Answering Jakarta: My step to move on
We, consciously, give the conservatives a chance to be a prominent player in our electoral democracy.
Since the defeat of Ahok in 2017, people are aware that religious sentiment is pivotal in determining political preferences in Indonesia. The degree of rationality of middle-class voters has been challenged since then. The Muslim conservatives have claimed that they are the main force behind the victory of Anies-Sandi, and they have been openly demanding pieces of power from political elites as gratitude. Even President Jokowi could not withstand their demand and eventually took one of the main actors behind the blasphemy verdict of Ahok as his vice-president candidate.
To put it frankly: Jakarta gubernatorial election has changed the rule of the game of our democracy by claiming the victory of Anies as the rise of religious conservatism, and almost every political elite buys this notion. What is the impact? We, consciously, give the conservatives a chance to be a prominent player in our electoral democracy.
Having all these matters, the big question still remains: did Ahok lose because of religious sentiment? It is a straightforward question, but apparently is also so hard to explain, scientifically.
I lived with that question in mind for at least one year after the election. As a voter, I was also emotionally attached to the anger of what Ahok’s supporters endured throughout the blasphemy trial. Living as a minority in some extent gave me a reason to sympathy the former governor. Yet, as a political scientist, I was also very curious to know exactly what is happening to our democracy.
I was not satisfied with all the explanations I had from the media and public debates. No single explanation is very sound or concise enough for me. Some of them are somewhat emotional, and some of them are only taking one unique perspective. I brought that curiosity to my post-grad study at the LSE, UK, and I tried to spend a whole year to get my result. And this article is a glimpse of the answer I have been studying for.
I chose the Jakarta gubernatorial election as my dissertation topic with two intuitions: there has to be something related to the runoff rule to make the election the way it was; stating that the voters were merely irrational is oversimplified.
Explaining all of my findings in a short article is difficult. In this piece, I just try to tell several results that I get from my research instead of emphasizing the whole story.
Using a quantitative approach that I learned from my program, I built a quantitative model to examine the election outcome with some observable variables that can be accounted for. I used observational data which I collected from some reliable sources, especially the open government data. Our election data is pretty comprehensive (with sub-districts level of data), and that is good for research. Though, most of them are not in an open format, which makes it challenging for researchers to use it. Fortunately, with the help of one online media institution, I could get the election outcome in a right format and easy to compute in the statistical software that I used.
My models are pretty straightforward. For the first round, I use the simple OLS regression to give me partial associations of the vote share with some explanatory variables such as the proportion of Muslims in one sub-district (to measure the religion factor) and incumbency area (where the Jokowi-Ahok’s main supporters are). I control with other sub-districts’ features which are a flood-prone area, population density and the proportion of high-educated people. What I want from my models is basically to prove similar characteristics of Anies-Sandi and Agus-Sylvi supporters that can be caught from the sub-district level. The results are quite striking.
I found that the proportion of Muslims is statistically significant for all candidates and only the incumbent who was not benefited from it. Controlling for other explanatory variables, the association is negative for Ahok-Djarot (-0.627), and instead, is positive for either Agus-Sylvi or Anies-Sandi. Although, the magnitude for Anies-Sandi (0.466) is larger than Agus-Sylvi (0.161). This result is interesting because we know that the population of Muslims reaches 80 percent approximately. It is logical to think that all of the candidates would get a positive association with this variable. Unfortunately, it demonstrates that Ahok-Djarot receives less vote in a sub-district with more Muslims. Yes, religion aspect does matter. But, still, how significant it was? To answer this, we ought to take a glance at the other partial associations.
It shows that the incumbency of Ahok-Djarot matters a lot, as most of their supporters come from the same group who supported Jokowi-Ahok.
The dummy variable, incumbency area, is the most interesting variable. The assumption of the variable is quite mild. A sub-district is an incumbent area if it is the area where Jokowi-Ahok beat Fauzi-Nachrowi in the previous election. I get the statistically positive association of this incumbent variable for the vote share of Ahok-Djarot (4.355), holding other explanatory variables as constant. The association of this variable in the model with Anies-Sandi and Agus-Sylvi as dependent variable are negative (-2.176 and -2.179). It shows that the incumbency of Ahok-Djarot matters a lot, as most of their supporters come from the same group who supported Jokowi-Ahok.
Meanwhile, the education, flood-prone area, and population density are not giving us similarity for both Agus-Sylvi and Anies-Sandi. Still, by taking those variables, we are still better off since they improve the model (reflected in the R-square I got), and hence we keep them both.
The main point of my dissertation is the second stage of the election. I compute the difference of vote share from two phases of the election for Ahok-Djarot and Anies-Sandi in each sub-district and produce a scatterplot from it by putting Agus-Sylvi vote share as the x axis. Therefore, the visualization from the distribution of sub-districts is what I am interested in. The assumption which can be taken is: if Ahok-Djarot did own solid supporters, the difference would be around 0. Most of the sub-districts are located around zero. But that is not the only thing I get.
From the scatterplots (above), I discovered that there are outliers. Their position on the distribution is very far from 0 (below zero). It reveals that Ahok-Djarot were actually losing their vote in the second round in those particular sub-districts. Surprisingly, by taking the difference of Anies-Sandi as the y axis, I also got those specific sub-districts as outliers.
I assume that these outliers play a vital role in the making of Anies-Sandi as the winner of the Jakarta Gubernatorial Election. To prove this assumption, the existence of statistical evidence is not enough. I require more observable facts to formulate a compelling argument. My first clue came from a paper by Savirani and Aspinall (2017), Adversarial Linkages: The Urban Poor and Electoral Politics in Jakarta.
The reason behind this is that the urban poor felt that they had been let down by the Ahok-Djarot and also Jokowi-Ahok government.
The paper states that in the Jakarta Gubernatorial Election 2017, there was a coordination between urban poor representatives and all the challengers in the election to beat Ahok-Djarot. The reason behind this is that the urban poor felt that they had been let down by the Ahok-Djarot and also Jokowi-Ahok government. The main issue here is the significant number of force evictions which happened during that period of government. The urban poor demanded a political agreement between the candidate and their representatives concerning some particular policy, primarily about land use. They were backed by a reliable team consisting of urban-planner and also some critical NGOs and activists. As an offering for the candidate, they would ensure to mobilize votes for the challengers.
However, they waited until they were sure that one of the candidates was more likely to go through to the second round. And as what we understood from earlier, Anies-Sandi performed better than Agus-Sylvi in public debates. We can examine the performance by solely glancing at the change of their electability from several pre-election polls after the public discussion had taken place. Eventually, Anies-Sandi is the beneficiary of all this. The political agreement did pass between both sides, and it makes the urban poor representatives can actually consult the city plan of Jakarta. Is it sufficient for them? That is not a problem that I attempted to discuss in this piece.
So, how does it relate to my outliers? I compare my data with a research conducted by LBH Jakarta in 2016. They gathered all sub-districts in Jakarta which become the area where the force evictions had occurred in 2016. The data itself is obtained from the news. Therefore, they may be omitting some sub-districts which are not recorded in the media. Nevertheless, it is sufficient to check the sub-districts’ profile.
And what do I find afterwards? Turns out that all the outliers from my scatterplots are actually sub-districts where force evictions were done.
Qualitatively speaking, the force eviction is strongly associated with the strategic coordination between the urban poor and Anies-Sandi. We cannot possibly calculate this premise with regression, but by connecting each puzzle that we could observe, we may get a plausible explanation.
Ahok-Djarot put themselves in a very extreme position with the urban poor. Following the majority principle by Anthony Downs (1985) in An Economic Theory of Democracy, an incumbent should please the majority on every issue to avoid such defeat output in the election. Without the principle, a challenger could assemble a coalition of minorities in the election process to beat the incumbent. In our situation, Anies-Sandi successfully made this coalition which consisted of the urban poor and conservative Islamist groups.
It is fair then to categorize Ahok as a policy-oriented politician. His utility is coming from the policy he executes, not only mere electability.
Hence, policy space should be taken into account. Perhaps, it is the one primary explanation that we have been asking for. We are just too afraid to speak it out loud. Whether we are afraid to lose the power of the majority we had or we are utterly too selfish to say that Jakarta is the city of ‘us,’ both sides are utilizing this controversy of the power of piety in our electoral politics.
So, the defeat of Ahok does not necessarily about religious sentiment alone. What would it be turn out if there was no blasphemy case occur? My intuition tells that Ahok would still found it difficult winning the election, for there was still a condition with three candidates and the runoff rule in it. Regarding the runoff rule, the system has been marked as much more favoring majority candidates rather than minority candidates. Empirically, minority candidates often find it tough to reach above 50 percent of vote share, though they may be the plurality winner. If we compare it with the US, our runoff threshold in Jakarta can be regarded as immense. Thus, from the very beginning, a double-minority candidate like Ahok is therefore hard to win a runoff election.
In addition to the electoral system, his position in policy space was too extreme to begin with. The force eviction, the Jakarta Bay project and many more; he chose not to stand on the median as what a typical politician would do. We may have our own standpoint of the policies, but it is the election game we are talking about. In political economy, we usually call this type of politician as policy-oriented rather than office-seeking. It is fair then to categorize Ahok as a policy-oriented politician. His utility is coming from the policy he executes, not only mere electability. Unfortunately, to win the game, sometimes you must follow the median.
Due to the lack of attention towards this strategic coordination between the urban poor and the candidates, the conservatives have been enjoying their part in our democracy. Still, it is not too late to realize that inequality and other economic variables do matter in the electoral game. If we crave to move on from Jakarta, or even reflect from it as a study case for a bigger electoral stage in the future, we have to understand it first for the least.
This article is just the small step.
Edbert Gani
Political Economist
*For further questions regarding the research, please do not hesitate to email me: edbert.gani@gmail.com
