Getting a fix on the situation in retail has been difficult of late. The monthly (nominal) retail trade survey has never been particularly good, and it seems to have been busted by the noise of the floods etc in Q1.
I think that a better picture of what’s going on can be had from the large retailer sub-sample (the Completely Enumerated sector). This sector showed real growth of ~1%q/q in Q1, which seems more consistent with the gain in retail jobs, and my assessment of the overall tenor of recent data.
There has lately been some disagreement where there ought not be — between company reports, the monthly and quarterly retail trade surveys (ABS 8501), the quarterly Business Indicators (ABS 5676), and quarterly employment in the retail sector (ABS 6291). I have put the level of the three (normalised in Q3’06) above, and the growth below.
Given the strength of employment growth, I have had a hard time swallowing the retail doom and gloom – especially as the sector is typically quick to cut jobs in a downturn (witness Q4’07). A possible explanation is that they had cut things too hard and are now playing catchup – however if this is the case, we should not be too worried about job losses if retail is indeed flat for a quarter or so.
So what does the data show? In Q1 the ABS estimated that nominal spending grew around 0.8%q/q (Ave since 2002 is 1.3%q/q nominal growth). This was decent nominal spending growth, but lousy real growth (~0.0%q/q).
The reasons real spending was weak are well understood, and they are unlikely to remain a headwind — the drop in real spending was partly because the food price deflator rose 1.5%q/q and partly due to the ‘fuel tax’ (fuel is not in the Aussie retail survey, however the increase in petrol prices would also have sapped spending power).
This does not solve the puzzle between the increase in employment and the weakness in real retail, however I think the answer is that the monthly retail trade survey has a few bugs in it.
We can separate the retail trade survey into the Completely Enumerated sector (a population measure of large retailers) and a sample of the rest. Re-weighting the implicit price deflators, to reflect the different sales mix of the larger CE retailers and the smaller sampled retailers, we can then look at what we know for sure (the CE sector) and what we’re just guessing about (the sample).
Nominal spending growth in the CE retail sector was 1.8%q/q in Q1, and price inflation was around 0.8%q/q, leaving us with a 1%q/q real increase in retail spending. This seems more consistent with the employment data. In the sample, we find that nominal spending fell by 0.8%q/q, with real spending down 1.3%q/q, thanks to a +0.5%q/q increase in retail price inflation.
I find the divergence in sample and CE sales a little hard to make sense of, and think that the CE data also comports better with my assessment of the broader economy – in particular it better fits the ongoing down-trend in the unemployment rate, and increase in job advertisements.
Thus, I conclude that the recent ‘weakness’ in retail sales probably has been overstated. Even if it hasn’t been overstated, wage gains and jobs growth appears to have been sustained, and the subtraction in spending power from the food and fuel tax will at least end, and probably (partially) reverse.
The dataflow must be killing you guys at the moment. Off the top of my head, in the past week we’ve had…
Weak retail sales.
Weak vehicle sales.
Weak employment growth.
Weak consumer sentiment
Weak wages growth.
(I’m sure I’ve forgotten something else)
Which bit of that don’t you understand?
And yet, if the labour force survey had been strong you would have been thumping the table for a rate rise. Ditto for wages growth, lending, retail sales etc etc. If the data doesn’t fit your view, you discredit the data or dismiss it as meaningless.
FWIW, I think you’re massively overplaying the significance of the mining investment boom. Even if it results in aggregate GDP growth, the growth is all happening in a very narrow sector of the economy. The services sector is flat at best, feeling the pinch from higher-than-necessary interest rates, and confidence is sagging with house prices. Everyone knows someone in manufacturing, education or tourism who’s doing it tough, but the vast majority of Australians have no connection with mining, which is something that happens half-a-continent away.
Monetary policy is being set according to economic conditions in the Pilbara and not Western Sydney, and last time I looked, a hell of a lot more people live in Western Sydney. If nothing else, this will have huge political implications if (when?) the RBA hikes again.
You guys really need to get out more.
P.S. Hi Chris. Why don’t you allow comments on your blog?
Good analysis, when data is completely enumerated is obviously much more reliable.
However, as far as I know, the unemployment report from the ABS has a large 95% Confidence interval. It is not completely enumerated.
In particular for “Total Employment” there is a 95% Confidence interval that the change may fall between:
April: -76 700 to 32 500
March: -16 800 to 92 400
Feb: -64 500 to 44 300
So we could have lost 100k jobs easily in the last 3 months or added 100k.
I believe this would apply to retail jobs too.
The Performance of Service Index has been weak for retail and flattish in the employment index.
Comments are closed.