Payment Behaviour of Electricity Consumers: Evidence from the Greater Accra Region of Ghana

Abstract

Electricity companies are typically businesses with significant credit sales. However, timely payment of bills on the part of customers, which is supposedly the most reliable source of cash flows for these utilities, has been generally very poor especially in the context of much of the developing world. The aim of the study is to identify a set of organizational and behavioural factors that influences the bill payment behaviour of customers of the Electricity Company of Ghana. Based on a survey of households in the Greater Accra Region of Ghana, our empirical analysis suggests that electricity utilities must work towards reducing the transaction time of customers at the bill collection centres and improving upon the quality of service and customer satisfaction in order to curtail customer bill payment period. These findings are robust to the influence of potentially extreme data points in our sample. We provide detailed discussion and policy implications of our findings.

Keywords

Payment behaviour revenue collection efficiency utility billing electricity utilities electricity customers

Introduction

Revenue is undoubtedly the lifeblood of companies both in developed and developing countries. The success of a company is chiefly dependent on how much revenue it is able to generate to cover its operational cost (Brealey et al., 2006). Therefore, it is imperative that electricity utilities take steps to maintain constant cash flows through sustained and improved bill presentment and payment. Timely bill presentment and collection of payment thereof show how efficient the utility is in utilizing its massive investment in assets.

Rajan (2002) reports that quite a few countries globally are implementing reforms and restructuring in the electricity supply industry to ensure sound financial and operational performance. In the context of Ghana, Aryeetey and Ahene (2005) note that, over the years, the Electricity Company of Ghana (ECG) has also undertaken several reforms initiatives including restructuring and corporatization. These reform measures aim at easing the financial burden on the Government of Ghana (GoG) and ensuring the operational and financial efficiency of the company. However, the fact remains that the GoG still needs to spend quite heavily on electricity subsidy due to an unsatisfactory revenue performance of the company. For instance, in 2012 alone, the government spent 179.7 million Ghana cedis in consumer subsidies for water and electricity, a move that effectively suspended the quarterly rate adjustment formula¹, which is often applied by the Public Utility Regulatory Commission (PURC; B&FT, 2013).

The ECG is unable to raise much needed revenue to meet its operational cost due mainly to unfavourable bill payment behaviour of its customers. The worsening revenue situation has even recently constrained the company’s efforts in financing investments through corporate bonds (B&FT, 2013). Although the government is spending huge sums of money in subsidizing the cost of utility consumed by the public (especially, the poor) through social block tariff system (Boakye and Nyieku, 2010), the company still remains unable to meet its revenue collection target. Even marginal increases in tariffs to improve revenue situation have always resulted in customer dissatisfaction and industrial unrest. In this backdrop, this study attempts to identify a set of potential factors and their effect on the bill payment behaviour of the electricity utility customers of the Greater Accra Region of Ghana (GARG). First, we undertake a thorough review of related literature in our effort to identify some factors that are potentially useful in explaining the payment behaviour of the electricity utility clientele. Subsequently, a regression analysis is carried out to understand the effect of these potential factors in the Ghanaian context.

The study is presented in seven additional sections. First, in the review of literature section, we review relevant literature on some organizational and behavioural constructs and their potential relationship to customers’ behavioural intentions. In the second section, research objectives of the study are specified. The third section provides the rationale of the study. Fourth, in the methodology section, we describe the sample, data collection and variable construction procedures and specify the postulated regression framework for the study. The fifth section presents the results of empirical analyses. In the sixth section, we present discussion of results and implications of our findings. The final section summarizes the study and draws conclusion.

Review of Literature

Utility Billing (UB) Issues

In spite of the low-income levels in developing countries, cost recovery is a prerequisite for the sustainability of electricity utilities. It is argued that, by 2015, 88 per cent of the global population will reside in urban centres (WHO and UNICEF, 2000). These newly developed urban centres are likely to pose a daunting challenge to the electricity utilities both in terms of service provision and cost recovery. The bill payment and collection efficiency to support adequate service provision are generally very poor in developing countries. For instance, Kayaga et al. (2004) argue that the bill collection efficiency, which is often referred to as the headline efficiency², in African urban water utilities is lower than 50 per cent. This situation is quite representative of a rather gloomy picture of other African utilities as well. An effective billing and payment mechanism may therefore help improve the cost recovery levels of these utilities.

Chipofya et al. (2009) argue that utilities do not achieve efficiency in billing because (i) bill packages fail to establish the customer base, (ii) bills delivery is irregular, often resulting in non-payment by registered consumers especially in slum areas, (iii) there are incidences of undercharging and overcharging due to billing errors, creating dissatisfaction among consumers and (iv) ineffectiveness of the billing system. They suggest that utilities should explore opportunities to hand deliver bills during meter reading and also outsource meter reading to subsidiary companies. Therefore, electricity utilities must focus on improving bill presentments instead of concentrating too much only on power generation and distribution in order to achieve better customers’ bill payment behaviour.

Ogujor and Otasowie (2010) argue that prepaid system ensures adequate and proper billing of customers. Under the prepaid system there is no debt accumulation. The dependence on huge material and human resources needed for disconnection and reconnection is avoided. In addition, the prepaid system may help to reduce the danger and inconveniences associated with such disconnection and reconnection. Ironically, in some jurisdictions, regulators and consumer advocates have expressed concerns that utilities might impose prepaid system on customers in low income areas, thus stigmatizing customers whose positive history of bill payment may equal or frequently exceed that of their wealthier neighbours. Consistent with this line of argument, Oracle (2009) suggests that most consumers are not in favour of the prepaid system because of its cost, fairness and health and safety concerns.

Customer Satisfaction (CS) and Behavioural Intentions

According to Cronin et al. (2000), customer satisfaction (CS) has a direct relationship with behavioural intentions. The survival of an enterprise in the long run depends ultimately on how well its customers are satisfied with services that are provided. If customers are not satisfied with the service provision, their dissatisfaction will manifest itself in their bill payment behaviour. Nimako (2012) argues that the satisfaction of customers is very vital since it can influence their behavioural intentions. In his study on service quality, CS and behavioural intentions in Ghana’s Mobile Telecommunication Industry, he finds a significant relationship between CS and behavioural intentions.

Drucker (1973; as cited in Ahmed et al., 2010) sees CS as the lifeblood of a business, which helps in creating a solid relationship between marketing and management. Moreover, he explains that there is a positive relationship between CS and behavioural intentions. Therefore, it is important for utilities to rely on CS to stimulate behavioural intentions of customers in order to expedite their payment of utility bills.

Service Quality (SQ) and Behavioural Intentions

SQ is a judgment that a customer makes when he compares his expectations of service performance prior to the service encounter and the perceptions of the service he really receives (Grönroos, 1984).

The study by Lai and Chen (2010) examines the factors that affect the behavioural intentions of the public to use the new public transit system in Taiwan. The results show that there is a significant relationship between SQ and behavioural intentions. It is argued that services can sometimes be effective if they are designed around the needs of the user, instead of the convenience of the provider. Also, interactions between a utility and its customers regarding service outages, emergencies, billing questions and billing disputes create a certain level of satisfaction among customers, which lead to prompt bill payment (Moss, 2007). Parasuraman et al. (1998) have, however, stressed that an assessment of overall SQ depends on the gap between expectation and perceptions of actual performance.

Cronin and Taylor (1992) investigate the concept and measurement of SQ and the relationships among SQ, CS and purchase intentions. Their results show that there is a positive relationship between SQ and purchase intentions. Similarly, Kuruuzum and Koksal (2010) examine the impact of SQ on behavioural intentions in hospitality industry. Using the structural equation modelling (SEM) and linear structural relationship (LISREL), they find that SQ has more impact on the behavioural dimensions of ‘loyalty’ and ‘pay more’.

Corporate Image (CI) and Behavioural Intentions

CI is the net result of the interaction of all experiences, beliefs, feelings, knowledge and impressions that people have about a utility company (Bernstein, 1984). The concept of corporate reputation is anchored on three elements – brand, stakeholder and organizational reputation. The brand recognition is the probable perception of customers about a particular brand. The stakeholder reputation emphasizes the reputation of a particular brand among stakeholders. And finally, organization reputation is the general perception of the public about the company relative to a particular brand (Dissanayake, 2012).

Maiyaki and Mokhtar (2012) in their literature review on the determinants of customer behavioural intention in Nigerian Retail Banks propose that CI plays a significant role in the formation of customer behavioural intentions. Similarly, Yu and Ramanathan (2012) investigate the relationships between SQ, CI, CS and behavioural intentions for a sample of 404 supermarket customers in China. Using a SEM they find a significant relationship between CI and behavioural intentions. Interestingly, however, in their study on bill payment behaviour in urban water utilities in Uganda using regression techniques, Kayaga et al. (2004) find that there is no significant relationship between CI and behavioural intentions.

Service Value (SV) and Behavioural Intentions

SV is the consumer’s total evaluation of the utility of a product based on perceptions of what is received and what is given (Zeithaml, 1988). The difference (or trade-off) between what is purported to be supplied by utilities and what customers actually receive is what customers consider as the value of service. Cronin et al. (2000) assess the effects of quality, value, satisfaction on consumer behavioural intentions in service environment. Their results reveal that SV has a direct impact on behavioural intentions. Similarly, Choi et al. (2004) examine the relationship among quality, value, satisfaction and behavioural intentions. Using a sample of 537 South Korean health care consumers, the result of the article shows that SV has a significant direct impact on behavioural intentions.

The study by Khan and Kadir (2011) on the impact of perceived value dimension on satisfaction and behavioural intentions among young adult consumers in the banking industry also finds that SV has a significant direct impact on behavioural intentions. These findings suggest that SV may indeed be an important determinant of consumer behavioural intentions.

In addition to the constructs reviewed in the previous paragraphs, Mugabi et al. (2007) and others (Kayaga et al., 2003; Waldron, 2011) find that certain organizational variables such as transaction time (TT) and monitoring and control (MC) may also affect customers’ behavioural intention to pay utility bills in a timely fashion.

Research Objectives

Based on the review of literature presented in the previous section, we can clearly see that there are certain organizational and behavioural factors that are potentially useful for an understanding of the bill payment behaviour of the ECG customers in the GARG. Having identified those potential factors, we proceed to assess their true effect on actual bill payment behaviour. In summary, therefore, the following two specific objectives guide our study:

Identifying principal factors that influence the bill payment behaviour of the electricity utility clientele in the GARG, and

Determining the direction, magnitude and relative importance of the effects of those factors on actual electricity bill payment behaviour of the customers of ECG.

Rationale of the Study

Electricity companies are businesses with significant credit sales. Therefore, timely collection of payments from the customers constitutes the most critical indicator of the company’s liquidity and future growth potential. In this study, we measure customers’ behavioural intention to pay electricity bills by what is termed as the mean bill payment period (MBPP)—the number of days, on average, it takes a company’s credit customers to pay their accounts. From the financial analysis perspective, this metric is essentially one of the most important activity ratios that reflect a company’s efficiency at using its assets in generating revenues. It also indicates the effectiveness of the pricing strategy and credit policy of the company. In evaluating and enhancing these aspects of overall company performance, knowledge of the factors affecting the payment behaviour of customers may prove extremely useful. In addition, although most of the behavioural constructs that we use in our analysis have been studied mostly in the context of the service sectors of mainly well developed economies, their relevance remains largely untested especially in the context of African utility sector. Therefore, this study is expected to contribute to the literature by providing first-hand empirical evidence on the effect of these factors on actual bill payment behaviour of electricity customers of a major African economy.

Methodology

Data Source

The data for the study are collected in a cross-sectional survey of 150 households in the GARG. For managerial purposes, the ECG has grouped the region into Accra East, Accra West and Tema regions. At the time of the study, the total customer population of ECG in the GARG was 926,197, with Accra East, Accra West and Tema regions having customer populations of 277,737, 379,084 and 269,376, respectively. In a random exercise, the Accra East region is selected for the study. The sample for the study is selected using a systematic sampling technique, which is an appropriate sampling technique in a household survey (Cooper and Schindler, 2002). The sample size of 150 respondents is considered appropriate for estimation purposes (Cooper and Schindler, 2002; Hyndman and Kostenko, 2007).

The Questionnaire

The instrument used for the data collection is questionnaire. And since it is developed and not adapted, we conduct a literature review to operationally define the constructs in order to come up with the scales for measuring the constructs. In order to come up with the first draft of the survey questionnaire, apart from using the literature, we engage the Customer Service Directorate (CSD) of the ECG in a focus group discussion and their suggestions aid the design of the instrument. The questionnaire is sent out for a pilot study with 60 customers of the ECG in the GARG. The results of the pilot study are analyzed in terms of reliability and factor analyses, on the basis of which the questionnaire is further refined. The reliability assessment of the five (5) scales of CS, SQ, CI, UB and SV yields alpha coefficients of 0.877, 0.925, 0.964, 0.867 and 0.953, respectively. Notably, all these alpha values are higher than 0.7, a minimum level considered as good and acceptable (Sekaran, 2000). Construct validity of the scales is established through exploratory factor analysis. Using the principal components method, the scales are factor analysed and subjected to orthogonal rotation in order to produce interpretable dimensions.

The questionnaire has a total of 48 items divided into 7 sections: TT (1 item), MC (1 item), quality of service (10 items), CS (10 items), CI (9 items), UB (5 items), SV (7 items) and background information (5 items). A 5-point Likert type scale with classifications from 1 (strongly disagree) to 5 (strongly agree) is used to capture data for the constructs of SQ, CS, CI, UB and SV. This is to enable the respondents express their levels of agreement or disagreement adequately to the statements under each construct.

Data Collection Procedure

For data collection, we send out the questionnaires to the respondents with the help of research assistants. While the literate respondents fill out the questionnaires themselves, the illiterate ones are assisted by the research assistants to fill out theirs by reading out and explaining the items to them (respondents). Out of the 150 questionnaires sent out, 127 questionnaires are returned, giving a response rate of 84.67 per cent.

Additionally, we obtain data for the MBPP from the ECG billing database for the period from January 2010 to December 2011. Billings to and payments made by respondents for each financial year is extracted. The MBPP at the end of each financial year is then calculated for each respondent and averaged over the 2-year period.

Regarding the demographic characteristics, of the 127 respondents who answer the questionnaire, 54.30 per cent are males while 45.70 per cent are females. While 66.90 per cent claim that they are educated, 33.10 per cent claim to be uneducated.³ Also, about 38.60 per cent of the respondents report they are aged between 18 and 35 (young adults), while 42.50 and 18.90 per cent of the respondents report they are aged between 36 and 60 (adults) and above 60 (elderly), respectively. Majority (42.50 per cent) of the respondents are in the middle income category confirming the recent World Bank classification of Ghana as a middle income country.⁴

Variable Construction

To estimate the model, we examine the constructs and their conceptual and empirical relationships. The detailed descriptions of the constructs and their potential influence on behavioural intentions are given in the proceeding paragraphs.

Bill payment behaviour: As it is mentioned earlier, MBPP is used as a proxy for customer bill payment behaviour, which is the behavioural intention of customers in this study. The bill payment behaviour of the electricity utility customers is measured on the basis of customers’ response to utility bills using the mean collection period, a credit control ratio, often used by utilities. According to Chardwick (as cited in Kayaga, 2002), the mean collection period is a measure of how long, on average, it takes an organization to collect its debts. Instead of the mean collection period on the side of electricity utilities, our study uses the mean payment period of customers since we are looking at the factors that influence bill payment and not collection. However, the concept underlying mean collection period is used.

Raw data for the MBPP, the criterion variable, is obtained from the ECG billing database for the period January 2010 to December 2011, across two (2) financial years. Billings to and payments by respondents for each financial year are extracted. The MBPP at the end of each financial year is then calculated for each respondent. To do this, we divide the outstanding arrears at the end of each financial year by the total billings done in the year to obtain the ratio. The ratio obtained thereof is then multiplied by 365 to obtain the values for the bill payment period for 2010 and 2011. We get the value of the MBPP by averaging yearly the bill payment period for each customer for 2010 and 2011 financial years.

Transaction time (TT): It is the time taken in minutes at revenue collection centres by customers to pay their bills in the GARG. It is sometimes referred to as ‘travel time’. It is a quantitative variable, which is measured on a ratio scale. To construct the model of the study, individual respondent’s TT is captured as appeared in the questionnaire. We expect the TT to have a positive relationship with the MBPP, which means that the longer the time taken by customers to pay their bills the higher their MBPP.

Monitoring and control (MC): MC is the average number of times the electricity utilities’ inspection units carry out routine inspections on customers’ facilities in the GARG in a year. Kayaga et al. (2003) refer to it as ‘inspection times’. It is believed that the frequency of inspection times prevents illegal connection and power loss and ensures frequent bill payment. In this study, we expect the MC to have an inverse relationship with the MBPP. If it is found significant, it means that it contributes to the reduction of customers’ MBPP. It is a quantitative variable which is measured on a ratio scale.

Customer satisfaction (CS): It is a measure of how services supplied by electricity utilities meet or surpass customers’ expectation. Kayaga (2002) reports significant effect of CS on customers’ water bill payment behaviour. If customers are satisfied with the services of a utility provider, naturally, they will be more willing to pay for their bills. Therefore, in this study, it is our a priori expectation that CS will have a negative effect on the MBPP. Since CS is a qualitative construct, we use the 5-point Likert type scale to transform it into a quantitative variable. The composite scores of the respondents obtained thereof are then used for the estimation.

Service quality (SQ): It is a judgment that a customer makes when he or she compares his expectations to the perception of the service he/she has received. It is also a measure of customers’ expectations of the services of utility companies. In Ghana, customers of utility service providers attach much importance to quality and are therefore willing to pay for services if quality is improved. So, we add this variable to the specified model to see whether or not it will have significant effect on the MBPP. We expect SQ to have an inverse relationship with the MBPP. Data for SQ is captured using the 5-point Likert type scale. The scores for the statements in the scale are added up for each respondent and the total score is used for the estimation.

Corporate image (CI): CI is the mental picture that springs up at the mention of a firm’s name. Our study looks at the public perception about the ECG. The concept of CI is anchored on three elements and they are brand, stakeholder and organizational reputation. In this study, we use organizational reputation, which is a measure of the general perception of customers about electricity utilities’ service offerings. CI is expected to have a negative effect on the MBPP, which means that an improvement in CI of the ECG will result in a decrease in the MBPP of customers. We capture data for the construct by using the 5-point Likert type scale with classifications from 1(strongly disagree) to 5 (strongly agree).

Utility billing (UB): UB in the context of this study captures tariff setting, connection fees, metering and billing. In Ghana, the lifeline tariff is used, which is computed using the marginal cost pricing concept. Bad pricing policies can affect customers’ responses to bills. To the extent that utility companies offer reasonable prices, unambiguous tariff structure and clear invoices for services provided, customers are expected to respond positively by paying their bills on time. We, therefore, measure UB as the perception of customers about tariff setting, metering and billing by utility companies. While a positive perception will result in a reduction in the MBPP, a negative perception will increase the MBPP. We expect an inverse relationship between UB and MBPP.

Service value (SV): SV is the measure of consumer’s total evaluation of the utility of a product based on perceptions of what is received and what is given. The difference (or trade-off) between what is given and what is received is the value that customers look for in utilities. We expect SV to have an inverse relationship with the MBPP. This is because, if customers get the value they are looking for in utilities, they will, naturally, be more willing to pay their bills promptly. To construct SV, we use a 5-point Likert type scale with classification from 1 (strongly disagree) through to 5 (strongly agree).

Model Specification

In order to understand the nature and extent of the effect of each of the previously defined constructs on behavioural intentions of the electricity customers as reflected in their bill payment behaviour, we use the following regression framework: MBPP_i(1)β₀β₁TT_iβ₂MC_iβ₃SQ_iβ₄CS_iβ₅CI_iβ₆UB_iβ₇SV_i 𝜀_i

where MBPP_i, TT_i, MC_i, SQ_i, CS_i, CI_i, UB_i and SV_i are MBPP, TT, MC, SQ, CS, CI, UB and SV corresponding to i-th customer in the sample, respectively. b_k is the k-th regression parameter of interest and f_i is the error term.

Empirical Results

Descriptive Statistics

We begin by providing summary statistics of each of the variables under consideration in Table 1. The dependent (criterion) variable in our regression framework, MBPP, has a mean value of approximately 80 days and it displays the highest level of dispersion among the variables presented in the table. A large value of coefficient of variation (CV) of about 80 per cent indicates that the MBPP of a randomly selected electricity customer may actually be very different from that of a typical customer. Among the predictor variables, all variables except for TT show only a moderate amount of variation relative to their respective mean values. Clearly, a high dispersion of TT is indicative of very different customer experiences in terms of waiting times at different revenue collection points in the GARG.

Table 1.

Descriptive Statistics

	N	Minimum	Maximum	Mean	Standard Deviation	Coefficient of Variation
TT	127	5.00	180.00	49.8189	39.6237	0.7954
MC	127	0.00	14.00	6.1102	3.8714	0.6336
SQ	127	13.00	50.00	32.2520	8.2836	0.2568
CS	127	10.00	49.00	32.0787	7.4286	0.2316
CI	127	9.00	45.00	31.1811	7.5671	0.2427
UB	127	5.00	25.00	16.8819	3.8496	0.2280
SV	127	10.00	92.00	24.8661	8.2012	0.3298
MBPP	127	1.06	346.74	77.9667	61.5537	0.7895
Valid N (list wise)	127

Source: Survey data, 2012.

Table 2.

Pearson’s Correlation Matrix

	MBPP	TT	MC	SQ	CS	CI	UB	SV
MBPP	1
TT	0.498**	1
	(0.000)	–
MC	–0.162	–0.224*	1
	(0.068)	(0.011)	–
SQ	–0.639**	–0.405**	0.064	1
	(0.000)	(0.000)	(0.477)	–
CS	–0.606**	–0.333**	–0.020	0.775**	1
	(0.000)	(0.000)	(0.827)	(0.000)	–
CI	–0.546**	–0.445**	–0.008	0.735**	0.687**	1
	(0.000)	(0.000)	(0.931)	(0.000)	(0.000)	–
UB	–0.475**	–0.32**	–0.117	0.597**	0.659**	0.566**	1
	(0.000)	(0.000)	(0.189)	(0.000)	(0.000)	(0.000)	–
SV	–0.442**	–0.395**	0.145	0.49**	0.475**	0.428**	0.386**	1
	(0.000)	(0.000)	(0.103)	(0.000)	(0.000)	(0.000)	(0.000)	–

Source: Survey data, 2012

Note: ** and * indicate statistical significance at 1 and 5 per cent level, respectively.

In Table 2, we present the correlation matrix showing the strength and direction of linear pairwise relationship between the variables under consideration. Most notably, the correlation coefficients between the dependent variable (MBPP) and each of the seven predictor variables have expected signs and, except for MC, they are all significant at 5 per cent level or better. The correlation matrix also facilitates an examination of the sample correlation between predictor variables to detect possible multicollinearity problem. When predictor variables are either perfectly or very highly correlated with absolute correlation coefficients being close to 1, the identification of true effect of predictors on dependent variables becomes problematic. We find that, although TT and MC are not as highly correlated with each other or with other predictors, there are some incidences of quite high and significant correlation coefficients among the rest of the predictor variables. For example, SQ shows a positive correlation of 0.775 with CS and 0.735 with CI.

Table 3.

Multicollinearity Diagnostics

	TT	MC	SQ	CS	CI	UB	SV
Tolerance	0.705	0.881	0.305	0.318	0.392	0.513	0.681
VIF	1.419	1.134	3.281	3.143	2.551	1.949	1.467

Source: Survey data, 2012

To investigate the multicollinearity issue further, we compute the variance inflation factor (VIF), and its reciprocal known as the tolerance, for each of the independent variables. These values are reported in Table 3. We find that VIF values range between a minimum of 1.134 (a maximum tolerance of 0.881) for MC and a maximum of 3.281 (a minimum tolerance of 0.305) for SQ. Therefore, judging by the popular rule of thumb of a maximum acceptable VIF of 10 (a minimum acceptable tolerance of 0.10), none of our predictor variables seems to pose any serious threat of multicollinearity to subsequent regression analysis.

Main Results

The results of estimating regression equation [1] are presented in Table 4. Although consistent with the results of simple correlation analysis presented in Table 1, all predictor variables have an expected sign of the relationship with MBPP, we find only TT, SQ and CS to have a statistically significant effect on customers’ MBPP at 5 per cent level or better. Specifically, the coefficient value on TT implies that, holding other effects constant, electricity customers’ MBPP increases by more than one-third of a day (0.368) for an increase in TT by 1 minute. Similarly, significant SQ and CS coefficients imply that the MBPP of a customer decreases by 2.076 and 1.972 days for each unit of increase in the SQ and CS indices, respectively. Reported standardized coefficients indicate that one standard deviation change in SQ has the highest effect on MBPP followed by CS and TT, respectively. Finally, the results in Table 4 suggest that the predictor variables together explain 48.80 per cent of the variance in customers’ MBPP (adjusted R-square), the significance of which is confirmed by the overall test of the goodness of model fit (F = 18.146, p < 0.001).

Though the results in Table 4 show that the model is a good fit for the dataset, only three predictor variables out of seven appear to have a significant influence on MBPP. Therefore, to obtain a more parsimonious presentation of the model of customers’ bill payment behaviour, we carry out a stepwise multiple regression analysis. The results of which are presented in Table 5. Consistent with the unrestricted model, we find that TT, SQ and CS retain their significance in explaining the MBPP of electricity customers. We observe that the parsimonious model shows some marginal improvement in the model fit in terms of adjusted R-square (48.80 vs. 49.20 per cent). We also find that, consistent with our earlier results, a unit standard deviation change in SQ has the highest effect on customers’ MBPP. In contrast to our previous findings, however, the effect of TT on MBPP now dominates that of CS in terms of unit changes in their respective standard deviations.

Table 4.

Summarized Ordinary Least Square (OLS) Regression Results

	Unstandardized Coefficients				Standardized Coefficients
	B	Standardized Error	t-stat	Significance	Beta
Constant	228.813	27.717	8.255	0.000**	–
TT	0.368	0.118	3.116	0.002**	0.237
MC	–1.509	1.08	–1.397	0.165	–0.095
SQ	–2.076	0.858	–2.419	0.017*	–0.279
CS	–1.972	0.937	–2.106	0.037*	–0.238
CI	–0.134	0.829	–0.162	0.872	–0.016
UB	–0.901	1.423	–0.633	0.528	–0.056
SV	–0.416	0.58	–0.718	0.474	–0.055
R-square		0.516	F-ratio		18.146
Adj. R-square		0.488	Significance		0.000
Durbin–Watson		1.424
No. of observations		127

Source: Survey data, 2012.

Note: ** and * indicate significance at 1 and 5 per cent level of significance, respectively.

Table 5.

Summarized Stepwise Multiple Regressions Results

	Unstandardized Coefficients				Standardized Coefficients
	B	Standardized Error	t-stat	Significance	Beta
Constant	203.448	21.553	9.439	0.000**	–
TT	0.435	0.108	4.032	0.000**	0.280
SQ	–2.375	0.770	–3.087	0.003*	–0.320
CS	–2.199	0.832	–2.642	0.009*	–0.265
R-square		0.504	F-ratio		41.731
Adj. R-square		0.492	Significance		0.000
Durbin–Watson		1.389
No. of observation		127

Source: Survey data, 2012.

Note: ** and * indicate significance at 1 and 5 per cent level of significance, respectively.

Finally, as part of an unreported exercise, we have also checked the robustness of our estimates in Table 5 to the effect of potentially extreme observations in our sample using two frequently used robust estimation techniques—the M-estimation method and median regression⁵. Although we have checked for the legitimacy of all seemingly extreme data points in our sample, robust regressions may still help us assess the efficiency of our point estimates especially in the face of wide variations in the values of MBPP reported in Table 1. In comparison to results reported in Table 5, we find that robust estimations lead to some deterioration in the point estimates of only SQ (–1.274 in median regression and 1.450 in M-estimation) and the overall model fit (pseudo R-square of 31.75 per cent in median regression and R-square of 32.25 per cent in M-estimation). More importantly, however, neither of the robust estimation techniques alters the signs and statistical significance of TT, SQ and CS in influencing customers’ MBPP.

Discussion of Results

Consistent with the conceptual framework and the a priori expectation, our regression exercises suggest that TT has a positive relationship with the MBPP of customers. This shows that the MBPP increases each time ECG customers take extra minute to pay their bills. The obvious implication of the finding is that, to ensure that customers of electricity utilities pay their bills promptly and on time, electricity companies must work towards reducing the TT. This can be achieved by, for example, employing an electronic bill presentment and payment system. Also, in the long run, customers could be gradually rolled onto the prepayment metering system as a way forward to minimizing the possibility of bill default. Currently, although revenue collection points can be found across communities in the GARG, they are regrettably inadequate. The long and widening queues at the electricity bill payment counters of the banks are certainly not helping the customers’ willingness to pay on time.

Congruent with Kayaga’s (2002) finding in relation to Ugandan water utility, we find SQ to have a negative relationship with the customers MBPP. This suggests that the electricity customers of Ghana put significant value on SQ and they will pay for their electricity bills if the electricity utility managers keep to their word of providing quality service. Indeed, the intermittent power supply, which is christened as ‘dum so’ locally, and the increasing delay in fixing faulty transformers are the very core concerns of customers. One might therefore argue that these are some of the key issues undermining SQ of electricity utilities, which in turn, serve as a disincentive for customers to pay their bills.

The results of this study show that CS is another significant predictor of the variance in the MBPP of electricity utility customers. Kayaga (2002) also reports similar findings in the context of water utility in Uganda. Customers, naturally, will not be willing to pay for services that they do not derive satisfaction from (Ahmed et al., 2011). Moreover, CS leads to positive behavioural intentions (Ahmed et al., 2011; Khan and Kadir, 2011; Nimako, 2012), which in our case is reflected in MBPP.

Contrary to the predictions underlying our conceptual framework (Figure 1 in the Annexure), the results of this study show that SV, MC, CI and UB are not statistically significant in explaining the variance in the MBPP of the electricity customers of Ghana. The insignificance of SV in our analysis does not rule out the possibility that the construct only indirectly affects MBPP through another consumer judgment variable such as CS. Although this finding is in contrast to the predicted direct effect in our analysis, it is consistent with other competing models in the literature that suggest only an indirect effect of SV on outcome variable (see, for example, Cronin et al., 2000). One possible explanation for the insignificance of UB, MC and CI variables could be that the electricity customers of Ghana are not much concerned about these issues in the face of generally unstable electricity supply and the lack of access to alternative sources of electricity. Another standard explanation for the insignificance of these variables could be that they might have been measured with errors due to ‘strategic bias’ on the part of respondents. A strategic bias often occurs when a respondent does not answer the question(s) truthfully and accurately (Evans, 1992).

Summary and Conclusion

The ECG has been going through waves of reform measures in order to break away with its chronic dependence on government subsidy and achieve sound operational and financial performances. However, like many other electricity utilities especially in developing countries, the ECG is finding it more difficult than ever to keep its customers current with their electricity bills. Due mainly to its poor revenue collection performance, it has become increasingly difficult for the company to maintain adequate service provision and meet an acceptable headline efficiency target (Hassanein and Khalifa, 2006). In this backdrop, the purpose of this study is to assess the effect of a set of behavioural and organizational factors on the bill payment behaviour based on a sample of electricity customers from the GARG.

We find that TT, SQ and CS are the three significant constructs affecting customers’ bill payment behaviour. More specifically, our empirical results suggest that, in order to expedite the payment of electricity bills, the electricity company should work towards reducing the TT at the collection points and improving upon SQ and CS. Based on these results, we argue that efforts to address the long and widening queues at the bill collection centres and frequent service interruptions may help greatly to improve the revenue collection performance of the electricity utility.

The results of the study also highlight that some other organizational and behavioural factors, such as UB issues, MC and the customers’ judgment of SV, that are often found to affect customers’ behavioural intentions in the literature may not be as relevant to the Ghanaian electricity customers in determining their willingness to pay bills on time. By the same token, however, one can legitimately argue that our results may not be representative of the utilities operating in another country. Further research can therefore be undertaken to test whether and to what extent the results presented in this article extend to the context of electricity utilities in other countries with similar socio-economic realities. The use of a different measure of customers’ behavioural intention in a similar research endeavour would be another interesting development.

Annexure

The framework in Figure 1 depicts the relationship between the predictor variables, TT, MC, CS, SQ, CI, UB, SV and BI, which in this study, is the bill payment behaviour of the ECG’s customers. It is our belief that the identified variables have some relationship with the behavioural intentions, that is, the bill payment behaviour of customers.

Figure 1.

The Conceptual Framework of the Study

Footnotes

Notes

Acknowledgements

The authors are grateful to the anonymous referees of the journal for their extremely useful suggestions to improve the quality of the article. Usual disclaimers apply.

References

Ahmed

Nawaz

M. M.

Usman

Shaukat

M. Z.

Ahmad

Iqbal

(2010). Impact of service quality on customer’s satisfaction: Empirical evidence from telecom sector of Pakistan. Interdisciplinary Journal of Contemporary Research in Business, 1(12), 98–113.

Aryeetey

Ahene

A. A.

(2005). Utilities regulations in Ghana. CRC Working Paper 111/2005, University of Manchester.

Awaworyi

S. K.

Danso

J. K.

(2010). Poverty reduction in Ghana: The role of microfinance & socio-economic education. Retrieved from sefakwame.files.wordpress.com/2011/03/poverty-reduction.pdf (accessed on 16 December 2012).

Bernstein

(1984). Company image and reality: A critique of corporate communications. Eastbourne: Rinehart and Winston Ltd.

B&FT (2013). ECG ‘switches-off’ bonds idea. My joy online: Retrieved from http://www.business.myjoyonline.com/pages/news/201301/100530.php (accessed on 27 January 2013).

Boakye

Nyieku

I. E.

(2010). Cost recovery analysis of Ghana Water Company: A Case study of Sekondi-Takoradi Metropolis. European Journal of Scientific Research, 46(1), 119–125.

Brealey

R. A.

Myers

S. C.

Allen

(2006). Corporate finance (ed. 8th). New York, USA: McGraw-Hill/Irwin.

Chipofya

Hoko

Gustaff

(2009). An assessment of effective metering and billing as a water demand management tool. The case of area 49 in Lilongue, Malawi. In Schouten

Hes

Hoko

(Eds), Innovative practices in the African water supply and sanitation sector (pp. 99–110). Stellenbosch: Sun Media Stellenbosch.

Choi

K-S.

Cho

W-H.

Lee

Kim

(2004). The relationships among quality, value, satisfaction and behavioural intention in health care provider choice: A South Korean Study. Journal of Business Research, 57(8), 913–921.

10.

Cooper

D. R.

Schindler

P. S.

(2002). Business research methods (8th edition). New York, McGraw-Hill/Irwin.

11.

Cronin

J. J.

Brady

M. K.

Hult

G. T. M.

(2000). Assessing the effects of quality, value and customer satisfaction on consumer behavioural intentions in service environment. Journal of Retailing, 76(2), 193–218.

12.

Cronin

J. J. Jr

Taylor

S. A.

(1992). Measuring service quality: a re-examination and extension. Journal of Marketing, 56(3), 55–68.

13.

Dissanayake

D. M. N. S.W.

(2012). Integrated communications, integrated marketing communications and corporate Reputation: Evidence from dell computer operations. Journal of Arts, Science and Commerce, III(1–3), 26–33.

14.

Evans

(1992). Paying the Piper: An overview of community financing of water and sanitation. Occasional Paper 18. The Hague: IRC International Water and Sanitation Centre.

15.

Grönroos

(1984). A service quality model and its marketing implications. European Journal of Marketing, 18(4), 36–44.

16.

Hassanein

A. G.

Khalifa

R. A.

(2006). Financial and operational performance indicators applied to public and private water and wastewater utilities. Engineering, Construction and Architectural Management, 14(5), 479–492.

17.

Hyndman

R. J.

Kostenko

A.V.

(2007). Minimum sample requirement for seasonal forecasting models. Foresight: The International Journal of Applied Forecasting, (6), 12–15.

18.

Kayaga

(2002). The influence of customer perceptions of urban utility water services on bill payment behaviour: Findings from Uganda. Unpublished PhD Thesis, Loughborough University.

19.

Kayaga

Calvert

Sansom

(2003). Paying for water services: Effects of household characteristics. Utilities Policy, 11(3) , 123–132.

20.

Kayaga

Franceys

Sansom

(2004). Bill payment behaviour in urban water services in Uganda. An empirical data from Uganda. Journal of Water Supply: Research and Technology—Aqua, 53(5), 339–349.

21.

Khan

Kadir

S. L. S. A.

(2011). The impact of perceived value dimension on satisfaction and behaviour intention: Young-adult consumers in banking industry. African Journal of Business Management, 5(16), 7055–7067.

22.

Kuruuzum

Koksal

C. D.

(2010). The impact of service quality on behavioural intentions in hospitality industry. International Journal of Business and Management Studies, 2(1), 9–15.

23.

Lai

W-T.

Chen

C-F.

(2010). Behavioural intentions of public transit passengers-the roles of service quality, perceived value, satisfaction and involvement. Transport Policy, 18(2), 318–325.

24.

Moss

H. K.

(2007). Improving service quality with the theory of constraints. Journal of Academy of Business and Economics, 7(3), 45–66.

25.

Maiyaki

A. A.

Mokhtar

S. S. M.

(2012). Determinants of customer behavioural responses in the Nigerian retail banks: Structural equation modelling approach. African Journal of Business Management, 6(4), 1652–1659.

26.

Mugabi

Kayaga

Smout

I. K.

(2007). Why water utility customers don’t pay their bills promptly. Water Utility Management International, 2(2), 17–19.

27.

Nimako

S. G.

(2012). Linking quality, satisfaction and behaviour intentions in Ghana’s mobile telecommunication industry. European Journal of Business and Management, 4(7), 1–17.

28.

Ogujor

E. A.

Otasowie

P. O.

(2010). The impact of the pre-paid meter on revenue generation in Nigeria. The Pacific Journal of Science and Technology, 11(1), 138–142.

29.

Oracle (2009). Serve prepaid customers without prepayment meters. Oracle White Paper (2009). Retrieved from: http://www.oracle.com/us/industries/utilities/046584.pdf (accessed on 10/03/2013).

30.

Parasuraman

Zeithaml

V. A.

Berry

L. L.

(1998). SERVQUAL: A multiple-item scale for measuring consumer perceptions of service quality. Journal of Retailing, 64(1), 12–40.

31.

Rajan

A. T.

(2002). Restructuring state owned power utilities: Qualitative observation from an Indian experience. Global Business Review, 13(1), 77–96.

32.

Sekaran

(2000). Research methods for business: A skills building approach (ed. 3rd). New York: John Wiley & Sons.

33.

Waldron

(2011). Normal best practice in revenue collection. ECLO Project 2010/235-011. Further institutional strengthening support to WWRO project EU-WATREG. Retrieved from: http://www.wwro-ks.org/English/Workshops_03_03_2011.html (accessed on 26/02/2013).

34.

WHO and UNICEF (2000). Global Water Supply and Sanitation Assessment 2000 Report. World Health Organization and United Nations Children’s Fund, Geneva and New York. Retrieved from: http://www.who.int/water_sanitation_health/Globassessment/GlasspdfTOC.htm (accessed on 03/03/2013).

35.

Ramanathan

(2012). Retail service quality, corporate image and behavioural Intentions: The mediating effects of customer satisfaction. The International Review of Retail, Distribution and Consumer Research, 122(5), 485–505.

36.

Zeithaml

V. A.

(1988). Consumer perceptions of price, quality and value: A means-end model and synthesis of evidence. Journal of Marketing, 52(3), 2–22.