Chapter Ten:
The Inner Metropolitan Republic

Malcolm Mackerras

As is clear from the above, I say that electorates voting 'Yes' were exclusively from Australia's inner metropolitan areas. That is why I am calling this paper The Inner Metropolitan Republic. It needs to be admitted, however, that the Australian Electoral Commission does not define all those 42 divisions as 'Inner Metropolitan'. Rather, it defines 27 as being such, while 13 (Aston, Berowra, Boothby, Bruce, Calwell, Deakin, Fowler, Hotham, Jagajaga, Maribyrnong, Menzies, Ryan and Scullin) are defined as 'Outer Metropolitan'. Two seats, Cunningham and Newcastle, are defined as 'Provincial'. For more information see Table 4, below.

In the cases of Cunningham and Newcastle the AEC description as 'Provincial' is very difficult to defend. I would have thought the description 'Inner Metropolitan Wollongong and Newcastle' would be appropriate. It is a striking fact that Cunningham and Newcastle voted 'Yes' while all the outer metropolitan Newcastle and Wollongong divisions (Charlton, Dobell, Hughes, Robertson, Shortland and Throsby) voted 'No'.

In the cases of the 13 AEC 'Outer Metropolitan' seats which voted 'Yes', I can see how Ryan in Brisbane (238 sq km), Berowra in Sydney (463 sq km) and Calwell in Melbourne (234 sq km) might merit that description. However, the AEC describes both ACT electorates as 'Inner Metropolitan'. Canberra has an area of 1,900 sq km and Fraser has an area of 535 sq km. For that reason my picture of the 42 seats voting 'Yes' is that every one of them may sensibly be regarded as inner metropolitan. No wonder they combine to cover only 0.1 of one per cent of Australia's land mass!

Altogether 14 seats classified as 'Inner Metropolitan' by the AEC voted 'No'. They were Banks, Blaxland (Paul Keating's old seat), Cook, Fremantle (Carmen Lawrence!), Hindmarsh, Lilley, Moreton, Parramatta, Perth, Port Adelaide (Mick Young's old seat), Reid, Stirling, Swan and Tangney. Still, while monarchists may rejoice at the result in a seat like Blaxland, it has to be admitted that the Queen was, as they would say, 'done like a dinner' throughout inner metropolitan Australia -- and in every State.

Meanwhile the republic was massively rejected everywhere else.

The above observations may sound rather partisan. Tables 1 and 2 record the situation in formal terms.

Since two questions were put to the people the Preamble result is also recorded here. The fact that the formal vote was some 11,000 less for the Preamble is another way of saying that the Preamble's informal vote was some 11,000 greater than for the republic. It can be seen from Table 3 opposite that its 45.1 per cent affirmative vote gave the republic a rank of 30 out of the 44 referendum questions put since Federation. In other words, there were 14 proposals getting less overall support than the republic which, in any event, was markedly more successful than the Preamble. The dismal defeat of the Preamble was shown by its 39.3 per cent affirmative vote, placing it at number 38 when ranked by national 'Yes' percentages. In not a single electorate (not even in Bennelong) was support for the Preamble greater than for the republic. Now please excuse me for writing no more about it. The less said about the Preamble the better!

If readers should wonder why only formal votes and percentages are shown, there is a good reason. It relates to the accepted interpretation of section 128 of the Constitution. The fourth paragraph of that section actually reads:

When I first read those words (as long ago as 1951) I assumed that the 'Yes' vote must exceed the combination of 'No' and informal votes both nationally and in four States. Otherwise the proposal would be deemed to have failed. Certainly the words create that impression.

However, I checked the records and found that in September, 1946 the social services amendment was deemed to have been carried in all six States. Yet in three States it was not true that 'a majority of the electors voting approve the proposed law'. In Queensland the affirmative vote was 299,205, the negative vote 284,465 with 28,500 informals. In South Australia the affirmative vote was 197,395, the negative vote 184,172 with 17,734 informals. In Tasmania the affirmative vote was 67,463, the negative vote 65,924 with 11,493 informals. On an apparent literalist interpretation the social services amendment would never have been carried. It 'failed' in three States.

The explanation is that the relevant part of section 128 has always been interpreted as though it reads as follows:

Constitutionally speaking, informal votes have exactly the same status as votes not cast at all. Therefore, let me ignore informal votes entirely. However, when I use the expression 'the accepted interpretation of section 128' (see above) it should be noted that at least one man will dissent from me vehemently. Mr Justice Ken Handley (Court of Appeal, Supreme Court of New South Wales) has given me details of a Scottish case in 1921 (Latham v. Glasgow Corporation) which he claims settles the argument. The judges wrote:

In brief, the judges decided the proposal had not been carried because the combination of negative and informal votes was high enough to counter the 'Yes' votes. When our High Court makes a similar finding in some future Australian case I shall change my analytical practices.

The 45.1 per cent affirmative vote means that the republic would have gained an overall national majority with a mere five per cent lift in its support. However, that would not have carried the republic proposal.

There are, in fact, seven different vote values in an Australian referendum. Seen from the pro-republic perspective the unfortunate fact is that, by and large, their support was greatest where vote values were least. By contrast, opposition to the proposal was strongest (again, by and large) in the States with the best vote values. A vote cast in either the Australian Capital Territory or the Northern Territory has the same value as each other. However, such a vote has the least value of all since it is counted nationally but not by State. Of the seven values, therefore, the Territory vote is the least valuable. The second least valuable vote is that cast in New South Wales. At the other extreme of value is the vote cast in Tasmania. If we combine the two votes (ACT plus Northern Territory) we get 58.7 per cent Territory support for the republic. In other words the only affirmative vote among the seven was the one with the least value! On the theory of uniform swing, a national 'Yes' vote of 54 per cent would have seen success for the republic. It would then have carried the four States needed for passage -- Victoria, New South Wales, South Australia and Western Australia. That would have left Queensland and Tasmania as the dissenting States.

It is often asked why the Australian Capital Territory was the only jurisdiction to record an affirmative vote. Throughout this paper readers will come to understand that the ACT has every feature which would predict its high republican vote. The three main characteristics are those of residence, socio-economic status and party. The republic was always a Labor cause, and the ACT is the most strongly Labor of the eight jurisdictions. The referendum result, however, was one in which the 'Yes' vote was essentially an inner metropolitan phenomenon with a link to high socio-economic status. As the most Labor, most inner metropolitan, jurisdiction, with high indexes of relative socio-economic advantage, the referendum vote in the ACT should cause no surprise.

Before I leave the ACT (for the time being) it is worth noting a point from Table 6 below. The combined votes of the adjoining seats of Sydney and Grayndler give a higher 'Yes' percentage than the combined votes of Canberra and Fraser. The combined votes of adjoining Melbourne and Melbourne Ports give an even higher affirmative percentage. Thus we can say that inner metropolitan Labor seats in Melbourne, Sydney and Canberra provided the heartland of support for the republic.

The point about place of residence is so clear from the aggregate data that recourse to opinion poll findings has not yet been necessary in my analysis. From now on that changes. I rely increasingly on the findings of the opinion polls. I contend that the second best predictor of the vote is by political party.

On the day (6 November, 1999) of the referendum itself The Weekend Australian carried the results of the Newspoll taken on 3-4 November. Overall they showed a 'Yes' vote of 47 per cent, a 'No' vote of 50 per cent and three per cent uncommitted. That was a moderately accurate prediction of the outcome, albeit an under-estimate of the magnitude of the republic's defeat.

However, the truly interesting finding is on page 8 of The Weekend Australian. It showed Labor voters as splitting 61-38 in favour of the republic (with one per cent uncommitted), while Coalition voters split 62-35 against (with three per cent uncommitted). When the votes were actually counted the results were fully consistent with such a finding.

A useful exercise is to translate the October, 1998 general election vote into the November, 1999 republic referendum. I estimate that, of those who gave their two-party preferred vote to Labor in 1998, the split in 1999 was 57-43 in favour of the republic. Of those who gave their two-party preferred vote to the Liberals in 1998 the split in 1999 was 65-35 against. Finally, I estimate that 80 per cent of the 1998 National Party vote was cast against the republic in 1999.

In the absence of opinion poll findings the data in Tables 7 and 8 might not be so persuasive. However, when combining the two I think the estimates of the preceding paragraph are highly plausible. The terms 'Safe Labor' and 'Safe Liberal' refer to all those seats above 10 per cent on the Mackerras pendulum. For example, on the Labor side the strongest 'Safe Labor' seat was Batman and the weakest Fremantle. For the Liberal Party the strongest 'safe Liberal' seat was Bradfield and the weakest Indi. (Note: reference to the pendulum is to the one published immediately after the 1998 general election. Thus Bradfield was the strongest Liberal seat. However, boundary changes made recently weaken the Liberal vote in Bradfield. Thus the strongest Liberal seat going into the 2001 general election is the Victorian Division of Murray. By the same process the strongest Labor seat is Fowler, to which further references are made below).

On the night of the referendum John Howard was made to suffer from continual sneering references by broadcasters to 'the Republic of Bennelong'. As Table 6 shows, it was indeed the case that his inner metropolitan Sydney seat of Bennelong was one of the 17 Liberal seats to vote 'Yes'. However, it is clear that the Prime Minister had good reason to be pleased by the results, taken overall. With the exception of 'Safe Labor' seats, all categories turned in an overall negative vote. It is true that 'Safe Liberal' seats, taken as a whole, did not vote as solidly 'No' in 1999 as they had voted Liberal in 1998, as may be seen by comparing Tables 7 and 8. However, taking all the 64 Liberal seats together we find that the Liberal two-party preferred vote in 1998 was 57.2 per cent, while the 'No' vote in those seats in 1999 was 55.3 per cent. I shall return to the case of Bennelong in due course.

Virtually every Labor member of Parliament and office-holder advocated a 'Yes' vote. Consequently it should have been possible for the Labor Party to do better than persuade only 57 per cent of its 1998 supporters to vote affirmatively. That failure is the essential reason why the republic was defeated. It is best illustrated by what happened in the safe Labor seats, as shown in Table 9. The right-hand column in that Table, 'Relative Socio-Economic Advantage Rank', is a concept to which I shall return. The correlation is clear. The greater the socio-economic advantage of the seat the more likely it was to vote 'Yes'.

Note: These aggregates are 65,223 votes higher than those of the AEC. The reason is that these totals include an estimate for the 65,223 formal votes cast in Newcastle. The AEC totals of 5,630,409 for Labor and Lib-Nat 5,413,431 are those for 147 contests only. The missing seat of Newcastle is caused by the fact that there was no Coalition candidate at the supplementary election on 21 November, 1998. A candidate for the 3 October election died before polling day.

The drop from the Labor vote of 67.3 per cent in safe Labor seats in 1998 (Table 7) to the 'Yes' vote of 51.4 per cent (Table 8) was very far from uniform. In the inner metropolitan 'Safe Labor' seats of Melbourne, Sydney, Grayndler, Fraser, Canberra and Kingsford-Smith the average drop was only two per cent. At the other extremity were the 15 'Safe Labor' seats set out in Table 10. Comparison of Tables 9 and 10 shows that the single most extreme case was Bonython, which turned in the highest 'No' percentage of all the 36 'Safe Labor' seats, as well as showing the biggest defection from the Labor vote. Bonython is also the most disadvantaged socio-economically.

Kim Beazley was made to suffer the indignity of losing the referendum as well as having a disastrous defeat in his own seat of Brand (see Table 9). John Howard had to put up with sneers about 'the Republic of Bennelong' as well as watch 17 of the 64 Liberal seats turn in 'Yes' majorities (see Tables 4 and 5). There was, however, one leader who had every reason to smile. John Anderson succeeded in getting every one of his party's seats to vote 'No'. The extent of his success is shown in Table 11B below. In only one National Party seat did the 'No' vote fall below 60 per cent. That was in Richmond (NSW) where the member, Larry Anthony, was a self-proclaimed republican and advocate for a 'Yes' vote.

Before I go on to the National Party seats, permit me an aside. Several monarchist friends of mine watched the referendum-night coverages on television. They claim to have been struck by the republican bias of the commentators. I cannot express any view because I did not see any television that night. One point should, however, be made in defence of the commentators. If it is true that repeated references were made to 'the Republic of Bennelong' and no references were made to 'the Kingdom of Brand', that could be easily explained by the fact that the polls in Brand closed three hours later than in Bennelong.

I said above that 'I estimate that 80 per cent of the 1998 National Party vote was cast against the republic in 1999'. A major problem with estimating the National Party separately from the Liberals is that opinion polls typically lump the two together under the heading 'Coalition'. Where polls do distinguish, there is a strong tendency to over-estimate the Liberals and under-estimate the Nationals because voters do not really differentiate the two. That leads typically to very small and, therefore, unreliable samples of National Party voters. In the light of Tables 11A and 11B and such opinion polling as has been done, the 80 per cent figure is highly plausible, if unprovable.