Thursday, March 1, 2012

Dangers in Periodic Growth Measures - But the Medium Term Outlook Could Be OK

As we approach the 3rd anniversary of the 6 March 2009 bottom of the Australian All Ordinaries let's look at actual and projected growth for the All Ords.

Consider the following tables. Each table includes a column which shows what percentile of performance since 1984 each period's performance falls within. For comparison, the right hand column shows the median percentile for each period taking into account all tops in the All Ordinaries since 1984. The performance which might cause concern that we are near a market top is highlighted red in the percentile column. We will see how it advances year by year because of the change in the market levels which are falling out of periodic performances as time goes by.

At 6 March 2012 if the market is at about present levels (I have assumed 4400):

All Ords 4400 6/03/2012 Med Top
Perf Over % Percentile Percentile
1 year -10% 14 68
2 years -8% 16 83
3 Years 40% 76 77
4 years -20% 4 54
5 years -23% 0 79

Three year growth looks quite high in percentile terms. You might feel that, based on that measure, the market shouldn't go much higher. But on each other period the market is at quite low levels of growth and could easily be higher.

Also, what will happen to 3 year performance after 6 March 2012 if the market stays at about current levels? It will fall significantly!

But, 4 year performance which is presently near all time lows in percentile terms will start to increase significantly!

This happens as the relatively high market levels of 2007 and early 2008 fall out of the calculation and are replaced by the lower values of the current market, even if the All Ords stays around current levels.

So lets look at a scenario for 6 March 2013. We will see that 3 year performance goes back to well below the mean (50th percentile) at 16th percentile, whereas 4 year performance is now in the 65th percentile and well above the median 4 year performance for market tops since 1984.

All Ords 4400 6/03/2013 Med Top
Perf Over % Percentile Percentile
1 year 0% 29 68
2 years -10% 12 83
3 Years -8% 16 77
4 years 40% 65 54
5 years -20% 0 79

So now we can look at 2014 and 2015 and you will see how the percentile performance would continue to change each year. These tables continue to use 4400 (around current market values, rounded for convenience) at 6 March 2014 and 2015.

First 2014:

All Ords 4400 6/03/2014 Med Top
Perf Over % Percentile Percentile
1 year 0% 29 68
2 years 0% 25 83
3 Years -10% 14 77
4 years -8% 12 54
5 years 40% 38 79

And now 2015:

All Ords 4400 6/03/2015 Med Top
Perf Over % Percentile Percentile
1 year 0% 29 68
2 years 0% 25 83
3 Years 0% 22 77
4 years -10% 9 54
5 years -8% 1 79

 It can be seen that by 6 March 2015 without further growth the All Ords would be well below median values for each performance measurement period, would be near all time lows in growth percentiles for 4 and 5 years and dramatically below the median levels of performance for each period at prior market tops since 1984.

In fact the All Ords could grow by 12% pa for each of the next 3 years (from my assumed level of 4400 at 6 March 2012) and all performance measurement periods would be below the median levels for the market tops since 1984. In the table below the rows are in descending order from 1 to 5 year performance. The top half of the table is percentage increase over the relevant period of years. The bottom half of the table shows the percentile since 1984 into which that performance falls, again in descending order from 1 to 5 years. Every percentile in 2015 is less than the median percentile for that period of performance for the market tops from 1984 to February 2012.

6/03/2013 6/03/2014 6/03/2015 Med Top
12% 12% 12% 15%
1% 25% 25% 37%
3% 13% 40% 41%
57% 16% 26% 33%
-11% 75% 30% 70%

61 61 61 68
26 70 70 83
26 38 77 77
83 33 44 54
0 85 26 79

The moral of the story is that after booms and busts, when looking at relative performance compared to historical records, care has to be taken to ensure an understanding of what is falling out of the periods as well as what is coming into the periods from recent changes.

This is not a prediction of market rises for a particular length of time or even at all. It is just to ensure that we understand how the performance of the market over time periods is affected by the index values falling out of the period being examined. It also gives some idea of what could be possible if past patterns were to repeat, but past performance is no guide to future performance. Remember Japan and look at China and consider our private debt levels.

Since 1985, the longest period of negative 1 year growth being maintained appears to be from January 2008 to September 2009. The shortest significant periods seem to last about 12 months. But to get back to 0 growth over 1 year the market has to have turned up from the lows which would seem likely to have occurred by 6 months after 1 year growth first went negative. If the 1 year growth went negative in around July 2011, then by December 2011 you would be expecting to see an upturn in the market, with 1 year growth getting back to positive by July 2012.

There are a lot of things indicating that unless this time is different, the current upturn is likely to be sustained for a year or two.
1. Coppock is looking like turning up from below 0 and has been down to 13th percentile
2. 1 year performance has been negative for almost 9 months and has been down to 14th percentile
3. 2 year performance has been down to the 8th percentile, currently 13th percentile
4. Market has been to a median bottom based on various measures such as percentage fall of 20%, sum of growth over 1 & 2 and over 1, 2 & 3 years and has recovered more than 10%
5. 10, 30, 50 and 100 day moving averages have all turned up.
6. The 3 quarter market performance compared to a long term average has started increase. It rarely breaks down again until getting well into positive territory (ie getting 3 quarters at significantly above 7% pa increase in All Ords)
7. RBA has moved away from forecast tightening to having eased and being aware of the contractions in the real, non-resource, high employment sectors of the economy.

So statistically speaking the probability is for the All Ords to rise over the medium term (unless we were to have a performance like Japan where the performances and percentile cureves changed dramatically during the unwinding of the 1990's bubble.)

The Caveats to this are:
1. Things can always get worse as they did leading into 1991 which fell back almost to 1987 lows.
2. China could have a hard landing
3. Greece will still probably default eventually and Portugal and Ireland are not out of the woods.
4. There are still some calls out for a US recession based on "ensembles" of data (ignoring monetary data which has been grossly distorted by FED intervention, although Hussman seems to have let his timeframe drift out from 8 weeks to 18 months and ECRI seem to have gone very quiet. The Conference Board revised its Leading Indicator series but it suggests that the danger may have passed.
5. Oil prices are back near recent highs and there is uncertainty regarding Iran supply.
6. There is virtually no growth in Australian employment and this often leads to recession, but the savings rate seems to have stopped rising so as incomes increase even at low rates it might now flow into spending.


Assigning probabilities I would say 70% chance of a reasonable period of growth continuing for up to 1 to 2 years, with the chance increasing to 85% if we get a positive Coppock signal and the market breaks out above previous resistance around 4500. The Coppock did not give a signal at the end of February, 2012.

12% pa growth from current levels for the next 3 years would not look extraordinary for 1,2 or 3 years, although the 4 and 5 year growths would look quite high in separate years as explained above.

An All Ords of 5300 in 2015 would look quite reasonable on most "growth over period" measures.

Whether that growth would be evenly distributed or would be higher in currently depressed sectors is a matter of conjecture.

No comments:

Post a Comment