Spatial Market Segmentation & Healthy Productive Landscapes Framework
Published by the Victorian Government Department of Primary Industries
Tatura, 29 April 2008.
ISBN 978-1-74199-971-6
| Authors: | |
|
Elizabeth Morse-McNabb and Andy McAllister |
Geoff Kaine, Ben Rowbottom and Chris Linehan |
Acknowledgments:
The authors would like to thank members of the project steering committee for their guidance, landowners of the North Central and Goulburn Broken catchments who gave up their valuable time and input into this research and Lisa Cowan for her contribution to the interviews.
Table of Contents
- Abstract
-
Introduction
- 2.1 Spatial Market Segmentation
-
Materials & Methods
- 3.1 Collation & Data Outputs
- 3.2 Descriptive Statistics & Spatial Representation
- 3.3 Water Market Segments
- 3.4 Landholder Interviews
- 3.5 Regression Analysis
-
Results
- 4.1 Changes in Perennial Pasture Across the Regions
- 4.2 Classification of Trading Segments
- 4.2.1. Permanent Sellers
- 4.2.2. Temporary Sellers
- 4.2.3. Small Traders
- 4.2.4. Small Temporary Buyers
- 4.2.5. Large Temporary Buyers
- 4.2.6. Permanent Buyers
- 4.2.7. Trading Segments Among Dairy Farmers in NCIR and SIR
- 4.3. Interview Results
- 4.3.1 Confirming Accuracy of Spatial Data
- 4.3.2 General Findings
- 4.3.3 Permanent Sellers
- 4.3.4 Temporary Sellers
- 4.3.5 Small Traders
- 4.3.6 Small Temporary
- 4.3.7 Large Temporary Buyers
- 4.3.8 Permanent Buyers
- 4.3.9 Conclusion
- 5.1. Relevance of Results to HPL Framework
- 5.2. Strength in Spatial Market Segmentation Approach
- 5.3. Limitations
- 5.4. Further Research
- 5. Discussion
- Conclusions
-
References
- 8.1. Appendix I - Assumptions
- Appendix
Figures
Figure 1: NC CMA Healthy and Productive Landscapes Framework
Figure 2: Flow chart of work process
Figure 3: Land use by enterprise type across irrigation district*
Figure 4: Change in perennial pasture cover between the1996/1997 and 2003/2004 irrigation seasons
Figure 5: Spatial distributions of water trade segments
Figure 6: Scatter plot with linear regression line of the relationship between water entitlement plus average net temporary water trade and perennial pasture
Tables
Table 1: Perennial pasture cover in the North Central and Shepparton Irrigation Regions
Table 2: Profiles for water trade segments for all enterprise types from 1996/97 to 2003/04
Table 4: Profiles for Shepparton Irrigation Region dairy enterprises between 1996/97 and 2003/04
Table 5: Profiles for the North Central Irrigation Region dairy enterprises between 1996/97 and 2003/04
Table 6: Regression model results for dairy farmers in the NC and GB CMA’s
1. Abstract
The aim in this project was to develop and pilot a method for classifying landholders into segments (that is, groups such that landholders exhibiting similar characteristics are in the same group, while landholders exhibiting different characteristics are in different groups) and to spatially map those segments. The expectation was that such a method would enable natural resource managers to target investment in policy instruments to greater effect. The project was funded through the North Central Catchment Management Authority (NC CMA) and the Goulburn Broken CMA (GB CMA) and involved researchers from the Spatial Sciences Group and Practice Change Research, DPI, Tatura.
The method was developed and piloted in an investigation of the relationship between landholder participation in the water market and changes in the area of irrigated perennial pasture in the Shepparton Irrigation Region (SIR) and North Central Irrigation Region (NCIR). This relationship was selected because the Spatial Science Group had recently developed a procedure for using spatial mapping techniques to distinguish between and measure, perennial and seasonal pasture. Furthermore, the team in Practice Change Research had developed a method for classifying landholders into segments based on differences in their farm context and for predicting differences in the behaviour of landholders.
The hypothesis to be tested in this project was that differences among landholders in the area planted to perennial pasture were associated with differences in the water available to landholders for irrigation and differences in the water available for irrigation were brought about by the water trading behaviour of landholders.
We have demonstrated that it is possible to use spatial data on land cover and water use to link variations in land cover with water use and water trading by dairy farmers. Importantly we demonstrated how quantitative data on spatial land cover and water use can be usefully combined with qualitative data to explain water trading by dairy farmers. This combined analysis allowed us to identify different types of trading, to map where the different types were occurring and to explore why they were happening.
The commonly held assumption that the area of perennial pasture is decreasing in the SIR and NCIR was supported by the findings. The results indicate that in the SIR and NCIR the area of perennial pasture decreased by at least 20 per cent between the 1996/1997 irrigation season and the 2003/2004 irrigation season, for the dairy industry and across all industries.
Landholders were classified into six water trade segments using data on water trading (permanent sellers, temporary sellers, small traders, small temporary buyers, large temporary buyers and permanent buyers). The area of perennial pasture declined in all segments; however, farmers in segments that sold water, either permanently or temporarily, appeared to have reduced their perennial pasture by a greater proportion than farmers in other segments.
Interviews with dairy farmers revealed that in general, farmers’ participation in the water market reflected either a need to purchase water to assure supplies of irrigation water where this was a constraint on the area of perennial pasture; or an opportunity to sell water where their water supplies were in excess of requirements for perennial pasture.
The results indicate that, where the behaviour of farmers is strongly influenced by elements in their farm context that are measurable and documented (such as water entitlements and/or area of perennial pasture) different farm contexts can be identified across landscapes using spatial mapping techniques. This means, in principle and in the right circumstances, the spatial effects of policy instruments can be identified.
2. Introduction
Natural resource managers, such as those in Catchment Management Authorities and government departments such as the Department of Sustainability and Environment (DSE) and the Department of Primary Industries (DPI), use policy instruments such as extension and incentives to encourage landholders to change their land management practices. Understanding how and where to target these instruments within the community and across the landscape, is a constant challenge.
For example, the North Central Catchment Management Authority (NC CMA) is the statutory authority responsible for the coordination of natural resource management programs within the North Central region. Under the Catchment and Land Protection Act 1994, the NC CMA is required to develop a Regional Catchment Strategy that establishes the structure for planning the management of land, water and biodiversity resources in the region. The NC CMA has established a framework, the Healthy and Productive Landscapes (HPL) Framework, for reconciling and integrating the targets set in the regional catchment strategy with the contextual reality of the region and the various policy options that are available (Figure 1).
The two main objectives for the HPL Framework are to assist the setting of priorities for action and helping in selecting policy responses and instruments that suit the contextual, including spatial, characteristics of the region. The starting point for the HPL Framework is the asset based approach enshrined in the regional catchment strategy. The protection of assets requires identification of critical areas that require managing, the identification of key threats to those areas and an assessment of the degree to which those areas are at risk. A critical component in the strategy is identification of policy instruments to manage areas where the degree of risk is unacceptable and so create healthy landscapes. Identifying threats and policy instruments that will efficiently and effectively combat those threats is an important challenge for the NC CMA, as it is for all CMAs.
This project was formulated as part of an effort to develop a method that allowed natural resource managers to understand the landscape and landholder context in which a policy instrument was to be implemented and to predict the spatial effects of the instrument in the landscape. Specifically, the aim in this project was to develop and pilot a method for classifying landholders into segments and to spatially map those segments. The expectation is that such a method would enable natural resource managers to target investment in policy instruments to greater effect. The project was funded through the NC CMA and the Goulburn Broken Catchment Management Authority (GB CMA) and involved researchers from the Spatial Sciences Group and Practice Change Research, DPI, Tatura.
Figure 1: NC CMA Healthy and Productive Landscapes Framework
At a joint meeting between representatives of the NC CMA, the GB CMA, the Spatial Sciences Group and Practice Change Research, a range of natural resource issues were identified that could be investigated in the project (see Appendix I). From these issues the relationship between landholder participation in the water market and changes in the area of irrigated perennial pasture in the two regions was chosen for the project. This issue was selected because the Spatial Science Group had recently developed a procedure for using spatial mapping techniques to distinguish between, and measure, perennial and seasonal pasture. Furthermore, the team in Practice Change Research had developed a method for classifying landholders into segments based on differences in their farm context and for predicting differences in the behaviour of landholders.
The hypothesis to be tested in this project was that differences among landholders in the area planted to perennial pasture were associated with differences in the water available to landholders for irrigation and differences in the water available for irrigation were brought about by the water trading behaviour of landholders.
2.1 Spatial Market Segmentation
The collaboration between Spatial Sciences and Practice Change Research developed from a study in which Practice Change Research investigated the types of information dairy farmers might to need in order to convert their irrigation system from border check to centre pivot technology (CPT) (Hill et al. 2004). In that project the factors that influenced farmers to adopt CPT were identified. These factors were: whether dairy farmers were likely to be redeveloping land that was not previously part of the farm milking area; the topography of the land; the type of soils on that land; the distance of the land from the milking shed and the type of forage to be sown.
These factors are primarily biophysical in nature and so measurable, in principle, in relation to specific dairy farms in a landscape. This raised the possibility that dairy farms exhibiting these factors could be located in the landscape. Discussions with the Spatial Sciences Group indicated that there was a strong possibility that existing Geographic Information System data could be used for this purpose. This proved to be the case.
The study into the adoption of CPT demonstrated that, by integrating social research methods with GIS data, the size of the population of potential adopters of an innovation could be estimated using GIS techniques and the location of potential adopters in the landscape could be mapped (Hill et al. 2004).
In this project we built on the methods used in the CPT study by extending these methods to classify a population of landholders into segments (that is, groups such that landholders exhibiting similar characteristics were placed in the same group, while landholders exhibiting different characteristics were placed in different groups) based on Kaine (2004) and to estimate the size of those segments and locate them in a landscape. We have termed this process ‘spatial market segmentation’.
3. Materials & Methods
A project method was established and used to identify work flow and process (see Figure 2). This process will be used to structure this section of this report. The principle stages in the study were:
- to obtain appropriate data on land use, water use, water trading and area of perennial pasture;
- to identify different patterns of trading in the water market and classify landholders in segments based on their trading pattern;
- to explore differences across the segments in water use and change in area of perennial pasture; and
- to investigate the factors that influenced patterns in water trading and the importance of maintaining perennial pasture as a factor in those patterns.
Figure 2: Flow chart of work process
3.1. Collation & Data Outputs
Changes in perennial pasture can be determined using satellite imagery. A process developed by the Spatial Sciences Group Tatura uses Landsat Thematic Mapper satellite imagery to identify areas with or without vegetation and whether those areas are cold or hot (Abuzar et al. 2008). This creates four classes of land cover: green/cold, green/hot, not green/cold and not green/cold. By measuring these four characteristics at least three times in any irrigation season it is possible to identify the land cover class of an area. Importantly, a sequence of green/hot observations indicates that the land cover consisted of perennial pasture. This information was available for the 1996/1997 irrigation season and the 2003/2004 irrigation season.
The two sets of data on land cover could be used to estimate changes in land cover and where those changes occurred. However, land use information was also needed to associate land cover class with agricultural enterprise type. A land use database for 2003/2004 was compiled from data used in a 2001/2002 catchment scale land use map and the Valuer-General database for 2005. There was no equivalent information for the 1996/1997 season. However, it was possible to identify changes in enterprise type between the 1996/1997 season and the 2003/2004 season using information from the Goulburn-Murray Water (GMW) census (Douglas et al. 1998)(Figure 3). This showed that the area of dairy enterprises remained relatively constant across all irrigation areas and increased over time in only two instances. We assumed, therefore, that the area of dairy enterprises would not be overestimated if the current land use map were used to retrospectively locate dairy enterprises in 1996/97.
This assumption allowed us to link the area of perennial pasture with different types of farm enterprises in each of the two irrigation seasons. The four enterprise types linked to perennial pasture were dairying, horticulture, mixed farming and grazing. As the reliability with which we could identify dairy farms was greater than for grazing or mixed farming, and as horticulture had very little perennial pasture cover, this project focused on perennial pasture change
Figure 3: Land use by enterprise type across irrigation district*
3.2. Descriptive Statistics & Spatial Representation
The area of perennial pasture for the dairy industry for each of the two seasons across the entire northern irrigation district was extracted using the combined land cover and land use datasets. The absolute change in area and percentage change in area were then calculated.
Water use and water trade information was made available by G-MW through a data sharing agreement. This information was linked to dairy enterprises that had the same service identification (service ID is linked to the property) between the two irrigation seasons. This allowed the water use and water trade data to be linked with the data on perennial pasture. The water data was taken from the 1996/1997 season and the 1998/1999 season through to the 2003/2004 season. The data for the 1997/1998 season was missing. This provided seven years of data on water use.
3.3. Water Market Segments
The next stage in the analysis was to classify primary producers (excluding intensive livestock producers) into segments based on their trading behaviour in the water market. Producers were classified into segments in the water market based on their water trading over the period from 1996/1997 to 2003/2004. We used producers’ water entitlement in 1996/1997 as a base and expressed their sales and purchases of temporary and permanent water over the sample period as percentages of their 1996/1997 entitlement. Only complete water trading records were used in the analysis. This meant that 5,819 of the 7,432 service ID holders for which we had records could be used (78 per cent). A further 2,154 service ID holders were excluded from the analysis because they could not be classified as dairy farmers, horticulturalists or mixed farmers (leaving 3,665 service ID holders).
Producers were classified into trading segments using cluster analysis. We used Ward’s procedure with squared Euclidian distance as the similarity coefficient (Aldenderfer & Blashfield 1989)*. Wards procedure is widely used and was chosen because the procedure creates hyper-spherical clusters that optimise the minimum variation within clusters while maximising the distance between clusters (Aldenderfer & Blashfield 1989).
A ‘scree’ test (Aldenderfer & Blashfield 1989) indicated that a minimum of four clusters were present in the data. The solutions for four to eight clusters were then inspected using descriptive statistics to determine whether the minimum cluster solution had aggregated clusters that were qualitatively different from each other. On the basis of this inspection we identified the solution involving six clusters as the most satisfactory. The resulting six market segments are described in section 4.2.
3.4. Landholder Interviews
The next stage in the analysis was to investigate the relationship between the involvement of dairy farmers in water trading and their farm context, in particular, their management of their perennial pasture. Having classified dairy farmers into trading segments we were able to identify, contact and interview dairy farmers in each segment. The aim in the interviews was to:
- determine whether the characteristics of each trading segment data accurately portrayed the actual trading behaviour of dairy farmers in them
- develop an understanding of the factors in the farm context that influenced the trading behaviour of dairy farmers
- compare and contrast the trading behaviour and farm contexts across the segments.
Following established procedures (Kaine 2004; Kaine et al. 2005; Kaine and Bewsell 2005), convergent interviewing (Dick 1998) and laddering (Grunert & Grunert 1995) were used in the interviews to identify the factors that influenced dairy farmers’ trading in water. Care was taken to interview dairy farmers in all segments and irrigation areas. A total of 35 interviews where conducted. Twenty one dairy farmers were interviewed in the SIR Irrigation Areas (Murray Valley, Shepparton, Central Goulburn & Rochester) while 14 dairy farmers were interviewed in the NCIR Irrigation Areas (Pyramid-Boort and Torrumbarry).
3.5. Regression Analysis
The final stage in the analysis involved investigating the relationship between water trading and management of perennial pasture using regression analysis to identify statistically significant relationships in the GIS database on water trading and perennial pasture. The results of this analysis provided a means of validating the findings from the interviews with dairy farmers.
4. Results
Changes in Perennial Pasture Across the Regions
The spatial imaging data indicated that the total area of perennial pasture in the SIR decreased from 163,087 hectares to 110,695 hectares between the 1996/1997 irrigation season and the 2003/2004 irrigation season, a reduction of 32 per cent (Table 1). The area of perennial pasture in the NCIR decreased from 40,944 hectares to 30,537 hectares between the 1996/1997 irrigation season and the 2003/2004 irrigation season, a reduction of 25 per cent (Table 1). There were no statistically significant differences in the decrease in the area of perennial pasture between the Shepparton and North Central regions in either absolute or percentage terms. The area of irrigated perennial pasture in the dairy industry was estimated to have decreased by 30 per cent and 20 per cent in the SIR and NCIR, respectively.
Table 1: Perennial pasture cover in the North Central and Shepparton Irrigation Regions
| Total Area Perennial Pasture 03/04 (ha) |
Total Area Perennial Pasture 96/97 (ha) |
Change Since* 96/97 (ha) |
Change Since* 96/97(%) |
|||
|---|---|---|---|---|---|---|
| All land use types | North Central | 30,537 | 40,944 | -10,407 | -25 | |
| Shepparton | 110,695 | 163,087 | -52,392 | -32 | ||
| Dairy | North Central | 13,524 | 16,930 | -3,406 | -20 | |
| Shepparton | 59,487 | 85,276 | -25,789 | -30 | ||
A map of the percentage change in perennial pasture on farms across the SIR and NC regions is presented in Figure 4. Inspection of the maps indicates that there may be areas of concentrated change; however, the distribution has not been spatially calculated.
Figure 4: Change in perennial pasture cover between the1996/1997 and 2003/2004 irrigation seasons
4.2. Classification of Trading Segments
As described earlier, six trading segments were identified in the cluster analysis. We based our interpretation and description of the segments on differences across the segments in terms of purchases and sales of temporary and permanent water. We labelled the six segments permanent sellers, temporary sellers, small traders, small temporary buyers, large temporary buyers and permanent buyers. The profiles of the different segments in terms of water trading are reported in Table 2. Note that some landholders in each segment may exhibit, to a limited degree, trading behaviours characteristic of other segments.
The profiles of the dairy farmers in each segment in terms of water trading are reported in Table 3. There was a significant difference between the proportion of dairy farmers in each segment and the proportion of all service ID holders in each segment+. The proportion of dairy farmers was lower than for all service holders in the temporary sales segment while the proportion of dairy farmers was correspondingly higher than for all service holders in both of the temporary purchase segments. We describe the segments below.
4.2.1. Permanent Sellers
The producers in this segment made up only four per cent of service ID holders in the sample. Permanent water sellers had, on average, sold the equivalent of 96 per cent of their 1996/1997 irrigation water right by 2003/2004 (Table 2). Dairy farmers in this segment were estimated to have reduced perennial pasture on their land by 51 per cent over the same period (Table 3).
4.2.2. Temporary Seller
Producers in this segment leased a substantially higher proportion of their water each year than producers in other segments. Temporary sellers made up 17 per cent of the service ID holders in our sample. These producers had, on average, leased 60 per cent of their 1996/1997 water right each year (Table 2). Dairy farmers in this segment were estimated to have reduced perennial pasture on their land by 72 per cent over the same period (Table 3).
4.2.3. Small Traders
On average, the producers in this segment had only traded small volumes of water, if they had traded at all (see Table 2). Small traders made up 53 per cent of all Service ID holders making this the largest segment. Dairy farmers in this segment had reduced perennial pasture cover on their land by 27 per cent over the period from the 1996/1997 irrigation season to the 2003/2004 irrigation season (Table 3).
Table 2: Profiles for water trade segments for all enterprise types from 1996/97 to 2003/04
| Permanent Sellers | Temporary Sellers | Small Traders | Temporary Buyers (Small) | Temporary Buyers (Large) | Permanent Buyers | F-value | p-value | |
|---|---|---|---|---|---|---|---|---|
| Cluster size as proportion of service IDs* | 4.5 | 16.8 | 52.9 | 19.4 | 5.3 | 1.2 | ||
| Permanent water bought** | 7.3 (0 - 161) |
0.4 (0 - 52) |
2.3 (0 - 79) |
0.2 (0 - 18) |
6.1 (0- 247) |
102.4 (32 - 429) |
577.3 | 0.0 |
| Average Temporary water bought** | 16.6 (0 – 135) |
2.2 (0 - 34) |
5.1 (0 - 53) |
32.9 (9 - 70) |
100.4 (60 - 268) |
32.6 (0 - 152) |
2,119.3 | 0.0 |
| Average Temporary water sold** | 16.1 (0 - 83) |
60.7 (24 - 218) |
5.2 (0 - 42) |
5.6 (0 - 95) |
10.5 (0 - 280) |
36.2 (0 - 204) |
1,010.5 | 0.0 |
| Permanent water sold | 96.1 (69 - 261) |
4.7 (0 - 100) |
2.6 (0 – 77) |
0.2 (0 - 55) |
1.7 (0 - 96) |
1.8 (0 - 76) |
2,060.5 | 0.0 |
Table 3: Profiles for all dairy enterprises from 1996/97 to 2003/04
| Permanent Sellers | Temporary Sellers | Small Traders | Temporary Buyers (Small) | Temporary Buyers (Large) | Permanent Buyers | F-value | p-value | |
|---|---|---|---|---|---|---|---|---|
| Cluster size as proportion of service IDs* | 5.2 | 6.6 | 48.9 | 29.7 | 8.7 | 1.0 | ||
| Permanent water bought | 8.3 (0 - 142) |
0.3 (0 - 30) |
3.7 (0 - 67) |
0.2 (0 - 14) |
5.5 (0- 66) |
101.1 (41- 159) |
228.0 | 0.0 |
| Average Temporary water bought | 23.4 0 – 101) |
2.6 (0 - 18) |
7.2 (0 - 53) |
33.2 (13- 66) |
95.0 (60- 253) |
41.3 (0- 126) |
808.0 | 0.0 |
| Average Temporary water sold | 10.8 (0 - 66) |
55.7 (27- 141) |
3.6 (0 - 37) |
3.1 (0 - 65) |
3.3 (0 - 24) |
24.6 (0 - 86) |
456.4 | 0.0 |
| Permanent water sold | 95.1 (69- 188) |
3.8 (0 - 81) |
3.0 (0 – 77) |
0.1 (0 - 12) |
1.0 (0 - 45) |
5.4 (0 - 76) |
1,216.3 | 0.0 |
| Change in Perennial Pasture, 96/97 area greater than 5ha† | -51.2 (-100- 54) |
-72.4 (-100- 138) |
-26.7 (-100- 310) |
-21.8 (-100- 150) |
-18.2 (-100- 113) |
-27.3 (-98- 18) |
39.9 | 0.0 |
4.2.4. Small Temporary Buyers
The producers that were classified into this segment made up 20 per cent of all Service ID holders in the sample. On average, small temporary buyers leased water equivalent to 33 per cent of their 1996/1997 water entitlement each year between the 1996/1997 and 2003/2004 irrigation seasons (Table 2). We estimated dairy farmers in this group had reduced perennial pasture on their land by a relatively modest 22 per cent over the same period (Table 3).
4.2.5. Large Temporary Buyers
Producers in the large temporary buyers segment made up five per cent of all Service ID holders in the sample. Large temporary buyers, on average, leased water equivalent to 100 per cent of their 1996/1997 water entitlement each year between the 1996/1997 and 2003/2004 irrigation seasons (Table 2). The dairy farmers in this segment had reduced perennial pasture cover on their land by a relatively modest 18 per cent over this period (Table 3).
4.2.6. Permanent Buyers
The permanent buyers segment made up only one per cent of the service ID holders in the sample. Permanent water buyers purchased, on average, the equivalent of 102 per cent of their 1996/1997 water entitlement by the end of the 2003/2004 irrigation season (Table 2). The dairy farmers in this segment had reduced perennial pasture cover on their land by 27 per cent over that period (Table 3).
4.2.7. Trading Segments Among Dairy Farmers in NCIR and SIR
The profiles in each trading segment are reported for dairy farmers in the SIR and NCIR in Tables 4 and 5 respectively 3. There were significant differences across the segments in the proportion of dairy farmers in each segment from the SIR and NCIR. A smaller proportion of dairy farmers in the NCIR were temporary sellers, large temporary buyers and permanent buyers compared to SIR. A higher proportion of dairy farmers in the NCIR were permanent sellers compared to SIR.
After classifying the irrigators into segments we were then able to link segment membership with data on property identities, allowing the spatial distribution of the segments to be mapped (Figure 5). Inspection of the figure reveals the segments appear to be randomly distributed across the SIR and the NCIR.
Figure 5: Spatial distributions of water trade segments
Table 4: Profiles for Shepparton Irrigation Region dairy enterprises between 1996/97 and 2003/04
| Permanent Sellers | Temporary Sellers | Small Traders | Temporary Buyers (Small) | Temporary Buyers (Large) | Permanent Buyers | F-value | p-value | |
| Cluster size as proportion of service IDs* | 4.9 | 7.3 | 47.8 | 29.7 | 9.3 | 1.0 | ||
| Permanent water brought | 6.5 (0 - 142) |
0.3 (0 - 30) |
3.5 (0 - 67) |
0.2 (0 - 14) |
4.4 (0 - 56) |
98.6 (41 - 159) |
221.7 | 0.0 |
| Average temporary water brought | 23.3 (0 - 98) |
2.6 (0 - 18) |
7.0 (0 - 53) |
33.3 (13- 66) |
95.3 (60- 253) |
44.4 (0 - 126) |
722.2 | 0.0 |
| Average temporary water sold | 11.7 (0 - 66) |
56.4 (27- 141) |
3.9 (0 - 35) |
3.4 (0 - 65) |
3.2 (0 - 24) |
26.2 (0 - 86) |
409.4 | 0.0 |
| Permanent water sold | 93.5 (69 - 142) |
3.9 (0 - 81) |
3.2 (0 - 77) |
0.1 (0 - 12) |
0.9 (0 - 45) |
5.8 (0 - 76) |
1,003.9 | 0.0 |
| Change in perennial pasture, 96/97 area greater than 5ha† | -50.3 (-100- 54) |
-72.1 (-100- 138) |
-28.2 (-100 - 310) |
-23.7 (-100- 150) |
-20.1 (-100- 113) |
-21.9 (-75- 18) |
33.7 | 0.0 |
*Percentage of service numbers in cluster ** Mean Percentage of 1996/1997 irrigation season water entitlement †Mean Percentage change 1996/1997 to 2003/2004. Yellow highlights indicate the major type of trade in each segment.
Table 5: Profiles for the North Central Irrigation Region dairy enterprises between 1996/97 and 2003/04
| Permanent Sellers | Temporary Sellers | Small Traders | Temporary Buyers (Small) | Temporary Buyers (Large) | Permanent Buyers | F-value | p-value | |
| Cluster size as proportion of service IDs* | 7.0 | 2.2 | 56.5 | 29.0 | 4.8 | 0.5 | ||
| Permanent water brought | 19.9 (0 - 98) |
0.0 | 5.0 (0 - 46) |
0.2 (0 - 7) |
20.0 (0 - 66) |
134.7 | 22.6 | 0.0 |
| Average temporary water brought | 28.0 (0- 101) |
4.2 (0 - 16) |
8.4 (0 - 36) |
32.6 (14 - 62) |
91.0 (53 - 121) |
1.4 | 82.4 | 0.0 |
| Average temporary water sold | 3.4 (0 - 33) |
39.4 (38- 42) |
2.0 (0 - 37) |
1.3 (0 - 17) |
4.3 (0 - 14) |
3.9 | 31.0 | 0.0 |
| Permanent water sold | 104.7 (69- 188) |
0.0 | 2.4 (0 - 63) |
0.3 (0 - 12) |
1.9 (0 - 17) |
0.0 | 206.2 | 0.0 |
| Change in perennial pasture, 96/97 area greater than 5ha† | -56.0 (-99- 18) |
-78.8 (-96- -50) |
-18.3 (-99 - 58) |
-8.5 (-85- 123) |
6.4 (-36- 55) |
-98.4 | 8.7 | 0.0 |
Notes:
*Percentage of service numbers in cluster ** Mean Percentage of 1996/1997 irrigation season water entitlement †Mean Percentage change 1996/1997 to 2003/2004. Yellow highlights indicate the major type of trade in each segment.
Interview Results5
4.3.1. Confirming Accuracy of Spatial Data
The interviews confirmed that, generally speaking, the data developed and used by the Spatial Sciences Group was reasonably accurate. However, some issues with the data were identified.
The GIS data on property characteristics were incorrect for four out of the 35 landholders we interviewed. For example, one interviewee was a horticulturalist who had leased land to a dairy farmer. Another interviewee had a different history of water trading than was indicated in the data because their service ID was incorrectly recoded in the data. There was one example of transfers in the ownership of land creating errors in the data linking land use and trading.
4.3.2 General Findings
Although the area of perennial pasture declined in the dairy industry between the 1996/1997 and 2003/2004 irrigation seasons, we found that dairy farmers still viewed perennial pasture as essential to feed management. The dairy farmers we interviewed suggested that a 10 to 20 per cent variation in the area of perennial pasture from season to season was normal when seasonal conditions and pasture renovations were taken into account. In general, dairy farmers were of the opinion that perennial pasture was the key to achieving high stocking rates on farm and that the maximum number of cows they could milk on their property was directly related to the area of perennial pasture they had available. Most dairy farmers told us that they would have to increase the area they had sown to perennial pasture if they were to increase stocking rates on their farm. Conversely, farmers indicated that if they were decreasing cow numbers for any reason, they would be likely to reduce their area of perennial pasture.
Farmers stated that the key determinant of the area of perennial pasture they sow is their access to irrigation water. There were a number of ways a dairy farmer could access water including farm water right, sales water, buying temporary water, buying permanent water, obtaining water from drainage or diversion of groundwater. Naturally, not all of these options were available to every dairy farmer. Consequently, dairy farmers determine the area of perennial pasture they can sustain based on their perception of the supply of water they can obtain from the sources available to them. We found that, in most cases, the water trading of dairy farmers in each segment reflected the strategies they followed to ensure adequate stocking rate were maintained through the use of perennial pasture.
4.3.3. Permanent Sellers
All but one of the permanent sellers we interviewed had left the dairy industry. There were many reasons why they had left the industry including retirement, career change or hardship. One interviewee had bought and sold significant amounts of permanent water. This farmer described their trading activity as “like playing the share market”. However, we believe this farmer was unusual and that most dairy farmers who sell their water permanently were likely to be leaving the dairy industry.
4.3.4. Temporary Sellers
Overall, dairy farmers in this segment had access to more water than was required given the size of their herd and area of perennial pasture. There appeared to be two main reasons why dairy farmers were in this situation. First, some dairy farmers had decided to significantly reduce the number of cows they milked and therefore had reduced the area of perennial pasture they irrigated. The decrease in the area of perennial pasture released water which they then leased. While some of these farmers were likely to remain in the dairy industry but with a substantially reduced capacity, others may have been planning to leave the industry. Hence, for some farmers, the temporary selling of water is a precursor to permanent sale.
Second, some dairy farmers had a large water allocation, or access to alternative sources of water such as groundwater, which allowed them to lease substantial volumes of water in favourable seasons. Hence, these dairy farmers were leasing water opportunistically and were unlikely to be reducing their cow herd or area of perennial pasture area.
Further investigation of this segment could be helpful in predicting the number of dairy farmers who are likely to leave the dairy industry.
4.3.5 Small Traders
These dairy farmers rarely engaged in trading compared to other segments. Dairy farmers in this segment had a relatively high water right to land ratio or they had limited access to other sources such as groundwater or drainage diversion. Generally speaking these irrigators had access to adequate supplies of water relative to the area of perennial pasture and would only buy or sell small volumes of water in response to variations in seasonal conditions.
4.3.6 Small Temporary
These dairy farmers leased, on average, an additional 33 per cent of their water right each year which enabled them to irrigate a larger area of perennial pasture than would otherwise have been the case. The farmers we interviewed who were in this segment suggested that this was a feasible strategy provided water was always available and the price of water was approximately $100 per Megalitre or less. Most of the irrigators in this segment believed that purchasing permanent water was not a sensible investment as they could obtain greater economic returns by investing in the farm in other ways.
4.3.7. Large Temporary Buyers
These dairy farmers leased an additional 95 per cent of their water right, on average, which enabled them to irrigate a larger area of perennial pasture than would otherwise have been the case. These farmers did not usually have access to other sources of irrigation water. These farmers argued that this was a feasible strategy provided water was always available and the price of water was approximately $100 per Megalitre or less. Most of the farmers in this segment also believed that purchasing permanent water was not a sensible investment as they could obtain greater economic returns by investing in the farm in other ways. However a few of the farmers in this segment had been concerned at the run of dry seasons and were thinking of increasing their water right.
4.3.8. Permanent Buyers
From our interviews the main reason the dairy farmers purchased permanent water was because they were increasingly uncertain that they would be able to obtain sufficient water on the temporary market to maintain the area of perennial pasture they currently irrigated. However, one farmer in this segment purchased water as an investment after selling one of their two farms.
4.3.9. Conclusion
Interviews with dairy farmers indicated that, on the whole, the spatial data used to identify the trading segments was reasonably reliable. The interviews also revealed that purchases of water by dairy farmers were largely related to obtaining access to supplies of water sufficient to irrigate their perennial pasture. Sales of water by dairy farmers were largely related to having supplies of water in excess of that required to irrigate their perennial pasture.
Note that the availability of data restricted the analysis and interviews to the period prior to the drought. However, we found in the interviews that dairy farmers were suggesting they had, or were, dramatically altering their pasture composition on their farms in response to the drought.
4.4. Regression Analysis
The interview results supported the hypothesis that there was a relationship between water availability and area of perennial pasture on dairy farms. Statistical support for this hypothesis was sought by conducting regression analyses on the combine land use, water trade6 and perennial pasture data set.
We assumed that the volume of irrigation water available to farmers in a season could be roughly approximated by water entitlement plus average net temporary water trade. Hence, the area of perennial pasture in a season was hypothesised to be a function of water entitlement (net of permanent purchases and sales) plus average net temporary water trade. The estimated regression was:
y = 0.084x + 12.393 R2= 0.424, F = 947.7, p = 0.00
Where:
y = Perennial Pasture area 2003/2004 in hectares
x = Water entitlement 1996/1997 plus net permanent trades plus average net temporary water trade 1996/1997 to 2003/2004 in Megalitres
The regression model is represented geometrically by the solid line in Figure 6 which shows a scatter plot of water entitlement plus average net temporary water trade and perennial pasture area. The regression results indicate that 42 per cent of the variation in the area of perennial pasture among dairy farmers in 2003/2004 can be explained by the variation in their water entitlement in 1996/1997 plus net permanent trades plus average net temporary water trades from 1996/97 to 2003/2004.
The inverse of the estimated coefficient for water entitlement and net trades gives an estimate of water use per hectare of perennial pasture. In this case the estimate is 11.9 Megalitres per hectare. This is a slight over-estimate compared to other estimates (Linehan et al. 2004), but seems reasonable given the data excludes water available from other sources.
Regressions were also plotted to determine the relationship between water availability and the area of perennial pasture area within each water trade segment (Table 6). The relationship between water availability and perennial pasture was strongest for the permanent buyers segment. This is consistent with the proposition that access to water was a constraint on production of perennial pasture for farmers in this segment.
Three interviews were conducted with dairy farmers from the permanent seller segment, four with dairy farmers from the temporary seller segment, ten with farmers from the small trader segment, eleven with farmers from the small temporary buyer segment, four with the large temporary buyer segment and two in the permanent seller segment. One farmer was interviewed that was not included in the classification analysis.This rough approximation excludes water available from other sources such as drainage diversion, bores, spear points and effluent.
Figure 6: Scatter plot with linear regression line of the relationship between water entitlement plus average net temporary water trade and perennial pasture
The relationship was weakest between water availability and perennial pasture for the temporary sellers segment (Table 6). This was expected as farmers in this segment reported having access to water in excess of their requirements for irrigating perennial pastures. Hence, water was not likely to be a factor limiting the area of perennial pasture for farmers the temporary sellers segment.
Table 6: Regression model results for dairy farmers in the NC and GB CMA’s
| Model | R2 | 1/b(ML) | F-value | p-value | |
|---|---|---|---|---|---|
|
Permanent Sellers |
y = 0.106x + 21.038 | 0.465 | 9.43 | 37.329 | 0.000 |
|
Temporary Sellers |
y = 0.058x + 15.428 | 0.136 | 17.24 | 6.196 | 0.016 |
|
Small Traders |
y = 0.085x + 12.411 | 0.420 | 11.76 | 473.281 | 0.000 |
|
Temporary Buyers (Small) |
y = 0.098x + 8.214 | 0.373 | 10.20 | 245.550 | 0.000 |
|
Temporary Buyers (Large) |
y = 0.077x + 9.633 | 0.544 | 12.99 | 141.844 | 0.000 |
|
Permanent Buyers |
y = 0.090x + 4.334 | 0.805 | 11.11 | 45.432 | 0.000 |
5. Discussion
5.1. Relevance of Results to HPL Framework
The main objectives of the Healthy and Productive Landscapes Framework were to assist in setting priorities for action and selecting policy responses that are relevant to the contextual characteristics of the region. The work in this pilot provides a new, quantitative method for describing the contextual characteristics of the region.We have demonstrated in this project that it is possible to use data on spatial land cover and water use to link variations in land cover with water use and water trading by dairy farmers. Importantly we have demonstrated how quantitative data on spatial land cover and water use can be usefully combined with qualitative data to explain water trading by dairy farmers. This combined analysis allowed us to identify different types of trading, to map where they were occurring and to explore why they were happening.
The commonly held assumption that the area of perennial pasture is decreasing in the SIR and NCIR was supported in this project. The results indicated that the area of perennial pasture decreased by at least 20 per cent between the 1996/1997 irrigation season and the 2003/2004 irrigation season, across all industries and for the dairy industry alone.
Landholders were classified into six water trade segments (permanent sellers, temporary sellers, small trader, small temporary buyers, large temporary buyers and permanent buyers). The area of perennial pasture declined in all segments; however, farmers in segments that sold water, either permanently or temporarily, appeared to have reduced their perennial pasture by a greater proportion than farmers in other segments.
To understand the reasons for the different water trade clusters and therefore the possible change in perennial pasture we interviewed 35 farmers. In general the interview results indicated that farmers’ participation in the water market reflected their need to purchase water to assure supplies of irrigation water where this was a constraint on the area of perennial pasture; or the opportunity to sell water where supplies were in excess of requirements for perennial pasture.
The results indicate that, where the behaviour of farmers is strongly influenced by elements in their farm context that are measurable and documented (such as water entitlements and/or area of perennial pasture) different farm contexts can be identified across landscapes using spatial mapping techniques. This links directly with the HPL Framework by allowing, in principle, the spatial effects of policy instruments to be identified.
5.2. Strength in Spatial Market Segmentation Approach
Land use data and land cover spatial mapping techniques can provide information on landscapes which may be useful in identifying farm contexts. In this project we have found that market segments identified using spatial data can provide insights into the distribution of farm contexts across the landscape. The extent to which this is generally possible for agricultural and natural resource policy issues in the landscape will depend on the extent to which the factors in the farm context that influence the behaviour of farmers are present in, or can be linked to, spatial data sets. In this case data on land use was available and this could be reliably linked with data on water use and water trade.
5.3. Limitations
There some limitations with the approach used in this project. First, as with all large datasets, errors occur in the compilation of data. Second, the currency of landscape scale data is often a problem. Some of the data that was used in this project is only updated every four years. Third, the reconciliation of records where the variables that are used to link databases are not completely reliable and can be problematic.
For example, water trade data was compiled using the GMW Service ID number which is linked to property information. We joined the data on water use and trading to the data on land use based on the link between service ID and property. Naturally, properties and water entitlements were bought and sold between the 1996/1997 and 2003/2004 irrigation seasons. It was possible to trace some of these transfers but not all, and the process of reconciling transfers can be very time consuming. Therefore we have only used data on land use and water trading for properties whose service ID remained the same over the period we analysed. This has potentially introduced an element of bias into the analysis.
5.4. Further Research
Further research could investigate land cover change over shorter intervals and link this with the appropriate water trading data. Further research could also explore water trading on an annual basis to establish seasonal trading patterns. This may provide a basis for understanding and identifying the conditions that motivate farmers to switch between the segments identified in this study over time. This may then be used to investigate how drought or water reform will affect water trading by farmers, or to identify trading patterns that indicate farmers are likely to leave the industry.
This study was confined to investigating the link between water trading and perennial pasture. Future research could investigate the link between water trading, annual pasture and cropping.
Through the interview process we discovered that access to alternative sources of water was an important contextual factor that determined whether a dairy farmer is likely to purchase or sell water. Hence, we can hypothesise that data on alternative water sources would allow us to develop a more accurate model of water trading by farmers. This could include, for example, adding drainage, pumping and bore water licences to water trade data, and overlaying groundwater or drainage maps with pasture change or water trade maps.
Other issues that could also be explored include:
- regional trends in land cover and market segments. While at a catchment scale water trade segments appeared randomly distributed this may not be the case at a local or pod7 scale;
- whether there is any relationship between landholder demographics and enterprise characteristics and water trade segments; and
- whether there are differences in market segments and sizes between different industries.
6. Conclusion
The aim in this project was to develop and pilot a method for classifying landholders into segments and to spatially map those segments. The hope was that such a method would enable natural resource managers to target investment in policy instruments to greater effect. The method was developed and piloted in an investigation of the relationship between landholder participation in the water market and changes in the area of irrigated perennial pasture in the SIR and NCIR. This relationship was selected because the Spatial Science Group had recently developed a procedure for using spatial mapping techniques to distinguish between, and measure, perennial and seasonal pasture. Furthermore, the team in Practice Change Research had developed a method for classifying landholders into segments based on differences in their farm context and predicting differences in the behaviour of landholders and based on the differences in farm context.
The hypothesis to be tested in this project was that differences among landholders in the area planted to perennial pasture were associated with differences in the water available to landholders for irrigation, and differences in the water available for irrigation were brought about by the water trading behaviour of landholders.
We have demonstrated that it is possible to use data on spatial land cover and water use to link variations in land cover with water use and water trading by dairy farmers. Importantly we demonstrated how quantitative data on spatial land cover and water use can be usefully combined with qualitative data to explain water trading by dairy farmers. This combined analysis allowed us to identify different types of trading, to map where the different types were occurring, and to explore why they were happening.
The commonly held assumption that the area of perennial pasture is decreasing in the SIR and NCIR was supported by the findings. The results indicated that in the SIR and NCIR the area of perennial pasture decreased by at least 20 per cent between the 1996/1997 irrigation season and the 2003/2004 irrigation season, across all industries and for the dairy industry alone.
Landholders were classified into six water trade segments using data on water trading (permanent sellers, temporary sellers, small trader, small temporary buyers, large temporary buyers and permanent buyers). The area of perennial pasture declined in all segments; however, farmers in segments that sold water, either permanently or temporarily, appeared to have reduced their perennial pasture by a greater proportion than farmers in other segments.
Interviews with dairy farmers revealed that, in general, farmers’ participation in the water market reflected either a need to purchase water to assure supplies of irrigation water where this was a constraint on the area of perennial pasture; or an opportunity to sell water where their water supplies were in excess of requirements for perennial pasture.
The results indicate that, where the behaviour of farmers is strongly influenced by elements in their farm context that are measurable and documented (such as water entitlements, area of perennial pasture), different farm contexts can be identified across landscapes using spatial mapping techniques. This means, in principle and in the right circumstances, the spatial effects of policy instruments can be identified.
7. References
Abuzar, M., McAllister, A., Whitfield, D., Morse-McNabb, E. and Savige, C., 2008, Remote sensing tools and approaches to integrated irrigation water management at farm and regional scales, 14th ARSPC/NARGIS, 29th Sept – 3rd Oct, Darwin
Aldenderfer M.S. and Blashfield R.K. (1989). Cluster Analysis – Series: Quantitative Applications in the Social Sciences. Sage Publications: London.
Dick B. (1998). Convergent Interviewing: a Technique for Qualitative Data Collection. Available at http://www.scu.edu.au/schools/gcm/ar/arp/iview.html.
Douglas W., Poulton D., Abuzar M. and Morris M. (1998). Results of Irrigated Farm Census 1998, December 1998. Goulburn-Murray Water, Tatura.
Grunert K. and Grunert S. (1995). Measuring subjective meaning structures by the laddering method: Theoretical considerations and methodological problems. International Journal of Research in Marketing, 12: 209-225.
Hill M., Linehan C. and Murdoch H. (2004). Efficient Irrigation Technologies to Match Soils and Dairy Farming Systems, Market Research Report. Department of Primary Industries, Tatura.
Kaine G. (2004). Consumer Behaviour as a Theory of Innovation Adoption in Agriculture. Social Research Working Paper, 01/04. AgResearch, NZ.
Kaine G and Bewsell D (2005). An innovative approach to irrigation extension in horticulture, Acta Horticulturae, 672:177-183.
Kaine G., Bewsell D., Boland A. and Linehan C. (2005). Using market research to understand the adoption of irrigation management strategies in the stone and pome fruit industry. Australian Journal of Experimental Agriculture, 45: 1181-1187.
Linehan C.J., Armstrong D.P., Doyle P.T. and Johnson F. (2004). A survey of water use efficiency on irrigated dairy farms in northern Victoria. Australian Journal of Experimental Agriculture, 44: 131-136.
8. APPENDIX
8.1. APPENDIX I - Assumptions
LAND
Land use change is occurring in our region
Decreasing area of perennial pasture
Productive land used by hobby farmers
Agriculture is becoming less intensive and more extensive
Rural community enterprises becoming more diverse
WATER
Water is leaving our regions
Water will leave productive land uses
Water will move to high value use
Water is being used more efficiently
Number of irrigation farmers declining
ENVIRONMENT
Natural resources are degrading
Land holders will have reduced capacity to undertake catchment works
SOCIO-ECONOMIC
Drought impacts on peoples decisions
Enterprise scale is important to viability
Off farm income supports low viability enterprise
SALINITY
Salinity has a negative impact on productivity and the environment
Salt disposal is becoming more expensive
Groundwater levels are falling as a consequence of drought
Water use change is salinity neutral
9. Footnotes
1 * Available in SPSS 12.0.1
2 + χ2 =350.9, p = 0.00
3 There were significant differences across the segments in the proportion of service ID holders from the SIR and NC (χ2 = 59.8, p = 0.00).
4 χ2 = 14.4, p = 0.02.
5 Three interviews were conducted with dairy farmers from the permanent seller segment, four with dairy farmers from the temporary seller segment, ten with farmers from the small trader segment, eleven with farmers from the small temporary buyer segment, four with the large temporary buyer segment and two in the permanent seller segment. One farmer was interviewed that was not included in the classification analysis.
6 This rough approximation excludes water available from other sources such as drainage diversion, bores, spear points and effluent.
7 A concept developed in the Pyramid-Boort irrigation area is used to describe the water system in terms of Pods, Trunks and Carriers. A ‘pod’ is a grouping, typically comprising 10 – 30 properties and the associated channel systems. A pod also identifies a group of customers with a common stake in the water system such as service levels and system infrastructure. A ‘trunk’ channel conveys water to two or more pods. A ‘carrier’ channel conveys water to two or more trunk channels.
