Author: Africa No Filter (ANF) is a donor collaborative that supports the development of nuanced and contemporary stories that shift stereotypical and harmful narratives within and about Africa.

Type of publication/ Type de publication: Report

Date of publication/Date de publication: October 2024

Site of the organisation/ Site de l’organisation: Africa no filter

Link to the original document

 

Extracts from pages/ Les extraits proviennent des pages : (9, 10, 11, 12, 13, 14, 15-16, 17, 18, 19, 20, 21)

How the Media portrays elections in Africa: A comparative quantitative analysis

Having developed the databases of media articles, an analysis was undertaken to determine differences in reporting across the election cycles for each country, and potential instances of bias. Biased media reporting occurs when news outlets present information in a way that systematically favours a particular viewpoint or perspective, often to the detriment of objectivity.

Common examples of biased reporting about Africa include:

• Overemphasis on negative stories: Focusing primarily on conflict, poverty, and disease, while neglecting to highlight positive developments and achievements.
• Stereotyping: Portraying Africa as a monolithic entity, ignoring the diversity of cultures, economies, and political systems across the continent.
• Omission of positive stories: Failing to report on the continent’s progress in areas such as education, healthcare, and technology.
• Ethnocentrism: Viewing African cultures and societies through a Western lens, often leading to misunderstandings and misrepresentations.
• The results are presented by highlighting the key themes that emerged. Comparison is done across each country, as well as between countries with similar political risk profiles.

The media provides extensive coverage of African elections

By analysing the frequency of reporting across countries, it is possible to gauge how important the events are to the media. From the sample, African countries received more attention from the media during elections compared with countries outside of Africa, even those with similar political risk scores. When looking specifically at Kenya versus Malaysia, Malaysia has fewer articles than Kenya, even though there were election fraud issues in both countries, and Malaysia experienced a corruption scandal related to the then Prime Minister Najib Razak during the same period. Furthermore, Egypt saw greater reporting compared to Thailand, a country with a similar political regime and a low level of media freedom. The difference in reporting frequency/number of articles could be attributed to an existing over-emphasis on covering Africa in media during potentially negative events, especially when focusing on Kenya and Malaysia, both important economies in their subregions with similar electoral index scores.

Negative issues are mentioned frequently where they occur but are disproportionately emphasised in Africa

Negative events are common during election cycles, however, the extent of coverage devoted to these is amplified for African nations. It illustrates the difference in the prevalence of negative terms in articles across the different countries, specifically looking at the percentage of articles that contained words relating to “rigging” (elections) and “corruption”. When looking at “rigging”, the word features more frequently in African than non-African countries. For example, both Kenya and Malaysia had “gerrymandering” (a form of election rigging) and election fraud concerns shaping their election period. However, there is more prevalence in terms related to “election fraud” in Kenyan articles. Additionally, when looking at the electoral democracy index which gives a score between zero and one, with a score of one being the most democratic, both countries had similar scores (0.48 versus 0.44 respectively).

Only 2% of Malaysian articles contain “rigging” (expressed through a variety of comparable terms to account for language idiosyncrasies) compared to 16% of Kenyan articles. Corruption as an issue is mentioned frequently in countries where it was a predominant issue in the election, regardless of the political risk rating of the country. For example, both Malaysia and South Africa had high levels of corruption-related issues during their elections and were therefore similarly reported on despite South Africa being lower risk, reflecting that the political reality and issues at the time strongly influenced the content in media. The difference in reporting about corruption in Denmark and South Africa is expected given corruption issues being a major theme about the government led by the South African party, African National Congress (A.N.C).

Violence and election fraud feature more prominently in coverage of African countries than in non-African countries with comparable political risk profiles

During the desk research for this study, it was noted that one indicator of media bias is the use of emotive language and negative event-focused headlines. Qualitative research revealed that headlines often conveyed negative sentiments, whereas the main text of articles did not always align with those perceptions. The findings indicate that articles about Africa exhibit signs of bias, characterised by misleading headlines.

Notably, violent incidents are rarely discussed in reporting on non-African countries, whereas they are frequently featured in Africa. When comparing countries with similar political risk scores the disparity is even more pronounced. For example, there are significantly fewer headlines about violence in Malaysia and Thailand (0.1% and 0% respectively) than in Kenya (5.8%), even though Kenya scored better than both countries in the electoral democracy index and there were comparable incidents of violence in the Thai and Kenyan elections. Moreover, Egypt, whose political dispensation and press freedom scores are similar to Thailand, had more headlines related to violence. The significant difference may be partially explained by more events of violence occuring during the period, but the significant difference between Malaysia (0.1%) compared with Kenya (5.8%) indicates a stereotype.

For example, both Malaysia and South Africa had high levels of corruption-related issues during their elections and were therefore similarly reported on despite South Africa being lower risk, reflecting that the political reality and issues at the time strongly influenced the content in media

Headlines about violence in South Africa are likely to be centred around the high violence and crime in the country, a major election theme and campaign issue for all political parties, and the difference seen between Denmark and South Africa is not necessarily an indication of negative bias. Issues about corruption are consistently represented in headlines in African media articles, and not mentioned at all in countries outside of Africa unless the specific election was embroiled in corruption scandals (e.g Malaysia). Broadly speaking, where corruption was a more prominent theme during an election, media headlines more frequently mention the term (e.g. South Africa and Malaysia).

Media sentiment is overly negative about African countries during elections, especially when compared with non- African countries of similar political risk

A sentiment analysis model13 was performed on the collected articles for each country to understand how many articles from each country expressed a negative, neutral, or positive sentiment. Each article was ranked from +1.00 to -1.00 with scores between +0.25 to +1 being positive sentiment, between -0.25 to +0.25 being neutral sentiment and between -1.00 to -0.25 being negative sentiment. For example, when looking at countries of medium political risk, both African nations (Kenya and Nigeria) have a higher prevalence of negative sentiment in media (55% and 50% respectively) when compared with their non-African counterpart Malaysia (40%). Similarly, when looking at higher political-risk countries, there is also a difference in the percentage of articles with negative sentiment – 47% of articles in Egypt compared with 34% in Thailand.

It is noted that Malaysia has particularly high levels of negative sentiment and this is likely due to the corruption scandal involving the then prime minister and the first lady. Denmark and South Africa have the lowest percentage point difference in negative sentiment articles. This is likely because South Africa has enjoyed more favourable coverage compared to other African countries and is often seen as the exception in Africa, especially during the Nelson Mandela, Thabo Mbeki, and 2010 FIFA World Cup periods. Although there was an increase in negative sentiment due to corruption, energy, crime, and other issues, Ramaphosa’s campaign was widely viewed as a positive change and improvement by the media and the business community. These factors explain why the negative sentiment difference between South Africa and Denmark is not as pronounced when compared with the other African countries and their Asian counterparts. However, despite the positive sentiment about Ramaphosa and better reporting about South Africa compared to their African counterparts, there are media narratives about South Africa becoming a failed state like other African countries.

There is also the prevalence of white expat media supposedly spreading Afro-pessimistic narratives about the country, which negatively impacts sentiment. Media outlets chosen for this analysis were Al Jazeera, the BBC, Bloomberg, CNN, Financial Times, Reuters, and The Economist. According to desktop research and insights from the KIIs, these are media outlets widely used by foreign investors to keep up to date with worldwide economic and political events. For example, a significant 88% and 69% of global news articles about Kenya and Nigeria respectively are negative, compared with 48% for Malaysia. When comparing high political risk countries the trend is similar. Global media sentiment regarding Egypt is twice as negative as Thailand’s.

The relationship between the media & financial flows

Having compared the media’s portrayal of various countries during election periods, it is clear that some of the assumptions of negative bias about reporting on Africa still hold. In this section, the relationship between the media coverage analysis we performed and financial flows is explored quantitatively and qualitatively and the findings are used to quantify the potential economic cost of biased reporting of African countries.

The relationship between media and sovereign bond yields

Media sentiment is a key determinant of investor sentiment and perception of risk. According to the KIIs and desktop research, media plays a significant role in shaping general perceptions of countries as investment destinations and, over long periods this has implications for their development trajectory. In particular, media sentiment is a key determinant of investor sentiment and perception of risk, which plays a critical role in influencing decision- making regarding the allocation of capital and the rate at which African countries can borrow. For example, a study by Cathcart et al. (2019) found that media sentiment is significantly correlated (10%) with sovereign credit risk. There are also findings on media and sentiment by Huang, Cook and Xie (2021) in a large-scale quantitative study of media impact on public opinion that shows 54% of the variance in the American public’s opinion on China is explained by media reporting. Borovka and Zhang (2022) investigated the relationship between media sentiment and corporate bonds, similarly showing that there is a statistical relationship between the two – positive media sentiment is associated with lower yields for those bonds.

For example, a significant 88% and 69% of global news articles about Kenya and Nigeria respectively are negative, compared with 48% for Malaysia. When comparing high political risk countries the trend is similar. Global media sentiment regarding Egypt is twice as negative as Thailand’s

Furthermore, as part of this research, we ran a Pearson’s correlation test across each of the countries, assessing the relationship between the bond yield spreads17 during election months and the media sentiment scores for the same month. The average aggregated score across all case study countries is -0.09 (scores can range between -1 and 1). This indicates a slight inverse correlation between the variables, and further substantiates findings from desktop research that media is one of the factors determining sovereign credit risk and bond yields. In other words, when media sentiment becomes more negative (decreases), bond yields increase. Alternatively, a fairly positive sentiment should be reflected in a fairly low-risk profile and lower bond yields. Criticism of Africa’s credit ratings and subsequently large repayments is not a new concern, and African governments are rejecting credit rating agencies as a result. In a Financial Times article by Ryder, 2024. it is argued that “When it comes to analyst discretion on risk factors that cannot be scientifically measured – such as political risk, the quality of institutions and policy effectiveness – the assessments are based on overly pessimistic assumptions, desktop reviews, virtual discussions or limited to publicly available information, omitting critical data that is often only obtained in-country”.

Based on the findings from this research, it is argued that media is one of these factors influencing global perceptions and credit ratings, particularly in the absence of in-country data to support more accurate ratings. With the above in mind, the remainder of this section explores the differences in bond yields and media sentiment levels between the case study countries with similar political risk profiles and estimates a revised bond yield that would result if the African countries had similar global media sentiment scores compared with their non-African counterparts. Egypt’s bond yields are consistently around 15% while Thailand’s is 2.5%, thus resulting in significantly higher repayment costs. At the same time, according to the sentiment model derived, 66% of Egypt’s news articles in global media were negative, compared with 32% of Thailand’s. More positive coverage in the media would result in proportionately lower bond yields. It is therefore estimated that if Egypt was covered similarly to Thailand (a country outside of Africa that is also ranked as “higher political risk” according to this study’s model), bond yields would decrease by 0.91 percentage points, resulting in major savings on interest repayments as shown in section 3.2.

Media is one of the key factors influencing sovereign bond yields – if media sentiment improves, sovereign bond yields will decrease, but African sentiment in the media is consistently more negative than it should be, making loans more expensive. According to the sentiment model derived, 88% of Kenya’s and 69% of Nigeria’s news articles in global media were negative, compared with 50% of Malaysia’s. It is estimated that if Kenya and Nigeria were covered similarly in global media to Malaysia, (a country outside of Africa that is also ranked as “medium political risk” according to this study’s model) bond yields would decrease by 0.68 and 0.29 percentage points respectively, resulting in major savings on interest repayments. In lower political risk countries, the bond yields are lower, but still high in African countries. South Africa’s average quarterly bond yields range from 8.3% to 8.5% while Denmark’s range from 0.5% to negative 0.2%. According to the sentiment model derived, 38% of South Africa’s news articles were negative, compared with 31% of Denmark’s. It is estimated that if South Africa was covered similarly in global media to Denmark, (a country outside of Africa that is also ranked as “low political risk” according to this study’s model), bond yields would decrease by 0.05 percentage points, resulting in major savings on interest repayments.

What does media bias for Africa potentially cost the continent?

Having undertaken a quantitative and qualitative analysis of media and financial flows during elections in various major African economies and comparative non- African economies, the study finds that media coverage in Africa is overly negative and still frequently paints Africa in a stereotypical fashion, for example by emphasising violence. Since media sentiment is one of the factors that influence the interest rate at which African countries can take out loans. it is concluded that if media sentiment levels were more realistic about Africa, the interest rate for borrowing money would also decrease. This would result in African countries paying much less money over time to repay their loans. They are therefore paying extra money towards debt which could instead be used to pay for public infrastructure and other critical expenditures.

To calculate the amount by which the interest rates would decrease for each African country in the study, it is assumed that:

1. Each African country should have the same percentage of negative articles in international media as the non-African country in the study in the same political risk category.
2. Media sentiment influences the interest rate at which countries can borrow 10% (per findings in desktop research and the Pearson’s correlation test), meaning that a 100% reduction in the number of negative news articles results in global media results in a 10% decrease in the interest rate at which African countries can borrow.

• Egypt is compared with Thailand,
• Kenya and Nigeria are compared with Malaysia, and
• South Africa is compared to Denmark.