Mode choice for long distance trips: Evidence from the Greater Accra Metropolitan Area of Ghana

Journal of Transport Geography 64 (2017) 150–157

Contents lists available at ScienceDirect

Journal of Transport Geography journal homepage: www.elsevier.com/locate/jtrangeo

Mode choice for long distance trips: Evidence from the Greater Accra Metropolitan Area of Ghana

MARK

Ernest Agyemang Department of Geography & Resource Development, School of Social Sciences College of Humanities, University of Ghana, P.O. Box LG 59, Legon, Accra, Ghana

1. Introduction Urban form in many cities of the developing world is generally monocentric, with high concentrations of important facilities and services at the city core. It has emerged that in Africa and South America, a third or more of formal jobs, for instance, are found in large concentrations in the city core (UN-Habitat, 2011; Cervero, 2013). Meanwhile, due to urban sprawl, the majority of urban dwellers who need to access these important facilities reside several kilometres away from the city core. Ghanaian cities are no exception to this type of urban form (Doan and Oduro, 2012; Owusu, 2013; Cobbinah and Amoako, 2014). The net result of this type of urban form is the high demand for travel. In other words, “travel demand does not derive its utility from the trip itself, but rather from the need to reach locations where activities take place” (van Wee, 2002, p. 260). Thus, as rational beings, trip makers make deliberate choices on the types of modes that can minimise their ‘travel resistances’ such as time and monetary costs (van Wee et al., 1997) while maximising their service quality. Consequently, modespecific attributes, such as speed, comfort, reliability, safety etc. which serve as proxies for what trip makers perceive as service quality are critical towards mode choice determination (Abane, 1993, 2011; AmohGyimah and Aidoo, 2013). Also, towards the determination of mode choice, it has been demonstrated that socio-economic and demographic circumstances of individual trip makers, including, age, gender, education, occupation type, and car ownership are very important (Rajamani et al., 2003; Schwanen et al., 2004; Dargay and Hanly, 2004; van Acker et al., 2007). The ever-dynamic nature of transport consumption and the multiplicity of factors which influence the selection of transport modes by trip makers call regular studies to update our knowledge on the theme. While the literature on transportation in Ghana does exist generally, as far as I am aware, limited attention has been given to trip behaviour and mode choice determining factors. Abane (1993) studied work-related trip patterns in Accra with a particular focus on formal sector workers who work in Accra's central business district (CBD). Later on, he studied travel behaviour in four major Ghanaian towns (Abane, 2011). The latter study is credited for providing a rare opportunity for inter-city comparative analyses of mode choice in the country. The former study is also praised for being a useful pioneering study on

travel behaviour in Ghana. However, the former study could not highlight the mode choice of informal sector workers who also visit Accra's CBD where his respondents were interviewed. In a fairly recent study on mode choice to the CBD in Kumasi, Ghana's second-largest city (Amoh-Gyimah and Aidoo, 2013), informal sector workers again were neglected. Thus, the need to fill the knowledge gap with the inclusion of informal sector workers is very important, at least for two major reasons. First, in terms of size, informal sector workers may constitute the majority of trip makers who regularly commute to the CBD. Second, dissimilarities between formal and informal sector workers with respect to trip purposes and specific destinations could make interesting revelations on mode choice in Ghana than we currently know. Consequently, this present paper has chosen to focus attention on mode choice for both informal and formal sector employees who visit the CBD of Accra. The paper also regresses personal characteristics of trip makers, including the car ownership status of trip makers, and other mode-specific attributes which potentially determines mode choice in the country's capital city. Specifically, the paper examines, first, the current residential locations where trip makers who visit the CBD to work or shop normally travel from; second, the transport modes used by these trip makers and third, key factors which influence their modal choices. In arriving at these specific objectives, the author took inspiration from Abane's (1993), (p.222) viewpoint that residential and office locations greatly “influenced to some extent the modes chosen for trips to work”. Also, knowledge on the current trip originating localities and mode choice in the city may provide evidence-based targeted interventions for policy and planning purposes. 2. Contextual issues This paper seeks to explore mode choice for long distance trips in the Greater Accra Metropolitan Area (GAMA). There is no universally accepted minimum physical distance threshold for categorising trips as either ‘long’, ‘medium’ or ‘short’. Different countries have adopted different distance thresholds for particular purposes in national travel surveys (Limtanakool et al., 2006). The author is unaware of any officially accepted distance thresholds in Ghana for such studies. Thus, the term ‘long distance trips’, as used in this paper, is defined arbitrarily to

E-mail address: [email protected]. http://dx.doi.org/10.1016/j.jtrangeo.2017.09.003 Received 22 May 2017; Received in revised form 3 August 2017; Accepted 11 September 2017 0966-6923/ © 2017 Elsevier Ltd. All rights reserved.

Journal of Transport Geography 64 (2017) 150–157

E. Agyemang

authorities, planners, and other stakeholders to anticipatively plan for the ease of mobility of urban dwellers.

include all trips bound for Accra's CBD which originate from within the much wider area of the Greater Accra region and its immediate functionally interlinked neighboring settlements, collectively referred to as GAMA (see Yankson et al., 2005; Oteng-Ababio et al., 2013). In terms of size, GAMA covers 1079 km2, or approximately 33% of the Greater Accra regional total land surface of 3245 km2 (Ghana Statistical Service, 2005). GAMA is home to about 3.7 million inhabitants, that is close to about 96% of the Greater Accra region's total population of around 4 million (Ghana Statistical Service, 2012). This makes GAMA the most densely populated and the single largest urban conurbation in Ghana. GAMA has sprawled considerably in recent years with its estimated built-up area soaring from around 133 km2 in the mid-1980s to above 344 km2 in 2000 (Owusu, 2013). The urbanisation of GAMA has been phenomenal, especially when one considers that its nucleus, simply referred to as ‘Accra’ was an inconspicuous ‘fish village’ at the end of the sixteenth century (Acquah, 1972). The transfer of the British colonial administration from Cape Coast to Christiansborg, Accra on 19th March 1877, among other factors, contributed to the urbanisation of the city. As the administrative capital, public amenities were provided in large numbers, especially in areas where the Europeans resided or had their headquarters, namely Victoriaborg (site of modern-day Bank of Ghana, the Ministries of state and the Independence Square) and Christiansborg (modern day Osu). In present-day Ghana, this geographical area is referred to as ‘Accra city proper’ (World Bank, 2015) or simply as ‘Accra Central’. Accra Central is bounded by the Accra Ring road and remains the most diversified economy in terms of the concentration of industries, administration, marketing, finance, insurance, transportation and tourism firms. These activities attract high levels of human and vehicular traffic. An estimated one million passenger trips are made daily in and out of Accra Central. Many of these trips, about 84% are made by public transport (Armah et al., 2010). Public transport is dominated by the informal paratransit services, popularly known as trotro,1 due to macro-economic reforms in the 1980s (Fouracre et al., 1994) and a government policy directive (see Ofosu-Dorte, 1992). Formal transit services are also provided, notably by the Metro Mass Transit (MMT) limited. In recent times, motor-taxi services, known as Okada is also fast becoming popular among commuters in GAMA (see Oteng-Ababio and Agyemang, 2012). For those who can afford, personal means of mobility may be used but for the poor and vulnerable, walking is the only option (UN-Habitat, 2013). GAMA's monocentric land use patterns and its inefficient transport system have negative implications for national growth and poverty reduction. In order to address this, Ghana has adopted a National Transport Policy that seeks to prioritise mass transportation ‘to move at least 80% of [urban] passengers’ (GoG, 2008, p.45). To achieve this objective, a quasi-public bus company known as Aayalolo bus rapid transit system has recently been incorporated. Since November 2016, the company has been running three services on the Amasaman-Tudu/ Accra Central corridor. However, the service is reportedly yet to attract optimum interest among commuters (Nyabor, 2017). Meanwhile, the Greater Accra Passenger Transport Executive which manages the Aayalolo BRT plans to roll out the Adenta-Tudu/Accra Central corridor in 2017 and subsequently, another service along the Tema Beach RoadTudu/Accra Central corridor. There are plans to replicate the BRT mass transit services in other major Ghanaian cities. Prior to the Aayalolo bus service, the Metro Mass Transit Limited, also a quasi-public bus service, has been providing intra and inter-urban mass transportation for Ghanaians. In the light of the above, a study of the current mobility patterns of urban dwellers, particularly those who are ‘forced’ to daily commute between the city core and the peripheries, is needed to assist city

3. Methodology The primary data which formed the basis of this present paper were derived in May 2014 through a cross-sectional survey of trip makers who are non-residents of Accra Central but who make regular trips to the area. In order to arrive at an objective choice of study sites within Accra Central, a two-day vox pop survey was conducted in March 2014. This involved asking respondents two simple questions: ‘what is the purpose of your trip today (work, shop, others)?’ and ‘which specific destination are you travelling to?’. On day one, drivers were systematically selected when they were observed to have completely stopped to observe the red light at traffic light intersections on four busy corridors leading to Accra Central. The shortness of the vox pop questions did not warrant the direct assistance of police officers to pull drivers over to participate in the interviews while ensuring smooth flow of traffic, as has been used in previous detailed surveys conducted in traffic (Odero, 1996; Popoola et al., 2013). The interview took two minutes, on average, per driver. In total, the survey lasted for a period of one hour at the respective study sites. On day two, users of public transport were also asked similar questions over a one-hour period through a convenience sampling technique at the popular public transport terminal called ‘Tema Station’ where commuters normally alight. Out of the 213 drivers and passengers who participated in the survey, the Makola Market (40.2%) and the Ministries Area (26.6%) emerged as the two leading trip destinations in Accra Central. Armed with this useful information, a structured questionnaire was administered in various study sites at the Makola Market area and in some selected government ministries, found at the Ministries Area, as annotated in Fig. 1. The structured questionnaire was sectioned into three parts; namely, the characteristics of trip generating facilities or activities in Accra Central, trip patterns, and the socio-economic characteristics of survey respondents. With respect to the characteristics of trip generating facilities, respondents were asked to indicate particular point(s) of activities that they visit while in Accra Central and how frequently they utilised these activities, as well as the reason(s) for the use of these points. Concerning the journey characteristics, the respondents were asked to indicate their main transport mode (i.e. the transport mode that covered the longest distance of their journeys) to various locations in Accra Central, the frequency of use of these modes, and mode-specific attributes, which influence their choice of modes. Finally, demographic and socio-economic characteristics of respondents, such as gender, age, educational attainment, household size, occupation type, and income as well as residential locations were elicited through the questionnaires. The Optimised Routes application in ArcGIS Network Analyst 10.4 extension was used to estimate the shortest possible commute distances along major corridors from these residential areas where trips originated and the CBD where such trips terminated. The shapefile-based road network dataset upon which the distances were calculated was obtained from CERSGIS, Legon. A major challenge observed was that the dataset had not been previously processed or formatted for use in the ArcGIS Network Analyst extension. Thus, a new network dataset had to be built in ArcCatalog to format the dataset to make it compatible with the network analyst software. For instance, a new field such as ‘Length (in metres)’ was calculated using the Calculate Geometry option and added to the dataset. Also, ‘One-way’ entries that specified road segments which obeyed the one-way rule and those that did not were created and added to the dataset. 4. Sampling framework

1 See Abane 2011, p.319 for the characteristics of trotro. This paratransit service is similar to the Matatus (Kenya), Combis (South Africa), Cars Rapides and Ndiaga Ndiaye (Senegal), Molues and Danfo (Nigeria), just to name a few.

A sample size of 400 was predetermined for the main survey. As was 151

Journal of Transport Geography 64 (2017) 150–157

E. Agyemang

Fig. 1. An annotated map showing GAMA and study sites in Accra Central.

seen in the vox pop survey, the number of Accra Central visitors who frequent the Makola Market were comparatively higher than those who visit the Ministries Area. Besides, the category of trip makers who visit the Makola Market was observed to be more than those who visited the Ministries Area. Therefore, as seen in Table 1, the total population sampled at the Makola Market (n = 320) is correspondingly higher than the total population sampled at the Ministries Area (n = 80). Table 1 demonstrates sampling sizes allocated to the various target population and the sampling techniques adopted in the generation of primary data.

Table 2 Summary statistics of sampled trip makers who visit the Makola Market and the Ministries Area of Accra Central. Variable

Elements

Makola Market (n = 268)

Ministries Area (n = 70)

Gender

Male Female < 20 years 20–45 years 45–60 years Above 60 years No formal Basic (Primary & Junior Secondary) Second cycle (Sen. Sec. & Voc./ Tech.) Tertiary (Polytechnic & University) < 5 km 5–10 km 11–15 km Above 15 km

9.3 90.7 3.0 71.3 21.6 4.1 16.0 26.5

70.0 30.0 0.0 68.6 30.0 1.4 0.0 0.0

46.7

22.9

10.8

77.1

7.1 45.2 16.0 31.7

10.0 25.7 11.4 52.9

Age

5. Data characteristics

Education

A total of 338 respondents, representing 85% of the total sample were found to reside in various localities in GAMA. The remaining 15% Table 1 Sampling framework adopted in the generation of primary data.

Trip distances

Main study site

Target population

Sample size

Sampling technique

Makola Market (n = 320)

Wholesalers of cosmetics, textiles & foodstuffs Retailers of cosmetics, textiles & foodstuffs Hawkers/itinerant traders Shoppers Formal sector employees Patrons of public services

40

Systematic

40

Systematic

40 200 50 30

Accidental Accidental Simple random Accidental

Ministries Area (n = 80)

were other trip makers who have travelled from all over Ghana to visit Accra Central. Accordingly, the analyses presented in this paper are based on only the 338 respondents who are known GAMA residents. As illustrated in Table 2, females (90.7%) dominated the informal trading activities performed at the Makola Market. The majority of informal traders and patrons (71.3%) are within the age bracket of between 20 152

Journal of Transport Geography 64 (2017) 150–157

E. Agyemang

homes to the Ministries. The use of formal bus services was also nil for this category of trip makers. The data was analysed based on the three most important transport modes. Females reported higher use of public collective transport, i.e. trotro and taxi while private car use was basically male dominated (Table 5). Trotro use appears to have an inverse relationship with age such that usage drops with increasing age of trip makers. For instance, while the majority of trotro users (87.5%) are below 20 years, the percent use of trotro reduces to about 58.3% as trip makers grow older. Instead, elderly trip makers who are above 60 years were found to commonly use taxis for their trips to Accra Central. Car use was reported high among the middle-aged working class. Trotro use is high among illiterates and also among trip makers with low education. As trip makers' education rise from the second cycle and beyond, their use of trotro dips significantly. Erudite trip makers, especially diploma and degree holders were by far the largest group of Accra Central visitors who undertake their journeys using private means of transport. In terms of the final trip destination, informal sector traders and shoppers alike in the Makola Market were found to make higher use of trotro and to a lesser extent, taxis. Comparatively, trip makers who frequent government offices in the Ministries Area made greater use of the car. With regards to commuting distances from homes of trip makers and their final destination, the data show that trotro use is highest (87%) within 5 km radius from Accra Central. Trotro use minimises with increasing distance such that at trip distances above 15 km, the percent use of trotro drops to 65.3%. The data did not reveal any marked variations in taxi use on account of trip distances except that taxi use is highest within 11 and 15 km radius from Accra Central. Car use was found to be highest (22%) for trip makers who reside beyond 15 km from the city centre.

Table 3 Origin-destination characteristics of sampled trip makers to Accra Central. Locality

Frequency (N)

Percentages (%)

Distance from Accra Central (km)

Status

La Dansoman Kaneshie Nima Odorkor Achimota Kwashieman Laterbiokorshie Bubuashie Chorkor Others Total

34 27 7 7 6 6 6 6 6 5 55 165

20.6 16.4 4.2 4.2 3.6 3.6 3.6 3.6 3.6 3.0 33.6 100

5.7 7.8 4.7 5.2 8.7 9.2 9.3 4.9 5.5 5.2 –

Urban Urban Urban Urban Urban Urban Urban Urban Urban Urban Urban

Kasoa Teshie Madina Nungua Adenta Tema Ashaiman Legon Weija Others Total

28 26 20 11 9 9 7 8 6 49 173

16.2 15 11.6 6.4 5.2 5.2 4 4.6 3.5 28.3 100

27.5 13 16.8 16.9 18.8 31.6 30.2 13.7 17 –

Peri-urban Peri-urban Peri-urban Peri-urban Peri-urban Peri-urban Peri-urban Peri-urban Peri-urban Peri-urban

and 45 years and the highest educational attainment is pegged at the second cycle. The data further shows that the majority of these informal traders and patrons reside from between 5 and 10 km away from the Makola Market. On the other hand, workers and visitors sampled at the Ministries Area were males (70%). None of the respondents were below the age of 20 years and those aged between 20 and 45 years dominated the sample (68.6%). A significant majority of trip makers who work or visit the Ministries Area (77.1%) had obtained university degrees and diplomas. This was followed by those who had obtained senior secondary school certificates (22.9%). Most of the sampled population in the Ministries Area (52.9%) was found to reside beyond 15 km from their workplaces or service points.

7.1. Determinants of mode choice among trip makers From a list of possible mode-specific attributes, surveyed trip makers were asked to identify all factors which applied to them when selecting their primary mode of transport. The results from a cross tabulation of mode specific attributes by the three most important modes are presented in Table 6. Among trotro users, the key attributes most considered were that trotros are simply available, highly affordable and presented the opportunity for passengers to alight at any place they desired. In addition, trotros had high departure rates. Trip makers who travelled by taxis considered, in order of importance, comfort, speed, convenience, and safety as paramount in their mode choice. Among car users, the key factors taken into consideration were convenience and high departure frequency as well as availability and safety, in that order.

6. Present trip originating localities in GAMA Table 3 summarises the major residential localities in GAMA where formal and informal sector workers, as well as shoppers and users of public service, reside. The Table further illustrates the trip distances from these localities to Accra Central where the Makola Market and the Ministries Area are located. The majority of trip makers (51.2%) reside in the peri-urban areas of the city. These are residential areas that lie beyond 10 km away from the city centre, similar to an earlier Ghanaian study (Cobbinah and Amoako, 2014). Notable among these peri-urban localities, in order of importance, are Kasoa, Teshie, and Madina located about 27.5 km, 13 km and 16.8 km respectively away from Accra Central.

7.2. Multinomial logit regression specification A Multinomial Logit (MNL) analysis was performed in order to test the main effects of selected predictor variables (socio-economic, demographic, spatial and mode-specific attributes) on the dependent variable (mode choice). The MNL has been used extensively in modelling urban travel demand in earlier studies (Ben-Akiva and Bowman, 2000; Dong et al., 2006; Abane, 1993, 2011). This is because the MNL as a generalised binary logit model is very efficient in modelling transportation data that are primarily categorical. The MNL is based on the Random Utility Theory which postulates that when provided with alternatives, an individual is most likely to select an alternative that is most likely to yield the highest utility (Dong et al., 2006). Utility, in this context, is a function of how trip makers perceive a particular mode's level or quality of service, measured among other things, by how comfortable, available, safe, convenient, fast and affordable the mode is. This utility measure is modelled against the backdrop of socioeconomic factors which are also hypothesized to influence modal use by

7. Mode choice for trips in GAMA The long commute distances between GAMA's peri-urban localities and Accra Central understandably results in a relatively high use of motorised transport modes (98.5%), compared to non-motorised transport modes which are a paltry 1.5%. In terms of modal share, the trotro dominates all other modes of transport with over 50% of all trips entering into Accra Central, irrespective of the trip purpose, being made in a trotro (Table 4). In addition, shared taxis and cars emerged as the second most commonly used transport modes by trip makers in the informal and formal sectors respectively. Significantly, none of the interviewed trip makers in the formal sector walked or biked from their 153

Journal of Transport Geography 64 (2017) 150–157

E. Agyemang

Table 4 Modal split of journeys to Accra Central (%). Transport mode

Category

Makola Market (n = 268)

Ministries Area (n = 70)

Walking Bicycle Motorcycle Trotro Taxi Private car Bus Total

Non-motorised/personal means Non-motorised/personal means Motorised/personal means Motorised/public transport Motorised/public transport Motorised/personal means Motorised/public transport

1.5 0.4 0.4 80.6 14.2 2.5 0.4 100

0 0 5.7 51.4 5.7 37.1 0 100

Table 5 Sample characteristics by mode of transport to Accra Central (%). Variable

Trotro

Gender Female Male Age group (years) < 20 20–45 45–60 Above 60 Educational status No formal education Basic (Primary & Jnr. Sec.) Second cycle (Snr. Sec. & Voc.,/Tech.) Tertiary (Polytechnic & University) Trip destination Makola Market Ministries Area Trip distances < 5 km 5–10 km 11–15 km Above 15 km

Taxi

Car

Table 7 Type and description of variables used in the Multinomial Logit Model on mode choice. Total sample size (n)

79.2 69.1

15.8 1.5

5.0 29.4

259 68

87.5 82.6 62.3 58.3

12.5 10.0 16.9 41.7

0.0 7.4 20.8 0.0

8 230 77 12

97.6 98.5 85.5 34.5

2.4 1.5 6.5 38.3

0.0 0.0 8.0 27.2

41 67 138 81

82.8 54.5

14.6 6.1

2.7 39.4

261 66

87.0 84.7 79.6 65.3

8.7 13.1 14.3 12.7

4.3 2.2 6.1 22.0

23 137 49 118

Variable name

Trotro (N = 252)

Taxi (42)

Car (N = 33)

Availability Affordability Convenience (i.e. alighting at place of choice) Frequent departure Safety Long distance Comfort Speed Evading traffic congestion Least polluting Lack of transport options Short distance

89.3 82.1 79.4

85.7 50.0 95.2

97 75 100

78.2 65.9 62.7 62.3 62.3 56.0 31.7 20.2 18.3

85.7 90.5 69.0 100 95.2 38.1 45.2 11.9 14.3

100 90.9 30.3 78.8 87.9 81.8 72.7 15.2 9.1

quality Avail Comf Conv

4 5 6 7 8 9 10 11 12

Afford Fast Freq Traffic Safe Pollution Dist_far Dist_near Options

Socio-spatial variables 13 Gender 14 Age 15 Educ 16 House 17 Occup

Table 6 Mode-specific attributes which influence mode choice (%). Variable

Service 1 2 3

18 19

Own car Dist

20

Dur

Description

Most available; 1 if Agree, 0 if Disagree Most comfortable; 1 if Agree, 0 if Disagree Convenience in alighting at location of choice; 1 if Agree, 0 if Disagree Most affordable; 1 if Agree, 0 if Disagree Fastest; 1 if Agree, 0 if Disagree Frequently departs; 1 if Agree, 0 if Disagree Easily dodges traffic; 1 if Agree, 0 if Disagree Safest; 1 if Agree, 0 if Disagree Very polluting; 1 if Agree, 0 if Disagree Distance is too far; 1 if Agree, 0 if Disagree Distance is too near; 1 if Agree, 0 if Disagree No other transport options; 1 if Agree, 0 if Disagree Sex; 1 if female, 0 if male Age of respondent Years of formal education Household size; 1 if 1–5, 2 if 6–10, 3 if 11–15, 4 if > 15 Occupation type; 1 if unemployed, 2 if self-employed, 3 if salaried/gov't, 4 if other Effect of car ownership on mode choice; 1 if yes, 0 if no One-way trip distance (Km); 1 if < 5, 2 if 5–10, 3 if 11–15, 4 if > 15 Trip duration

Mode-specific variables 21 Trotro 22 Taxi 23 Car Reference category

next stage of analysis. Socio-economic and spatial factors including gender, age, education status, occupation type, car ownership status and trip distances which were found in earlier analyses to influence mode choice were included in the model. Also, mode-specific attributes such as comfort, availability, convenience, and affordability which are known to influence individual trip maker's perception of service quality were all included in the model (Table 7). The Independence of Irrelevant Alternatives (IIA) which simply assumes that adding or deleting alternative outcome categories does not affect the odds among remaining outcomes in the MNL model was tested using the Hausman test type (Hausman and McFadden, 1984). This was performed using the STATA 13.1 software. The output from Hausman, with a Chi-square statistic (25.50) and a Likelihood Ratio Chi-square statistic (0.06) confirms no systematic change in the coefficients when the base category excluded one of the outcomes from the model. This means that the IIA assumption is validated. Again, as illustrated in Table 8, the MNL model is statistically robust and shows a good fit with a Pearson Chi-square value of 240.20 which was significant at 99% confidence level. A Nagelkerke R(Abane, 2011) of 0.76 shows that the model correctly predicts 76% of the variance in mode choice among long distance trip makers in GAMA. In addition, the dataset used for the model did not over-disperse as seen in the deviance chi-square value which was not significant.

trip makers. Following after Abane (2011), the utility measure, including the co-efficient and odd ratios, as well as the statistical significance or otherwise of the variables is illustrated in the formula:

Pjq = 1 M ∑ fi (X q )q, where Pjq represent the probability of q selecting mode j; Xq is a set of variables influencing mode choice; M represents the sample population of the individuals; and fi the choice function. The MNL model parameters were estimated through an iterative maximum-likelihood algorithm. This first stage of the analysis was to identify statistically significant variables with which to carry out the 154

Journal of Transport Geography 64 (2017) 150–157

E. Agyemang

two main socio-economic variables that were found to be significant predictors of mode choice. Similarly, the perception that taxi drivers can re-route through secondary road networks that are less congested was the only modespecific attribute that was found to be a significant predictor of choosing a taxi among survey respondents. A unit increase in the years of formal education of a trip maker decreases his/her odds for travelling by taxi by 47%. This means that compared to the choice of a car, a trip maker of higher educational status was 47% less likely to choose a taxi. Trip makers who have their personal cars or have the possibility of ridesharing with other car owners have a 98% decreased odds for commuting by taxis. Trip makers who perceive the use of cars to be more flexible in terms of the possibility of dodging congested road corridors were less likely to travel by taxis by as much as 94%. Finally, the Table further demonstrates that the car ownership status of GAMA trip makers has a statistically significant effect on modal use.

Table 8 Multinomial Logit analysis estimating the effects of the predictors on modal choice among trip makers. B (SE)

Trotro Intercept Socio-economic Age (years) Education (years) Car ownership Yes No Service quality Convenience Disagree Agree Departure frequency Disagree Agree Pollution Agree

Wald

Socio-economic Education (years) Car ownership Yes No Service quality Traffic Disagree Agree

95% CI for odds ratio Lower

Upper

59.058 (6.972)***

71.759

− 0.104 (0.051)* − 0.72 (0.229)**

4.131

0.901

0.815

0.996

9.897

0.487

0.311

0.762

− 4.471 (1.354)**

10.903

0.011

0.001

0.163

1.00

8. Discussion and conclusions − 15.231 (1.023)***

221.751

2.43E-07

3.27E-08

1.80E-06

1.53E-09

0.4.169E-008

0.006

0.784

The purpose of the paper was threefold –to identify localities where trips which end at the city core originate; modal choice of trip makers and key explanatory variables which influence mode choice in the Greater Accra Metropolitan Area. The study revealed that a majority of trips headed towards Accra Central originate several kilometres away in the peri-urban areas of the city, confirming Abane's (1993) study. Two factors may help explain this. First, land use patterns in most developing cities, including Accra, are monocentric in nature with all the important trip generating public and private facilities and services which offer job, shopping, and opportunities to access government services concentrated in the city core (UN-Habitat, 2011; Cervero, 2013). Second, uncontrolled sprawl-like development in GAMA has resulted in the creation of dormitory towns located several kilometres from the city core. Consequently, functional interlinkages between the city centre and its peri-urban areas emerge as commuters who are resident in these dormitory towns are encouraged to regularly visit the city centre to work, shop or access public services (see Abane, 1993; Yeboah, 2003; Owusu, 2013; Cobbinah and Amoako, 2014). The sprawl of Accra vis-àvis its monocentric land use pattern has often resulted in longer travel distances. The high use of motorised transport modes, as opposed to non-motorised modes, as was found in this study is quite understandable and in sync with earlier studies that have consistently found a strong relationship between long distance trips and the high use of public transport and private cars (Boapeah, 2005; van Acker et al., 2007; Cervero, 2013; Cobbinah and Amoako, 2014). Specifically, trotro paratransit services are commonly utilised for commuting in the study area, lending credence to earlier studies (Lartey, 1977; Adarkwa, 1991; Abane, 1993, 2011; Amoh-Gyimah and Aidoo, 2013). Formal bus services were seldom used by the surveyed trip makers who visited Accra Central, corroborating an earlier study (World Bank, 2015). Respondents were specifically asked to indicate their use or otherwise of the intra-urban bus services being operated by the Metro Mass Transit limited (MMTL) when visiting Accra Central, for two specific reasons. First, government policy has promoted MMTL with technical, logistical and financial support to make the company run efficiently and attract more commuters in GAMA. Second, the MMTL has a major bus terminal located at Kimbu, right at the heart of Accra Central. However, this study found evidence to support an earlier study (Abane, 2011, p. 321), the “company attracts only a small proportion of the travelling public because of operational problems”. Compared to male trip makers, females were by far the largest group of trip makers who visit Accra Central by public transport, i.e. trotro and taxis, albeit with an interesting variation. While trotros are generally used by female informal retail traders and shoppers, the youth as well as trip makers with little or no education, taxi users, on the other hand, were found to be female informal wholesale traders and

1.00

− 18.647 (0.844)***

488.49

7.98E-09 1.00

− 2.673 (1.24)*

4.649

Disagree Taxi Intercept

Odds ratio

0.069 1.00

40.176 (2222.562)

0

− 0.633 (0.237)**

7.16

0.531

0.334

0.844

− 6.262 (3.259)*

3.693

0.002

3.21E-06

1.132

0.004

0.802

1.00

− 2.863 (1.348)*

4.509

0.057 1.00

Note: R = 0.76 (Nagelkerke); Model fit χ2 = 240.202***; Goodness-of-Fit (df = 540), Deviance = 157.158 (df = 540); −2 Log χ2 = 253.612 Likelihood = 397.361; *p < 0.05; **p < 0.01; ***p < 0.001.; Car is used as reference category. Also, the reference categories for the various categorical independent variables have been italicised in the model. 2

Table 8 shows only statistically significant variables as non-significant variables in the initial model (see Table 7) were dropped (similar to Abane, 1993). For instance, it is seen in Table 8 that age, education and ownership/access to personal means of transport were the main socio-economic variables that were found to be significant predictors of choosing trotro among long distance trip makers. Similarly, perceptions of convenience, the frequency of trip departures and less pollution/emission of smoke (i.e. an indication of better vehicular performance) were found to be key predictors of the mode choice among trotro users. Compared to car use, Table 8, for instance, indicates that a unit increase in a person's age and years of formal education result in some decreased odds for commuting to Accra Central in a trotro by 10% and 51% respectively. Also, the odds of choosing trotro among commuters who either personally own cars or have the possibility to rideshare in someone else's car when travelling to Accra Central decrease by as much as 99%. Among taxi users, education and ownership of personal car were the 155

Journal of Transport Geography 64 (2017) 150–157

E. Agyemang

which emerged was that obvious factors such as safety and comfort were not considered so high compared to convenience and frequent departure rates. Perhaps, in Ghana where road accidents are an everyday occurrence, trip makers who use personal means of transport may perceive themselves as equally vulnerable to accidents like other any other road users. In a nutshell, the crowding of GAMA's road space with low occupancy vehicles has resulted in chaotic traffic congestion with its attendant socio-economic costs. These include traffic accidents, usually involving vulnerable road users, loss of productive hours, air and noise pollution. In addressing these challenges, the Ghanaian government has vowed to prioritise mass transportation. The study finds this policy initiative laudable and further recommend that mass transit services could be operated successfully between the city core and Kasoa, Teshie, and Madina. This in view of the high number of Accra Central trips which originate from these three specific peri-urban localities. It must be noted that the Kasoa-Tudu/Accra Central pilot BRT system has been shelved temporarily, according to government sources, due to project under-costing and insufficient funding. However, in addition to the ongoing Aayalolo BRT services on the Amasaman-Tudu/Accra Central corridor, the Adenta-Tudu/Accra Central corridor (which passes through Madina township) and the Tema Beach Road-Tudu/Accra Central corridor (which passes through Teshie township) have been earmarked for mass transit. If and when they become operational, this paper posits is that the high population of Accra Central trip makers who reside in these localities will produce the requisite customer base for the transit service to be successful. Equally important factors such as making the transit service highly available, affordable and comfortable will also promote patronage of the transit service. This targeted intervention will not only address the mobility challenges of a majority of urban commuters but will also ensure that transport indeed becomes the facilitator of economic growth and progress, as envisioned Ghana's (2008) National Transport Policy, the African Union Commission's Agenda 2063 and indeed the United Nations' Sustainable Development Goals (2016–2030). This paper did not examine the role income as a factor plays on mode choice in GAMA, even though other studies have found empirical evidence that links income to mode choice. In the survey, respondents were asked to indicate their incomes. However, the data on income was found to be inaccurate and therefore, inappropriate to report on. Thus, going forward, it is recommended that future studies adopt Abane's (1993) approach of estimating incomes of survey respondents based on their expenditure, rather than on their stated incomes, as was initially used in the conduct of this study.

shoppers, older trip makers (above 60 years) and those who have received higher levels of education. This is partly explained by the fact that while taxis fares may be relatively higher trotro fares, the turnover from large-scale wholesale trading activities and the possibility of hiring taxis to transport bulky products make taxi use worthwhile among female wholesalers. Also, personal conversations the author had with some market women at Makola revealed that some of them employed their daughters and wards in the market after they had completed tertiary education and have had no luck in securing formal sector employment. This segment of Accra Central trip makers may have a higher demand for travelling by taxis or even personal cars if the family owns one. The high use of taxis among older trip makers may be explained by their inability to cope with challenges associated with using trotros, including, for instance, rush hour competition with young able-bodied trip makers for seating space in the trotros. Rising levels of education appear to have an attenuating effect on trotro use by trip makers. Higher education may translate into better jobs and better working conditions, including the possibility of having access to official cars (for those in the formal sector) or purchasing one's personal means of mobility. This may help explain why males, particularly those aged between 45 and 60 years, dominate when it comes to car use, especially for trips made to the Ministries Area. Due to gender and educational disparities, many of the directors, deputy directors and senior administrative staff of the various ministries, departments, and agencies of the state are males. The above findings support earlier studies that found that age (Rajamani et al., 2003; Dargay and Hanly, 2004; Schwanen et al., 2004; van Acker et al., 2007) and levels of education which invariably determines a person's occupation type, income and car ownership status greatly influenced modal use (Abane, 1993, 2011; AmohGyimah and Aidoo, 2013). It was revealing to note that trip makers who make work-related trips into Accra Central resort to faster means of transport such as private cars and taxis, as the trotros are usually bogged down with delays due to frequent stops. This highlights an earlier study which found that travel time was considered paramount among government employees in Kumasi (Amoh-Gyimah and Aidoo, 2013) and also in Accra (Abane, 1993). In terms of commute distance on mode choice, the study found that collective public transport was very popular among trip makers who lived closer to the city centre and up to about 15 km radius from the city centre. Trotros and taxis were used mostly for relatively shorter distances. This is understandable given that at longer distances, trip makers will be required to pay comparatively higher fares and have their trips broken into several legs before reaching their intended destinations. Therefore, for trips made from over 15 km radius away from the city centre, cars use was found to very high among surveyed trip makers. The high use of cars for such long distance trips could be explained by the large number of middle and high-income trip makers who find it convenient to live in the emerging gated communities dotted across the peri-urban areas with all its health and aesthetic benefits, but also the possibility of commuting back to the city centre to work, shop or access government services on improved transportation networks, as have been found in earlier studies (Grant and Yankson, 2003; Bryceson, 2006; Doan and Oduro, 2012). With respect to the effect of mode specific characteristics on mode choice, trotros are not only perceived to be highly available, but they are relatively affordable, as found earlier (Addo, 2002; Agyemang, 2015). These two factors alone may explain the popularity of trotro paratransit services in the study area. Interestingly, comfort as an attribute was ranked higher among taxi users who frequent Accra Central. In recent times, the keen observer will notice the use of newly-registered vehicles for either the traditional taxi services or the Uber taxi services in GAMA, particularly during daytime trips. Perhaps the idea of being driven around in one of these newer and possibly comfortable cars may explain why all surveyed trip makers thought taxis were preferred comfortable option to travel with. Among car users, an unusual finding

Acknowledgement The author wishes to acknowledge the support of the Danish International Development Agency (DANIDA) (BSUPHD-UG) in the form of a PhD Research Grant to successfully complete the research whose findings are published here. The author further wishes to acknowledge the advice and moral support he received from Senior Researchers Pia Frederiksen and Anne Jensen of the Department of Environmental Science, Arhus University, Roskilde, Denmark in putting this paper together. References Abane, A.M., 1993. Mode choice for the journey to work among formal sector employees in Accra, Ghana. J. Transp. Geogr. 1 (4), 157–168. Abane, A.M., 2011. Travel behaviour in Ghana: empirical observations from four metropolitan areas. J. Transp. Geogr. 19, 313–322. http://dx.doi.org/10.1016/j. jtrangeo.2010.03.002. Acquah, I., 1972. Accra Survey: A Social Survey of the Capital of Ghana, Formerly Called the Gold Coast, Undertaken for the West African Institute of Social and Economic Research, 1953–1956. Ghana Universities Press. Adarkwa, K., 1991. Urban consumer needs in the transport sector and government policy in Ghana. J. Adv. Transp. 25 (1), 42–53.

156

Journal of Transport Geography 64 (2017) 150–157

E. Agyemang

Nyabor, J., 2017. Aayalolo Buses to ply Adenta-Accra route – Minister. http:// citifmonline.com/2017/03/22/aayalolo-buses-to-ply-adenta-accra-route-minister/ (Accessed date 18/7/2017). Odero, W., 1996. Conducting a roadside survey of drivers in Kenya: methods and experiences. Health Policy Plan. 11 (3), 329–331 (Oxford University Press). Ofosu-Dorte, D., 1992. Options for Using Mass Transportation Facilities to Reduce Vehicular Fuel Consumption and Traffic Congestion in Urban Areas. (Report for the Ministry of Energy). Oteng-Ababio, M., Agyemang, E., 2012. Virtue out of necessity? Urbanisation, urban growth and Okada services in Accra, Ghana. J. Geogr. Geol. 4 (1), 148–162. Oteng-Ababio, M., Melara Arguello, J.E., Gabbay, O., 2013. Solid waste management in African cities: sorting the facts from the fads in Accra, Ghana. Habitat Int. 39, 96–104. http://dx.doi.org/10.1016/j.habitatint.2012.10.010. Owusu, G., 2013. Coping with urban sprawl: a critical discussion of the urban containment strategy in a developing country city, Accra. J. Urban. 1 (26), 1–17 (ISSN 17230993). Popoola, S.O., Oluwadiya, K.S., Kortor, J.N., Denen-Akaa, P., Onyemaechi, N.O.C., 2013. Compliance with seat belt use in Makurdi, Nigeria: an observational study. Ann. Med. Health Sci. Res. 3 (3), 427–432. http://dx.doi.org/10.4103/2141-9248.117950. Rajamani, J., Bhat, C.R., Handy, S., Knaap, G., Song, Y., 2003. Assessing the impact of urban form measures in non-work trip mode choice after controlling for demographic and level-of-service effects. In: Paper presented at the Transportation Research Board Annual Meeting, Washington, DC, January 2003. Schwanen, T., Dieleman, F.M., Dijst, M., 2004. The impact of metropolitan structure on commute behavior in the Netherlands: a multilevel approach. Growth Change 35 (3), 304–333. UN-Habitat, 2011. Cities and climate change. Global report on human settlements 2011. In: United Nations Human Settlements Programme. Earthscan, London, Washington D.C. UN-Habitat, 2013. Planning and Design for Sustainable Urban Mobility: Global Report on Human Settlements. Routledge, New York. Van Acker, V., Witlox, F., Van Wee, B., 2007. The effects of the land use system on travel behaviour: a structural equation modeling approach. Transp. Plan. Technol. 30 (4), 331–353. Van Wee, B., 2002. Land use and transport: research and policy challenges. J. Transp. Geogr. 10, 259–271. Van Wee, B., Baker, T.D., Van der Hoorn, T., 1997. Office suites suit the railways: the effects of office locations to public transport nodal points on passenger transport. In: PTRC European Transport Forum, Annual Meeting, Proceedings of Seminar E: Transportation Planning Methods, 1. World Bank, 2015. Rising through Cities in Ghana: Ghana Urbanization Review Overview Report. Indiana University Press, Washington DC. Yankson, P.W.K., Kofie, R.Y., Moller-Jensen, L., 2005. Monitoring urban growth: urbanization of the fringe areas of Accra. Bull. Ghana Geogr. Assoc. (24), 1–12. Yeboah, I.E.A., 2003. Demographic and housing aspects of structural adjustment and emerging urban form in Accra, Ghana. In: Africa Today. EBSCO Publishing, pp. 107–119.

Addo, S.T., 2002. Provision of Urban Transport Services in Accra. SSATP Annual Conference and Stakeholders' Meeting. http://www.ssatp.org/en/search/node/ Urban%20Transport%20Services%20in%20Accra (Date accessed on 01/06/2015). Agyemang, E., 2015. The bus rapid transit system in the Greater Accra Metropolitan Area, Ghana: looking back to look forward. J. Geogr. 69 (1), 28–37. http://dx.doi.org/10. 1080/00291951.2014.992808. Amoh-Gyimah, R., Aidoo, E.N., 2013. Mode of transport to work by government employees in the Kumasi metropolis, Ghana. J. Transp. Geogr. 31, 35–43. http://dx.doi. org/10.1016/j.jtrangeo.2013.05.008. Armah, F.A., Yawson, D.O., Pappoe, A.A.N.M., 2010. A systems dynamics approach to explore traffic congestion and air pollution link in the city of Accra, Ghana. Sustainability 2, 252–265. Ben-Akiva, M.E., Bowman, J.L., 2000. Activity-based disaggregate travel demand model system with activity schedules. Transp. Res. A 35, 1–28. Boapeah, S.N., 2005. The Dynamics of the Growth and Spatial Distribution of Population in Kumasi Metropolis. vol. 9. Bi-annual Journal of the Building and Road research Institute (CSIR), Ghana, pp. 27–33. Bryceson, D.F., 2006. Fragile cities: fundamentals of urban life in east and southern Africa. In: Bryceson, D.F., Potts, D. (Eds.), African Urban Economies: Viability, Vitality, or Vitiation. Palgrave Macmillan, Basingstoke. Cervero, R., 2013. Linking urban transport and land use in developing countries. J. Transp. Land Use 6 (1), 7–24. http://dx.doi.org/10.5198/jtlu.v1.425. Cobbinah, P.B., Amoako, C., 2014. Urban sprawl and the loss of peri-urban land in Kumasi, Ghana. Int. Sch. Sci. Res. Innov. 8 (1), 313–322. Dargay, J., Hanly, M., 2004. Land Use and Mobility. Paper Presented at the World Conference on Transport Research, Istanbul, Turkey. Doan, P., Oduro, C.Y., 2012. Patterns of population growth in peri-urban Accra, Ghana. Int. J. Urban Reg. Res. 36 (6), 1306–1325. http://dx.doi.org/10.1111/j.1468-2427. 2011.01075.x. Dong, X., Ben-Akiva, M., Bowman, J.L., Walker, J.L., 2006. Moving from trip-based to activity-based measures of accessibility. Transp. Res. A 40, 163–180. Fouracre, P.R., Kwakye, E.A., Okyere, J.N., Silcock, D.T., 1994. Public transport in Ghanaian cities: a case of union power. Transp. Rev. 14 (1), 45–61. Ghana Statistical Service (GSS), 2005. 2000 population and housing census: greater Accra region. Analysis of district data and implications for planning. Republic of Ghana. Ghana Statistical Service, 2012. 2010 population and housing census: summary of report of final results. The Republic of Ghana. Government of Ghana (GoG), 2008. National Transport Policy. Ministry of Transport, Accra. Grant, R., Yankson, P., 2003. City profile: Accra. Cities 20 (1), 65–74. Hausman, J., McFadden, D., 1984. Specification tests for the multinomial logit model. Econometrica 52 (5), 1219–1240. http://www.jstor.org/stable/1910997. Lartey, E., 1977. Traffic within our cities. In: Proceedings of the Ghana Academy of Arts and Sciences. vol. xv. pp. 119–123. Limtanakool, N., Dijst, M., Schwanen, T., 2006. The influence of socioeconomic characteristics, land use and travel time considerations on mode choice for medium - and longer-distance trips. J. Transp. Geogr. 14, 327–341. http://dx.doi.org/10.1016/j. jtrangeo.2005.06.004.

157