Papua New Guinea’s latest attempt at redistribution

Papua New Guinea’s parliament is elected from single-member electorates which are meant to be regularly reviewed and redrawn, as they are in Australia. Yet the current boundaries have not been significantly changed since the 1977 election. The Boundaries Commission is having another go now, but the proposed changes have limited relationship to the actual population imbalances.

I haven’t previously analysed Papua New Guinea politics, and I won’t claim to be an expert in general, but this blog post is narrowly focused on the population statistics and analysing how the current proposal matches up against the data we do have. I have been wanting to make a digital boundaries file for a while now so I could get a better sense of the political geography of PNG and make more maps as the election (due in mid-2022) gets closer. I’ll explain at the end how I made the map file so you can also use it if you wish.

Papua New Guinea’s parliament consists of two different types of MPs. 89 members represent “open” electorates, which cover the whole country. Theoretically they should cover similar numbers of people. Then there are 22 members who represent the 22 provinces (including the Autonomous Region of Bougainville and the National Capital District). So everyone gets two votes: one for your local member and one for your provincial member. It’s a bit like having a Senate, but the senators sit in the same chamber.

The provincial seats are designed to be malapportioned, so I won’t be looking at them today. Instead I’ll be focusing on the level of malapportionment in the open seats, and the attempts to change that.

Papua New Guinea achieved independence in 1975, but the parliament was already in existence at that time. There was a significant redistribution conducted prior to the 1977 election, the first since independence. Since that time, there was a slight change in 2012 when two new provinces were created, but that is it.

The PNG constitution does specify that redistributions should take once a decade, and they should be conducted by a Boundaries Commission. Legislation also specifies that each seat’s population should be within 20% of the average population per seat, and that there should be between 110 and 120 open seats.

There have been numerous attempts to redraw the electoral map, but ultimately they have failed to pass the parliament. This used to be a problem in Australian redistributions, and is discussed in this episode of the Tally Room podcast from September 2020.

Gathering data for this post has been more difficult than it would be in Australia. The PNG Electoral Commission has been “under construction” for a while, so I haven’t been able to get any data directly from them. The census was due to be conducted in 2021, but has been delayed until 2024 due to the current COVID-19 outbreak.

So I have enrolment data from 2017, and estimates of population for each province as of 2021. Bear in mind that the redistribution is calculated on population, not on enrolment, and there are big differences in enrolment rates across the country.

I won’t dive into the reasons why, although there are allegations of “roll inflation“, particularly in the Highlands region. For whatever, reason, enrolment rates are much higher in the Highlands, which can make the malapportionment look worse than it is.

This table breaks down the key statistics for the four regions of Papua New Guinea.

So while the Highlands have more registered voters per (open) seat, that is washed out when you look at population totals, and ultimately that is the standard for redistribution.

Still, we only have enrolment data at the electorate level, so we need to look at that now, while taking it with a grain of salt.

This map shows the number of registered voters per electorate. You can click on each electorate to see it’s name, the sitting mp, the number of registered voters and the number of quotas this makes up out of a total of 89.

Another way to look at the distribution of seat enrolments is with this chart:

The average enrolment as of 2017 was just over 56,800. Yet one seat had over 140,000 registered voters (Lagaip-Porgera in the Highlands province of Enga) and numerous others had over 100,000. Less than 23,000 voters were enrolled in Rabaul.

But we know that these figures are effected by differential enrolment rates, so we can also look at average population per electorate for each province, as you can see on this next map:

This doesn’t look quite as dramatic, but there’s still some big outliers. The average province has a population of 105,710 people per seat. No provinces fall too far below that level – the smallest are Manus and Chimbu which have about 77-78,000 people for each of their seats. But West New Britain has 171,500 people per seat, and the National Capital District has over 157,000 people per seat.

If you were maintaining the current 89 seats, West New Britain would easily qualify for a seat, while Port Moresby would narrowly qualify for two extra seats. The Southern Highlands and Western Highlands would each be entitled to one extra seat each. Four other provinces would be entitled to one less than their current number, with Chimbu dropping from six seats to four.

Now we come to the first proposal we’ve seen from the current Boundaries Commission, as published by Loop PNG.

Their proposal would add 22 extra seats to the parliament for a total of 111 open seats, which fits with the legislative requirement for 110-120 open seats.

The proposal doesn’t suggest changes to every electorate. Instead they’ve nominated 22 seats to be split in half, with the other 67 left as they are. Which is probably not a recipe for ensuring all electorates fit within the quota, but if the extra seats were concentrated in the provinces most in need of extra representation, it could work okay. It avoids aggravating too many MPs, and the addition of a quarter more seats means that no province needs to lose a seat to achieve equality.

But the list of seats proposed for duplication doesn’t appear to have much to do with which seats or provinces are most in need of greater representation.

This chart is the same as the one above, but I’ve highlighted seats that are marked for duplication.

Most of the most populous seats are on the list – six of the top seven seats are proposed for duplication. But not all of them are on the list. It’s worth noting that two of the three Port Moresby electorates are ranked in the top eight, and they are the only ones on that list which aren’t earmarked for redistribution.

Meanwhile, the other seats marked for redistribution are scattered all over the list. The strangest choice is Manus, which is earmarked to be split in two. Manus has just 80,000 people and about 30,000 voters. West Sepik is due to gain two new seats despite already being over-represented.

Almost every province receives one new seat. Just four provinces are nominated to receive two new seats, while four others receive no new seats. It’s particularly strange that Port Moresby receives no new seats, which is the second most under-represented region by population. This map shows which seats have been marked for duplication:

This proposal would widen the gap between the most over-represented and under-represented provinces with 157,000 people per seat in Port Moresby and just 39,000 per seat on Manus.

This approach would almost certainly fail to meet with the requirement that all seats fall within 20% of the quota. I don’t have the data to calculate each individual seat, but at the provincial level it would leave three provinces with an average seat size above the quota, and another four below the quota. These seven provinces contain a population of about 2 million people out of a total of about 9 million.

At this point I thought it would be interesting to look at how many extra seats each province would be entitled to if you added the 22 extra seats proportionally. With an estimated 2021 population of 9.18 million, the quota drops from 103,146 to 82,703.

No province would lose a seat by this methodology. Manus falls just short of a full quota, and Chimbu comes close to losing its sixth seat with 5.586 quotas. Eight provinces retain their existing seats, seven provinces gain a single extra seat, five gain two seats, and two provinces gain three. Those two provinces are the National Capital District, which would go from three seats to six, and the Southern Highlands, which would go from five to eight.

This final chart compares the proposed new allocations from the Boundaries Commission to the increase in quota above the current seats for each province under the proportional method.

While most provinces are proposed to gain one seat, regardless of how many they need to equalise the electorates, West Sepik is proposed to gain two despite already having the right number of seats in an expanded parliament, and the National Capital District is the opposite.

One final thought: if PNG doesn’t have precise population data in the absence of a census, it’s hard to know how they could possibly draw electoral boundaries based on population. It seems more practical to use the Australian approach, where you use population data to allocate a number of seats to each province, and then use more fine-grained voter registration data to draw the lines within the province. But that’s not what the current legislation says.

If you compare enrolment in each electorate only to the seats in the same province, the amount of variation does drop significantly, although there are some provinces where the population is very lopsided between neighbouring seats. The electorate of Lae includes over one fifth of the voters in the Morobe province, which has nine seats. Meanwhile Rabaul makes up less than 14% of the voters in the East New Britain province, which has four seats. Rabaul was a much more significant town prior to the 1994 volcanic eruption, but the electoral boundaries haven’t kept up.

It’s not clear whether there is the political will in Papua New Guinea to finally redraw the boundaries, and whether the current proposal will go anywhere, but there does remain a need to do something to ensure equality in who gets to choose the leaders of the country.

A note on methods and thank you

This analysis required more assistance from others than I normally require.

If you are looking to do your own analysis of Papua New Guinea and want to use the electoral boundaries, most of the boundaries are available here. There is a map of the provinces and the “districts”. The districts file mostly aligns with the 89 open electorates, but for some reason the three Port Moresby electorates are missing. You can download a KMZ file of the Port Moresby electorates from my maps page.

Thanks to Francis Markham for tracking down the Port Moresby electorate boundaries and to David Barry for finding the formula to convert the geographic data into the correct projection.

I’d also like to thank Maholopa Laveil from the University of PNG for helping me understanding the political and constitutional context and tracking down registered voter data.

Liked it? Take a second to support the Tally Room on Patreon!