Did Floor-Crossing Alienate South African Voters? Evidence from Municipal Legislatures

Abstract

Floor-crossing, the practice of defecting from one political party to join another, is common in democracies. While empirical research has answered a number of questions about the determinants of party switching and the motivations of party switchers, little is known about the consequences of floor-crossing in electoral democracies, especially for the voters who watch their elected representatives ‘switch uniforms’ in the middle of the game. From 2002 to 2008, floor-crossing was legal in South Africa and defections at all levels of government were numerous. Critics of the country's floor-crossing regime often speculated that rampant defections would drive voters away from the polls and undermine voters' trust and political engagement. This paper uses data cataloguing defections in municipal legislatures to assess the extent to which floor-crossing in South Africa may indeed have alienated voters. The paper finds strong evidence that floor-crossing may have suppressed voter turnout, at least in immediately subsequent elections, but only weak evidence to suggest that floor-crossing had any direct or lasting effect on voters' attitudes about government and the politicians who run it.

Notes

1. This logic played out in South Africa where, at the ward level, councillors elected as ward representatives were much less likely to defect during the 15-day windows than their colleagues who were elected on party lists (McLaughlin 2012).

2. This is not to say that there are not strong cross-national analyses in the literature. For an example, see Owens (2003).

3. Information on the wards was gathered from a variety of Internet sources. Electoral data are available from South Africa's Independent Electoral Commission at www.elections.org.za. Information on social demographics (from the 2001 Census) is available for all wards from the Municipal Demarcation Board at www.demarcation.org.za

4. The quality of service delivery was found by averaging respondent's answers to two questions about services. The first question asked respondents about their municipality's success at maintaining roads and keeping the community clean.

5. The maximum number of possible ward crossings is 2, since each ward has only one seat and there were only two floor-crossing windows between 2000 and 2006.

6. This database can be accessed at www.demarcation.org.za. The matching was performed by the author in 2007.

7. If the geographical database returned multiple results (or no results) or if the author could not determine the municipality of residence with certainty, the observation was simply dropped in the interests of accuracy.