The graphs MIS9201.CHT through MIS9207.CHT inclusive and MIS9211.cht through MIS9218.cht inclusive show average benchmark prices corresponding to real data for the 3 markets 1989 through 1994. For each sex and market, the shape will be the same over the same period, but the values will be higher or lower depending on the weight and grade used. These graphs are composed in the following way:
- for each market day, the data points for each sex are analysed to get a best fit straight line on graph of the log(price obtained) vs. log(animal edible weight). This is done using mathematical formulae and also discounting up to 33% data points which are outside the envelope of +/- 2.5 standard deviations (to discount stray or erroneous measurements).
- if for this market-day and each of the 3 sexes, the value of the slope of this best fit straight line is between 0.5 and 3.0, then the slope value (B) and the y intercept value (K) are used in a formula to calculate the price for the 7 benchmark weights and grades which I chose from averaging historical data.
- thus for every market day there may be 0, 7, 14 or 21 values of benchmark prices corresponding to the 21 benchmark weights and grades.
- each of the graphs above chooses only 2 of these 21 benchmark prices and shows them for 2 or all 3 of the 3 major markets.
The graphs themselves show what we are getting used to seeing:
- mis9201 to mis9003 - for the top steer benchmark - 350 kg grade 0 - show that prices between the 3 markets are in general very close to each other.
- mis9204 to mis9207 - for the lowest steer benchmark - 160 kg grade 4 - show that Dar is rather higher than Moshi which in turn is higher than Arusha. Note that when comparing Arusha in particular with the real data points it appears that the theoretical curve fit understates the prices at low weights and grades for high B values. I recognise as a legitimate weakness of the log-log approach, but am totally convinced that the benefits of the approach easily outweigh this weakness, which only shows at this kind of benchmark.
- mis9211 to 9214 - for steer 275 kg grade 1 - show that Dar and Moshi are pretty much on par, but with a Dar bulge upwards in August 1994 due to the rail disruption mentioned above; Moshi is generally a bit above Arusha, but with a significant bulge upwards during the period December 1993 through May 1994 (reason unknown).
- mis9215 through mis9218 - for steer 240 kg grade 2 - shows similar behaviour to the 275 kg graphs above, but with Dar prices spacing out above Moshi and Arusha numbers.