Tuesday, 14 October 2008

The need to get a grip on percentages

It is of course a matter of some satisfaction to economists that issues of fascination to them, but which are normally considered unutterably boring by almost everyone else, have suddenly become intensely fashionable in the media. When normally would interbank lending rates feature in the first item on the TV news? When else would people be commenting on theories of debt-deflation? In what other circumstances could TV presenters terrify their viewers and children with talk of movements in credit default swaps?
Doubtless during hiatuses in the crisis it will all change and the news programmes and most newspapers will get back to serious matters such as Celebrity Big Brother - or perhaps not, it all depends what happens next. However, in order to interpret what is going on, in particular today's claim of 'bounce' in the markets, it is important to understand percentages - maths not usually being a favourite subject at school it is time to brush up a bit in order to follow the news. This is particularly important when reading headlines such as Japan Stocks Gain on Bank Support Plans; Nikkei Rises by Record - commenting on the fact that as the Bloomberg article put it 'The Nikkei climbed 14.2 percent, rebounding from the worst week in its 59-year history'. Similar articles will doubtless appear on British share markets if they rise today.
If something falls by 90 per cent and then the same thing, for example a share index, rises by 90 per cent someone might, without thinking about it, imagine that they had got back to where they started - that certainly seems to be the thinking of journalists who write such headlines. Unfortunately it is not the case - such a movement would mean that something remained 81 per cent below where it started - do the maths and it will be obvious.
The drop of 24 per cent on the Nikkei between Friday 3 October and Friday 10 October means that the it would actually have to rise by 32 per cent to reach its previous level. What actually occurred is shown below. As can be seen the 'bounce' on 14 October only made a small dent in the loss. This shows why SEB concentrates most on long term trends.
Flaky use of percentages regarding individual days, either up or down, or selective use of baselines, creates seriously misleading impressions and is an example of dodgy statistics - for another illustration remember that if you look at GDP starting in 1933 the 1930s appear as a decade of successful growth, the problem is that you have cut out the fall that took place after 1929.

No comments: