During the great Stimulus Debate there was some argument about whether the cash bonus ($900) would be saved or spent. The government argued that the money would be spent. In our Senate submission Ash de Silva and I referred to this debate (links added):
The Australian Treasury relied on an unpublished study by Christian Broda and Jonathan Parker to support their argument that the tax rebate announced as part of the federal government stimulus package would be spent and not saved.
This particular study investigates the 2008 US$950 tax rebate by comparing spending in households that had received the rebate to spending in households that were eligible to receive the rebate but had not yet actually received the rebate. The econometric test determines whether or not consumption is higher in those households that have received the tax rebate compared to those households that have not received the rebate. It does not test whether the entire tax rebate has been consumed or saved. The study finds that those households that had received the rebate consumed more than those that have not. Unsurprisingly, they find that low income households and liquidity constrained households that had received the rebate spend more than similar households yet to receive the rebate. (They claim that their test shows that low income households have spent more than higher income households, but it is not clear from the reported table that this is the correct interpretation of their results.)
It is important to note that they have found that some of the tax rebate is spent. Of course nobody is suggesting that some of the tax rebate wouldn’t be spent. The debate is about (1) how much would be spent and (2) was this the best way for the government to stimulate the economy? Indeed, Broda and Parker could very easily expand their econometric analysis to determine how much of the US$950 was in fact spent, but do not do so. Rather they provide survey evidence of additional spending. The survey results indicate that US$448 was spent in additional purchases – approximately 48 percent of the US$950. That implies the other 52 percent was saved (at least temporarily). They concluded that ‘the stimulus payments are initially being spent at significant rates’. This is, of course, true; it is also fair to say that the Economic Stimulus Act of 2008 has not succeeded.
Two things have happened since then: Broda and Parker updated their analysis, and an Australian version of the Broda and Parker paper has appeared (the ANU paper).
Judith has a discussion of that paper in The Australian.
The main result of the ANU paper is summarised by the authors in the following way: “The household consumption response to the bonus payment is insignificant. It is also quantitatively small. For example, the average household, which received a payment of $900, spent in the week of receipt of the payment an additional $1 on non-durables or less than 1 per cent.”
Yes – you read that correctly – out of $900 an addition $1 was spent on non-durable goods. I don’t want to concentrate of Judith’s piece – she can post more on that. I want to point to another scandal – how did the government (and Treasury) come to rely on the original Broda and Parker paper?
If you examine the original five page paper it is very preliminary. The latest version of the paper is 27 pages long plus tables. The original paper contains a (very) incomplete description of the methodology and very basic econometric results. There is a statistically significant difference between spending in households that have received the stimulus payment and those that have not. The revised version is much more sophisticated. The basic result is unchanged. We now know that the there is a statistically significant increase in spending in the week that the household receives the stimulus payment. But, of course, the question is how much additional spending?
On average, we find that household spending rises on receipt of an ESP and remains elevated for some time. Across specifications, we find that households raise their spending on NCP-measured household goods in the week of receipt by roughly 10 percent of average weekly spending, about 1.5 percent of the average ESP, or around $14. This spending effect decays slowly over the following weeks, so that over seven weeks, the receipt of a payment causes 3 to 5 percent more spending, roughly 4 percent of the ESP, or 30 to 50 dollars spent on covered items. There is no significant change in spending prior to receipt, lending support to our methodology.
In this paper the average stimulus payment (labelled ESP) is $902 (in the original version they reported a $950 payment) – so we are looking at $50/902 over a seven week period. In the original study Broda and Parker spoke of a survey augmenting their results showing additional spending. In the second paper they use the survey to provide additional controls but do not refer to additional spending. The other interesting finding is that the ANU paper finds an announcement effect while the Broda and Parker finds none. Both paper claim that result bolsters their argument.
Okay – so here is the rub: Treasury claimed at the original Broda and Parker paper showed that spending would increase but didn’t and couldn’t say by how much. Set aside that Treasury would rely so heavily on what can, at best, be described as an extended abstract and not a peer-reviewed or published paper. The problem here is what Deirdre McCloskey calls Oomph. Broda and Parker claim that the spending they find is economically significant. I’m not convinced but at least they made the argument. I wonder, however, if politicians had been told …
Quick. We have to get people spending and if we give them $900 they’ll rush out during a seven week period and blow $50.
… whether they would have been so keen to spend the money?
As an aside I notice that the ANU paper has not referenced either of the Broda and Parker papers despite using the same methodology and data source (the Australian version of the American consumer panel). That is an oversight that should be corrected.
Update: Barnaby Joyce raised the issue in the Senate.