We have estimated that 90 percent of New Zealand farms had some form of debt as at their 1998 balance date, and the debt of the average New Zealand farm in 1998 was $285,957. This is up 91 percent in real terms on the estimate made 10 years ago. We have estimated total New Zealand agricultural sector debt at $18.7 billion. This is up 53 percent in real terms on the estimate made 10 years ago. The increase for the whole sector is smaller than the increase for individual farms because there were fewer farms in 1998 than in 1988. These estimate have been made using a sample of farms drawn from five farm sectors (sheep and beef, dairying, horticulture, cropping and "other" farms) and calculating from the sample the mean liabilities per farm. The national estimate was made by multiplying the estimated debt per farm by the number of farms in New Zealand. Because the percentage increase in debt between 1988 and 1995 is so much greater than between 1995 and 1998, most of this increase in debt occurred between 1988 and 1995.

The average increases in farm debt between 1988 and 1998 mask substantial differences between sectors. Average debt on sheep and beef and horticulture farms changed to a small degree, but debt on cropping and dairy farms increased to a very large degree. This suggests that there have been huge structural changes in the cropping and dairy sectors in recent years.

The estimate of total agricultural sector debt is dependent on knowing the number of farms in each sector. The various available estimates of the number of farms in New Zealand are not directly comparable because the method of estimation has varied in each year. The last agricultural survey was in 1996, when it was estimated that there were 61 512 farms in New Zealand (excluding plantations and tobacco); the figure was based on the number of farms registered for GST (D. Golby, Statistics NZ, pers. comm.). The estimated number of farms in New Zealand in 1988 was 82 063; this was based on a land-based census, which would be expected to capture all farms with crops or livestock. Another reason why the estimates differ is that there is uncertainty in deciding the point at which a farm should be considered a lifestyle farm.

Even if agreement could be reached about how many farms there are in New Zealand, the figure would be changing constantly, as some farms are amalgamated and others subdivided. We have used the farm numbers in the AgriQuality New Zealand Agribase in 1998/99, modified by removing small lifestyle farms, to calculate total agricultural sector debt for 1998. These appear to be the best available figures, and it is this sampling frame that Statistics New Zealand has chosen for the national agricultural production survey. However, because it is difficult to know how many small farms to remove from the sample, estimates of total sector debt should be considered to be less reliable than estimates of average farm debt.

The estimates of the percentage of farms with liabilities had small 95 percent confidence intervals. Because of the large variance in the data, 95 percent confidence intervals for the estimate of total liabilities were large. We also calculated 95 percent confidence intervals for the estimate of percentage equity of farms. Because percentage equity is a standardised calculation for farms (the range being limited from 0 to 100 percent for farms of all sizes) the confidence interval is much smaller for percentage equity than for estimates of total liabilities.

In 1988 no confidence intervals were calculated for the data, but the 1988 survey analysed data from 6314 responses out of a total survey sample of 11 000 farms. In their report on the 1988 survey, Pomeroy & Reynolds commented that their sample had the advantages of being selected from the Statistics New Zealand sampling frame (allowing rigorous selection criteria) and that the sample was particularly large. However, increasing the sample size would not necessarily decrease the variance of the data (a major influence on the standard error, from which confidence intervals for normally distributed data are calculated). Also, a large sample may not improve the level of accuracy of the data if survey questionnaires are completed incorrectly.

Our own estimates are based on a sample of only 881 responses (the survey having been conducted separately from the national agricultural production survey). Our sample size was determined largely by the funding available for the survey, but an advantage of the relatively small, manageable sample was that we were able to check all questionnaires and, if necessary, contact farmers again for clarification of the data. Further data accuracy and a higher response rate would be achieved if enumerators visited farmers to collect the data. Because standard errors vary according to the square root of the sample size, halving the confidence interval (to provide more precise estimates) would require four times the sample size.

While all the uncertainties we have mentioned remain, accurate and precise estimates of New Zealand’s agricultural sector debt will be difficult to make. Nevertheless, we have used the best available information on farm numbers, and our estimates are as good as we could obtain in a small survey.