To Contact Us
to page back through Previous Editions click here Index is in right hand column

October 2009 Edition ----

We propose a simple lesson in data analysis this month. As you may guess it is based upon data Australian carpet size data as presented by Exftx Ltd (in receivership and liquidation) formerly known as Feltex Carpets Ltd on page 37 of its 2004 IPO prospectus. We believe this data is referred to on the top of page 91 of the prospectus where they say an assumption upon which its projection for the 2005 year is based is that “ The size of the carpet market in New Zealand and Australia, measured by volume of linear metres sold, will grow over the projected period by approximately 1%, which is below the average growth rate over the past 10 years. “

Well it there refers to linear metres while page 37 refers to square metres. If carpet is one metre wide then both would be right but we think not. And page 91 refers to sale (by using the word “sold”) whereas we have been told by the Carpet Bureau that the page 37 data is of carpet supplied to the market and differs from sales in that it does not account for stock on hand. It assumes everything is sold when it is available to be sold. Perhaps Feltex have chosen to put one set of data in their prospectus and use another set for making their page 91 statement but we can see absolutely no justification for doing this.

We will now show the page 37 Australian data as it might be placed on a Microsoft Excel worksheet. Here it is on columns A and B. This commentary continues at the bottom of the table.

If you put those columns in the same place on an Excel spreadsheet the highlight some other cell, the construct for getting a prediction for 2005 based upon the last 10 years of known results is:

=FORECAST(2005,A5:A14,B5:B14)

Key it in and push “enter” and you should get the result 51482. This formula is based upon least squares regression analysis.

In column C we have put the predictions for 2005 using data up to 2003 from each starting year. To do this we used the formula +FORECAST(2005,A1:$A$14,B1:$B$14) into cell C1 and copied it down column C. It can be noted that starting at 1994 gives the third lowest with that from two more starting years being only a little higher.

In column E we have plotted the predictions for each year using the 10 years of data 1994 to 2003. This shows “growth” of 0.305%. Key into a cell =49788*1.00305^9 and press enter and you will get a result of 51171 which proves this (49788 is the prediction for 1994 while 51174 is the prediction for 2003).

Then in column F we have the difference between the predicted and actual figure for each year. The source data at the extremities tends to have the greatest effect on the slope of a least squares regression line. In a time series used to predict the future this is appropriate with respect to the latest data but not appropriate for the earliest data since the relevance of data generally diminishes with age. It can be seen that that 1994 data is close to the average line and so has had little effect on the slope of that line. 1995 actual data is well above the line and will have had a big influence on it but not near as much as it would if it was the 1994 data. The validity of the line from this aspect is quite good. Column G is the predictions based on nine years of data 1994 to 2002. As can be seen the slope of the line is negative. Instead of the prediction for 2005 being 3% below the 2003 actual size the 2005 prediction is 5.5% below the 2003 actual size. The idea here is to demonstrate that the 2003 figure has had a significant influence on the line as being the most recent measurement suggests that it should. We have no argument with this. It should be noted however that a sharp increase in size over the previous year has not really resulted a permanent increase since 1994.

Feltex has obviously got its 1.7% “growth” by applying it to the 1993 market size figure like so =44687*1.017^10 and getting 52892 which is close to the 2003 figure of 53076. Its projected Total Operating Revenue for 2005 was $348m which is 10.75% higher than the $314m which was declared for its 2003 year. It would seem this is made up of 2 years of inflation at 3%p.a., two years of increasing market size at 1% p.a., and two years of gains in market share yielding a 1%p.a. increase to operating revenue.

Please note that the excel forecast function can be used with just the two years, 1993 and 2003 and it will give a very similar answer to what Feltex got. The distort is in choosing only the remote year 1993 instead of the ten most recent years 1994 to 2003.

We don’t think the word growth is applicable to this data which is up and down so much. As we demonstrated it is possible to find a nine year segment where the average “growth” is negative. Growth is an applicable term for data which is onward and upward like the length or height of a young plant or animal. Generally it is not possible to get these measurements to go negative. The organism will probably die if that is attempted. The concept of negative growth is almost a contradiction in terms.

The 1.7% growth as “calculated” by Feltex comes about because the size of the earlier year was very low and that of the latest known year was quite high.

We return to the part sentence on page 91 of the prospectus reading “ 1%,which is below the average growth rate over the past 10 years”. We believe that this can be paraphrased as saying “which is below that which is logically predicted when only the market size data of the past 10 years is considered”. The reason for making this comparison is because it is a simple and objective measure of what a person with no knowledge of the market might expect so they can gauge to what extent they are relying on the experts. The Securities Commission mention that Feltex stated its 1% growth assumption to be below the average growth rate of the past 10 years presumably because the Commission thought this to be a very significant statement as do we. The Commission purports to have made no attempt to verify the claim however. It is quite obvious that straight line between a low point and a high point does not constitute a fair measure. While the formula which calculates the least squares line is not particularly simple a graph of the line so generated along with the data which derived it does generally make sense to the average person.

We say that the prediction based on past years of more than a 1% p.a. increase in market size above the 2003 figure is wrong and the 3% decrease from the 2003 figure is the correct such prediction of the 2005 market size. To be conservative a greater decrease than 3% is required, probably a further 2% since actual market size has been going about 2% above and below the average line. The difference between these approaches is more than 5% or $17m of Feltex revenue. With a 7% reduction, to be conservative like they were claiming to be, the revenue overprojection is $23m. By their wrong statement Feltex have wrongly given the impression that they were making conservative assumptions as well as wrongly inferring what past year’s figures indicate the future market sizes will be, and this will have greatly increased the uptake of the offer. If the projections had been prepared with an 3 or 5% decrease in 2005 market size instead of a 2% increase then the number of shares offered would have been far smaller as would have been the total size of the first dividend and the company may well have survived. Nothing said here is modified by what is said in the next paragraph.

What we must concede is that the actual ‘market sizes for 2004 and 2005 were in fact greater than what Feltex indicated by way of its 1% p.a. growth assumption. Its prediction of market growth was in fact conservative with the benefit of hindsight. Investors were not mislead as to future market size. Feltex might have had good information at the time the prospectus was published fully justifying the 1%p.a. increase assumption adopted. But they were not justified in effectively say there market size assumption was less than that of market size predicted from the size of years prior to 2004. That was wrong and misleading regardless of how justified they were in adopting the assumption they did.

While this large increase in Australian market size occurred in 2004 and was largely maintained in 2005 we doubt that it was of a type that Feltex was able to participate in. We think the increase was of a very unusual nature. We think there was probably a large push from importers during this period. Imported carpet does not need to be sold to be included in ‘market size”. We do not claim to know any details of Feltex’s purchase of the Australian factories of Shaw Industries in the year 2000 but we think it would be normal for there to be clauses restraining Shaws from competing with Feltex in Australia for a set period of time. There might also have been agreements allowing Feltex to manufacture and market Shaw’s brands of carpet for a set time since the factories would be set up for this. The agreement with Shaws which was current at the time the prospectus was published did not seen to contain any such understandings We think this may have recently expired and this has contributed to an import influx. Or perhaps there was just a verbal understanding which expired. We think Shaws as an importer probably re-established trading relationships with retailers they had had when they manufactured in Australia, and this has had a major effect on Feltex sales.

While the assumed 1% increase in “market size” was quite accurate from the point of view of hindsight the 1% increase in market share assumption was appallingly bad and the large reductions in years up to 2003 continued or got worse. Feltex made no attempt to disclose its recent market share history despite market share being one of its critical assumptions.

We say that if people are lured into an investment on a false claim such as what the latest known 10 years of total market size data predicts the market size of a future year to be, then if that investment goes bad those making that claim should be liable for the loss regardless of whether the falsely claimed figure turned out to be correct in reality. The people should not have had the investment so they should be compensated for losses.

Advertising section

------------------------

Suitable Objectives

Register of Members

Members Forum - Topical
* open to all meantime:

Plenty of Opinion

Magazine Plans

Need an Accountant?
or Prepared to Change?

Users Forum
* have your say

Ready to Join?

Knowledge Tests
* being developed

What's Wrong with
the existing 
accounting body?

So called BNZ Audit
an extensive case study

 Visit our Advertising
page from 
Here