Published March 18, 2009

COMMENTARY
Time to turn attention to q-o-q figures

They identify turning points much faster during uncertainty

By ANNA TEO
Email this article
Print article
Feedback

ANOTHER month, another debilitating plunge in output and exports. In this season of economic anxiety, the next key indicator on the watch list is the first quarter's gross domestic product (GDP) growth - or rather, contraction, in these recessionary days.

Economists expect the economy to shrink a record 8.5 per cent in Q1, according to a poll by the Monetary Authority of Singapore late last month. That would be more than twice the 4.2 per cent contraction in Q4 2008, and would probably reinforce the feeling on the ground that things may well get worse before they get better.

Those GDP figures are year-on-year (y-o-y) rates that track the change compared with the corresponding quarter in the previous year. This is the primary headline measure in Singapore - as well as most other Asian countries, including Korea, Hong Kong and China.

The y-o-y rate is also the conventional measure for international comparison, although the key short-term GDP measure in major economies such as the United States, Japan, the United Kingdom, the euro area and Australia is the quarter-on-quarter (q-o-q) rate - the change since the previous quarter, either the raw measure or annualised.

Singapore compiles and publishes both y-o-y and q-o-q figures - the latter more as a supplementary measure. The y-o-y figure is the conventional measure for good reason - it is straightforward, transparent and objective. But the q-o-q measure is arguably the more indicative figure in these recessionary times.



Let's see why. Quarterly GDP data - or for that matter, monthly retail sales figures, too - have a strong seasonal component. So Q4 data covering the Christmas and year-end festive period would typically be higher than Q3's. The y-o-y measure inherently removes such seasonal effects.

Less comparable

By the same token, q-o-q rates need to be de-seasonalised. And just how - and how far - the data is seasonally adjusted in each country makes q-o-q rates less comparable internationally. Also, while y-o-y figures are directly comparable with annual growth rates because they are of the same scale, q-o-q rates need to be 'annualised' if we are to compare apples with apples. The annualised figure basically assumes that the economy will grow at that quarterly pace over a one-year period. (Hence the clunky 'saar' for 'seasonally adjusted annualised rate' when referring to q-o-q growth.)

But y-o-y growth rates are not without disadvantages. They do not correct for differences in the number of working days arising from moving holidays, for instance, and, following an unexpected shock or one-time boost in a period, will give rise to base effects the next year.

More importantly - or most pertinently for now - y-o-y growth rates are slower in identifying turning points. As a measure of immediate momentum, q-o-q rates - or more precisely, 'saar' - show up the underlying trend faster.

This is most evident in the latest downturn. Last year, the Singapore economy was in a steady downswing - from around Q2, as it turned out. But compared with Q2 2007, the y-o-y pace was still 'positive' - and indeed remained flat through Q3.

The 'saar', however, turned negative in Q2 2008, and stayed in the red through Q3 and Q4. The red ink in the revised y-o-y figures appeared only in the final quarter.

Typically, when the economy is on a steady growth path, the y-o-y rate, being a less volatile figure, is more than a good measure.

But in times such as now - and especially when the economy is under close watch for further downswings or the first signs of a rebound - the q-o-q figures are particularly relevant, and warrant greater attention.

Chances are, the q-o-q rate could start to turn positive soon, possibly in the current Q1, while the y-o-y figures continue to languish in the red for another quarter or two. But give it at least a second quarter of positive q-o-q before calling a turning point, and declaring a tentative rebound.