In 2008-09, I was seconded to Treasury. It was an extraordinary time to be there, seeing how a good response to fiscal stimulus gets crafted. And while I was an SES officer, my short stint meant that I did much more listening than talking.
But one thing I can claim credit for is the suggestion that when the $900 tax payments were being delivered in 2009, that the timing should be randomised by postcode, so as to allow the possibility of a subsequent randomised evaluation.
I’m pleased to report that such an evaluation has now been done, with my former ANU colleagues Emma Aisbett and Ralf Steinhauser (along with Markus Brueckner and Rhett Wilcox) combining a list of random postcodes with household spending data from AC Nielsen’s Homescan survey.
While their research finds little impact on some types of spending, it’s not hard to see why this is at odds with other studies – including mine- which find a large impact of the Australian fiscal stimulus on expenditure.
The problem is that while the Homescan dataset is the only one that lets you measure week-to-week spending patterns, it only captures groceries. So if someone spends their cheque on a washing machine, bicycle or restaurant meal, it gets missed.
The other factor is that this new study is very short-term. So if groceries spending went up after a two-month delay, it wouldn’t be captured.
So it’s perfectly consistent to note that the stimulus had a big overall effect, while also observing more limited impacts on narrow categories of spending. And I think the totality of the evidence is useful – unlike some evidence from the US, Australian households are typically not so badly off that their initial response to receiving a $900 payment is to stock up the fridge.