Will This Year’s Contraction Be Much Worse Than Last Year’s?
This post was written by Karl Whelan
Media coverage of the current Irish recession has popularized two ideas. The first is that the recession came about very suddenly. The second is that forecasts (such as the ESRI’s prediction of GNP growth of -4.6% for 2009 compared with -2.6% for 2008) imply that economic performance this year is likely to be far worse than last year. It turns out, however, that neither of these ideas are correct. The source of the misconception, in both cases, turns out to be the lack of use of the CSO’s quarterly national accounts (QNAs) in popular discussions of the Irish economy.
Take the first idea — that the recession occurred quite suddenly. The swing from GNP growth of 4.1% in 2007 to a projected decline in GNP of 2.6% in 2008 has been widely cited to illustrate a violent switch from excellent economic performance in 2007 to a slump in 2008. However, the truth is more subtle. These figures for economic growth are the “annual” growth rate, defined as the percentage difference between the average level of output in a year and the average level of output in the previous year. However, this calculation doesn’t always give an accurate indication of economic performance during a year because it is heavily influenced by what are known as “base effects”. To give an example of what this means, consider a case in which GNP expands rapidly one year and then stays constant throughout the next year. We might intuitively view the second year as one in which there was no growth. However, the “annual” growth rate will appear to be quite positive because the average level of GNP in the second year is higher than its average level in the previous year.
A closer look at the quarterly data shows a gradual slide into the current recession. The QNAs show seasonally adjusted real GNP alternated between expansions and declines from 2007:Q1 onwards. GNP turns out to have peaked in 2007:Q3 and the graph below clearly shows the economy was essentially flat over the period from 2007:Q1 to 2008:Q1 before a more pronounced recession set in during 2008:Q2. So, far from being sudden, the quarterly data show the economy stalling for about a year before the public realized a recession was upon us. (My UCD colleague Colm McCarthy made this point in his paper with Rossa White presented at the DEW in Kenmare).
Now consider the second idea, that a swing from -2.6% growth last year to -4.6% this year implies that things will be getting much worse. Again it turns out that this swing is completely due to the dreaded base effects. The ESRI’s figure of -2.6% growth for 2008 is consistent with a 1% decline in seasonally adjusted GNP in 2008:Q4. Combined with the declines from earlier in the year, the implied level of GNP at the end of 2008 is below the average for 2008 as a whole. So suppose, for example, that GNP managed to remain flat at this level for the whole year (a highly unlikely achievement). In this case, the average-over-average calculation for GNP growth for 2009 would give -2.1%. A dramatic turnaround from sharp contraction to flattening out would not be picked up by this measure.
Against the background of a severe worsening of the global economy and serious domestic fiscal problems, perhaps a better starting point is the observation that the year after 2008:Q3 is unlikely to be better than the previous year. GNP declined 4.8% over the year ending in 2008:Q3. Projecting this average decline of -1.2% per quarter through 2009:Q4, gives a figure for annual (average over average) GNP growth for 2009 of -5.1%.
These calculations show that, rather than a severe worsening, forecasts such as the ESRI’s projection of -4.6% for 2009 actually represent a slight improvement relative to the economic performance seen in the year ending in 2008:Q3.
My point here is not to question the ESRI’s forecast, which may turn out to be a good one, but to argue that economists and media commentators should make more use of the QNA figures when discussing the performance of the Irish economy.
One small change that could be made is to switch from discussing the average-over-average figure for output growth to instead discussing Q4-over-Q4 figures, which reflect the level of output at the end of the year relative to the start of the year. From my time working at the Federal Reserve Board, I know that this is the measure of growth that the Fed staff uses when describing its forecasts to the FOMC. In the above example of flat Irish GNP this year, this Q4-over-Q4 figure would show the economy improving from -4.7% in 2008 to 0% in 2009 rather than the marginal improvement cited above of going from -2.6% to -2.1%.
A caveat to this suggestion, however, relates to the use of GDP and GNP figures for projecting tax revenues. Because Ireland’s tax year corresponds to the calendar year, the year-average level of output is the appropriate figure for budget calculations. The calculations above show how far we are now from the projections underlying October’s budget. Taking the ESRI’s figure for 2008, achieving the budget’s projection of a 1% (average over average) decline in GNP in 2009 would require GNP to actually grow at a 2% annual average rate throughout 2009.
January 6th, 2009 at 11:07 am
Nice.
I’d be inclined to use GDP for analysis of quarterly output trends as the exact timing of the profit outflows that account for most of the difference seem largely unrelated to output.
Interestingly, real GDP peaks even earlier: in the first quarter of 2007. (That’s the same as when nominal house prices peaked. Wouldn’t you know it.)
But if you do use the GDP figures, then the 2009 quarterly falls will need to be steeper than you say to get to the ESRI figure of -4% for that year. Maybe things are getting steeper after all.
January 7th, 2009 at 5:43 am
I am not happy using the CSO quarterly national accounts to time turning points in the Irish economy too precisely. The article by Patrick Quill of the CSO in the Autumn QEC shows that substantial revisions are made to the quarterly accounts over the first two years after publication. While substantial revisions are also made to the annual numbers by the CSO these numbers have a firmer base, being subject to smaller, though still substantial, revisions in the years after first pubication.
I presume what we all try and do is forecast/estimate the final “true” figure.
January 7th, 2009 at 9:53 am
John’s point about the preliminary nature of the QNAs is correct (and posted impressively early in the morning!)
It raises some interesting questions about the usage of the QNAs in assessing the state of the economy. On this, I’d make a couple of observations.
While the revisions to quarterly GDP are indeed large, Quill’s article also reported that the revisions were not statistically significant. So, despite the obvious caveats, it could be argued that the QNAs still represent as accurate a picture of economic trends as we can get.
Also, in terms of picking out turning points, the revisions are largely due to the information added on the release of detailed annual national income and expenditure accounts. These revisions tend to change all of the quarters within a year more than the intra-year pattern. So, with the caveat that even the final quarterly estimates are not perfect, it appears that one can still use the intra-quarter movements in the current QNAs to assess, for instance, whether the economy strengthened or weakened over the course of a particular year.
March 26th, 2009 at 2:58 pm
[...] I noted before, the Irish media and forecasters tend to focus on the year-average over year-average figure for GDP [...]
July 15th, 2009 at 1:35 pm
[...] Central Bank forecast actually implies, there is the more general issue, which I have referred to before, of the difficulty in mapping forecasts based on year-average over year-average into commentary [...]