on observed changes in milk price, pasture and milking area fodder costs per tonne DM, concentrate costs per tonne DM, purchased fodder costs per tonne DM and costs per FTE38. For example, in farm 1, concentrate costs increased from 2006/07 What we found Given the large increase in milk price from 2006/07 to 2007/08, milk EBITD per cow would be expected to have increased without management changes if unit prices of inputs did not increase. But it was likely that some managers also made management changes over this time. Table 7: Distributions of contributions to change in milk EBITD per cow ($) from 2006/07 to 2007/08. Variable All farms (n=65) Concentrate feeding category in 2006/07 <1t (n=20) 1t to <2t (n=24) ≥2t (n=21) Min 25 p’ile Median 75 p’ile Max Median Median Median Change due to increased milk price 173 681 1,011 1,310 1,946 641 1,196 1,265 Change due to changed litres per cow at 2007/08 milk price -573 -106 76 183 1,627 11 133 102 Change due to changed total operating costs per cow -1,147 -576 -330 163 191 -206 -451 -423 Change due to changed feed costs per cow -1,094 -505 -334 -185 147 -183 -372 -442 Change due to changed labour and management costs per cow -341 -21 15 89 540 9 46 10 Change due to changed herd costs per cow -250 -48 -11 18 80 -5 -2 -21 Change due to changed shed costs per cow -68 -17 -6 7 68 -6 -12 1 Change due to changed overhead costs per cow -220 -44 6 22 137 -12 -9 15 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 31 1000 1500 0 500 1,000 500 Expected change in milk EBITD per cow ($) −500 Actual minus expected change in milk EBITD per cow ($ ) 1500 1000 500 Actual change in milk EBITD per cow ($) 0 0 <1 >=2 Figure 33: Expected and actual changes in milk EBITD per cow from 2006/07 to 2007/08 in farms where <1t concentrates per cow (green squares), 1t to <2t (blue triangles), and ≥2t (gold diamonds) were fed; expected changes were those expected had there been no changes in feed or labour inputs, or in yields per cow. Figure 34: Distribution of differences between actual and expected changes in milk EBITD per cow from 2006/07 to 2007/08 by concentrate feeding category; expected changes were those expected had there been no changes in feed or labour inputs, or in yields per cow. to 2007/08 by $129 per tonne DM. Had the amount of concentrates fed in 2006/07 (1.33t DM) not been changed in 2007/08, milk EBITD per cow would have fallen by $172 (1.33*129) due to increased unit price of concentrates. This method assumes that changes in these unit costs did not result in altered milk yields. on the majority of farms, the increase in milk EBITD per cow in 2007/08 was similar to that expected had no major management changes been made. Given the high milk prices in 2007/08, it was likely that opportunities to substantially increase farm profitability were not taken on many farms. What we found The relationship between expected and actual change is shown in Figure 33. For farms close to the yellow line, the actual changes were similar to that expected given changes in milk price, pasture and milking area fodder costs per tonne DM, concentrate costs per tonne DM, purchased fodder costs per tonne DM and costs per FTE38. Management changes did not contribute to the increased milk EBITD per cow in these farms. For farms below the line, actual increases were less than expected in the absence of management change, while for farms above the line, actual increases were greater than expected. Distributions of differences between actual and expected are shown in Figure 34. In about half of the farms, milk EBITD per cow increased by more than expected; actual increases were more than $200 per cow above that expected in more than 20% of farms. Distributions were similar across 2006/07 concentrate feeding categories. These findings indicate that, on some farms, milk EBITD per cow increased by substantially more than expected had no management changes been made. However, 32 1 to <2 As discussed on page 24, there is a need to better understand farmer decision making in the face of sudden substantial increases in milk price, and to identify practical management strategies that can be rapidly implemented to capitalise on this situation. With an increasingly volatile operating environment (climate and markets), dairy farmers do need to ‘make hay while the sun shines’ in the good years to increase equity and/or farm performance so they cope better during more difficult years. What this means fo r industry Most dairy farms do not make large syste mic management change s from year to year in response to changing market co nditions. It is critical that farms have a stable and co nfident platform from which to respond or change . Change from instability is often eroded or unprofitable . Farmers need more support in identifying practical management strategie s that can be rapidly implemented to take greater advantage of high milk price years when the y occur. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Effect of enterprise size on farm profit What we did Within each study year, we assessed milk EBITD per cow for enterprises of different size. Enterprise size was described as number of cows, number of hectares, and litres of milk and kilogram milk solids produced for the year. We assessed milk EBITD per cow in the same way. Operating costs per cow were assessed for various herd sizes. Milk EBITD per cow and per hectare increased a little with number of hectares in 2006/07 but not in 2007/08 or 2008/09. Operating costs per cow and per hectare did decrease with increased herd size but, again, these decreases were small. The TasMilk60 study did not allow conclusions to be drawn about whether larger herds had lower operating costs through greater "buying power" due to increased enterprise size nor whether they had higher operating costs associated with their greater size and scale involved. What we found Milk EBITD per cow and per hectare increased with herd size, litres of milk and kilogram milk solids in 2006/07 and 2007/08. However these increases were relatively small and no such increases occurred in 2008/09. For example, in 2006/07 milk EBITD per cow increased by $72 for every extra 100 cows. However, milk EBITD per cow in that year varied from -$660 to $757, a range of $1,417, so an increase of $72 for every extra 100 cows was quite minor. What this means fo r industry Beneficial effects on farm profitability of inc reasing enterprise size are ge nerally relatively small. Any beneficial effects of 'dilution' of costs per cow with more cows are not large, due to the rel atively low levels of fixed co sts available for diluti on. Increases in total farm profitability with increa sed enterprise size are like ly to be due to scale (m ore cows / hectares / mi lk), not efficiency (op era ting cost per cow / hectare). Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 33 Degree of consistency of profit outcomes achieved year to year What we did Consistency of milk EBITD per cow was assessed within study farms between years. Fifty-six farms studied in all three years were included in these analyses. Consistency of relative profit outcome was assessed by: • looking at each farm’s absolute profitability across the three years; and What we found Both absolute and relative farm profitability were markedly inconsistent between years, as illustrated in the two ‘fiddle stick’ charts (Figures 35 and 36). Correlations between milk EBITD per cow remained weak within Fonterra suppliers. This indicates that differences in milk prices between manufacturers did not explain these weak correlations. Correlations between consecutive years were closer on farms where less than 1t concentrates were fed, but even on these farms, correlations were only moderately close. • using deviation from within year medians and tertile within year Because milk EBITD per cow was so inconsistent within farms between years, it was also important to assess milk EBITD per cow cumulatively over the three study years for each farm. Of particular interest were the medians and upper values achieved within concentrate feeding categories, and the extent of variability within each category. Three-year cumulative milk EBITD per cow values were calculated as the unweighted sum of milk EBITD per cow in each of the three study years. $ 2000 2000 1500 1500 1000 1000 500 $ 500 0 0 -500 -500 -1000 06/07 07/08 -1000 07/08 08/09 Figure 35: Line graphs of EBITD per cow in 2006/07, 2007/08 and 2008/09 for 56 farms studied in all three years; each farm is depicted by the same line colour between both pairs of years. 34 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Table 8: Distribution of 56 farms by number of study years in each of upper, mid and lower tertile of milk EBITD per cow within year. No. years in No. farms Upper Mid Lower Expected* Observed 3 0 0 2.1 7 2 1 0 6.2 4 2 0 1 6.2 4 6 1 2 0 6.2 1 1 1 12.4 10 1 0 2 6.2 4 0 3 0 2.1 1 0 2 1 6.2 11 0 1 2 6.2 3 0 0 3 2.1 6 These findings indicate that, across years with widely varying milk prices, farm profitability, in both absolute terms and relative to other farms in the same region, is highly inconsistent and demonstrates that high (or low) relative farm profitability should not be viewed as a repeatable attribute when milk price fluctuates substantially. 1,500 1,500 1,000 1,000 500 500 $ diffference from median $ diffference from median *Expected numbers under random variation, i.e. if each farm’s tertile in each year was determined randomly. As shown in Table 8, deviations from within year medians changed substantially between years. Of the 19 farms initially in the highest tertile for that year, only seven (37%) were in the highest tertile in both subsequent years. Of the 19 farms initially in the lowest tertile, only six (32%) remained in the lowest tertile in both subsequent years. Almost two-thirds of the 56 farms were in the top tertile in at least one study year, but only seven (13%) were consistently high (in the upper tertile in all three years) and only six (11%) were consistently low. Eighteen per cent of farms (10/56) had one year in each of the lower, mid and upper tertiles. 0 -500 0 -500 -1,000 -1,000 -1,500 -1,500 06/07 07/08 07/08 08/09 Figure 36: Line graphs of milk EBITD per cow deviation from median within year in 2006/07, 2007/08 and 2008/09 for 56 farms studied in all three years; each farm is depicted by the same line colour between pairs of years. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 35 2,000 1,000 What we did 0 Consistency of milk EBITD per cow between years was assessed by concentrate feeding category. Selected indirect and direct profit determinants of milk EBITD per cow were assessed in the seven consistently high relative profitability farms and the six farms that had consistently low relative profitability. −1,000 3−year cumulative milk EBITD per cow ($ ) 3,000 Determinants of consistent relative profit performance over the three study years <1t 1t to <2t >=2t Figure 37: Distribution of three-year cumulative milk EBITD per cow by amount of concentrates fed for all three study years. Medians were similar in each concentrate feeding category (Figure 37), indicating that, among farms that did not change concentrate feeding categories during the study, no category had generally higher three-year cumulative milk EBITD per cow. Variances were also similar in each concentrate feeding category. s for industry What this mean ) is not very tability (high or low ar. Relative farm profi te widely year to ye milk prices fluctua repeatable when ing level or concentrate feed or There is no ‘best’ concentrate level ing system. Any y an in production / feed e profitabl ing system can be production / feed agement, milk an m te mix of ria op pr ap an n ve year gi sts. price and input co What we found Among these consistent relative profitability farms, high profitability was due in part to consistent relatively high milk price, milk yield per cow, and consistent relatively low pasture and milking area fodder cost per tonne DM and purchased fodder costs per tonne DM. Lower labour and management costs per cow also contributed to high profitability among consistent relative profitability farms, due in part, to more cows per FTE38. Total costs per cow were consistently low more often among the high profitability farms. Importantly, no consistently high relative profitability farm was more than 10% better than median in all three years for all determinants of profitability. This emphasises the need to assess all key direct and indirect determinants of profitability collectively when analysing farm performance, rather than assessing performance against achievable or target values separately for individual indices. Note that because these findings are based on small numbers of farms, they should be regarded as preliminary, and require confirmation in larger studies. Given the infrequency of consistent relative profitability, a case-control design may be an efficient study design to identify determinants of consistent relative profitability across years. What this me ans for indus try Farmers that ha ve consistently higher profits not to be outs tend tanding perform er s for any partic determinants ular of profit, but ar e consistent al l-rounders. Determinants of profit should be assessed co not separately, llectively, when analysin g farm perform ance. 36 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Correlation between farm profitability and ‘farming style’ What we did To explore the question ‘Do decision makers in more profitable dairy farm businesses tend to have a particular management style or set of attitudes and beliefs?’ the Derived Attitudinal Farmer Segmentation (DAFS) method was used. This method was developed by Thomson (2001), and selected by Waters, Thomson and Nettle (2006) as the preferred segmentation method for dairy farmers. It segments farmers on the basis of their perceptions of a wide range of situational and individual characteristics and has been successful in explaining patterns in a wide range of behaviours across industries and geographic locations. Dairy Australia has used the method to characterise ‘farming styles’ to help the industry develop more effective and targeted extension and communication strategies within the different segments of the dairy farm sector. The Client Stocktake project, completed in 2009, involved a telephone survey of 450 dairy farmers from around Australia using a survey form of 35 attitudinal statements, each of which respondents were asked to indicate the extent to which they agreed or disagreed. Six DAFS groups were derived from a K-Means clustering process using the farmers’ responses to these attitudinal statements. Eight indices were used to interpret the attitudinal profile of these groups, these indices being ‘Business orientation’, ‘Aversion to risk’, ‘Sustainable improvement’, ‘Knowledge & self-reliance’, ‘Inter-generational orientation’, ‘The dairy way of life’, ‘Financial pressure’ and ‘Farming tradition’. Though useful in its own right, the Client Stocktake survey collected no hard data about the physical and financial performance of each respondent’s dairy farm business, so the ‘farm styles’ could not be aligned to a farm’s commercial standing or farming practices. The TasMilk60 study offered the opportunity to conduct a DAFS study about a group of farmers for whom their farm business’s physical and financial performance was known. During the data collection phase, it was a 10-minute addition to have participating farmers record their opinions about the extent to which they agreed or disagreed with the 40 statements. Sixty people from 33 TasMilk60 farm businesses completed a valid survey form. These data were passed to Warwick Waters (Waters Consulting) and Don Thomson (Landscape and Social Research) for analysis. What we found Primary decision maker respondents on the TasMilk60 farms were most commonly in one of three DAFS groups (‘Family first’, ‘Open to change’ and ‘Growing for the kids’). The distribution of respondents was similar to that for the sample of 450 Australian dairy farmers who participated in the Client Stocktake project, except the percentage in Group 4 (‘Established and stable’), which contained fewer respondents. Neither DAFS group nor attitudinal indices varied markedly by concentrate feeding category, indicating that high profitability occurred on farms with primary decision makers with diverse attitudes and farming styles. Table 9: Distribution of Derived Attitudinal Farmer Segmentation group of primary decision maker by concentrate feeding category in 2006/07 on 32 farms. DAFS group <1t 1t to <2t ≥2t Family first (Group 1) 43% (3) 17% (2) 8% (1) Love farming (Group 3) 8% (1) 8% (1) Established and stable (Group 4) 17% (2) Winding down (Group 2) 15% (2) Open to change (Group 5) 14% (1) 42% (5) 31% (4) Growing for the kids (Group 6) 43% (3) 17% (2) 38% (5) Total 100% (7) 100% (12) 100% (13) dustry What this means for in style or a certain management Farmers don’t require ble. liefs to be highly profita set of attitudes and be The original DAFS survey form of 35 statements was expanded to 40, with the added statements geared around cow nutrition, feeding, risk and wealth creation. Note: During the analysis of the DAFS survey data, a weakness in method was identified – the necessity of nominating the “primary decision maker” from within a farming family. By default, since “there can be only one”, the process tends to assign this role to the senior male of the family, even when it was a fact known to the survey team that husband and wife were joint decision makers with different knowledge and expertise. If the DAFS method is to become more reliable and accurate, means must be found to integrate the views of the members of a farming family’s decision-making team. Sons, daughters, brothers and sisters may all be part of this team as well, each taking primary responsibility for making decisions related to different aspects of the farm business, e.g. pasture management, finances, herd husbandry. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 37 38 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Section 4 Risk – Insights from TasMilk60 This section describes findings from the TasMilk60 study about risk, and explains what these mean for industry. The three year spread of data across a wide range of feeding approaches within the TasMilk60 group of farms lent itself to a long-overdue objective examination of risk, risk management and risk measures. Risk and uncertainty Risk managment Risk and uncertainty are central to all dairy farming decisions. Malcolm (2009) offers this view, “Imagining the future with rigour means adopting the structured approaches to decisions developed in modern decision theory that emphasises probability analyses and alternative futures, with sound appreciation of the pervasiveness of uncertainty”. Attempting to avoid risks is often futile or counterproductive. Instead, risk exposure can be reduced by spreading, selling or shifting risk, or by risk averaging. While there is broad agreement about the nature of the risks facing dairy farmers (production, price or market, financial, institutional and human), farmers themselves define risk in several different ways: 1. The probability of an adverse outcome or financial loss. 2. The size of the loss if it were to occur. 3. The expected value of a potential loss (i.e. 1 multiplied by 2). Many of the risks in farming are inter-dependent. Individual risks may correlate with other risks positively or negatively, e.g. weather and fodder prices. Farming systems are most severely tested by the relatively rare but most severe circumstances that may occur when drought (low yields) coincides with high feed costs, low milk prices and high interest rates. Statistical approaches have been used to define risk as equal to the variance of the probability distribution of all possible consequences of a risky course of action. In a farming systems context, risk can be defined as the variability of the profit outcomes attributed to that system given a range of values of influential parameters over time. Uncertainty is characterised by lack of complete information about an event, such as the probabilities of frequency, size, severity and duration. For the dairy farmer, there is always a trade-off between risk and reward. Those with a low risk tolerance will seek options where little risk is involved and will require a very high reward for the risk involved. Those with a higher tolerance for risk will be willing to accept risk without such a big potential reward. What constitutes an acceptable risk-reward trade-off is also coloured by the farmer’s perception of several factors – familiarity, control, catastrophic potential (‘dread factor’), equity and level of knowledge (Slovic 2000). For all production/feeding systems, there is a mix of risk, performance and management principles whose understanding can improve the chances of successfully balancing risk and reward. Since the risk environment faced by farmers is not static or fixed, risk management is, “… a continuous adaptive process that needs to be integrated into all relevant aspects of the decision making procedures of the farm” (Hardaker et al. 1997). Topp and Shafron (2006), in their comprehensive review of risk, note that, “Where a farmer can gain control or at least strongly influence his risk-reward trade-off, he may take the opportunity to spread risk by diversifying, selling or shifting risk or risk-averaging. Where a higher than average risk in one aspect can be offset by a lower than average risk, the resulting combined risk may mean less overall risk exposure”. Against the background of a decade of major market and climatic volatility, the dairy farm sector has not yet rigorously accommodated risk and uncertainty into its decision-making and profit-planning process, despite sound warnings of the logical consequences being delivered at the start of that decade. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 39 Malcolm (2001) noted that, “Another misplaced emphasis in farm management economics has been the traditional emphasis on risk/volatility as something to be minimized. This has been at the expense of a more realistic emphasis that would have been on risk/volatility not only as something farmers manage live with, but equally important, risk/volatility as a major source of the opportunities for management and of the opportunities to prosper and grow”. Risk measures To summarise, uncertainty characterizes the decisionmaking environment that exists when a farmer decision maker does not have perfect information; and risk is the economic consequence on the welfare of the farm business as a result of this uncertainty. Risk management is the process by which the probability of unfavourable outcomes is reduced and favourable outcomes increased through guiding the farmer decision maker to make timely choices among alternative options. Risk measures should provide the probability of unfavourable and favourable outcomes occuring. Unfortunately, many so-called risk measures currently used in the dairy industry do not provide information about risk and its consequences. These erroneous risk measures include: • Pasture as a % of total feed consumed • Unit cost of milk production (¢ per litre or $/kg MS) • Operating profit margin Pasture as a % of total feed consumed is a technical measure of the herd feedbase, not a risk measure, and is biased towards feeding system 1 (low bail). It incorrectly assumes that a higher EBITD and return on assets (ROA) is positively correlated to higher % pasture in the cow’s diet, and that increasing pasture’s share of the diet must therefore be a means of lowering risk. These assumptions are unsound and are not supported by the TasMilk60 dataset. Unit cost of milk production is an economic measure, not a risk measure, and provides no information about the probability of adverse cost events or enterprise profitability. The TasMilk60 data show that achieving a low unit cost has no effect on the risk exposure of the farm business at all. Since operating at the point of lowest unit cost does not maximise profits, farmers making risk-reducing decisions based upon the unit cost “risk” measure may expose their farm business to increased risks through lost opportunity. 40 Operating profit margin is also an economic measure, not a risk measure, and provides no information about the probability and repeatability of particular enterprise profit outcomes, as TasMilk60 has demonstrated. Commercial reality has higher rewards associated with higher risk, so it is illogical to assume that achieving higher dairy profits automatically results in lower risk exposure. Furthermore, profitable farmers who choose to apply surplus profit to debt reduction may reduce their exposure to financial risk but profitable farmers who choose instead to buy a boat do not. Therefore, risk reduction hinges on the spending decision, not on the level of profitability achieved. Many technical, financial and economic measures provide information about financial efficiency, liquidity, solvency or repayment capacity but are not risk measures. For well-established definitions of these terms, refer to those provided on the USA Farm Financial Standards Council website http://www.ffsc. org/Files/FarmFinancialGuidelinesRatios.pdf The TasMilk60 study has shown that the dairy industry needs to re-examine its use of risk measures as part of a new approach to risk and risk management in dairy farm decision making. Monte Carlo simulation study using TasMilk60 data What we did As part of the TasMilk60 study, Courtney Gronow (AGFS), Gordon Cleary (AGFS) and Steve Little (Dairy Australia) considered the different qualitative, semi-quantitative and quantitative methods of assessing risk that could be applied to the three years of physical and financial performance data collected from the TasMilk60 farms. Gronow, Cleary and Little chose to pursue a quantitative analysis because it assigns specific values and probabilities to risks. However, they were also mindful of three limitations which apply even to quantitative risk analysis, as highlighted by Hardaker and Lien (2007): 1. There is a common failure to deal with the decision maker’s risk aversion. 2. There is too strong a focus on downside risk. 3. There is too little attention to the place of individual risks in the total risk portfolio. To address these three limitations, a quantitative method is required which considers both the downside and upside consequences of risks, calculates probability distributions for operating profits and accounts for risk aversion through use of certainty probabilities. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study A probabilistic budgeting approach was therefore applied in the TasMilk60 study. Agrilink FarmStats used Crystal Ball® (Oracle Corporation, UK) to carry out Monte Carlo simulations on an integrated dairy biophysical and financial model (CUD,® Cowdata Pty Ltd, Australia) and generate profit distributions for three model farms (A, B and C), respectively feeding a low, moderate or high level of grain-based concentrate. The fixed inputs and the variable inputs (each with minimum, maximum and most likely values) used in these simulations (Table 10 and Figures 38a, b and c) were drawn from the physical and financial data from the TasMilk60 farms in each of the three feeding categories for the three-year period 2006/07 to 2008/09. The Monte Carlo simulations generated profit distribution curves for the three model farms, which enabled comparison of the variability of farm EBITD and showed the probability of achieving with certainty a forecast EBITD amount. A target EBITD value of $200,000, nominally capable of supporting personal drawings and debt servicing, was selected to test the probability that each feeding system could achieve the target sum with certainty in each of the three years. The simulation also generated Tornado charts that enabled the influence of the key profit drivers on farm EBITD to be compared in each case and across the three concentrate feeding levels. What we found Milk price was found to have a greater impact on profit variability at higher levels of concentrate feeding. Based on general discussions with farmers during Year 2 of the study (2007/08), when the grain price peaked, it is also likely that many farmers, especially those feeding low levels of concentrates, perceive grain/concentrate price to be a more important profit driver than it really is for their farm business, and therefore are inclined to over-react to changes in grain/concentrate price relative to other profit drivers. The full results of this simulation work have been reported in ‘Managing production system risk: An application of Monte Carlo simulation to Tasmanian dairy farm data’, by C.S. Gronow, S.B. Little and G.V. Cleary, in the proceedings of the 2010 Australasian Dairy Science Symposium held at Lincoln University in New Zealand. Table 10: Fixed inputs used in Monte Carlo simulations for each model farm. Farm A Farm B Farm C Concentrate feeding rate Low <1t Moderate 1 to <2t High ≥2t Effective area (ha) 180 165 177 Figures 38a, b and c show the profit distribution curves and Tornado charts generated for the three model farms (A, B and C), respectively feeding a low, moderate or high level of concentrate in 2006/07. Herd size 393 402 420 Stocking rate (cows/ ha) 2.18 2.44 2.37 Base case milk yield (L/cow) 4,271 6,400 8,421 Base case (milk solids /cow) 333 478 608 The profit distribution curves showed that feeding low levels of concentrates tended to result in less profit variability but the lowest mean profit. Fat percentage 4.4% 4.1% 3.9% Protein percentage 3.4% 3.4% 3.4% Milk solids percentage 7.8% 7.5% 7.2% Based on the Tornado charts for each model farm, on-farm profit drivers of pasture utilisation, pasture quality and core cost had a greater impact on profit variability than did the off-farm profit drivers of milk price, concentrate price and purchased fodder price. Bodyweight (kg) 479 529 579 Feed Conversion Efficiency 0.92 1.12 1.26 Protein to fat ratio 0.76 0.83 0.87 ECM 4484 6452 8247 ECM FCE 0.96 1.13 1.23 Total feed 4.66 5.71 6.68 Concentrates 0.47 1.47 2.31 Body condition change 0.25 0.25 0.25 Weather effect 10% 10% 10% Rumen fill 100% 100% 100% Min. fodder 0.2 0.2 0.2 Home grown fodder 0.28 0.42 0.52 Other feeds 0.12 0.12 0.12 Fodder efficiency 80% 80% 80% Pasture management was confirmed as a key profit driver in all pasture-based feeding systems, regardless of the level of concentrates fed (low, moderate or high). Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 41 Farm A (low concentrate feeding level) 2006/07 Profit distribution curve ➔ Mean EBITD = $110,948 12% chance of $200,000 EBITD Variable inputs used in simulations for model farm A: Variable Pasture MJ ME/ kgDM Pasture utilised (t DM /Ha) Core Cost Per Cow Milk Price ($/kgMS) Pasture NDF Quality Concentrates $ / t DM Purchased Fodder $/t DM Input Downside Upside Base Case 11.2 10.5 12.2 8.90 6.41 11.56 $1,055 $889 $1,221 $4.46 $4.07 $4.85 45% 43% 48% $425 $339 $511 $147 $93 $222 Tornado chart $134,362 $121,679 $12,683 Farm EBITD -$100,000 $0 Pasture MJ ME/ kgDM Pasture utilised (t DM /Ha) Core Cost Per Cow $100,000 $200,000 10.5 12.2 6.41 11.56 $1,221 Milk Price ($/kgMS) $4.07 Pasture NDF Quality 48% Concentrates $ / t DM Purchased Fodder $/t DM $889 $4.85 43% $511 $222 Figure 38a: Monte Carlo simulation for low concentrate farm in 2006/07. 42 Farm EBITD Downside Upside Range $64,402 $221,751 $157,348 -$11,967 $128,984 $140,952 $194,388 $63,581 $130,806 $77,086 $180,882 $103,796 $174,980 $78,402 $96,579 $144,843 $113,126 $31,718 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study $339 $93 $300,000 Farm B (moderate concentrate feeding level) 2006/07 Profit distribution curve ➔ Mean EBITD = $145,326 30% chance of $200,000 EBITD Variable inputs used in simulations for model farm B: Tornado chart Input Variable Pasture utilised (t DM /Ha) Core Cost Per Cow Milk Price ($/kgMS) Pasture MJ ME/ kgDM Concentrates $ / t DM Pasture NDF Quality Purchased Fodder $/t DM Farm EBITD Downside Upside Base Case Downside Upside Range 11.56 6.41 8.90 -$12,624 $205,085 $217,709 $1,524 $1,108 $1,316 $234,672 $67,768 $166,903 $4.85 $4.07 $4.46 $73,450 $228,990 $155,541 12.2 10.5 11.2 $89,644 $239,668 $150,025 $511 $339 $425 $201,957 $100,483 $101,474 48% 43% 45% $170,927 $127,398 $43,529 $222 $93 $147 $163,938 $133,944 $29,994 Farm EBITD -$100,000 Pasture utilised (t DM /Ha) Core Cost Per Cow Milk Price ($/kgMS) Pasture MJ ME/ kgDM Concentrates $ / t DM Pasture NDF Quality Purchased Fodder $/t DM $0 $100,000 $200,000 6.41 $300,000 $400,000 11.56 $1,524 $1,108 $4.07 $4.85 10.5 12.2 $339 $511 48% $222 43% $93 Figure 38b: Monte Carlo simulation for moderate concentrate farm in 2006/07. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 43 Farm C (high concentrate feeding level) 2006/07 Profit distribution curve ➔ Mean EBITD = $160,210 36% chance of $200,000 EBITD Variable inputs used in simulations for model farm C: Variable Pasture utilised (t DM /Ha) Milk Price ($/kgMS) Core Cost Per Cow Concentrates $ / t DM Pasture MJ ME/ kgDM Pasture NDF Quality Purchased Fodder $/t DM Downside 6.41 $4.07 $1,321 $339 10.5 43% $93 Input Upside Base Case 8.90 11.56 $4.46 $4.85 $1,569 $1,817 $425 $511 11.2 12.2 45% 48% $147 $222 Farm EBITD Downside Upside Range -$13,225 $226,530 $239,755 $58,883 $267,048 $208,164 $266,915 $59,015 $207,900 $246,265 $79,666 $166,599 $98,430 $255,664 $157,234 $183,652 $137,960 $45,692 $177,208 $143,617 $33,591 Farm EBITD Tornado chart -$100,000 Pasture utilised (t DM /Ha) $0 $100,000 $200,000 6.41 11.56 Milk Price ($/kgMS) $4.07 $4.85 Core Cost Per Cow $1,817 $1,321 Concentrates $ / t DM Pasture MJ ME/ kgDM Pasture NDF Quality Purchased Fodder $/t DM $511 $339 10.5 12.2 48% $222 43% $93 Figure 38c: Monte Carlo simulation for high concentrate farm in 2006/07. 44 $300,000 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study $400,000 Where to from here? The dairy sector must accept that for all feeding systems there is a mix of risk, performance and management principles whose understanding can improve the chances of success. The traditional and outdated mindset is that risk is to be avoided. (For example, if you’re worried about the rising cost of grain / concentrates, not using that feed type completely avoids exposure to price risk). This needs to be replaced by the concept of risk-swapping as a far more practical solution to risk mitigation than a futile attempt to avoid risk. In conjunction with other stakeholders, particularly the banks, the dairy sector needs to agree on standard risk measures and to explore new ways (or old ways newly used) to quantitatively assess risk so that coping with risk can be better handled and appreciated. Extension messages to farmers about risk need to be sound, consistent and unambiguous. Risk should not be used inappropriately by champions of one production / feeding approach to argue that theirs is better or lessrisky than another. The TasMilk60 study has demonstrated how a probabilistic budgeting approach to risk analysis can be useful at an industry level in comparing different farm production / feeding systems on the basis of their probability distributions of operating profit, rather than on single values of operating profit for each system (as discussed by Chapman et al. 2007). However, through the TasMilk60 study, Dairy Australia’s Grains2Milk program and Agrilink FarmStats have also recognised the value a probabilistic budgeting approach may also provide at the individual farm level in the annual forward planning process used by dairy farmers. The conventional annual forward planning process usually begins with an annual budget. Single, “mostlikely” unit values for key financial drivers such as milk price and supplementary feed costs and for key biophysical drivers such as milk yield and concentrate and fodder feeding levels are selected and used to derive key income and cost values. Prior year spending on non-feed operating costs and debt-servicing costs (if any) is used as a guide to plan spending in the upcoming year(s). For many, if the budget result is a surplus, that is the end of the planning process. Others extend the process to include a sensitivity analysis, often on milk price alone, but sometimes including selected costs such as grain or fertiliser. However, sensitivity analyses that place equal weight on the outcomes generated by pairing variable input measures (e.g. milk price and concentrate price) deny reality, which is that certain combinations of input values have a higher or lower likelihood of occurring in the farmer’s real world. Unfortunately, despite this methodological weakness, many dairy farm businesses rely on this approach to prepare for the shock of the unknown and unpredicted. Such an annual forward planning process can lead farmers to make decisions that do not use all the best information available (Malcolm et. al. 2005). If increased market and climatic volatility are now a constant in the dairy farmer’s world, such sub-optimal planning procedures could have high real costs and avoidable negative impacts in rural communities. As an extension of the TasMilk60 study, Dairy Australia’s Grains2Milk program has commissioned Agrilink FarmStats to undertake a pilot study to assess the benefits of an alpha version of a new probablistic risk assessment tool for farmers dubbed ‘Risk Tamer’. At the time of writing this booklet, the pilot study was still in progress. What this means for industry Risk and uncertainty rule all dairy farming decisions. Attempting to avoid risks is often futile or counterproductive. Instead, risk exposure can be reduced by spreading, selling or shifting risk, or by risk averaging. For all feeding systems there is a mix of risk, performance and management principles whose understanding can improve the chances of successfully balancing risk and reward. Pasture utilisation, pasture quality and core costs per cow are key profit drivers in all pasture-based feeding systems, regardless of the level of concentrates fed. Milk price has a greater impact on profit variability at higher levels of concentrate feeding. Farmers are inclined to over-react to changes in grain/concentrate price. A farmer has more control over risk than first meets the eye. One of the Australian dairy industry’s strengths is the diverse range of farmers and production/feeding systems used – we should celebrate this rather than promote or defend particular systems! Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 45 46 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Section 5 Other uses of TasMilk60 data to date As previously described, Grains2Milk’s TasMilk60 dataset has been used to: • estimate the financial value of incremental change in feed conversion efficiency (see page 20) ; • explore whether more profitable farmers tend to have a particular farming style (see page 37); and • assess risk using a probabilistic budgeting approach (page 40). In addition to these, the TasMilk60 farm dataset has also been used in two other ways: DairyMod pasture study (The University of Melbourne) What was done • animal physiology and production; and Brendan Cullen, Institute of Land and Food Resources, The University of Melbourne, carried out a range of simulations using DairyMod, with a view to establishing the degree of difference between actual pasture growth and non-limited potential pasture growth across the TasMilk60 farms. These simulations used data from 35 TasMilk60 farms for 2006/07 only. • pasture management, irrigation and fertiliser application. DairyMod is a simulation model used for predicting and understanding the consequences of weather and management decisions on dairy grazing systems. The model predicts outcomes according to: • pasture growth and utilisation by dairy cows; • water and nutrient dynamics; Daily climate data is required for each model and details about soil types, pasture species, fertiliser use, irrigation, animal type and grazing systems must be specified. The model produces annual summaries and daily data of pasture and animal production, water balance and soil nutrients. Figure 39 is an example of DairyMod output. The method and main assumptions used in the DairyMod simulations of the TasMilk60 farms were: 1. Climate data was obtained from Bureau of Meterology DataDrill using latitude and longitude of nearest town. Pasture: • biomass • species composition • quality Livestock: • animal intake • liveweight and milk production • stocking rate Climate: • rainfall • temperature Soil characteristics: • moisture • nitrogen All paddocks Farm summary Figure 39: Sample DairyMod output. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 47 2. Soil type was specified by the farmer, with clay loam as the default. 3. Herd size and calving pattern were specified by the farmer. 4. All pastures were assumed to be perennial ryegrass/ white clover. 5. Farm area was as specified by the farmer, but all were given 50 paddocks with an allocation to dryland and irrigated on a pro rata basis. 6. Grazing management was the same on all farms, i.e. graze at three-leaf stage to a residual of 1.4t/DM/ha. 7. The supplementary feeding rules used were the same for all farms. 8. Lactation curves were the same for all farms. 9. N application – total N applied was split between irrigated and dryland areas at a 3:1 ratio of kg/N/ha. Timing and rate were defined as: Dryland – generally one application in autumn and 2-3 in late winter/spring with individual application rates up to 45kg/N/ha; and Irrigated – generally 4-6 weeks between applications, except none applied in June-July and individual application rates of 30-60kg/N/ha. Pasture growth rate (kg DM/ha.day) 140 Pasture growth rate (kg DM/ha.day) 140 120 100 ◆ ■ ■ ◆ 40 20 Figure 40 shows the simulated outcomes for pasture growth rates where water and N fertiliser were limited (farm actual) and unlimited (optimal). A geographical representation of the regional differences in pasture growth and consumption across northern Tasmania is given in Figure 41. These data show that the predominantly irrigated ‘Easterners’ have a greater potential to increase pasture growth rates than the pre-dominantly rain-fed ‘Westerners’, even though both groups have similar single-digit pasture utilisation rates. ■ ◆ ◆ ■ Mean +/- SE (t DM/ha) Eaten: 9.72 +/- 0.86 Potential: 11.53 +/- 0.87 growth ■ ■ ◆ ■ ◆ Mean +/- SE (t DM/ha) Eaten: 8.28 +/- 0.53 Potential: 15.66 +/- 0.84 growth ◆ ■ Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun ▼ ▼ ◆ Farm N and water 120 100 ◆ ■ 80 ◆ ■ 60 ■ ◆ ◆ ■ ■ Optimal N and water ■ ◆ ■ ◆ 40 ■ ■ ◆ ◆ ■ ■ ◆ ◆ ■ ◆ ◆ ■ 0 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Figure 41: Geographical differences in pasture growth and consumption across northern Tasmania. Figure 40: Modelled pasture growth rates for farm (actual) and optimal water and nitrogen fertiliser applications. 48 Easterners (59% irrigated) ■ ◆ ◆ 0 20 What was found Westerners (17% irrigated) ■ ◆ ◆ ■ 60 If the farm applied less irrigation than the model applied, pasture production was scaled back according to the rule 1/ML/ha = 1t/DM/ha. ◆ Farm N and water ■ Optimal N and water ◆ ■ 80 10. Irrigation was applied in response to soil water deficit of 40mm (calculated over top 50cm soil), with water volume applied to fill to field capacity: Irrigation season – 1 October to 30 April; minimum of three days between irrigation events. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Greenhouse gas emissions study What was found (TIAR) The study found that GHG production increases linearly as farm milk production increases (Figure 43). However, there is much to be discovered about the link between GHG and milk production, particularly at the higher end of the scale. What was done Karen Christie and Richard Rawnsley, TIAR climate change researchers, examined the greenhouse gas (GHG) emissions of the TasMilk60 farms using the Dairy Greenhouse Gas Abatement Strategies calculator (DGAS) and demonstrated differences in GHG emissions between farms with different feeding approaches (Figure 42). A paper about this study, A whole farm systems analysis of the greenhouse gas emissions of 60 Tasmanian dairy farms, co-authored by Karen Christie, Richard Rawnsley and Richard Eckard, was presented at the 4th International Greenhouse Gases and Animal Agriculture Conference in Banff, Canada, in October 2010, and will be published in a special issue of the Animal Feed Science and Technology Journal in mid-2011. 10.0 4,000 8.0 3,000 6.0 2,000 4.0 1,000 0 Total farm intensity emissions (t CO2-e/t MS) Total farm GHG emissions (t CO2-e) Figure 42: GHG emissions and GHG per tonne milk solids for TasMilk60 farms. 2.0 0.0 3,000 2,000 1,000 0 450,000 5,000 4,000 400,000 12.0 350,000 6,000 300,000 14.0 250,000 7,000 5,000 200,000 16.0 6,000 150,000 8,000 7,000 100,000 18.0 The TasMilk60 dataset has provided a useful insight into how different feeding and stocking rate strategies can affect GHG emissions at the herd level. 50,000 9,000 There is no strong relationship between pasture utilisation (t DM/ha) and GHG emissions (Figure 47). 0 20.0 It is conjectured that these effects are a direct consequence of some TasMilk60 herds choosing to rely more on higher starch diets to lift production through the use of grain/concentrates. Total farm GHG emissions (t CO2-e) 10,000 GHG production per unit of MS produced declines as feed conversion efficiency improves (Figure 44), as milk yield per cow rises (Figure 45) and as the proportion of concentrates in the cow’s diet increases (Figure 46). Milksolids per farm Figure 43: GHG emissions in relation to farm milk production for TasMilk60 farms. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 49 20.0 18.0 19.0 16.0 18.0 GHG intensity (t CO2-e/t MS) GHG emissions intensity (t CO2-e/t MS) 20.0 14.0 y = -0.1676x + 28.084 12.0 R2 = 0.5516 10.0 8.0 6.0 4.0 2.0 0.0 17.0 16.0 15.0 14.0 13.0 y = -0.0091x + 18.801 12.0 R2= 0.3894 11.0 50 55 60 65 70 75 80 85 90 95 10.0 200 100 250 300 350 400 FCE (kg MS/t DM intake) 20.0 20.0 19.0 19.0 18.0 18.0 17.0 16.0 15.0 14.0 13.0 12.0 5 10 650 700 y = -0.0609x + 14.931 R2= 0.0073 17.0 16.0 15.0 14.0 13.0 12.0 0 15 20 25 30 35 40 % concentrate in diet Figure 46: GHG emissions in relation to percentage concentrate in the diet for TasMilk60 farms. 50 600 10.0 R = 0.3545 2 0 550 11.0 y = -9.9911x + 16.691 10.0 500 Figure 45: GHG emissions in relation to milk yield per cow, expressed as kg MS/cow for TasMilk60 farms. GHG intensity (t CO2-e/t MS) GHG intensity (t CO2e/t MS) Figure 44: GHG emissions in relation to feed conversion efficiency expressed as kg MS/t DM for TasMilk60 farms. 11.0 450 kg MS/cow 45 2 4 6 8 10 12 14 16 18 Pasture consumption (t DM/ha) Figure 47: GHG emissions in relation to pasture utilisation, expressed as t DM/ha for TasMilk 60 farms. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Section 6 Conclusions The findings from Grains2Milk’s TasMilk60 study support the following conclusions: 1. Farm performance Despite fears of substitution, there is no simple relationship between amount of concentrate fed per cow and pasture utilised per hectare. Both good and poor pasture utilisation is seen at all levels of concentrate feeding. Farms that feed lower levels of concentrate per cow get their cows to eat more pasture. However, there is a larger variation in pasture intake per cow between farms feeding the same level of concentrate than there is between farms feeding different levels of concentrate, so herd and grazing management are critical. Milk protein to fat ratio is generally higher on farms where higher rates of concentrate are fed. This is largely due to reduced fat concentrations found in milk with higher daily milk volumes per cow. Feed conversion efficiency (FCE) is highly variable at all concentrate feeding levels. It is generally better on farms where higher amounts of concentrate per cow are fed. This is largely due to the higher total intakes in high concentrate-feeding farms and, hence, the greater proportion of nutrients that are used for milk production and the lesser proportion used for maintenance. Total feed intake per cow explains more than half the variability in FCE. However, there are also several other important factors known to help optimise FCE, including maintaining high feed quality and good rumen function, and minimising feed gaps, feed wastage and energy losses. protein concentrations, and the seasonality of milk supply patterns. Farmers aiming to increase profitability by reducing costs should review and monitor feed costs, but labour and management costs, as these generally constitute the highest proportions of operating costs and vary widely between farms. Conversely, herd, shed and overhead costs are relatively small components of operating costs and are less variable between farms. 2. Farm profitability In farm populations such as that examined by the TasMilk60 study, milk EBITD per cow, hectare, litre, and kilogram milk solids, and return on capital (based on milk EBITD) are closely correlated with each other, so are likely to tell the same story about farm profitability. There is no ‘best’ concentrate feeding level or production / feeding system. Any concentrate feeding level or production / feeding system can be profitable in any year, given an appropriate mix of management, milk price and input costs. The differences in average and median profits achieved between farms using low, moderate and high concentrate feeding levels are small compared with the variability between farms within each concentrate feeding level. When a dairy farm changes its production / feeding system, adjustments are needed (cows, people, farm infrastructure) to realise the full benefits. It may take several years before the benefits are fully realised. Dairy farms are complex systems and there are many ways to make a profit (or a loss). Determinants of profit should be assessed collectively, not separately, when analysing farm performance. Farms that have consistently higher profits usually have a relatively higher milk price, higher milk yield per cow, lower fodder costs and lower labour and management costs. However, they tend not to be outstanding performers for any of these particular determinants of profit – they tend to be consistent all-rounders whose efforts for these profit determinants collectively are superior. Income from milk sales is as important a component of profit as are costs – higher feed costs may be justified if they generate extra profit by lifting milk income. Milk prices paid to farmers (even those supplying the same manufacturer) vary widely due to on-farm factors under their control, including milk volumes, milk fat and Most farms maintain a ‘business as usual’ approach in high milk price years and do not take opportunities to substantially increase farm profit above what the milk Increasing FCE within a farm’s chosen feeding system can be a powerful lever for increasing farm profit, particularly in higher milk price years. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 51 price alone delivers. Few farms make large systemic management changes year to year. Beneficial effects on farm profitability of increasing enterprise size are generally relatively small and are likely to be due to scale, not efficiency. High or low relative farm profitability is not very repeatable when milk prices fluctuate widely year to year, so results in a single year may not reflect profit performance over the longer term. Farmers do not require a certain management style or set of attitudes and beliefs to be highly profitable. 3. Risk Risk and uncertainty are central to all dairy farming decisions. Attempting to avoid all risk is often futile or counter-productive. Instead, risk exposure can be reduced by spreading, selling or shifting risk, or by risk-averaging. For the dairy farmer, there is always a trade-off between risk and reward. Those with a low risk tolerance will seek options where little risk is involved and require a very high reward for the risk involved. Those with a higher tolerance for risk will be willing to accept risk without such a big potential reward. 52 For all production/feeding systems, there is a mix of risk, performance and management principles whose understanding can improve the chances of successfully balancing risk and reward. Risk measures should provide the probability of unfavourable and favourable outcomes occuring. Unfortunately, many so-called risk measures currently used in the dairy industry do not. These erroneous risk measures include pasture as a percentage of total feed consumed, unit cost of milk production and operating profit margin. Pasture utilisation, pasture quality and core costs per cow are key profit drivers in all pasture-based dairy feeding systems, regardless of the level of concentrates fed. These three on-farm profit drivers can have a greater impact on profit variability than the off-farm profit drivers of milk price, concentrate price and purchased fodder price, over which farmers have less control. Steve Little Grains2Milk program leader for Dairy Australia Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Section 7 Appendices – Further findings from TasMilk60 Tips and traps when collecting farm data By Gordon Cleary (Agrilink FarmStats P/L) The TasMilk60 project team had to deal with several common account allocation errors as part of its quality assurance process. The points below may be helpful for anyone collecting farm data for analysis. Labour costs • Costs of off-farm work by farm employees, e.g. sons as off-farm contractors. Passive ‘pigeon-hole’ errors • Internal cost transfers between related entities/family members, e.g. kids ‘wages’. These include: • Construction labour costs that should be capital works. • Value of imputed labour and management excluded. Herd costs • Non-dairy animal health costs included. • Leased land or agistment for replacements missing. • Concentrates and fodder fed to young stock missing (buried in milker feed costs). • A livestock breeding enterprise with costs above and beyond that needed for normal milker replacements (e.g. for Asian heifers). Shed costs • Electricity uses not related to the dairy facility, e.g. houses, irrigation pumping, workshop, all running off one meter. • Hibitane/teat dip/teat spray buried in ‘chemicals’. • Livestock requisites (herd costs) buried in ‘dairy supplies’. Feed costs • Cost of feedstuffs for replacement animals included. • Feed costs for other enterprises, e.g. beef, alpaca, sheep, emu, ostrich, pigs. • Use of hay or silage in one year that was made and paid for in the previous year, that is, an inventory issue. Overhead costs • One-off repairs that will not recur in the coming year. • Overhead costs that do not relate to the dairy enterprise. Active ‘pigeon-hole’ errors Farmers may be inclined to expensing capital spending within their farm accounting systems, leaving it to their tax accountants to remove the more outrageous spending items from operating costs. It is a trap for the unwary that farm development works do not always show up as lumpy big outlays. For instance, the wages for a labourer constructing a new calf shed should be capitalised as it is not a recurrent operating cost incurred in producing milk. In the TasMilk60 study, engaging the farmers in a discussion about farm capital works almost always allowed necessary adjustments to be identified. Unless capital works spending is removed, those farm businesses that are sufficiently profitable to support capital spending out of cash flow appear to have a high unit cost of production and low calculated profitability. • Feed dollar spend in the year is different from the dollar value of the feed used in the year, usually through end-of-year pre-purchases. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 53 Six principles for identifying determinants of profitability by analysing data from groups of dairy farms By John Morton, Epidemiologist (Jemora P/L) 1. Don’t observe only high (or low) profitability farms When identifying determinants of profitability, observations of just high profitability farms can be misleading. For example, suppose that TasMilk60 had studied just the 15 farms with milk EBITD per cow above $300 in 2006/07. In assessing whether milk yield per cow is a determinant of profitability, we could describe the milk yields in these 15 high profitability farms; the median litres per cow was 6,058. This is high compared to Tasmanian industry means of just over 4,000 litres per cow around the time that the study commenced, inviting the conclusion that higher milk yield per cow results in higher farm profitability. This conclusion would be erroneous. On the remaining 53 farms (those with milk EBITD per cow of $300 or less), the median litres per cow was 6,674, higher than for the high profitability farms. In fact, as discussed in Section 3, there was no simple relationship between milk yield per cow and farm profitability, and both farms with low and high milk yields per cow could be highly profitable. The same limitation applies with observations of just low profitability farms. (shown on page 28), milk price (¢/L) was proposed to be determined by milk protein concentration, milk fat concentration, calving system (as a surrogate for seasonality of milk supply), manufacturer and year. Accordingly, all five variables were accounted for when assessing determinants of milk price. 3. Incorporate unexplained variability into interpretation The human mind tends to place insufficient emphasis on unexplained variability (or ‘scatter’); this can result in us attributing importance to numeric patterns that may well be due simply to random variability. Even the most apparently convincing relationships on visual analysis should be assessed statistically. A strong or large effect is meaningful only if it is strong or large relative to the scatter; an apparently strong or large effect among study subjects is quite compatible with truly no effect if the scatter is extreme. For example, consider the relationship between stocking rate and pasture utilised per hectare in 2006/07 (Figure 48). This relationship is quite consistent with important indirect effects of stocking rate on pasture utilised per hectare. But we would be much less certain about these Multivariable analytical methods and restriction of analyses to sub-groups were used extensively in the TasMilk60 study, primarily to remove confounding. Path modelling allowed potential confounders to be defined (in the null path models), and multivariable and restricted analyses were performed to account for these potential confounders. For example, in the null path model 54 10 5 Pasture utilised per hectare (t DM) Results of univariable associations should be viewed only as preliminary reflections of reality, and to identify relationships worthy of more detailed investigation. A major reason for this caution is confounding. 0 Determinants of profitability on dairy farms are sometimes studied using univariable (or crude) associations. Indeed, such associations were used extensively in this report. For example, in the TasMilk60 study, potential determinants of changes in key physical and financial changes between years were assessed using univariable associations, as were associations between attitudinal indices and measures of farm profitability. 15 2. Consider confounding 0 1 2 3 4 Stocking rate (cows per hectare) Figure 48: Associations between pasture utilised per hectare and stocking rate in 2006/07 in farms where irrigation was used that were feeding <1t concentrates per cow (green squares), 1t to <2t (blue triangles), and ≥2t (gold diamonds). Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study effects if, instead, the relationship appeared as in Figure 49. This relationship is identical to that in Figure 48, except that additional random ‘scatter’ or unexplained variability has been added. Statistical methods such as regression analysis and analysis of variance explicitly assess possible effects in relation to this ‘scatter’ or unexplained variability. P-values and confidence intervals were used throughout this study to incorporate unexplained variability into interpretation. Note that this is an example of a univariable association and, as discussed above, potentially confounded. However, a positive association was also present on TasMilk60 farms where <2t of concentrates were fed after adjusting for several potential confounders. 4. Account for similarity of observations from the same farms between years in statistical analyses In the TasMilk60 study, management and profitability on dairy farms were studied using physical and financial data collected from the same farms over multiple years. The unit of analysis for most of this work was the farm year. Each farm contributed one farm year of data for each year in which it was enrolled. This raises a possible complexity as observations in one year may well be similar to those from another year on the same farm. 15 Three strategies were used in the TasMilk60 study to account for the similarity of observations from the same farms between years. Some analyses were conducted separately within each study year, so avoiding including repeated observations from the same farm in the same analysis. Second, each farm was fitted as a ‘random effect’ to account for any such similarity of observations from the same farms between years. Third, when analysing proportions (or percentages), robust standard errors adjusted for clustering were used. 5. Do not aggregate data from individual farms to produce means etc for groups of farms When studying management and profitability on dairy farms, it is possible to aggregate data from individual farms to produce means etc for groups of farms. For example, to assess the relationship between milk yield per cow and milk EBITD per cow, it would have been possible to calculate means of milk EBITD per cow for farms grouped into various milk yield per cow categories and to compare these means using linear regression. This approach would be valid if the research question was to describe this relationship for groups of farms (i.e. do groups of farms with higher mean milk yield per cow have higher mean milk EBITD per cow). However, when studying as management and profitability on dairy farms, inference is usually desired at the individual farm level. For example, if the study results are to be used to help farm managers understand what may happen on their farm with a particular series of management changes, inference is clearly being made to the individual farm or farm year level. Given this, such studies should be conducted with the farm (or farm year) as the unit of analysis. 5 10 6. Consider the stories behind the regression models 0 Pasture utilised per hectare (t DM) If repeated observations within a farm are more similar to each other than would be expected based on the overall variability in the data, we would say that the data are ‘clustered’ within farms. Such repeated observations from the same farm should not be considered as statistically independent observations. 0 1 2 3 4 Stocking rate (cows per hectare) Figure 49: Associations between pasture utilised per hectare and stocking rate in 2006/07 on farms where irrigation was used that were feeding <1t concentrates per cow (green squares), 1t to <2t (blue triangles), and ≥2t (gold diamonds) with added random ‘scatter’ or unexplained variability. Given the physical and financial complexity of dairy farm businesses, most practically achievable analyses are unlikely to fully explain all differences in profitability between farms or farm years. Regression models essentially describe general relationships across the study population, and individual farms will not fit these general patterns. This was demonstrated in the TasMilk60 study where some high profitability farms were shown to have quite high feed costs per cow, even though, all else being equal, profitability would have been higher when feed costs per cow were lower. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 55 A similar approach was used to explore differences between consistently high profitability farms and consistently low profitability farms. There are many ways to ‘skin a cat’ and numerous ways to run a profitable dairy farm. These approaches can add to the results of regression models to build a richer picture of why profitability varies between dairy farms. 56 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study Section 8 References Beever, D.F. and Doyle, P.T. (2007), Feed conversion efficiency as a key determinant of dairy herd performance: a review. Australian Journal of Experimental Agriculture. 47: 645–657. Malcolm, L.R. (2001), Farm Management Economic Analysis: A Few Disciplines, a Few Perspectives, a Few Futures. Australian Agribusiness Perspectives, Paper 42. Chapman, D.F., Malcolm, L.R., Neal, M. and Cullen, B.R. (2007), Risk and uncertainty in dairy production systems: Research concepts, tools and prospects. In: D.F. Chapman, D.A. Clark, K.L. Macmillan, D.P. Nation, eds. Meeting the Challenges for Pasture-Based Dairying, Proceedings of the Australasian Dairy Science Symposium, Melbourne, Australia, National Dairy Alliance. pp 476-491. Malcolm, L.R., Makeham, J. and Wright, V. (2005), The farming game: Agricultural management and marketing. Cambridge University Press, Cambridge, UK. Christie, K., Rawnsley, R. and Eckard, R. (2011), A whole farm systems analysis of the greenhouse gas emissions of 60 Tasmanian dairy farms, Proceedings of the 4th International Greenhouse Gases and Animal Agriculture Conference, Banff, Canada, October 2010. Gronow, C.S., Little S.B. and Cleary G.V. (2010), Managing production system risk: An application of Monte Carlo simulation to Tasmanian dairy farm data, Proceedings of the 2010 Australasian Dairy Science Symposium, Lincoln University, New Zealand. Malcolm, L.R. (2009), Managing Uncertainty Pragmatically in Private and Public Decision-making about Investment, Australian Agribusiness Perspectives, Paper 81. Pasture Consumption and Feed Conversion Efficiency Calculator Instruction Manual (2010). Dept. of Primary Industries (DPI) Victoria. http://new.dpi.vic.gov.au/agriculture/dairy/pastures-hay-silage/calculator. Slovic, P. (2000) ed., The Perception of Risk. Earthscan, Virginia. 2000. Thomson, D. (2001), As if the landscape matters: the social space of ‘farming styles’ in the Loddon Catchment of Victoria. Unpublished PhD thesis, The University of Melbourne. Hardaker, J.B., Huirne, R.B.M. and Anderson, J.R. (1997), Coping with risk in agriculture, CAB International. Topp, V. and Shafron, W. (2006), Managing Farm Risk, ABARE eReport 06.6. Hardaker, J.B. and Lien, G. (2007), Rationalising Risk Assessment: Applications to Agricultural Business. Australasian Agribusiness Review 15:75-93. USA Farm Financial Standards Council website http://www.ffsc.org/Files/ FarmFinancialGuidelinesRatios.pdf Little, S.B. (2010) Feed Conversion Efficiency – A key measure of feeding system performance on your farm. Dairy Australia Grains2Milk program. http://www.dairyaustralia.com.au/Animals-feed-and-environment/ Feeding-and-nutrition Waters, W., Thomson, D., Nettle, R. (2006) Derived attitudinal farmer segments: A method for understanding and working with the diversity of Australian dairy farmers, Extension Farming Systems Journal, Volume 5, Number 2. Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 57 58 Performance, Profit and Risk in Pasture-based Dairy Feeding Systems – Findings from the TasMilk60 study 4 Performance, Profit and Risk in Pasture-based Dairying Systems – Findings from the TasMilk60 study ISSN/ISBN/REF 0000 Dairy Australia Ltd ABN 601106227 Level 5, IBM Centre 60 City Road Southbank VIC 3006 T + 61 3 9694 3777 F + 61 3 9694 3888 www.dairyaustralia.com.au