Abbott’s new cuts to Australia Research Council funding

Update: It appears that this money will be redirected to medical research projects, hurting social sciences and perhaps other ARC projects. See here for instance.

A bill currently being debated in the Australian parliament will set the funding of the Australian Research Council (ARC) for the next four years. The ARC is the main body that administers funding for the Australian research community. The National Health and Medical Research Council also administers a large amount of funding specifically for medical research.

The bill sets the funding planned for the next four years. I will quote the actual amendment, but the important thing to focus on here are the dollar amounts.

(n)  for the financial year starting on 1 July 2013—$883,959,000; and

(o)  for the financial year starting on 1 July 2014—$853,110,000; and

(p)  for the financial year starting on 1 July 2015—$783,253,000; and

(q)  for the financial year starting on 1 July 2016—$716,205,000.

It appears that Tony Abbott is planning a dramatic funding cut, and he is, but it would be interesting to see the historical context.


The CPI is the consumer price index. If the ARC is to to maintain its funding then it should roughly grow at the same rate as the CPI.

The previous 5 bills introduced under John Howard, Kevin Rudd, and Julia Gillard’s prime ministerships are shown along with the figures from the most recent bill.

It looks like in 2006 John Howard planned to track the ARC back to the 2004 funding level. With the global financial crisis Kevin Rudd planned an increase in funding with the 2008 and 2010 bills. Julia Gillard began to promise a return to surplus, and this is reflected in two more bills, one in 2010 and another in 2011. It is interesting to see that the actual funding has always been adjusted upwards from that planned under the previous 5 bills. If someone knows why this is, then please comment below.

Like Gillard’s bills, the new legislation increases the funding for the immediate future, but also sets future reduction. However, the cuts in this bill are much more dramatic than in previous bills. They represents a 20% reduction in funding over 4 years, returning total funding to the same level that it was in 2005 (after adjusting for CPI growth). This removes all the increases that were made under the Labor governments and restores the ARC’s funding to the level that was set by John Howard in 2005.

Data: You can collect the raw data by going to the Australian Parliament website and search for Australian Research Council Amendment Bill.

The “physics” of the Australian House of Representatives

Why is it that every election is referred to as a “landslide”? It’s due to the mathematical rules that govern the House of Representatives.

Specifically, it’s related to the relationship between the national “two party preferred vote” (2PP) and the local 2PP vote in each individual seat around Australia. What is this relationship?

Imagine what would happen if a party won 90% of the national 2PP. Would that party get 90% of the seats in the House of Reps? No. They would likely win 100% of the seats. This is because a party that gets more than 50% of the vote in every seat  will win every single seat. What if a party wins 51% of the national 2PP vote? Do they then win 100% of the House of Representatives seats? Well, if every seat around Australia was made up of similar people, then that is exactly what would happen!

As it is, different parts of Australia do have different demographics so that even if a party wins 70% of the national 2PP then there would still be a few seats where that party gets less than 50% of the local 2PP vote. The general relationship is shown in the lower right of the image below with data from the 2010 Australian election. I re-plotted the original data, which gave the % vote in every seat, so that it now shows the expected fraction of seats in the House of Representatives won by the Labor party, given a particular national 2PP.  You can see that at 80% national 2PP then the Labor party would win 100% of the seats, and with 20% of the national vote, none of the seats.

But I can also repeat this analysis in a different way and with completely different data. Let’s get the data of every election result since 1937. This is shown in the upper left. Each blue dot is an election result showing the fraction of seats won, given a particular national 2PP vote. The thick black line shows the mathematical relationship between the two numbers.

Putting these two different approaches together (the large middle plot) shows amazing agreement. The 2010 spread of seats fit very very well into the typical election results since 1937. Apparently, the demographic disparity around Australia has stayed pretty constant since 1937.


Upper Left: Election results since 1937
Bottom Right: Seat spread at the 2010 election
Middle: Combined data.

What does all this mean? Well when you see a poll in the newspapers, it gives the national 2PP vote, not the number of seats in the House. With the mathematical relationship shown, you can now very easily work out the actual result in the parliament. See the 2.99 in the equation? This means you multiply the national 2PP vote’s difference from 50% by a factor of 3. For instance, 52% of the national 2PP vote will give that party about 56% of the seats in the House. 47% of the national 2PP vote is 3% away from 50% and so will give 9% less than 50% of the seats, 41%. You can also get the same result by using the graph to trace out the percentage of seats that will result from any given national 2PP vote.

The curve shown in the lower right graph is really interesting (yes, I am a nerd). It has an amplifying effect on the movements in public opinion. In other words, small changes in opinion around the 50% mark are amplified in the final number of seats in the House. “Landslide” is actually a pretty good word to capture this “physics” because small changes give a very big result. Depending on your perspective, this might represent a design flaw, or it may represent a design feature. I actually consider it a feature. It means that the House is very sensitive to the opinions of the electorate, which in a democracy should be a good thing right? It also means that the possibility of a hung parliament is reduced. Although, in the analysis I completely ignored other parties. And it is really that number of small parties represented which gives the likelihood of a hung parliament.

Note: Although everything I said is true and useful it remains hard to predict particular seat results. Sometimes that is because the Australian Electoral Commission changes the boundaries, and sometimes it is because you have someone like Sophie Mirabella.

Data sources: The election results since 1937 are available here. The 2010 seat results are here.