Non-payment culture and the financial performance of urban electricity utilities in South Africa

ABSTRACT

Non-payment for services continues to challenge sustainability in municipal service delivery across South Africa. Literature provides that the culture of non-payment stems from the apartheid era where mass civil disobedience manifested through boycotting the payment of rates. This study examines the impact of the non-payment culture on municipal financial performance in South Africa. Panel data for 28 municipalities for the years 2005–19 is used, and the random-effects model is employed to estimate the relationship between municipal financial performance and non-payment. Results confirm that non-payment has a negative impact on financial performance. For every R1000 increase in bad debts written off, financial performance is reduced by R291. Further, grants from the national government, the number of consumers, and the number of household units receiving free basic electricity positively affect financial performance. These revelations warrant the need for more innovative approaches that transform non-payment into a culture of payment.

KEYWORDS:

• Electricity provision
• municipal financial performance
• non-payment culture

JEL CLASSIFICATION:

• H26
• H63
• H68
• R58

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The Hausman test is used to choose between the FE and RE models. The null hypothesis of this test is that FE coefficients are not statistically different from RE coefficients. If the test statistic is significant, the decision would be to reject the null hypothesis, thus FE coefficients are consistent and would therefore be the right model to choose, and otherwise. The Breusch and Pagan Lagrangian Multiplier (BPLM) test is used to choose between the pooled OLS and RE models. The null hypothesis for this test is that there is no random effect. If the BPLM value is insignificant, the null hypothesis is rejected. Thus, the RE model would be chosen over the pooled OLS model.

2 On 15 March 2021, US$1 = R15.01.

3 While it may be argued that FBE and GRANT may be used interchangeably because the number of poor households in a municipality determines the grant received by each municipality, it is important to note that some grants are not determined by the number of poor households in a municipality. The variable GRANT in this study is inclusive of both conditional and non-conditional grants. Thus, it does not only capture the grant that is based on the number of poor households. As such, GRANT and FBE cannot be used as proxies of each other in this context. Further, the correlation coefficient of these variable is very low at 0.202 as shown in Table 2, indicating no evidence of multicollinearity. Thus, the two can be included as explanatory variables in the same model.