Electoral System Options

This chapter evaluates electoral system options and presents recommendations for Parliament’s consideration. Three options are evaluated against the research evidence to explain the expected effects of different electoral system options. Two options are recommended for Parliament’s consideration based on the assessment of electoral systems and input emerging from consultations with communities and sectoral stakeholders.

The chapter starts with an explanation of the seat allocation process used in compensatory systems, which applies to all options discussed. This is followed by an evaluation of common systems proposed for elections to the National Assembly. This includes three options: retaining the current system, introducing smaller multi-member constituencies, or adopting single-member constituencies while retaining overall proportionality. The three options are subsequently compared to explain where the options differ in their attainment of the guiding principles to inform our final recommendations for reform. The chapter ends with a discussion of options for provincial elections and their implications in relation to reforms considered at the national level.

Seat Allocation in Compensatory Systems

Before explaining the different electoral system options considered by the ERCP, it is helpful to return to the seat allocation process in compensatory systems, as this is a common feature of all three options. As explained in Chapter 4, the compensatory nature of the options identified means that all three options will result in the same overall political outcomes with the same parties represented to the same degree. In effect, if any of these three systems had been used in the 2024 election, the final seat allocation in the National Assembly would be the same. The seat allocation process for the National Assembly follows two main steps set out below and would apply to all three options.

Step 1. Constituency seats

• Seats are assigned to constituencies in proportion to the number of registered voters in an area.

• A separate regional ballot paper is used for voters to choose either a political party or an independent candidate.

• In Options 1 and 2, seats are allocated to political parties and independent candidates in proportion to their share of the vote. This is determined using a Droop quota and the largest remainders method. In Option 3, where single-member constituencies are used, the person or party that receives the most votes wins that seat.

• If an independent candidate wins a seat, they are elected. Votes for independent candidates are not considered in the next stage, because only parties qualify for compensatory seats.

Step 2. Compensatory seats

• There is a separate national ballot contested by parties. If someone voted for an independent candidate in the first ballot, they can, in this instance, elect a party for the overall national results.

• The compensatory allocation considers total votes across the whole system to assign seats proportional to the overall results. In effect, the outcome is identical to a straight nationwide allocation.

• A new Droop quota is determined across all 400 seats based on the votes received on both the regional and national ballots by qualifying parties.¹³⁶ The quota and largest remainder method are used to determine how many seats each party should be given in the 400-member Assembly,¹³⁷ based on the total votes for qualifying parties.

• Compensatory seats are then allocated to make up the difference between the total number of seats a party is due and the number of constituency seats it has already won. For example, if a party is entitled to 40 of the 400 seats in the National Assembly and it wins 18 constituency seats, it will then be compensated with 22 seats to make up the difference and ensure overall proportionality.

National Assembly Elections

Electoral system options for the National Assembly are assessed below in terms of the guiding principles adopted by the ERCP. This is based on the evidence for the effects of different aspects of electoral system design as set out in Chapter 4. Our final recommendations are guided by the research evidence and concerns raised during public consultations.

Option 1 – The Current System

Option 1 maintains the current electoral system structure based on a two-tier system with 200 constituency seats and 200 compensatory seats. Nine multi-member constituencies follow provincial boundaries and the 200 constituency seats are allocated in proportion to the number of registered voters in each constituency. The size of constituencies ranges from five to 47 seats. The sole reform proposal considered by the ERCP under Option 1 is to change the process dealing with remaining seats following the first allocation based on the quota. Under the proposed reform, only the largest remainder method would be used to assign the remainder of the seats rather than combining it with the highest averages method as is currently done.

Evaluation Against Guiding Principles

Inclusivity: The use of compensatory seats and high levels of proportionality allow for inclusion and diverse representation in terms of gender, political parties and demographic groups across geographic locations. The use of party lists will encourage political parties to put forward diverse lists, including female representatives.

Fairness: All options are equally proportional, and all votes hold equal value due to the compensatory nature of the electoral system.

Accountability: Electoral accountability is primarily exercised on a collective basis where voters can make their choice to reject or reaffirm the electoral mandate of parties every five years. Closed lists mean that voters cannot identify individual candidates on the ballot, nor can they vote for individual candidates across party affiliation. This reduces individual accountability of representatives to voters.

In terms of vertical accountability, the use of such large constituencies weakens the accountability relationship between voters and their representatives. Large constituencies make it more difficult for voters to determine individual responsibility for their area and distinguish between representatives. This also makes it more difficult for voters to assess and monitor the individual performance of representatives and reward or reject representatives at the next election. Large party lists make it difficult for voters to show their support for particular candidates and to reject representatives for poor performance. Weak or poor performing candidates can be protected through a high party list position making it difficult for voters to reject them. This makes individual representatives primarily accountable to party leaders rather than directly to the electorate as parties nominate candidates and voters have no leverage in determining the re-election prospects of individual representatives. As a result, there are limited incentives for public representatives to be responsive to voters.

In terms of horizontal accountability, large constituencies and the closed list system tend to skew the balance of power in favour of party leaders, who decide the compilation of party lists. This disincentivises MPs from holding the executive accountable. As a result, public representatives are incentivised to place the interest of party leaders ahead of their constituents.

Simplicity: The system is straightforward and easy for voters to use and find their party or independent candidate on the ballot. The number of spoilt ballots is relatively low compared to international standards.

Electoral Manageability: This system has been used previously without any issues. The introduction of measures to minimise non-competitive participation by political parties and independent candidates with limited prospects of electoral success should be considered.

Transparency: There is no need to delimit new constituencies as this option uses the existing provincial boundaries. Changing population trends over time will only require changes to the number of seats in regional constituencies and this is based on a transparent process.

Stable and Effective Government: As all three options will result in the same overall electoral outcomes, there will be no expected difference among the various options in terms of their effect on party system fragmentation. The adoption of an electoral threshold for compensatory seats could be used to reduce fragmentation in the party system.

Advantages:

No delimitation of constituency boundaries is required. This will prevent disputes over electoral boundaries.

Political parties and voters are familiar with the system.

Advances equitable gender and minority representation.

Disadvantages:

Option 1 presents the weakest accountability relationship between voters and their representatives. The use of large constituencies creates a substantial distance between voters and representatives, undermining the association between the two. Public consultations indicate that this leaves voters feeling unrepresented as they do not know their representatives, and representatives do not account to them.

The link between voters and elected representatives is through arbitrary constituency offices established after elections. This weakens the direct link between constituency services and the ballot. The ballot is the primary mechanism through which voters reward or reject candidates based on their performance and responsiveness to community needs. If the connection between the constituency office and the electoral district is broken, voters will find it harder to hold representatives accountable for their performance, thereby diminishing the effectiveness of democratic representation. This blurs lines of accountability and weakens the value of the ballot as a mechanism for holding MPs responsible to their communities.

The use of large, closed party lists means that voters do not have any influence in the election of individual representatives. As a result, voters lack an immediate mechanism to penalise individual representatives for poor performance or misconduct. This disincentivises the responsiveness of public representatives to the public as voters lack direct influence on their re-election.

The vast size of these provincial constituencies makes it difficult for independent candidates to campaign effectively. This also does not allow independent candidates to compete on similar terms against the merits of individual candidates from political parties.

Option 2 – Smaller Multi-Member Constituencies

This option is a contemporary variant of the proposal put forward by the Electoral Task Team led by the late Frederik Van Zyl Slabbert. It is a two-tier system with compensatory seats and smaller multi-member constituencies. In this option, multi-member constituencies are delineated at the local level following existing municipal boundaries to combine local municipalities into multi-member constituencies. The largest metros¹³⁸ are subdivided into smaller multi-member constituencies based on combinations of ward boundaries to ensure that all constituencies are similar in size.¹³⁹ Constituencies do not cross provincial boundaries. The constituency delimitation considered by the ERCP results in 41 multi-member constituencies using combinations of local municipal boundaries, but it is possible to consider other variations with more constituencies.

Local government boundaries have changed in the two decades since the Electoral Task Team’s 2002 report. This means that the same 69 multi-member constituencies considered in their model are no longer possible. The overall approach is, however, still valid. The model considered by the ERCP combined local municipalities rather than district council boundaries as used by the Electoral Task Team. The use of local municipality boundaries provides more flexibility in forming constituencies of a similar population size; but the use of district council boundaries or their combination would accomplish the same overall purpose and alternative delimitations of multi-member constituencies would be equally valid.

This option has two variants, 2A and 2B. They are similar in most respects. The only difference is the ratio of seat distribution between constituency and compensatory seats. Variant 2A has a 200/200 seat distribution for constituency and compensatory seats, while 2B offers 300 constituency seats and 100 compensatory seats. The two variants will lead to the same final results. In 2A, the number of seats varies between three and six per constituency (five on average). In 2B, this ranges from four to ten seats (seven on average).

Variant 2B, using 300 constituency seats, is considered the stronger version. Option 2A presents no additional advantages that are not accomplished by Option 2B, which achieves the same objectives while bringing a greater number of representatives closer to voters without any trade-offs in proportionality or other principles. The primary justification put forward by panellists supporting the Alternative Report for a 200/200 split between constituency and compensatory seats rather than 300/100 is to reduce the risk of overhang.

Monte Carlo simulations presented by panellists supporting the Alternative Report to test the robustness of electoral system options with simulated results, however, show that a negligible risk of overhang only arises in Option 2B in the most extreme cases of fragmentation. The simulations presented show that the potential for overhang only reaches more than 1% in 1,000,000 simulations in cases where the Effective Number of Parties (ENP)¹⁴⁰ is greater than 14. This level of fragmentation is significantly higher than seen in real-world circumstances, where the highest ENP is 9.9 (Brazil) compared to an average of 3.9 among all PR and mixed systems, and 4.2 in South Africa’s last election (its highest yet).

In effect, the simulations presented show that a perceptible potential for overhang will only arise in cases where fragmentation has reached levels vastly higher than currently seen anywhere in the world. In the South African case, it will mean worse than three times the level of fragmentation currently seen in election results. In short, the potential for overhang should be regarded as a vague possibility that could arise if the South African party system were to fragment to an extent significantly worse than anything seen in the real-world.

Regardless, the potential for overhang can easily be addressed with appropriate legal provisions and as the Constitutional Court has indicated previously, is not an “insurmountable challenge.”¹⁴¹ There is precedent for this in the local government electoral system. The Municipal Structures Act,¹⁴² states that in circumstances where a political party wins more constituency seats (excess seats) than it is entitled to (based on the overall seat allocation calculation), it retains its constituency seats. In such a scenario, the number of party list seats is accordingly reduced for allocation among the remaining parties. The effect on proportionality would be no different to a single independent candidate winning a seat. In this context, panellists supporting the Original Report consider the potential for overhang negligible and the effects manageable based on similar legal provisions in the local government electoral system.

Evaluation Against Guiding Principles

Inclusivity: The use of compensatory seats and high levels of proportionality allows for inclusion and diverse representation in terms of gender, political parties and demographic groups across geographic locations. The use of party lists will encourage political parties to put forward diverse lists, including female representatives. International evidence shows no correlation between the size of multi-member constituencies and rates of gender representation.

In fact, the smallest regional constituency in South Africa currently has the highest female representation. Consequently, Options 1 and 2 should result in similar rates of gender representation. The smaller size of the compensatory tier in Option 2B likewise should not affect gender representation. Under the current system, more women were elected in regional constituencies than in compensatory seats in the 2024 election.

Fairness: All options are equally proportional and all votes hold equal value due to the compensatory nature of the electoral system. The increase in the number of constituency seats allows independent candidates to compete on a more equitable basis, as there are more seats to be contested. More constituency seats will also increase the chances of smaller parties winning such seats.

Accountability: The system creates a closer link between voters and their representatives than is the case in Option 1. Of the options, Option 2B will bring the most representatives closer to voters through the use of 300 constituency seats. The use of smaller constituencies will make it easier to determine the responsibility of individual representatives to voters in a specific area, better incentivising responsiveness to communities. Smaller constituencies are better at “putting a face” to individual representatives and clearer lines of accountability will make it easier for voters to monitor and assess the performance of representatives elected to their area.

Smaller party lists and greater line of sight of candidates make individual candidates more electorally answerable for poor performance (compared to large party lists where poor candidates can be protected by a high list position). This gives voters greater leverage in re-electing individual representatives, and more power in rejecting or rewarding candidates based on the performance. This incentivises responsiveness and accountability to voters.

The use of multi-member constituencies rather than single-member constituencies (as used in Option 3) will mean that multiple representatives, seven on average, are accountable to a specific area. In most cases, this would likely result in the largest parties winning more than one seat in a constituency and would also allow for smaller parties to win constituency seats. This would encourage competition between party members as well as between candidates from different parties at the constituency level. This competition will incentivise more responsiveness to serving communities.

Simplicity: There is little difference from the current electoral system and voters will therefore understand it easily. This makes it unlikely to increase the number spoilt ballots. The change may require some voter education, but voters are familiar with the idea of local representation in a specific area.

Electoral Manageability: The system is largely similar to the current one and has been implemented successfully in the past. The introduction of individual ballots for 41 smaller constituencies will initially require some changes to the electoral administration process. The introduction of measures to minimise non-competitive participation by political parties and independent candidates with limited prospects of electoral success should be considered.

Transparency: This option makes use of existing municipal boundaries as constituencies. The use of fixed constituency boundaries limits the need for new delimitation every election. The only changes between election would be to the number of seats if population changes are substantial enough and this determination is based on an objective and transparent process. Only if municipal boundaries or population trends change very substantially would changes to constituency boundaries be necessary.

Moreover, the compensatory nature of the system means that gerrymandering results to favour parties or candidates is not possible as all votes count towards the final nationwide allocation. The option is clear in its intentions, and the overall concept of linking representatives to smaller geographic areas is well understood by the electorate given their experience of the local government electoral system.

Stable and Effective Government: As all three Options will result in the same overall electoral outcomes, there will be no expected difference among the various options in terms of their effect on party system fragmentation. The adoption of an electoral threshold for compensatory seats could be used to reduce fragmentation in the party system.

Advantages:

There is a closer association between voters and public representatives, making it easier for voters to identify their public representatives. This encourages constituency service and accountability to voters by providing voters with identifiable representatives, whom voters can easily monitor.

Option 2B would be able to bring the most representatives closer to voters without a trade-off in the attainment of other objectives.

Replaces party-centrism with a potentially responsive system, where public representatives prioritise their constituents.

Advances equitable gender and minority representation to the same extent as Option 1.

Voters and parties are familiar with the basic system, as it is somewhat similar to the current national system.

The use of existing municipal boundaries would mean that limited delimitation is required for constituencies compared to Option 3.

The use of smaller multi-member constituencies would mean that multiple parties would be represented in constituencies, compared to single-member constituencies, where only a representative from one party would represent an area.

The use of smaller constituencies would allow independent candidates to contest on more equitable terms against other candidates put forward by political parties.

Disadvantages:

Multiple representatives in a constituency can make lines of accountability more blurred compared to single-member constituencies, where there is one representative solely accountable for an area. Conversely, this can also encourage healthy competition between representatives in a constituency.

Some delimitation would be required to initially establish the system, following which fixed boundaries could be used in subsequent elections.

The use of closed party lists would make it more difficult to penalise a specific representative for poor performance or misconduct than would be the case in single-member constituencies. Smaller party lists would, however, make representatives more electorally answerable for non-performance than the large party lists used in the current system. In the longer term, the introduction of open list voting would be more viable in Option 2 due to the use of shorter party lists.

There is a practically negligible potential for overhang in the most extreme cases of fragmentation. This can, however, be dealt with through appropriate legal provisions as is currently done in the local government electoral system.

Option 3 – Single-Member Constituencies and Overall Proportionality

Option 3 would introduce a similar system to the one currently used at the local government level, where some representatives are elected in single-member constituencies while others are elected from party lists. The option is similar to the proposal adopted by the majority of the Ministerial Advisory Committee led by Valli Moosa. This option is a type of mixed-member proportional representation system. The option includes 200 single-member constituencies and 200 compensatory seats to restore proportionality to results.

Constituencies are based on combinations of municipal wards, maintaining a voter deviation of no more than 15%. Again, constituencies do not cross provincial boundaries, and the total number of seats within a province would be the same as in Options 1 and 2. In single-member constituencies, the party or independent candidate winning the most votes wins that seat. The large compensatory tier will, however, restore proportionality, and the overall results would be the same as in Options 1 and 2.

Evaluation Against Guiding Principles

Inclusivity: The use of compensatory seats and high levels of proportionality also allow for inclusion and diverse representation in terms of gender, political parties and demographic groups across geographic locations. The use of single-member constituencies means that constituencies are likely to be male-dominated as parties generally prefer male candidates for single-member constituencies, which might limit women’s representation. However, the use of a compensatory list would help restore gender representation. Internationally, mixed-member proportional systems are seen to lead to similar overall levels of gender representation to PR systems.

Fairness: All options are equally proportional and all votes hold equal value due to the compensatory nature of the electoral system. The use of single-member constituencies will allow independent candidates to compete on more equal terms to individual candidates put forward by a political party.

Accountability: Option 3 proposes the smallest constituencies of all the options and would bring individual representatives closest to voters (although Option 2B would bring more representatives closer to voters). The direct election of public representatives and the closer relationship between them and voters will strengthen the accountability relationship at the constituency level.

The use of single-member constituencies means that there is a clear identification of the elected representative for the constituency. As a result, voters will be able to assess candidates and make informed choices, and reward responsive representatives, while rejecting poor performers. Voters will have greater leverage in the re-election prospects of a candidate, which strengthens incentives for responsiveness. Constituency vacancies must be filled through a by-election. This will enable voters’ continuous assessment of the parties’ performance.

Simplicity: Voters are familiar with the system based on its current use at the local government level, which shows in the marginal number of spoilt ballots. The system is straightforward and easy for voters to find their party or independent candidate on the constituency ballot.

Electoral Manageability: This option is similar to the system used for local government elections, which has been administered successfully. The introduction of measures to minimise non-competitive participation by political parties and independent candidates with limited prospects of electoral success should be considered.

Transparency: The option is clear and well understood. The proposed use of municipal ward boundaries to determine single-member constituencies will require some delimitation. This would likely need to align and expand on the ward delimitation/review process which takes place every five years ahead of local government elections. The use of compensatory seats leading to the overall proportionality of results means that gerrymandering of constituencies to influence the overall election results is virtually impossible. Nevertheless, it will be important to ensure that the process for determining the boundaries of single-member constituencies is done in a transparent and consultative manner.

Stable and Effective Government: As all three Options will result in the same overall electoral outcomes, there will be no expected difference among the various options in terms of their effect on party system fragmentation. The adoption of an electoral threshold for compensatory seats could be used to reduce fragmentation in the party system.

Advantages:

There is a clear link between the constituency and a single elected representative. Voters will easily understand the system, given their experience of local government elections, including both ward and PR seats.

Encourages constituency service by providing voters with identifiable representatives whom voters can easily monitor.

Presents strong prospects of individual accountability as the re-election of a public representative is directly tied to a group of voters.

The retention of compensatory seats on a party list allows for equitable gender and minority representation.

By-elections arising from vacancies in single-member constituencies allow for the evaluation of parties’ performance in Parliament between regular elections.

The use of smaller constituencies would allow independent candidates to contest on more equitable terms against other candidates put forward by political parties.

Disadvantages:

Political parties’ candidate selection processes may lead to biases in putting forward male candidates in single-member constituencies, which may lead to over-representation of men in constituency seats. The use of a party list for compensatory seats should help restore overall gender balance. The use of gender quotas could be considered to address concerns over gender balance.

The system will likely lead to changes in the electoral boundaries of single-member constituencies every election, which may lead to disputes. While it would be near impossible to manipulate overall results through the delimitation process, the winner-takes-all nature of single-member constituencies may heighten public concerns over the manipulation of the delimitation processes.

It is likely that single-member constituency seats are only won by a few large parties with smaller parties only represented through compensatory seats.

The plurality factor in single-member constituencies will mean that the winner may be elected with less than 50% of the vote in a particular constituency and so may not have the support of the majority of a community.

The use of single-member constituencies will lead to some level of deviation from proportionality in the first tier, but compensatory seats will return overall proportionality. If the disproportionality in constituencies seats is too high, there is potential for a party to win excess (overhang) seats – more constituency seats than they should be entitled to in the overall allocation. Appropriate legal provisions would consequently be necessary to deal with such a scenario, as is currently used in the local government system.

Constitutional and Legal Implications

Amendments to the Constitution are not necessary. All three options would require amendments to Schedule 1A of the Electoral Act. Options 2 and 3 will require assigning responsibility for the delimitation process for constituencies.

Financial Implications

The financial implications of the Options include changes required to business processes related to counting votes and communicating reforms to the public.¹⁴³ Many of these costs would be a one-off expense to change the system. Demarcation costs under Option 2 would only be for the initial delimitation of constituencies following which these would be used in subsequent elections. In Option 3, this process would be required for each election and could build on the Municipal Demarcation Board’s five-yearly process to delimit ward constituencies ahead of local government elections.

Comparing Options and Recommendations for Reform

On most measures the three options will achieve the guiding principles to the same or similar extent. The primary difference is the extent to which each option fosters an effective accountability relationship between voters and representatives. Option 1 to retain the current system presents limited advantages that other options do not offer to the same degree or greater. In terms of inclusivity, the three options would accommodate a diversity of views and demographic groups to the same extent, as all systems would be equally proportional. This allows the same parties to compete and gain seats on equal terms across all three systems and would lead to the same overall outcomes.

The use of party lists across all three options should facilitate gender representation. International experience shows similar levels of gender representation across proportional representation and mixed-member proportional systems (although in the case of Option 3 there is a likelihood that single-member constituencies will skew towards male candidates). There should be no expected difference in the level of gender representation between Options 1 and 2. Both international and South African experience show no correlation between the size of multi-member constituencies and levels of gender representation.

Option 1, however, presents the weakest accountability relationship between voters and their representatives based on the use of such large constituencies. Public consultations raised concerns that voters do not know their representatives, that voters have no say in choosing their representatives and that representatives primarily account to party leaders rather than voters. These concerns are not merely speculative. This lived experience is confirmed by the empirical evidence on the effects of closed list PR systems with large constituencies. This evidence shows that large constituencies disincentivise contact with voters, reduces candidate recognition among voters and diminishes the ability of voters to vote out undesirable candidates.

Reasons for Rejecting Option 1

Unnecessarily large constituencies: The combination of large constituencies and a large compensatory tier is unnecessary from an electoral system design perspective. Large constituencies are used to increase the proportionality of electoral systems. But as the size of constituencies increases, so does the distance between voters and representatives, weakening the accountability relationship between the two. While proportionality is crucial for the South Africa context to achieve political diversity and representation, such large constituencies are unnecessary in a context where compensatory seats are used.

Other Options put forward by the ERCP, using smaller constituencies to bring representatives closer to voters, are equally proportional and lead to the same political outcomes and allocation of seats. Large constituencies therefore present no advantages in relation to improving proportionality and the representation of diverse groups. The primary effect of these large constituencies is that it creates a substantial separation between voters and representatives – a common complaint heard throughout public consultations. South Africa has the largest constituencies among all PR systems including those that do not use a compensatory tier.

Options 2 and 3 present the opportunity to bring representatives significantly closer to voters and build a stronger linkage between the two without any trade-off in proportionality or the inclusivity of the electoral results. This retains the best aspects of the current system while responding to voters’ concerns and growing voter apathy. Figure 10 shows the substantial differences in the size of constituencies proposed by the three Options based on the number of registered voters in a constituency. In Option 1, constituencies range from 656,826 to 6,540,567 registered voters in a constituency. In Option 2, this would be between 423,930 and 906,815 voters in a constituency. In Option 3, each single-member constituency would have between 118,548 and 151,373 voters.

Unnecessarily large compensatory tier: The use of a large compensatory tier (200 seats), as proposed in Options 1 and 2A, is also unnecessary for overall proportionality in a PR system using multi-member constituencies. The use of 100 compensatory seats in Option 2B will lead to the same proportional and electoral outcomes while bringing more representatives closer to voters at the constituency level. Concerns over the potential for and implications of overhang in a two-tier PR system, such as Option 2B, are overstated.

Other countries with PR systems that include compensatory seats use both smaller constituencies and smaller compensatory tiers and still achieve comparable levels of proportionality. South Africa’s combination of large constituencies and a large compensatory tier makes it an outlier compared to similar PR systems using compensatory seats (Figure 11). As shown in the figure, South Africa has both the largest constituencies and the largest compensatory tier among equivalent PR systems – despite this making no contribution to the overall proportionality of electoral results.

In Option 2B, the use of multi-member constituencies in the first tier will already lead to high levels of proportionality. These multi-member constituencies would consist of seven seats on average. As explained in Chapter 4, improvements in the proportionality of electoral results diminish as the size of multi-member constituencies increases. Any increase in the size of constituencies beyond eight seats shows very limited improvement in overall proportionality.¹⁴⁵ A large compensatory tier in Option 2B is consequently unnecessary for restoring proportionality in the same way that this is needed in Option 3. Where single-member constituencies are used, the winner-takes-all nature of the system will usually result in a deviation from proportionality. This is why mixed-member proportional systems, such as Option 3, need more compensatory seats to restore proportionality.

Based on the assessment of the Options against the guiding principles, it is clear that Option 1 presents no advantages in the attainment of the guiding principles that other Options do not achieve to a similar or greater degree. The primary advantage of Option 1 over alternatives is its simplicity for the Electoral Commission to manage, albeit at the expense of a closer linkage and stronger accountability relationship between voters and their representatives. Based on the frustrations expressed in public consultations and growing dissatisfaction with South African democracy, electoral reform is imperative.

The Imperative of Electoral Reform

South African democracy is struggling. The 2024 election saw the country’s lowest voter turnout yet with only 58.6% of registered voters casting their ballots. More concerningly, the gap between the number of eligible and registered voters is growing. Many South Africans no longer bother registering to vote. As a result, less than 40% of the voting age population voted in the 2024 election down from 86% in 1994. In the 2024 election, South Africans had a choice between 52 political parties and 11 independent candidates. Instead, the majority chose not to vote. Some argue that this should be understood in the context of a broader global trend of declining voter turnout. In reality, South Africa now has the 162^nd lowest turnout rate of 200 countries.¹⁴⁶ Our problems lie at home, and support for our hard-fought democracy is rapidly disappearing. The latest Afrobarometer data shows that 71% of South Africans are dissatisfied with the way democracy works in our country (Figure 12).¹⁴⁷ Worryingly, the majority of South Africans now think that it doesn’t matter what kind of government we have (30%) or that non-democratic systems are sometimes preferable (24%) (Figure 13).

During the ERCP public consultations across the country, we heard the same concerns over and over again. Across provinces, languages, stakeholder groups, rural or urban locations – the lack of accountability was a common theme. Voters complained that they do not know their MPs, they have limited say in choosing their representatives, and MPs are indifferent to their concerns. Overwhelmingly, consultations were dominated by the topic of accountability and the sense of distrust towards political parties and politics was clear. In several cases, when participants were told that the ERCP is only advisory and that Parliament will make the final decision, citizens expressed their frustration, saying that the consultation process was then meaningless.

This dissatisfaction with the electoral process and frustration with politics is likewise reflected in public opinion data.¹⁴⁸ Only 28% of the public believe that elections enable voters to remove unrepresentative leaders from office “very well” or “fairly well.” Only 31% believe that elections ensure that National Assembly representatives reflect the views of voters. For the vast majority of South Africans, elections to the National Assembly are seen as ineffective in holding representatives accountable or ensuring the representation of citizens.

Successive reports have identified this deficiency in the electoral system. In 2002, the Electoral Task Team commended the 1994 system for its fairness and inclusivity but indicated that it could be improved to strengthen individual accountability. In 2009, the Independent Panel Assessment of Parliament found that the system disincentivised MPs from holding the Executive accountable using examples of delayed action on the HIV/AIDS crisis and investigations into the Arms Deal. In 2017, the High-Level Panel led by former President Motlanthe found that the system made MPs beholden to party leadership rather than voters at the expense of service delivery. In 2021, the majority of the Ministerial Advisory Committee on Electoral Reform recommended reforms to address public aspirations for greater local representation and accountability. The following year, the Zondo Commission again found that the system undermined parliamentary oversight and recommended that Parliament consider electoral reform to strengthen accountability.

It should be remembered that these were all formal processes established by government or Parliament. Both the public and successive enquiries have found the current system wanting. Apart from a handful of small countries,¹⁴⁹ every electoral system in the world seeks to establish some form of strong constituency and accountability relationship between voters and their representatives. This is based on an established consensus among electoral system scholars that the size of constituencies is fundamental to shaping the accountability relationship.

A common line of argument against reform is that electoral systems “cannot guarantee accountability.” This is true. In the same way that democracy does not guarantee accountability or good governance. Institutions do not guarantee anything, rather they shape incentives to encourage behaviour that we want and penalise bad behaviour. Regarding electoral systems, we could also say that an electoral system cannot guarantee inclusivity or gender representation. There are many factors that shape representation, including parties’ candidate selection processes, who chooses to run for office or how voters choose to cast their ballot. Parties could choose to put forward all-male candidate lists regardless of the system and unless there are formal gender quotas in place, the system cannot guarantee that parties will put forward diverse candidates.

However, based on research of different systems, we can see that PR and mixed-member proportional systems have higher rates of female representation than FPTP systems. In these cases, we can study how different institutional rules shape incentives differently. In multi-member constituencies, parties are incentivised to put forward a variety of candidates that could win the support of different types of voters. In single-member constituencies where only one candidate can be put forward for an area, parties tend to put forward male candidates, believing that they would perform better. Based on these observations of how different institutional rules shape incentives, we can say with confidence that certain design features tend to result in better outcomes for gender representation. We cannot guarantee that this will result in the desired outcomes, as parties can choose to put forward a specific type of candidate regardless. But we can be confident that the design of institutional rules will create the incentives to encourage the type of behaviour that we want.

In the same way, research on electoral systems shows that certain design features strengthen the public’s ability to hold representatives accountable and incentivise MPs to be more responsive to citizens. Systems that give voters more of a say in the re-election prospects of representatives incentivise MPs to be more responsive to the concerns of citizens. When voters are better able to identify clear lines of accountability and responsibility to specific representatives, they can monitor the performance of representatives and their responsiveness to the concerns of citizens. When this is combined with giving voters greater power over re-electing or rejecting individual candidates, representatives are incentivised to be more responsive to the interests and concerns of voters.

Representatives that are exposed to direct electoral pressures from voters know that they may lose their seat in the next election if they engage in unethical conduct, act against the public interest or fail to respond adequately to the concerns of voters. Political parties know that if they put forward weak or unethical candidates, voters will reject them. Expanding the public’s influence in choosing their representatives and holding representatives to account does not remove the role or importance of parties. Instead, it makes representatives accountable to both their parties and voters. The evidence for the effects of electoral system design on the accountability relationship between voters and their representatives is discussed in detail in Chapter 4.

Global experience in electoral reform among PR systems shows a clear trend towards bringing representatives closer to voters and giving them more say in choosing their representatives – whether by introducing open lists, smaller constituencies or the adoption of a mixed-member proportional system. Debates over electoral system design have always revolved around a tension between the proportionality of electoral systems and the closeness of the relationship between voters and their representatives. South Africa’s existing use of a two-tier compensatory system overcomes this conventional trade-off in electoral system design. This presents a unique opportunity for electoral reform presented by Options 2B and 3 to bring representatives significantly closer to voters and strengthen the accountability relationship without any trade-off in proportionality or the inclusivity of the electoral results.

We recommend that Parliament consider electoral reforms to introduce smaller multi-member constituencies (Option 2B) or single-member constituencies (Option 3) to foster a closer and more direct relationship between voters and public representatives. Option 2B, introducing smaller multi-member constituencies and 300 constituency seats would bring the most representatives closer to voters and strengthen the accountability relationship between voters and the most representatives. Option 3’s 200 single-member constituencies create a clear link between voters and a single elected representative for their area.

In considering these two systems, Parliament should, however, also consider whether the same electoral system could be practically implemented concurrently at the provincial level. In the case of Option 2, the same or similar multi-member constituencies could be used concurrently for both national and provincial elections. In Option 3, the delimitation of single-member constituencies at both the national and provincial level would introduce complications as the two elections would need to use distinct boundaries for their constituencies. This introduces significant complexity for voters and election administration. This issue is examined in greater detail in the final section dealing with provincial elections and should be carefully considered in choosing between options.

Additional Considerations

Open Lists

The prospect of introducing open lists was widely supported by political parties as well as the broader public. The practical implications of open list voting would, however, need to be strongly considered before such a reform is introduced. We caution against this reform without due consideration of the conditions necessary for its effective rollout.

The practicality of introducing open lists is dependent on three issues: the number of seats in a constituency, the number of parties and independent candidates contesting, and the ability of voters to effectively engage with new electoral rules. Open lists would be highly impractical to introduce to the existing system (Option 1) due to the large size of multi-member constituencies. In Gauteng, for instance, each party would be required to put forward up to 47 candidates for voters to choose from on the regional ballot for Gauteng. Smaller multi-member constituencies, such as used in Option 2, will reduce the number of candidates each party would need to put forward on an open list and may be more manageable in terms of the size of ballot papers and accessibility to voters to assess different candidates.

The large number of parties contesting elections would, however, still remain a challenge in both Options. In the 2024 election, 52 political parties contested and even in Option 2’s smallest constituency with four seats, this would present voters with a bewildering number of candidates to choose from. For this reason, open list voting is more common in systems where electoral rules are in place to contain the number of parties contesting an election. There are two possible approaches to address this issue. The first is raising ballot access requirements to discourage frivolous contestation. This can include raising the signature requirements for registration or the election deposit required of parties. Alternatively, an electoral threshold would discourage parties with limited prospects of success from contesting.

The third question to consider is the extent to which voters could effectively engage with open list voting. As has been mentioned, the number of spoiled ballots in South African elections is low by international standards and South African voters should not be underestimated. The introduction of multiple electoral reforms simultaneously may, however, present challenges with election management and with voters’ understanding of reforms. For this reason, we strongly advise considering the importance of reform sequencing and emphasise that electoral reforms often span multiple electoral cycles. Open list voting will likely only become practically implementable in the longer term if issues related to the size of constituencies and the number of participants are addressed, and once voters have become familiar with such reforms.

Electoral thresholds

Panellists supporting the Alternative Report propose the introduction of a quota threshold. This would effectively use the seat quota for compensatory seats as the legal threshold for representation. Only parties meeting the seat quota in the first instance are considered for further allocations. In the South African context, the quota is determined by the formula: Quota = All valid votesNumber of seats + 1 + 1.

A quota threshold would consequently change based on fluctuations in voter turnout (the number of total valid votes) and the number of seats won by independent candidates (the number of seats available for compensation). This floating quota can present challenges for voters and new political parties to easily understand and consider the likely prospects of a successful run for office for a very small political party. Should Parliament deem a legal threshold necessary, a percentage-based threshold would be significantly easier for voters and political parties to understand over a quota threshold. It is fixed, clear and not dependent on various variables that voters and parties may not always understand.

The possibility of an electoral threshold was a contentious issue, particularly among smaller political parties. This is a delicate issue to balance between the inclusion of a wide range of political contenders and the importance of political stability for effective governance. There is a well-established relationship between the permissiveness of electoral systems and the extent of party system fragmentation. In turn, the extent of fragmentation in a legislature is a key indicator in determining the prospects of coalition survival – again, as shown by a strong body of research evidence discussed in Chapter 4.

Regulating coalition agreements may contribute to reducing coalition instability once coalitions are formed, but it would not address fragmentation in legislatures and the complexity that it introduces to government formation in the first instance.¹⁵⁰ The extent of fragmentation will inform how many coalition partners are needed to form a majority, increasing the complexity of coalition formation in finding common ground and agreeing on policy issues across multiple actors with distinct interests. Even in cases where coalition agreements are in place, complex and fragmented coalition partnerships increase the likelihood of collapse. Some level of legal threshold is likely necessary to contain excessive fragmentation. Currently, South Africa has the most permissive electoral rules in the world without an electoral threshold, and again, it is unique in this regard.

If Parliament considers a legal threshold necessary, a phased approach to its introduction over successive electoral cycles should be considered. This could start with a very small threshold in the first election with a clear timeline for incremental increases in subsequent elections. This would establish clear expectations to provide time for political parties to adjust to changes and for voters to become familiar with the principle.

Provincial Elections

The provincial electoral system received limited input in the course of public consultations, and few written submissions addressed this aspect of reform. However, the same concerns raised about the national system are equally applicable. In the current system, representatives in Provincial Legislatures are elected in province-wide constituencies ranging from 30 to 80 seats. This presents even larger party lists than currently used for the regional ballot, and concerns about the relationship between voters and their representatives are similarly applicable.

In our constitutional framework, the main functions devolved to provinces relate to healthcare and education, taking up the vast majority of their budgets. Provinces have a range of concurrent responsibilities shared with the national government, but most importantly, have a key role in supporting and overseeing local government through Provincial Treasuries and local government departments. These functions mean that provincial governments are a critical part of the service delivery ecosystem, including their role in healthcare and education, and in overseeing and supporting local government service delivery. In this context, effective oversight and constituency service should form a major focus of Provincial Legislatures.

The options for the provincial electoral system align with the three options presented for the National Assembly. Implications for the attainment of the guiding principles, and the advantages and disadvantages of different models remain the same. These need not be repeated as the same effects apply.

Provincial Option 1 – The Current System

Representatives are elected in a single province-wide constituency ranging from 30 to 80 representatives. The sole reform relates to the process for allocating seats left over after the first quota allocation. This would see seats left over allocated on the basis of largest remainders.

Provincial Option 2 – Smaller Multi-Member Constituencies

This option would introduce smaller multi-member constituencies within provinces, following the same boundaries used in Option 2 of the national system. For Provincial Option 2, an equal split (50:50) between constituency and compensatory seats is recommended for provincial legislatures on the basis that there are fewer seats in Provincial Legislatures which affects overall proportionality.

Provincial Option 3 – Single-Member Constituencies and Overall Proportionality

This option includes a combination of single-member constituencies and a compensatory party list. There would be an equal split in constituency and compensatory seats. The delimitation of single-member constituencies at the provincial level introduces a level of complexity, and it would be difficult to use single-member constituencies for the National Assembly and Provincial Legislatures concurrently.

For example, National Assembly elections in Option 3 would include 47 single-member constituencies in the Gauteng region. Under Provincial Option 3, there would be 40 single-member constituencies (as half of the 80 seats in the Gauteng Provincial Legislature would be set aside for constituencies). Consequently, different boundaries would be needed to maintain similar population sizes per seat. Similar complexities would be applicable to other provinces based on their unique circumstances. Adopting single-member constituencies at both the national and provincial levels simultaneously would be very complex. These implications would need to be considered very carefully.

The same challenges would not apply to adopting Options 2 concurrently as the number of seats in multi-member constituencies can be adjusted based on the number of available seats and population size. Consequently, the same boundaries can be used. This would be easier for voters to understand and support election manageability as the same set of ballots would be used in a particular area.

Introducing a Second Provincial Ballot

Both Provincial Options 2 and 3 would require the introduction of a second ballot for Provincial Legislatures as is currently the case for the National Assembly. One ballot would be for constituency representatives (including independent candidates) and the second ballot for overall compensation. This will require voter education but is not insurmountable. The introduction of the third ballot in the 2024 election did not have a significant impact on voters’ ability to effectively engage as evidenced by the low number of spoiled ballots and the 2024 Election Satisfaction Survey.¹⁵¹

Constitutional and Legal Implications

Amendments to the Constitution are not necessary. All three options would require amendments to Schedule 1A of the Electoral Act. Provincial Options 2 and 3 will require assigning responsibility for the delimitation process for constituencies.

Financial Implications

Increased costs related to Options 2 and 3 are driven by the introduction of a second ballot, increasing printing costs, as well as communication costs to explain reforms and the second ballot to voters.