abare.gov.au - research report 08.05

Appendix

Projection procedure

The projections of wine grape production contained in this report are generated from estimates of yields and wine grape bearing areas. Projections are made for up to 47 different varieties of wine grapes in each of 89 wine regions in Australia. These wine regions are defined by the Australian Bureau of Statistics (ABS), which collects detailed statistics on the Australian grape growing industry each year from wine grape growers (ABS 2007). The annual ABS wine grape census collection provides information about the current state of Australian viticulture that is integral to the projections of future wine grape production reported here.

As part of the wine grape census, the ABS collects data for each grape variety on the grape bearing area, non-bearing area and the quantity harvested for specific purposes such as wine making and drying. Information is also obtained about the amount of the non-bearing area that was planted or grafted-on in the previous twelve months as well as the non-bearing area that was non-bearing for more than a year.

As grapes are perennial, the bearing area of grapes next year is determined mostly by the area of grape vines that are currently bearing fruit. It is also affected by the area of grapes planted or grafted-on in previous years that will bear fruit next year (a portion of the current non-bearing area). In addition, it is possible that growers might take grape vines permanently out of production between the current and next year’s harvest, which is a practice known as grubbing.

Based on this intertemporal relationship, the following model is used to estimate future production:

G_i proj_t+1 = (BA_i,t+ PL_i proj_t-k – RM_i proj_t+1) * Y_i proj_t+1

G_i proj_t+1 represents projected production of grape variety _i , next year; BA_i,t is the current bearing area; PL_i proj_t-k is the portion of the current non-bearing area that was planted or grafted-on in an earlier year t–k, but is projected to be bearing next harvest (k = 3 for warm climate regions and k = 4 for cool climate regions); RM_i proj_t+1 is the projected area of vines of grape variety i that will be removed or grubbed next year, and Y_i proj_t+1 is the estimate of next year’s yield per hectare for wine grape variety i. The portion of the current non bearing area that is expected to come into bearing in the future is determined by estimating an age profile of the current non-bearing area that is more than one year old — that is, the area that was not planted (or grafted-on) last year.

The age profile of the non-bearing area that is two years of age and older is assumed to reflect the relative size of new plantings (and grafting-on) over recent years. For the cool climate regions it is assumed that it takes four years for newly planted vines to reach a commercially bearing age, while in the warm inland regions it is assumed to take three years. Hence, for warm climate regions, for example, the proportion of each variety planted two, three and four years ago of total plantings over that period is used to apportion the total non-bearing area that is currently reported to be at least two years old. The apportioned areas provide estimates of PL_i proj_t-k, the areas that are to become bearing in the future. For cool climates the plantings of each variety two, three, four and five years previously would be used in a similar manner to that for the warm climate regions to determine the additions to bearing area in the future.

Ideally, the area grubbed next year should be subtracted from the bearing area. However, given the very small area of grapes that have historically been grubbed or grafted-off according to ABS data, and that future decisions by growers to grub are too difficult to estimate with any objectivity, they were assumed to be zero and excluded from the projections model.

According to the model provided above, estimated wine grape production for each variety in 2007-08 is calculated as the sum of 2006-07 bearing area and the portion of non-bearing area reported in 2006-07 that is expected to reach commercial bearing in 2007-08 (four years old in cool climate and three years old in warm climate regions), multiplied by the projected yield in 2007-08. To forecast grape production for the following year, 2008-09, the area expected to become bearing in 2008-09 is added to the estimate of the bearing area in 2007-08 and the total area multiplied by the projected yield.

Calibration

While the ABS viticulture census is the only official source of statistics on grape bearing area, there are other sources that report estimates of wine grape production. For example, while the viticulture census collects data on grape production for wine making from growers, the Australian Regional Wine Grape Crush Survey collects estimates from the wineries of the quantity of grapes crushed for winemaking, by variety and ‘geographical indication’ (GI) region. This survey is conducted under the National Utilisation Project (NUP) and coordinated by the Australian Wine and Brandy Corporation. The ‘geographical indication’ is the official industry description that is similar to the ‘appellation’ system used in Europe but less restrictive in terms of viticultural and winemaking practices.

Before 2006-07 the NUP estimate of the Australian wine grape crush was consistently higher than production estimates reported by the ABS viticulture census. The NUP is regarded by the wine industry to be the more accurate measure of wine grape production. Because of this, the ABS bearing areas have historically been calibrated to ensure that wine grape production is equal to the NUP estimate of wine grape production. However, because of changes in coverage, the production estimate reported by the ABS viticulture census in 2006-07 was 2 per cent higher than the NUP intake estimate.

From 2006-07 there has been a general consensus that the ABS data are now more accurate than the NUP estimates. Hence, for this year, ABARE has used the ABS data as the benchmark for production of each variety of wine grapes in GI regions where there is a large discrepancy between the ABS and the NUP. Therefore, in this report, where the ABS and NUP report production of a grape variety in a GI region, the historical ABS bearing area data (BAi,t) is calibrated to ensure that wine grape production by variety in each region is equal to the ABS viticultural census figure. In cases where individual grape varieties are recorded by the ABS in a GI region but are missing in the NUP, the original ABS bearing areas are included in the projection model.

In contrast, the ABS statistics for non-bearing area and plantings plus grafted-on area are not calibrated to either the NUP or ABS production data because they do not affect the current bearing area. Hence, non-bearing area is unrelated to the ABS or NUP estimates of current production or winery intake.

The production of multipurpose and minor variety grapes (see tables) are estimated rather than model generated. Production is estimated by the required intake by wineries for each grape variety in every GI region as reported in the Australian Regional Wine grape Crush Survey. This is done because multipurpose and minor grape varieties are commonly used for purposes other than making wine and therefore model projections of the production of these types of grapes would typically far exceed their use for wine. Consequently, estimates provided by wineries of their requirements for multipurpose and minor grape varieties are considered to provide more accurate estimates of their production for wine making. These estimates more closely reflect expected demand for such grapes. In effect, this approach is based on the assumption that the quantity of multipurpose and minor grapes available for other purposes, such as drying and fresh table sales, is a residual after winery demand for multipurpose grapes is met.

Yields

Industry experts provided the yields that were used to project production. Historical yield data for each variety of grape for the previous five harvests and the average yield for that five-year period were provided as base information to the industry experts for every ABS region of relevance. These experts then provided their own estimates of yield for the 2007-08 harvest and each of the projection years on the assumption that there would be average winter rainfall in their regions and irrigation catchment areas following the 2008 harvest. It was not uncommon for industry experts to expect the yields at the end of the projection period to be similar to the five-year average yield.

Aggregation of regional projections

For reporting purposes, the projections of wine grape production are presented at the national, GI zone and regional levels. The vast majority of ABS regions are identical to the GI regions but in some cases the ABS collects data at a more disaggregated subregional level, particularly where GI regions or zones cross state borders.

In these instances, ABS region projections were aggregated to the GI region (zone) level. In some instances, such as for the Murray-Darling – Swan Hill GI, the GI zone is also the region. As a result, the 89 ABS regions used in the projections model, become aggregated into 83 GI regions (zones). The tables for wine grape production for each of the 27 GI zones that are presented in appendix B are simply aggregations of the projections generated at the GI region (zone) level. Projections of wine grape production, by variety, for each of the 83 GI regions can be accessed from ABARE’s website, abare.gov.au.

Table B1 lists the GI zones in each state, and the regions within each zone. Table B2 lists the wine grape varieties included in these projections. Wine grape production for each of the 27 GI zones is presented in appendix B3. Queensland, Tasmania and the ACT each have only one zone and one region. The Murray Darling – Swan Hill zone lies in both Victoria and New South Wales.

A full list of data relating to wine grape production, by variety and zone can be found at the Australian wine grape production projections to 2009-10 interactive website.