• Tue
  • Sep 23, 2014
  • Updated: 11:52pm

Consumer confidence a predictive tool when used upside down

PUBLISHED : Monday, 07 February, 2005, 12:00am
UPDATED : Monday, 07 February, 2005, 12:00am

I FACE A dilemma. Only two weeks ago, I added to my portfolio in the stock market (more Hang Seng Bank and some CLP Holdings) and now, Sod's Law, along comes one of the most reliable 'sell' signals I could ever expect. What shall I do?


Let me explain. Every three months, pollster ACNielsen conducts a consumer confidence survey and, as the first chart shows, the one undertaken last month through interviews with 1,016 people says that economic confidence in Hong Kong is at a record 12-year high.


Although this should be good news, I have my doubts about confidence indices. I think they tell you what confidence has been, not what it will be, and it is one of my golden rules that confidence is never greater than at the peak of an economic cycle and never worse than at the bottom of one. What we may have here is one of these contrary indicators that I so value.


Let us test it out against a hard benchmark, the stock market. We shall assume that you put $100 into the market in January 1993, which is as far back as I have data from ACNielsen. We shall then say that you sold the market at the major peaks of the confidence index and bought it back again at the major troughs.


We shall also say that you ignored minor one and two quarter reversals of the trend as representing only statistical volatility and that you put your money on three-month deposit with your bank during the periods when you were out of the market. The 'buy' and 'sell' notations on the first chart show you when you would have got into and out of the market on this basis.


Right, to the second chart. The green line along the bottom shows you how you would have done if you had just put your $100 on three-month deposit for the entire 12 years. At the end of last month, interest income would have made that deposit worth $153, which is not really that bad when you consider that inflation over those 12 years pushed consumer prices up by only 22 per cent.


The blue line shows you what would have happened if you had put that $100 into the Hang Seng Index throughout the period and never took it out. Your holding would now be worth $239, which is better than a three-month deposit. Take note, however, that if we had made April 2003 the ending date of this exercise you would have actually done better with the deposit than with the Hang Seng Index.


And now look at the red line. If you had followed those 'buy' and 'sell' signals in the first chart, your initial $100 would have grown to $1,486 at the end of last month. Yes indeed, an excellent contrary indicator, spot on at almost every major upturn and downturn of the market. A crystal ball that really works.


So did I call my broker as soon as I saw it and tell her to sell all? No, I did not. Maybe the index will go even higher before it goes down again, I told myself, and anyway this may be the one time that it does not work. Do not expect me to concede my mistake in print if I'm proved wrong. I will go only so far in confessions to having been a fool.


Nonetheless, it is true that you cannot actually regard the first chart as being quite so accurate a predictive tool as I have made it out to be. How indeed can you know at any one time that the index will not go higher or lower? How can you be sure in advance that any one or two-quarter reversal will last for only one or two quarters?


It does show you, however, that this sort of confidence index generally tells you the exact opposite of what it purports to tell you. The fact that everyone else is confident is a good reason not to be confident yourself.


Thus, while I am grateful to the people at ACNielsen for having sent me a copy of the underlying data for the index, I think they owe it to their clients to emphasise that the 1,016 interviewees have probably got it wrong in forecasting their own expenditure intentions this year.


This predictive tool unfortunately has predictive value only in reverse.


Share

For unlimited access to:

SCMP.com SCMP Tablet Edition SCMP Mobile Edition 10-year news archive
 
 

 

 
 
 
 
 

Login

SCMP.com Account

or