'... They (retail sales statistics) should not be regarded as a comprehensive indicator of overall consumer spending.' Data health warning Census and Statistics Department I'M PROUD OF that cautionary note published by the Census and Statistics Department every month when it releases the retail sales statistics. I think I had a part to play in having it published after repeatedly pointing out to our statisticians that these figures are seriously flawed as an indicator of consumer activity. But does this health warning ever show up in newspaper reports the following day? No, it does not. I cannot speak for the Chinese-language press as I cannot read it, but the English-language press, including this newspaper, yesterday again billed April's 8.5 per cent year-on-year increase in the value of retail sales as better than expected and a sign of a robust consumer recovery. The first chart should set the picture straight. The value of retail sales represented in these statistics amounts to barely 25 per cent of consumer spending in our domestic market and over the years that trend has been steadily down. Missing entirely in these figures is spending on consumer services, the biggest single component of consumer activity. It accounts for about 55 per cent of spending in the domestic market. Also missing is a fair proportion of spending on consumer goods. If you order something over the internet, for instance, it does not show up in the retail sales statistics. In addition, the retail sales statistics include spending by visitors to Hong Kong but exclude spending by Hong Kong residents abroad. The reason they do so is that they are intended as a short-term indicator of how much classically defined retailers are selling, not of how much consumers are buying, and it is increasingly an outdated definition. But, if the figures are to be taken as an indicator of what Hong Kong consumers are buying, then things should be the other way round with spending by visitors excluded and spending by Hong Kong residents abroad included. All of this has implications for the reliability of these figures. Look again at that long-term declining trend up to the year 2000 in that first chart. It tells you that throughout this period the retail sales statistics consistently understated the true level of consumer spending. It may have been only a small error from month to month, but in aggregate it was a large one. Then look at the right side of the chart. For the past two years, the trend has reversed itself and turned up. The retail sales statistics are now capturing more of total consumer spending. This tells that they are now overstating the real trend. Once again, it may not seem that they are doing so by much, but the second chart tells a different story. The red line shows you a three-month moving average of the year-on-year growth rate of retail sales. As of March, that figure was 8.9 per cent. The blue line represents the growth rate for spending on consumer services and for the same month the figure here was only 2.3 per cent. Let me emphasise again that none of this spending on consumer services is captured in the retail sales statistics and yet it accounts for twice as much consumer spending as they do. In short, that health warning is well merited. These figures are collected and published in the way they are because they always have been, not because they are of any real value to anyone. In fact, I think that publishing them unless they are drastically revised is possibly more a disservice than a service to the public. But I have said it before and it makes no difference. The health warning is routinely ignored. You can be sure that next month the figures will again be misrepresented.