The recession we had to deny

Once I thought I was wrong, writes Gavin R. Putland. (Updated July 2.)

In the June 2 edition of Crikey, quoting the previous day's post on this blog, I wrote:

The bounce in employment in Apr.'09 probably represents a pull-forward of activity due to the anticipated expiry of the First Home Owners' Boost — in which case the unexpected extension of the boost ... will presumably spread the activity into the second half of the year. On account of this stimulus, positive GDP growth for Q2 of 2009 is not out of the question. But a contraction in Q1 looks inevitable.

When I wrote those words I did not have the March-quarter balance-of-payments figures (released on June 2), which showed an improvement, prompting various forecasters to raise their GDP estimates for the same quarter. But the balance of payments improved only because a fall in exports was outweighed by a fall in imports. And while the fall in imports by itself was positive for GDP, it suggested less capacity to purchase imports and less scope for adding value thereto, both of which were negative for GDP, making me wonder whether the said forecasters should rather have revised their GDP estimates downward.

So when the +0.4% GDP figure was released on June 3, my first reaction was that I had underrated the temporary effect of the Federal Government's stimulus measures, or misjudged the distribution of that effect between the March and June quarters, or both. And I may have. But, as I noted in a letter in Crikey on June 4, there are other possibilities:

In the latest National Accounts spreadsheet from the ABS, the column that everyone watches is headed “Gross domestic product: Chain volume measures - Percentage changes ... Seasonally adjusted”, and declares that growth for the last two quarters was −0.6% and +0.4%.

Two steps to the right is a column headed “GDP market sector: Chain volume measures - Percentage changes ... Seasonally adjusted”, in which the figures for the corresponding quarters are −0.3% and −1.4%. That represents not only two consecutive quarterly contractions, but a year-on-year contraction of 1.1%. This column presumably should, and usually does, roughly agree with the first one. But the discrepancy for the March quarter is 1.8 percentage points — the biggest since December 1982, i.e. since the recession that finished Malcolm Fraser.

A further six columns to the right, we find changes in GDP at current prices. If we adjust these for the CPI (ABS 6401.0), the results for the last two quarters are +0.1% and −0.7%. That's an expansion followed by a contraction, whereas the first column has it the other way around.

I interpreted these discrepancies as evidence that in these “exceptional times,” official statistics were more “noisy” (i.e. uncertain) than usual.

But wait: there's more. In the balance-of-payments results, the ABS reported that goods imports were down 11% and goods exports down 10% in current-dollar terms. In the corresponding GDP results, it reported that goods imports were down 8.6% and goods exports up 2.8% in volume terms. (All changes are with respect to the December quarter, seasonally adjusted.) These figures imply that between the December and March quarters, import prices fell about 2.6% and export prices fell about 12.5%.

It was therefore easy to believe Gerard Minack when he wrote in the Age on June 4:

The surprise was due to the Australian Bureau of Statistics changing the way it tracks bulk commodity prices. Usually the bureau waits for the major contracts to be settled, and factors in the price changes in the June quarter (because the contract prices are set from April 1). Yesterday it announced that it was factoring in lower prices in the March quarter. Factoring in a lower price implied a higher volume for exports, which lifted GDP. Consequently, net exports added 1.4 percentage points to growth in the quarter. Forget this statistical jiggery-pokery...

This was denied by the ABS's Peter Harper on June 5:

The statement in the article that “Yesterday it (the ABS) announced that it was factoring in lower prices in the March quarter” is not correct. No such announcement was made. Furthermore, given the methodology that the ABS uses to measure the volume of bulk commodity exports, such a claim could not possibly have been correct.

The volume measures of exports of bulk commodities recorded for the latest quarters in the ABS Balance of Payments and National Accounts Statistics are calculated by multiplying the quantities of such exports, as reported by exporters in tonnes or some other unit of quantity, by the average price of such commodities, as reported by exporters in the reference year. For the March quarter 2009 volume estimates, 2006-07 is the reference year for prices. The reference year prices are updated annually, in the September quarter accounts. For example, the September quarter 2009 accounts will use average prices reported in 2007-08. This methodology assures that movements in the volume of bulk exports from one quarter to the next reflect only changes in actual volumes and are not influenced by changes in prices.

Except, apparently, between the June and September quarters. So why does the above comparison between the balance-of-payments and GDP figures imply that export prices changed between the December and March quarters? My guess is that Harper's statement applies to the production measure of GDP while Minack's applies to the income measure, and/or that Harper's statement was restricted to “bulk commodities” while Minack's should not have been. I therefore expected Minack to have some sort of comeback. But he offered only an abject apology (quoted by Steve Keen), which he seemed to contradict in his penultimate sentence:

As an aside, my mistake doesn't change the conclusions in my original comments.

Why not? Because while the ABS may not have “announced” that it was factoring in lower prices for “bulk commodity exports” in the March quarter, the percentage changes in export values and export volumes indicate a fall in prices for exported goods in general. And while the procedures common to the Balance of Payments figures and National Accounts figures may not have involved any change in “bulk commodity” prices between the December and March quarters, some procedure unique to the National Accounts figures certainly involved a change in export prices.

Whether that change in export prices has any foundation in reality is the two-consecutive-quarter question. If it doesn't, the overestimation of GDP in the March quarter will cause an underestimation of the change in GDP for the June quarter, making it more likely that official figures will show a contraction for that quarter. In the mean time, I stand by my conclusion in the June 4 edition of Crikey:

... I have not seen anything to change my view that (a) the economy is sick because property prices have fallen from absurd heights, (b) prices have further to fall, and (c) the various stimulus measures to date have merely delayed the inevitable.