Improvements in the production and dissemination of GDP data now mean that output and growth figures are now released within three months of the end of each quarter. Of course there is a great deal of interest in these important data, which can potentially reveal a great deal about the level and growth rate of economic activity. However, the data need to be interpreted carefully. The recent release of data for the first quarter of 2011 has given rise to headlines such as “Economy falls 2.2% in first quarter” (Mmegi, July 5, 2011), referring to the fact that real GDP was 2.2% lower in 2011Q1 than in the previous quarter (2010Q4). The reported negative growth would seem to be bad news, perhaps indicating that the economy is heading back into recession. But is this the correct interpretation?
GDP growth can be reported in various ways from any given set of GDP output data. The main ways of reporting GDP growth used internationally are:
(i) quarter-on-quarter, annualised growth (qoqar): the growth in real GDP from one quarter to the next – e.g. from 2010Q4 to 2011Q1 - expressed at an annual rate;
(ii) quarter-on-same-quarter-a-year-earlier (qo4q): e.g. the growth in real GDP between 2010Q1 and 2011Q1.
(iii) annual GDP growth (yoy): growth in GDP over full year (4 quarters) compared to the previous 4 quarters;
Each of these methods has various advantages and disadvantages, and the choice of the most appropriate measure depends on both the circumstances and requirements of the reporting.
In any period, GDP is determined by both underlying developments in economic growth and short-term events. Sometimes the short-term events can be the result of economic changes, such as an economy entering a recession or a recovery period (which is important information), but they can also be the result of random events or volatility (which is not so useful or important). Interpretation of GDP data, as with much other economic data, is in part a question of distinguishing “signal” from “noise”.
So which measure is most helpful for Botswana? One point to note is that the measure that Mmegi reported on is that it does not fall under any of the categories (i) – (iii) above. This is because the growth figure, which is reported by Statistics Botswana, is not “annualised” – i.e. it is simply quarter-on-quarter growth, which is not comparable with the more conventionally reported annual growth rates. Annualising the q-o-q figure in line with international practice results in a GDP growth rate in 2011Q1 of minus 8.6%, which sounds altogether more serious than the reported minus 2.2%.
The qoqar measure of growth is widely used internationally, as it can provide a timely snapshot of economic developments that responds quickly to underlying changes. It does, however, have a problem in that it is heavily influenced by short-term volatility or “noise”, which can detract from the more important underlying developments. Hence, it has to be used with caution. It is also influenced by seasonal factors, and hence the measure is most appropriately used when “seasonally adjusted” GDP measures are available.
In Botswana’s case, GDP tends to be very volatile from one quarter to the next. For instance, real GDP growth according to the qoqar measure was 80% in 2009Q4, but was minus 35% in the following quarter. This volatility is partly because movements in GDP are dominated by changes in mining output, which is highly variable on a quarterly basis. But other sectors also show volatile growth on a quarter-to-quarter basis. For instance, in the first quarter of 2011, using the qoqar measure, growth rates were minus 31% for agriculture, -15% for mining, -22% for finance & business services, -27% for government, and +36% for construction. Because of this volatility, and the fact that seasonally-adjusted GDP data are not provided for Botswana, the quarter-on-quarter measures of GDP growth that are conventionally reported are not very useful.
The qo4q measure ((ii) above) is perhaps more useful, in that it is less affected by seasonal factors, but it still dependent upon a small amount of data (two quarters, like the qoqar measure), and hence is still very volatile and the signal-to-noise ratio is low.
The annual (yoy) measure of GDP growth ((iii) above) has the advantage of being based on much more data (eight quarters) and hence is much less affected by short-term volatility. But it is less effective at detecting economic turning points quickly. Nevertheless, in our opinion it is the most useful measure of GDP growth, especially for monitoring longer-term growth trends. Unfortunately it is not regularly calculated or reported by Statistics Botswana, which only reports the qoq and qo4q measures on a quarterly basis – the annual figure is only reported for full calendar years. Nevertheless, it is easily calculated from published data, and Econsult does so each quarter.
GDP growth measures – quarterly since 2010Q1
| | 2010 Q1 | Q2 | Q3 | Q4 | 2011 Q1 |
| qoqar | -34.5% | 16.1% | 15.9% | 4.3% | -8.6% |
| qo4q | 17.3% | 4.2% | 12.1% | -2.1% | 6.4% |
| annual (yoy) | 2.3% | 3.8% | 10.5% | 7.2% | 4.9% |
So what does this all mean? The most recent GDP growth rate according to our preferred measure was 4.9% in 2011 Q1 – i.e. GDP in the year to March 2011 was 4.9% higher than over the previous 12 months. However, this reflected quite different growth rates in the mining and non-mining sectors of the economy – mining grew by 0.6% over that period, while the non-mining private sector grew by 7.7%. While the overall growth rate is healthy enough and broadly in line with expectations for the year as a whole, the mining result is disappointing given expectations that mining sector recovery would lead an upturn in growth. However, the outturn for the non-mining private sector is very positive.
In our view, Statistics Botswana should place less emphasis on the qoq growth rate that currently receives so much attention – at least until they can also produce seasonally adjusted GDP data. And even then it should be presented at an annual rate, in line with international practice. But more importantly, SB should calculate and publish the yoy GDP growth rate on a quarterly basis, and not just at the end of each calendar year.
See the figures below and decide which of the three measures presents the most useful information.
