Wanted – an economic rationalist that likes a punt. Hillary needs a hero who’s into shit-stirring and econometrics.
Me and my big mouth… A reader has taken me to task for the items
that appeared last month calling new Sydney Lord Mayor Clover Moore a
He disputes my claim that the new Sydney City Council regime will soon
be presiding over desperate scenes as unemployed suits press copies of
The Big Issue into the hands of the few remaining CBD workers up and
down the length of Martin Place.
Worst of all, he reckons that he’ll have figures that will support his
belief that Clover Moore’s election has done nothing to harm the Sydney
economy – and has offered me a bet on it.
Now, as a good economic rationalist, yours truly believes that free
choices made by informed individuals deliver the best possible outcomes.
The reader has got a methodology for his challenge – and an
adjudicator, University of Queensland academic Professor John Quiggin.
Our reader suggests we:
- Access the last four quarters of non-farm GDP for NSW and all other State pre-Clover
- Express the non-farm GDP for NSW as a percentage of the average of all non-farm GDP for other States for the pre-Clover period
- Access the next four quarters of non-farm GDP for NSW and all States post-Clover
- Express the non-farm GDP for NSW as a percentage of the average of all non-farm GDP for other States for the post-Clover period
- Compare the pre-Clover and post-Clover results in 2 & 4 respectively
If 4 is less than 2, yours truly will be asking our correspondent to
hand over eighty bucks to, say, the Centre for Independent Studies or
If 2 is greater than 4, however, I’ll be shelling out a donation to the Evatt Foundation or something similar.
Being a good capitalist, of course, I hope this will be tax deductible
– but I still need the help of a fellow economic rationalist who likes
“The Right is scared stiff of Quiggin…” his website – johnquiggin.com – boasts. Blud oath!
Is there anyone out there who can act as my second in this economic
duel? Anyone who can look at the methodology, suggest changes
and, when we’ve settled on what we’re going to do, pass verdict on the
figures 12 months down the track?
While yours truly appreciates the virtue of selfishness, it would be
good if such a person could participate out of a sense of enlightened
self-interest to prove that a certain hand isn’t just invisible, but
pretty deft, too – although it gets all sclerotic when the markets are
Some like minded subscribers have already come forward with good advice:
“If you accept the bet on those terms you will get growth figures
attributable to Lucy Turnbull, not Clover Moore,’ writes one.
“Well, probably growth figures attributable to neither of them, because
most of the growth will have nothing to do with who runs Sydney City
“If she has an impact it will be on building approvals and maybe
roadworks. So, the Moore legacy would probably be best reflected
in inflation in CBD office rentals, but that won’t happen for a
while. By the time she’s stuffed things up Crikey will be
profitable and she will be in retirement.”
Another Crikey fan is sceptical of the proposed methodology:
They say “It would appear your correspondent on the Clover Moore matter
either does not have an economics degree, or has one but has forgotten
what they learned in statistics and econometrics.
“Their proposed statistical method is patently flawed, evident in their
seeming misunderstanding of the use (or in their case misuse) of index
numbers in modelling exercises, such as the one they propose.
“Similarly, the proposal to compare to two outcomes is flawed, as two
point estimates can be practically different but not statistically
significantly different (your correspondent must have slept through
that lecture.) Anyone half-competent in statistics or
econometrics knows how to distinguish the two concepts.
“The correct method for assessing a Clover Moore effect is to use what
in time series econometrics are sometimes called ‘binary instrumental
variables’, using 0 for the pre-Moore period, and then 1 for the Moore
“You can also use scenario modelling for counterfactuals, but your correspondent probably did not understand that lecture.”
All Australians know what a bush lawyer is – and how they are bores best avoided.
Yours truly is something worse – a bush economist. We may not
know much about figures, but we know which ones make our bosses look
That’s what bush economists do. It’s a sad state that many political staffers find themselves in.
We find ourselves reduced to fiddling with the scale on X and Y axes on
graphs to produce tables that will look better – or trying to forget
that certain indicators even exist.
We will discuss statistics with lucidity while forgetting that people
who have had sixteen gin and tonics tend to be lucid, too – but short
on the coherence.
All of which makes this bet a difficult proposition.
If I’m going to accept it, I need to be able to isolate Clover Moore’s
effects from economic changes that are state wide or even played out
across Greater Sydney.
Commercial vacancies in the Sydney City Council local government area is one variable we could look at.
Other include rent (as a proxy for vacancy rates), advertising revenue
within the city, measures of retail vacancies and rents and counts of
people coming through the city.
If I am to accept this challenge – on behalf of economic rationalists
everywhere – what I really need to do is devise a “Clover Moore effect”.
That should really be quite easy to do in econometric terms – but a bush economist like me needs a bit of a hand.
What I want is some time series data – a collection of data across what is now the Sydney City Council area across time.
I need to find out if this data is reliable and available on, say, a monthly or quarterly basis at the very least.
With this data I can build what we bush economists call a “lagged
dependent variable model” and do what the econometrics bods call a
“trial run” and review the data.
If all that seems OK, I can then do one of two things.
I can create an impulse response model, where we look at the immediate
and near term impact of Moore on the chosen indicators, or I can insert
dummy variables to account for the pre-Moore and Moore periods, and
then run some econometric tests to examine their statistical
If these tests “fail” – a technical term I thought sounded good to use – then there will have been no “Moore effect”.
If they “pass”, however, then there is a “Moore effect” – one that can actually be measured.
As a mere bush economist, identifying the data to use, let alone doing
the modelling, may be a bit difficult – hence my cry for help.
However, there are some advantages. I will need at least six to
eight months of time series data under the Moore incumbency to try and
do this properly – plus around three years of monthly data before her
All of which might be a good excuse – if no white knight appears – to
spend the eighty bucks on a night out with Boilermaker Bill at the
Nippon Club and tell the challenger to piss off back to Leichhardt.
Hillary Bray can be contacted at [email protected]