Are Malaysians happy with Belanjawan 2021? — a Twitter analysis

Soung Low
8 min readFeb 11, 2021
Image by author

Belanjawan 2021: Malaysia’s Official Budget for 2021

source: Budget 2021 Official Website

In November 2020, the government of Malaysia has released its official budget for 2021: Belanjawan 2021. This budget represents the largest public expenditure in Malaysia’s history (RM322.5 billion) with 3 integral goals¹:

  1. Well-being of People (Kesejahteraan Rakyat)
  2. Business Continuity (Kelangsungan Perniagaan)
  3. Economic Resilience (Ketahanan Ekonomi).

As a Malaysian, I was motivated to understand the public opinion towards Belanjawan 2021 given that it would have a massive impact on the economic recovery from the global COVID-19 pandemic and therefore the livelihoods of the people who had been severely affected. This article is an extension of an interactive dashboard I have created.

Data Collection

I collected tweets containing keywords like Belanjawan2021, Budget2021, and Bajet2021 for the period between 6 November and 21 December 2020. This period is determined by the Google search interest for Belanjawan 2021, which could be found here.

Data Analysis

Is Twitter activity related to the discussion of Belanjawan 2021?

To answer this question, let’s examine whether the trend in the frequency of tweets correlated with the process of budget passage in the parliament.

Image by author

Despite the fact that Twitter is not the most popular social media among Malaysians², we can infer from the plot above that Twitter activity was linked to the passage of Belanjawan 2021. This is because the three peaks (marked by light blue markers) in the plot on the left corresponded with three important dates for Belanjawan 2021.

1st reading on 6 November: budget released by the Ministry of Finance
2nd reading on 26 November: budget passed at the policy stage
3rd reading on 15 December: budget passed at the committee stage

Nonetheless, it is clear that the discussion lost its momentum over time as the numbers of tweets for the last two peaks are significantly lower than that for the first peak.

What did the discussion involve?

Let’s explore the content of the tweets. In this step, tweets are tokenized, in which I convert tweets into lowercase, remove stopwords and punctuations, and break them down into unigrams, before visualizing the frequency distribution of words.

Image by author

The most common issues arising from the discussion of Belanjawan 2021 include covid19, followed by tax and jasa.

For non-Malaysian readers, the word jasa is an acronym of the Special Affairs Departments under the Ministry of Communications and Multimedia, Malaysia. In this budget, the government planned to allocate RM85.5 million (US$20.7 million) to resurrect this department, which was actually dissolved by the former government in 2018.

This allocation has been severely criticized by the opposition, who claimed that the department is a propaganda tool for the government to promote pro-government political messages.

What about hashtags?

Image by author

Most of the top hashtags appeared in the previous plot, with two noteworthy exceptions:

  1. #kitajagakita : 'Kita jaga kita' means 'We help ourselves' in Malay. It is a one-stop user-friendly mutual aid platform that matches individuals with trustworthy NGOs and verified charities in Malaysia³.
  2. #muhyiddinout : This is a hashtag that is used to call for the resignation of the Prime Minister of Malaysia, Muhyiddin Yassin. This might be a hint of the negative feelings among Malaysians towards Belanjawan 2021.

So, are Malaysians happy with Belanjawan 2021?

Before we start analyzing the sentiment of tweets, it should be noted that news media companies in Malaysia are using Twitter as well. To better capture the public opinion, we ought to separate tweets by news media from the tweets dataset. To do so, I collected a list of news media companies in Malaysia⁴ and their Twitter information using web scraping.

Twitter information of each news media in Malaysia

With this data, I successfully removed tweets by 61 news media companies from the main dataset.

Then I applied VADER⁵, a dictionary method for sentiment analysis which has proved to perform exceptionally well in the social media domain, to the personal tweets. Let’s see the result:

Image by author

The VADER compound score ranges from -1 (most negative) to 1 (most positive). We can see from the line chart that the compound score (purple line) of personal tweets concerning Belanjawan 2021 is positive for most of the days within the period. There were only 7 days where the compound score fell below zero (i.e. when negative score is greater than positive score).

Image by author

On the other hand, the sentiment analysis of news media tweets revealed a more positive result. The compound score not only illustrated less fluctuations, but also fell below zero for fewer days, which were 3 days.

Overall, it can be said that the sentiment of Malaysians towards Belanjawan 2021 based on the tweets analysis is generally positive!

What were the public concern?

While the general sentiment concerning Belanjawan 2021 was positive, it is interesting to know why some Malaysians were not happy with it. Let’s begin with a word cloud visualization.

The larger a word, the higher the frequency it appeared in tweets; Image by author.

From both word clouds above, we can easily identify belanjawan2021 and its equivalents, such as budget2021 and bajet2021. If you look closely at the negative word cloud on the right, words associated with widely discussed issues like covid19, tax, jasa, and pandemic are noticeable.

Interestingly, we can also see political figures not only from the incumbent government but also from the opposition. For instance, both anwar andanwaribrahim refer to Anwar Ibrahim, who is the current opposition leader. This can possibly be attributed to the public’s dissatisfaction with the fact that he instructed MPs from the opposition to pass the budget on 26 November without a bloc-voting⁶.

To delve deeper into the reasons why some opposed to Belanjawan 2021, we leveraged a topic modelling tool: the latent Dirichlet allocation (LDA)⁷, which allows us to extract topics from tweets. I obtained the following 4 topics along with their respective top 10 salient terms:

Image by author

Here is what we can observe from the above chart:

  • While Topic 1 is lack of distinctive terms, Topic 2 is clearly related to the fact that the opposition (e.g. anwar, mps) did not agree to the re-establishment of jasa (the Special Affairs Departments under the Ministry of Communications and Multimedia, Malaysia).
  • Topic 3 is also likely related to the opposition, according to the presence of terms like anwaribrahim and opposition.
  • Topic 4 contains terms such ascovid19 and health, indicating the concern of people over the pandemic.

Conclusion

To sum up, this tweet analysis reveals that the public sentiment with regards to Belanjawan 2021 is generally positive. Having said that, there are a couple of issues that may have concerned the public:

  1. Economic recovery from the pandemic
    The prevalent presence of both hashtags (e.g.#covid19 and #kitajagakita ) and words (e.g.pandemic and health) suggests that Malaysians are most concerned with how the government deals with the ongoing pandemic.
  2. The re-establishment of JASA
    I would like to mention that this controversial agency has been rebranded as the Community Communications Department (J-Kom) on 25 November 2020⁸. The Communications and Multimedia Minister Datuk Saifuddin Abdullah has also promised not to use it as a propaganda machine.

Last but not least, I would like to mention two major caveats that might undermine the findings of this analysis:

  1. The multilingual nature of tweets
    Most of the original tweets were actually in Malay. As this analysis is solely based on English tweets, it may not have captured the whole picture of discussions on Twitter.
  2. Twitter being not the most popular social media in Malaysia
    Consequently, I would like to re-emphasize that tweets may not be sufficiently representative of Malaysians’ opinion.

On a side note, this analysis is excerpted from a larger project I have conducted for my master’s course. If you’re interested to know the detail, feel free to get in touch with me via LinkedIn or Twitter. The code that produced the plots above is available here.

References

[1]: Ministry of Finance, Malaysia (2020). Budget 2021 Speech. Available at: https://belanjawan2021.treasury.gov.my/pdf/speech/2021/bs21.pdf (Accessed: 11 February 2021).

[2]: Malaysian Communications and Multimedia Commission (2020). Internet Users Survey. Available at: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/32225/11-515-bigger-better-business-helping-small-firms.pdf (Accessed: 8 February 2021).

[3]: Veit, Cooper (2020). ‘Grad student-organized COVID-19 response lab mobilizes teams to tackle coronavirus response’, The Stanford Daily, 12 April 2020. Available at: https://www.stanforddaily.com/2020/04/12/grad-student-organized-covid-19-response-lab-mobilizes-teams-to-tackle-coronavirus-response/ (Accessed: 8 February 2021).

[4]: ‘马来西亚报刊列表’ (2020) Wikipedia. Available at: https://zh.wikipedia.org/wiki/%E9%A6%AC%E4%BE%86%E8%A5%BF%E4%BA%9E%E5%A0%B1%E5%88%8A%E5%88%97%E8%A1%A8 (Accessed: 8 February 2021)

[5]: Hutto, C.J. & Gilbert, E.E. (2014). VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text. Eighth International Conference on Weblogs and Social Media (ICWSM-14). Ann Arbor, MI, June 2014.

[6]: Razak, Radzi (2020). ‘Anwar confirms he told Opposition to let Budget 2021 pass first hurdle’, Malay Mail, 26 November 2020. Available at: https://www.malaymail.com/news/malaysia/2020/11/26/anwar-confirms-he-told-opposition-to-let-budget-2021-pass-first-hurdle/1926438 (Accessed: 11 February 2021)

[7]: Blei, D. M., Ng, A. and Jordan, M. (2003). ‘Latent Dirichlet Allocation’. Journal of Machine Learning Research, 3, pp. 993–1022.

[8]: Bernama (2020). ‘J-KOM: Breathing new life into Govt’s information delivery system’, The Edge Markets, 26 November 2020. Available at https://www.theedgemarkets.com/article/jkom-breathing-new-life-govts-information-delivery-system (Accessed: 11 February 2021).

--

--

Soung Low

MSc Applied Social Data Science @ LSE | BA Economics @ FCU | Text Analysis | Causal Inference | Machine Learning