Detecting gerrymandering in GE14
Redelineation is the process of drawing electoral boundaries to form voting districts/constituencies. Redelineation can serve legitimate purposes. For example, we may want to better represent the underserved, or, “give” a Member of Parliament (MP) fewer constituents if the constituents are spread over a large area.
Since there are legitimate reasons for redistricting, its use as a political tool can be difficult to detect and easy to explain away. In this post, I quantify the extent of the redelineation prior to Malaysia’s GE14 and offer evidence that it was done for political reasons.
What does redelineation look like?
The following figures provide a visual example of redelineation between GE13 and GE14 in the Klang Valley area. Kelana Jaya (GE13) is now called Subang. PJ Selatan now includes what was previously Sungai Buloh, Kelana Jaya and Shah Alam. Such redrawing of the electoral map is present all over Malaysia.
How much redelineation was there?
One measure of how much a constituency has been modified is the fraction of the constituency that is composed of voters who previously voted in other constituencies. I looked at the electoral roll and considered voters who were present in the same constituencies in both the GE13 and GE14 electoral rolls. I then looked within each of the GE14 constituencies to gauge the percentage of voters who were moved between GE13 and GE14.
On average, 8.6 out of 100 Malaysians found themselves voting in a new parliamentary district. Given the legitimate reasons for redelineation mentioned above, this seems like a reasonable number. However...
Small, rural constituencies remained intact. Among the smallest 50 constituencies (representing about 10% of all voters), 4 of 100 Malaysians voted in new parliamentary constituencies.
Urbanites were shuffled. Among the largest 50 constituencies (representing about 40% of all voters), 14 of 100 Malaysians voted in new constituencies. Urban voters were moved around disproportionately more than other voter groups.
The very largest of constituencies are hardly recognizable. Malaysians in the 5 largest constituencies of Damansara, Bangi, Klang, Subang and Petaling Jaya (representing about 6% of all voters) have to deal with a whopping 44 new members in every group of 100. This makes it difficult for voters organize and make demands. An aspiring MP will have a harder time getting to know her constituents.
In some cases, the shuffling is extreme. Kelana Jaya (from GE13) has been entirely wiped out; it was chopped up so much that there is no constituency in GE14 in which a majority of voters were former Kelana Jaya voters in GE13. Kelana Jaya is now officially Subang (GE14) but Subang contains more voters from GE13’s version of Puchong than the former Kelana Jaya.
The following figure shows the percentage of new voters in each constituency. The larger circles represent constituencies with more eligible voters. There is a positive relationship between constituency size and how many voters were moved, suggesting that larger (usually urban) constituencies were subjected to more redelineation.
Was there gerrymandering?
The redelineation prior to GE14 was obviously targeted at particular constituencies. The big question is whether it was carried out for political reasons (i.e. gerrymandering). To answer this, I partitioned constituencies into:
- Safe Barisan Nasional seats, won by BN in GE13 with a margin > 15%
- Safe Pakatan Harapan seats, won by Pakatan in GE13 with a margin > 15%
- Marginal seats, won by either coalition with a margin < 15%
The following figure plots the proportion of voters taken out and new voters brought in for each constituency as a result of the redelineation. The x-axis in each panel shows the percent of voters removed from a constituency, while the y-axis shows the percent of voters added to a constituency.
If redelineation was carried out fairly, there should be similar patterns of voters being moved across all constituencies regardless of which political party is the incumbent. Instead, it is obvious that BN safe seats have very little changes in their voters. In contrast, marginal constituencies and PH safe seats show large changes – many of these constituencies have significant numbers of voters being moved around (both brought in and taken out). The analysis above suggests gerrymandering was carried out to benefit BN.
We can also look for evidence of gerrymandering in a different way. The following figure plots the majority of constituencies (I have removed a few outliers). The x-axis represents the support for BN amongst voters who got to remain in the same constituency (from GE13). The y-axis plots the influence on BN’s support levels from the moving of voters via redelineation.
The figure above shows that in areas where BN support was already strong (the right side of the graph), the moving of voters reduced BN support – in other words, BN supporters were likely moved. In areas where BN support was weak (the left side of the graph), the moving of voters increased BN support. This is a sign of “cracking” (a strategy used in gerrymandering) – taking supporters from places with a surplus and moving them to areas where they could help you win a marginal or opponent’s seat. While the effect of the redelineation may only be a few percentage points, it may be sufficient to swing some contests.
The election is over, why does this matter?
Even after the rakyat have spoken through the ballot box, it is important to understand how gerrymandering limits the accountability of the government to the rakyat. To be clear, gerrymandering is bad for the rakyat regardless of who you support. Those who seek to be elected leaders offer themselves for the rakyat to choose. They should not choose who gets to vote for them.
There are many components of a functioning democratic process. Civic-minded Malaysians showed up in GE14 to ensure high voter turnout and a fair electoral process on election day. The next step is to understand how more subtle manipulations (prior to the actual election day) can affect the outcome of an election. These manipulations can be invisible, and even explained away as legitimate. However, the data reveals otherwise.
Since the process of redelineation can be manipulated for political gain, how do we ensure that those who draw the lines do so fairly? What is fairness?
A radical proposal
Aside from adhering to legal requirements placing constraints on how much redelineation can occur, a necessary ingredient for ensuring the government of the day serves the rakyat is to ensure the electoral process is competitive.
How do we ensure that any redelineation exercise is competitive? One proposal is to let the opposition coalition lead the efforts to redraw the lines. This idea has roots in game theory.
Parliament can provide the opposition coalition with sensible, quantifiable guidelines to prevent gerrymandering. With clear guidelines like a fixed number of seats, limits on the number of people within each constituency, population density, racial composition, and convexity (constituencies cannot be too irregularly shaped), an opposition can draw lines to help them regain power subject to constraints preventing them from doing so unfairly.
The government of the day will undoubtedly be unenthusiastic. What can they do to win after relinquishing power in the redelineation process? They become incentivized to serve the rakyat regardless of geography.
While this proposal is radical, there are more conventional approaches. A committee headed by the opposition coalition can be tasked with appointing the EC which in turn, is tasked with making elections as competitive as possible. Such setups are present all around the world and Malaysians should start thinking about how their participation in this exercise can lead to a more vibrant democracy.