Road crashes in Namibia: Challenges and opportunities for sustainable development

ABSTRACT

Within a broad framework for comprehensive advancement, the Sustainable Development Goals (SDGs) directly address the importance of safe transport to human development. We analysed some 29 000 crashes recorded in Namibia between 2012 and 2015 to identify connections between road safety and sustainable development there and, by extension, throughout Sub-Saharan Africa and other developing countries. The results indicate that the heavy reliance on transport services for basic mobility, especially in rural areas, contributes to negative development outcomes in terms of significant burden of injuries and death, many of which affect younger people of prime working age. The results also suggest that development pressures such as increased commercial traffic and more private vehicles increase the burden of roadway crashes. Many of the underlying factors contributing to these crashes are attributable to driving behaviours and safety cultures that must be addressed as part of programmatic socioeconomic development in order to mitigate the negative effects of transportation and more fully realise its potential to support sustainable development. We document additional ways in which the provision of safe transport is interrelated to successful attainment of the broader SDGs including those not explicitly addressing safe transport such as 4, 5, 8, 9, 16, and 17.

KEYWORDS:

• Namibia
• road safety
• sustainable development
• latent class analysis

1. Introduction

Over the course of the twentieth century, the global community began focusing its attention on the idea of international development, much of which centred around the concept of economic development. After important events such as the 1972 United Nations Conference on the Human Environment, and the 1992 first UN Conference on Environment and Development (UNCED) in 1992, the international development community begin to focus on the concept of ‘sustainability’. This momentum let to milestones such as Agenda 21 and the 1997 Kyoto Climate Agreement that further integrated the sustainability concept into the international development priorities, policies, and programmes (Allen & Thomas, 2000). In 2015, the global development community followed the lead of the United Nations Development Program (UNDP) as it introduced the Sustainable Development Goals (SDGs) to follow the original globally-focused Millennium Development Goals (MDGs). Building on its foundational pillars (social, environmental, economic), the UNDP set out 17 goals articulated though 169 targets by which progress towards each would be measured. The SDGs represented an integrated approach to development in which multiple, sometimes competing, efforts must be balanced to achieve comprehensive progress (Sachs, 2012; United Nations, 2015).

Within a broad framework for comprehensive advancement, the SDGs recognised safe transport as essential to development – whereas the original MGDs did not do so. In particular, SDGs 3 and 11 directly address the importance of safe transport to the human development endeavour as set out in the following targets:

• Target 3.6 – ‘By 2020 halve global deaths and injuries from road traffic accidents’

• Target 11.2 – By 2030, provide access to safe, affordable, accessible and sustainable transport systems for all, improving road safety, notably by expanding public transport, with special attention to the needs of those in vulnerable situations, women, children, persons with disabilities and older persons. (United Nations, 2015)

The language of these targets clearly emphasises the importance of safe transport in achieving sustainable development, partly in recognition of the social and economic burden associated with road deaths and injuries (Kopits & Cropper, 2005; Bishai et al., 2006; Azestop, 2010; Kudebong et al., 2011; World Bank, 2017). Indeed, Le Blanc (2015) notes that goals 3 and 11 are linked to eight and six other goals, respectively, further underscoring the interrelated nature of the SDGs. We present an analysis of road safety data from Namibia with a view to go beyond the typical analytical process of determining crash cause and effect, identifying countermeasures, etc. Rather, our aim is to identify connections between what is happening on the roads of Namibia and how that relates to development there and, by extension, elsewhere throughout the developing world.

2. Development context

It is well documented that transport plays an essential role in human development (e.g. Stewart, 1988; Turner & Kwakye, 1996; Gannon & Liu, 1997; Gakenheimer, 1999; Limao & Venables, 2001; van de Walle, 2002; Bryceson et al., 2008; Banjo et al., 2012; Laird & Mackie, 2014; Porter, 2014; Beuran et al., 2015; Martincus et al., 2017). Several recent high-level studies, however, have documented relationships between the global increase in motorised travel and road crashes (Sperling & Claussen, 2004; IHME, 2013; Lamont, 2013; GRSF, 2014; World Bank, 2017). The World Health Organization (WHO) recently reported that some 1.3 million people die in traffic crashes each year (WHO, 2018). Thus, the need for Target 3.6 and the explicit reference to ‘safe’ transport in Target 11.2 within the SDGs.

2.1. Sub-Saharan Africa

Globally, the burden of death and injury caused by crashes is disproportionally borne by low- and middle-income countries (LMICs). LMICs bear some 90 percent of global road deaths and sub-Saharan Africa (SSA) is particularly impacted (Nantulya & Reich, 2002; Nantulya & Reich, 2003; Ameratunga et al., 2006; Lagarde, 2007; Chen, 2010; IHME, 2013; Onywera & Blanchard, 2013; GRSF, 2014; Wegman, 2017). In many cases, young people bear the burden of crashes most (Adesunkanmi et al., 2000; Nantulya & Reich, 2002; Okeniyi et al., 2005; Ackaah, 2010; Burrows et al., 2010; Kiser et al., 2012; Zimmerman et al., 2012; Alonge & Hyder, 2013) with an unfortunate impact on otherwise economically productive youth (e.g. Odero et al., 1997; Odero et al., 2003; Naddumba, 2004; Hyder et al., 2008; Nakitto et al., 2008). Indeed, the WHO (2018) reported that crashes are the leading cause of death among people 15–29 years old and thus have a significant negative impact on the human capital and potential of developing economies. In recognition of the broad challenges of sustainable development in Africa, Njenga & Odero (2014) documented the relevance of transport to meeting the SDGs, noting the particular importance of providing safe mobility and accessibility. Njenga & Odero (2014) mapped the relevance of transport to all of the SGDs and individual targets defining each. In doing so, they emphasised the explicit connections to targets 3.6 and 11.2 noted above. They went considerably farther, however, to describe how the provision of safe transport is essential to supporting socioeconomic activity through access to markets, jobs, education, etc. that reduce poverty and hunger. They report how transport is essential to accessing healthcare and that accessing socioeconomic activity through transport is essential to individual opportunities that support gender equality and community stability, of which contribute to the comprehensive sustainable development set out in the SDGs.

2.2. Namibia

The WHO estimates that the annual death rate due to crashes in Namibia is 30.4 per 100 000 persons. Indeed, from a traffic safety perspective, Namibia is one of the most dangerous countries in SSA where the average fatality rate across the region is 26.6 road deaths per 100 000 persons – by comparison the U.S. is around 12 and Norway is less than 3 annual road deaths per 100 000 persons (WHO, 2018).

Table 1 presents a snapshot of road safety across SSA compiled from the most recent WHO data (WHO, 2018). It can be seen that only eight (shaded in grey) of the countries listed have higher road deaths per 100 000 population than Namibia. Interestingly, the experience in North Africa is quite varied with Libya at 26.1 road deaths per 100 000 population (just at the average for the continent), Morocco at 19.6 and Egypt at 9.7.

Table 1. Comparison or road safety across sub-Saharan Africa (WHO, 2018).

Country	Road deaths per 100 000 pop.	% road deaths in cars and light duty vehicles (drivers/passengers)	Direction of 5-year trend in road deaths
Angola	23.6	27/33	Decreasing
Benin	27.5	8/9	Flat
Botswana	23.8	23/40	Increasing
Burkina Faso	30.5	No Data	Decreasing
Burundi	34.7	No Data	Decreasing
Cameroon	30.1	No Data	Increasing
Central African Republic	33.6	No Data	No Data
Chad	27.6	No Data	Flat
Comoros	26.5	13/52	Increasing
Congo	27.4	No Data	No Data
Côte d’Ivoire	23.6	4/7	Increasing
DRC	33.7	1/35	Increasing
Equatorial Guinea	24.6	No Data	No Data
Eritrea	25.3	10/26	Decreasing
Eswatini	26.9	24/30	Decreasing
Ethiopia	26.7	No Data	Increasing
Gabon	23.2	No Data	Decreasing
Gambia	29.7	No Data	Increasing
Ghana	24.9	5/7	Decreasing
Guinea	28.2	No Data	Increasing
Guinea-Bissau	31.1	No Data	Increasing
Kenya	27.8	12/25	Decreasing
Lesotho	28.9	No Data	Increasing
Liberia	35.9	No Data	No Data
Madagascar	28.6	No Data	Increasing
Malawi	31.0	6/25	Flat
Mali	23.1	16/12	Decreasing
Mauritania	24.7	No Data	Decreasing
Mauritius	13.7	9/8	Flat
Mozambique	30.1	No Data	Decreasing
Namibia	30.4	No Data	Increasing
Niger	26.2	No Data	Increasing
Nigeria	21.4	No Data	Decreasing
Rwanda	29.7	No Data	No Data
Sao Tome and Principe	27.5	No Data	Decreasing
Seychelles	15.9	20/27	Increasing
Senegal	23.4	No Data	Decreasing
Sierra Leone	No Data	No Data	No Data
Somalia	27.1	No Data	Increasing
South Africa	25.9	21/24	Decreasing
South Sudan	29.9	No Data	Decreasing
Sudan	25.7	No Data	Decreasing
Togo	29.2	No Data	Decreasing
Uganda	29.0	5/17	Increasing
Tanzania	29.2	8/30	Decreasing
Zambia	No Data	No Data	No Data
Zimbabwe	34.7	12/14	Increasing

The table also shows the percentage of road deaths among drivers and passengers of cars and light duty vehicles. While this does not tell the whole story, it is interesting to see the contracts across the continent. For example, some countries exhibit relatively high percentages of road deaths attributable to these modes (e.g. Angola, Botswana, South Africa). By contrast, countries in other parts of the continent where two- and three-wheelers are more prominent exhibit more deaths attributable to these modes.

The table also shows mixed results for whether the five-year trend in road deaths is increasing or decreasing – Namibia is one of the 17 nations with an increasing trend reported. Local sources report that the number of road deaths increased by some 22% from 598 to 731 between 2012 and 2016 (MVA, 2016). In addition to the human tragedy associated with deaths and injuries, crashes impact the economic potential of Namibia through lost productivity and elevated insurance costs (Forsyth, 2016; MVA, 2016).

3. Methodology

We apply statistical analysis to examine underlying relationships among factors contributing to crashes across Namibia. Statistical analysis of crash data is typically conducted using some form of logistic regression such as: multinomial logit models, probit models, tobit models, mixed logit models, latent class models, etc. (Savolainen et al., 2011; Mannering & Bhat, 2014). Crash records are complex and typically contain a wide range of parameters describing the event in addition to the conditions and circumstance contributing to its occurrence including: crash type, location, time, weather conditions, vehicles involved, demographic information of persons involved. With such a wide range of data and the potential for correlations among them, recent advances in safety analysis have focused on the need to account for unobserved heterogeneity¹ (Mannering et al., 2016). Latent class multinomial (LC-MNL) analysis has been shown to effectively account for unobserved heterogeneity across a range of traffic safety data (e.g. Greene & Hensher, 2003; Cerwick et al., 2014). While superior to the popular random parameters MNL in its ability to address unobserved heterogeneity, the LC-MNL approach used in this study did not account for random parameters within the latent classes as none of the individual variables were found to significantly vary within classes. Nonetheless, the ability to group explanatory data exhibited by the LC-MNL allowed for a richer interpretation of the estimated model results than would be derived from other MNL models. As such, we apply the LC-MNL modelling approach to Namibian crash data to investigate relationships among a range of factors defining individual crash events and the type of transport modes involved.

3.1. Data

Crash data from 2010–12 was obtained from the Namibian National Road Safety Council (NRSC). The crash data comprised records of individual crashes expressed as 60 individual parameters. These crash parameters (i.e. variables) provide specific information across ma range of conditions related to the crash. For example, demographics (age, gender, nationality) of individual involved in the crash are recorded along with any individual parameters that may have contributed to the crash such as a driver being under the influence of alcohol or driving without a valid license. Other parameters convey the conditions under which the crash occurred such as weather, roadway type (paved or unpaved), whether or not it was dark or light conditions and, if dark, whether lighting was present, etc. The data also contained information on what the responding officer determined to be the factors contributing to the crash such as whether or not a vehicle was speeding or violated a traffic control device (e.g. ran a stop sign). All of the parameters in the data were examined for quality (missing values, etc.) and cleaned to ensure consistency in the coding of parameters among record years. In some cases, entire parameter categories (e.g. drink driving, license validity, seat belt usage) were removed from the data set due to inconsistent reporting. In other cases, parameters were grouped logically to increase the overall number of usable crash observations – for example, road surface types such as dirt, sand, and gravel were combined into a single parameter category defined as unpaved. Additionally, information was obtained from the Namibia Weather Network (2017) to replace missing data in the crash records. In other words, the date and time of an individual crash was used to look up the weather conditions (e.g. raining or not) when the crash occurred. The weather data was then used to identify the ‘rainy season’ in Namibia, which typically occurs in January, February, and March.

Vehicle type data were aggregated to create three vehicle mode categories; privately-owned vehicles (PV), transport service vehicles (TSV), and goods vehicles (GV). In most cases, the categorisation of vehicle types was obvious (e.g. motor cars were designated as privately-owned vehicles, while sedan taxis and buses were assigned to the transport service category). Inspection of the crash data, however, indicated that commercial vehicle types defined as ‘light delivery vehicle’ and ‘panel van’ were often recorded carrying passengers. In other words, the crash records indicated that one or more passengers (in a few cases as many as 10) were injured when a light duty vehicle or panel van crashed. This phenomenon of passengers riding in small goods-carrying vehicles (often along with whatever goods are being transported) is typical across sub-Saharan Africa (e.g. Starkey et al., 2013; Jones et al., 2016; Jones et al., 2016). As such, these two vehicle types were included in the transport service category while the goods vehicle category accounted for only large trucks more typically used for commercial purposes such as hauling commodities between cities and to and from the port facilities on the coast. The final transport mode categories are summarised in Table 2.

Table 2. Transport modes by Assigned Category.

Mode	Vehicle types from Namibia crash records
PV	Motor car/Station wagon, Caravan/Trailer
GV	Truck: Articulated, Truck: Articulated Multiple, Gross Vehicle Mass greater than 3500 Kg
TSV	Sedan taxi, Minibus taxi, Minibus, Midibus, Bus, Light delivery vehicle, Panel van

The three vehicle types also allow the capture of interesting development-related information. For example, examination of crashes involving private vehicles is linked to the rapid motorisation that is currently happening in emerging economies as incomes rise and these vehicles become increasingly available. Goods vehicles are obviously important to development as increased truck traffic is typically associated with increased levels of economic activity. And finally, transport service vehicles are of critical importance in sub-Saharan Africa as they are a primary means of transport, especially for low income persons, women, children, elderly, and those not able to access private vehicles or participate in freight-based transport sectors of the economy (Bryceson & Howe, 1993; Porter, 2014; Berg et al., 2017). It is also worth noting that, although important in other parts of Africa, motorcycles were excluded from the current analysis as they did not represent a significant percentage of crashes in the Namibian data.

The final dataset contained a total of 29 437 individual crash observations including 1849 (6%) serious crashes (comprising fatalities and incapacitating injuries) distributed among the three types of transport modes. The data indicated that PV were involved in 49% of reported crashes while TSV and GV accounted for 47% and 4% of crashes, respectively. The crash data contained over 60 explanatory variables for each individual crash observation. All were initially in the included modelling process to determine which individual variables relate significantly to crashes involving each of the three transport modes.

3.2. Modelling

An LC-MNL model was estimated from the data to ascertain what factors contributed to crashes involving the three vehicle types. The LC-MNL model formulation assumes that outcomes depend on unobserved heterogeneity which is captured by the membership of distinct classes (Mannering & Bhat, 2014). As applied herein, the model allows each transport mode to have C different classes so that each distinct class will contain parameters with the probability given as:(1) $P_n\left(c\right)=\frac{\mathrm{e}\mathrm{x}\mathrm{p}\left({\alpha }_c{Z}_n\right)}{{\sum }_{\mathrm{\forall }C}\mathrm{e}\mathrm{x}\mathrm{p}\left({\alpha }_c{Z}_n\right)}$(1) where $Z_n$ represents a vector containing probabilities of $c$ for crash $n$, $C$ is the possible classes $c$, and ${\alpha }_c$ represent the estimable parameters (class specific parameters). The probability of a crash involving transport mode $i$ is given by:(2) $P_n\left(i\right)=\sum _{\mathrm{\forall }C}{P}_n\left(c\right)\ast P_n\left(i/c\right)$(2) where $P_n\left(i/c\right)$ is the probability of crash involving transport mode $i$ for crash $n$ in class$c$. Based on the Equations (1) and (2), we estimate the LC-MNL model for the three transport modes in l for class $c$ as follows:(3) $P_n\left(i/c\right)=\frac{\mathrm{e}\mathrm{x}\mathrm{p}\left({\text{β}}_{ic}{X}_{in}\right)}{{\sum }_{\mathrm{\forall }I}\mathrm{e}\mathrm{x}\mathrm{p}\left({\text{β}}_{ic}{X}_{in}\right)}$(3) w here $I$ represents the possible number of transport mode types and ${\text{β}}_{i\mathrm{c}}$ is a class-specific parameter vector that takes a finite set of values.

Seventeen of the 60 original crash parameters were found to be significant in the final model. The descriptive statistics of the modelled variables are presented in Table 3. The values presented in Table 3 indicate the proportion (i.e. percentage) of the data coded as 1 in the dataset and the standard deviation gives an information of the degree of balance in the data – this format is commonly employed to provide information on the characteristics of modelled binary variables (e.g. Adanu et al., 2017).

Table 3. Descriptive statistics of modelled variables.

Variable	Mean (Standard Deviation)
Serious injury crash (1 if serious injury, 0 otherwise)	0.063 (0.243)
Young driver (1 if driver ≤ 30 years old, 0 otherwise)	0.739 (0.439)
Female driver (1 if female, 0 otherwise)	0.170 (0.376)
Non-Namibian driver (1 if foreigner, 0 otherwise)	0.086 (0.280)
Built-up area (1 if built-up, 0 otherwise)	0.745 (0.436)
Non-intersection (1 if true, 0 otherwise)	0.548 (0.498)
Night (1 if true, 0 otherwise)	0.129 (0.335)
One-way road (1 if true, 0 otherwise)	0.118 (0.323)
Unpaved road surface (1 if true, 0 otherwise)	0.149 (0.357)
Weekday (1 if Monday to Thursday, 0 otherwise)	0.570 (0.495)
Weekend (1 if Friday to Sunday, 0 otherwise)	0.430 (0.495)
Single-carriage way (1 if true, 0 otherwise)	0.351 (0.477)
Rear-end crash (1 if true, 0 otherwise)	0.255 (0.436)
Rainy season (1 if January to March, 0 otherwise)	0.233 (0.423)
Sideswipe crash (1 if true, 0 otherwise)	0.146 (0.353)
Concrete road surface (1 if true, 0 otherwise)	0.048 (0.214)
Dawn/Dusk (1 if true, 0 otherwise)	0.025 (0.157)

As indicated previously, some 6% of crashes recorded were serious injury crashes as reflected in Table 3 (Mean = 0.063). Further inspection of Table 3 yielded additional observations about the nature of crashes in Namibia. For example, 74% of total crashes between 2010 and 2012 were attributable to drivers under the age of 30 and only 17% of crashes were caused by female drivers. More than 8% of crashes were caused by drivers from outside Namibia. Other particularly interesting observations include:

• Three quarters (Mean = 0.745) of crashes occurred in built-up areas. In other words, these crashes happened in and around cities, towns, and villages. Such crashes are often related to increased traffic levels in built-up areas and are typically less severe. Indeed, it appears that 25% of all crashes were rear-end crashes that are typically associated with traffic congestions or stop-and-go driving near intersections and most often do not result in serious injuries.

• Over half of crashes (54.8%) were coded as non-intersection crashes indicating that they occurred on sections of roadway in between major intersections and junctions. Intersection-related crashes include very serious right-angle crashes associated with turning movements within the intersection and/or drivers violating traffic control devices (i.e. running stops sigs and red lights). They also include rear-end crashes as noted above. Non-intersection crashes typically include such serious events such as run-off-road, head-on, and rollover crashes as well crashes where objects or animals in the roadway are involved. Unfortunately, crashes involving vulnerable road users (pedestrians, cyclists, motorcyclists, and animal-powered modes) occur under both intersection and non-intersection conditions and often result in serious injuries.

• More than 13% of crashes occurred under low lighting conditions (i.e. night, dawn/dusk). Lack of roadway and intersection lighting, coupled with a lack of adequate signage, could possibly explain the nearly 12% of crashes happening on one-way roads – although driver behaviour (i.e. lack of awareness, risk taking) is likely a factor as well.

• Around 15% of total crashes occurred on unpaved roads. And perhaps not unrelated, almost a quarter of crashes were recorded during the rainy months of January through March.

In order to more fully appreciate the significance of these parameters and the interactions among them, the LC-MNL model results must be inspected.

4. Results

The final LC-MNL model yielded three separate classes of parameters contributing to crashes among the three transport modes (PV, GV, and TSV). The model variables were found to be significant at the 0.05 level and the overall model yielded a McFadden Pseudo R-Square of 0.289, indicating reasonably good explanatory power for a latent class model. The results of the LC-MNL model (estimation parameters and fit statistics) are summarised separately for private vehicles, goods vehicles, and transport service vehicles in Tables 4 though 6, respectively. It can be observed that the probability of relationships in Class 1 is 0.62, whereas the probabilities for Class 2 and Class 3 are 0.27 and 0.11, respectively. The results of a latent class model can be interpreted such that relationships found in Class 1 are more likely to occur (i.e. within 62% of observations), whereas those in Class 3 are the least likely (11%). When latent class parameter estimators exhibit different signs (i.e. positive and negative), that gives additional confidence in the results implying there is no remaining unobserved heterogeneity for that parameter. Key observations of factors influencing crashes among the three transport modes are illustrated below and a general discussion within the context of development in sub-Saharan Africa follows.

The positive estimators in Class 1 and 3 in Table 4 indicate that female drivers contributed to about 38% of crashes involving private vehicles. Furthermore, the female driver parameter estimates found to be significant (i.e. Class 2 and Class 3) are both positive. Whereas the single-carriage way (i.e. two-lane road) parameter was found to positively contribute to crashes in 73% (Class 1 and Class 3) of private vehicle crashes, it also appears to reduce the occurrence of crashes in 27% of cases as suggested by Class 2. Perhaps this is attributable to the fact that single-carriage ways outside of built-up areas (i.e. rural areas) may contribute more to crashes as a result of higher speeds generally observed on such roads – while single-carriage ways in or in the immediate vicinity of built-up areas may exhibit lower speeds and thus less propensity for crashes. A similar pattern of unobserved heterogeneity can be seen for the rear-end crash parameter. First of all, it is interesting to note that Table 2 showed that rear-end crashes comprised 25% of all crashes while Table 4 indicates they account for 73% of crashes caused by private vehicles. Table 4 then goes on to show that driving during the weekend contributed to 73% of private vehicle crashes and that driving on a paved road reduced the chance of private vehicle crashes by 27%.

Table 4. Private Vehicle crash model estimation results (standard error in parentheses).

Parameter	Class 1 62%	Class 2 27%	Class 3 11%
Female driver		8.33 (2.20)	3.56 (0.46)
Single-carriage way	0.22 (0.06)	−0.95 (0.41)	0.42 (0.17)
Rear-end crash	0.34 (0.08)	−1.18 (0.53)	0.365 (0.17)
Weekend	0.17 (0.06)		1.59 (0.74)
Paved road		−2.22 (0.88)

Another very interesting aspect of latent class models is what can be inferred by the intra-class grouping. Class 1 indicates a relationship among driving on single-carriage way, driving on the weekend, and rear-end crashes for private vehicles in 62% of modelled observations. Whereas, Class 3 contains the same parameters with the addition of female drivers across 11% of modelled observations.

Perhaps the most striking result from Table 5 is that drivers from outside Namibia contributed to nearly 90% of goods vehicle crashes. Younger drivers contributed to 11% of goods vehicle crashes. Being in a built-up area contributed 27% percent of goods vehicle crashes implying that the majority of these crashes were on the open roads outside of cities, towns, and villages. The non-intersection parameter appears to contribute to goods vehicle crashes in 11% of observations. As might be expected, the majority of goods vehicle crashes occur during the work week (Monday – Thursday) and night conditions appear to only contribute to 27% of goods vehicle crashes.

Table 5. Good Vehicle crash model estimation results (standard error in parentheses).

Parameter	Class 1 62%	Class 2 27%	Class 3 11%
Non-Namibian drive	1.62 (0.91)	3.67 (0.73)
Young driver (age ≤ 30)			0.65 (0.18)
Built-up area		4.06 (1.76)
Non-intersection			0.50 (0.18)
Weekday	1.17 (0.66)	0.55 (0.20)
Night		3.87 (0.96)	−1.08 (0.62)
Sideswipe accident			2.46 (1.39)

Table 6 reveals a very interesting finding in that crashes involving transport service vehicles resulted in serious injury in almost 90% of observations (Class 1 and Class 2). It is interesting to note that young drivers contributed to 62% percent of TSV crashes and that low-lighting conditions (i.e. night or dusk/dawn) contributed to the same percentage. Built-up conditions contributed to only 27% (Class 2) of transport service vehicle crashes indicating that the majority of crashes involving transport services likely happen in rural areas of Namibia. The relative dangers of rural transport services in Namibia is further underscored by the fact that unpaved conditions contribute to 90% of transport service vehicle crashes and the open road (i.e. non-intersection) conditions contribute to 73%. The unpaved roads that are typically found in rural areas may be impacted by rainy conditions as these are shown to contribute to 73% (Classes 1 and 3) of transport service vehicle crashes. And crashes that occur on the open road in rural areas are more likely to happen at higher speeds than those in built-up areas or at intersections. Such conditions may contribute to increased serious crashes especially in light of the reduced access to emergency response and medical services common to rural areas.

Table 6. Transportation Service Vehicle crash model estimation results (standard error in parentheses)

Parameter	Class 1 62%	Class 2 27%	Class 3 11%
Serious injury crash	0.55 (0.11)	0.58 (0.35)
Young driver (age ≤ 30)	0.56 (0.09)
Night	0.59 (0.11)
Dawn/Dusk	0.58 (0.17)	−1.58 (0.64)
Built-Up area	−0.32 (0.10)	3.52 (1.73)	−3.38 (1.05)
Unpaved road	0.94 (0.11)	2.01 (0.49)
Not a junction	0.24 (0.06)	−0.819 (0.37)	1.98 (0.68)
Increased rainfall	0.26 (0.06)	−0.79 (0.21)	1.28 (0.56)

5. Development relevance

The following sections present a discussion of how the results of the Namibian crash analyses relate to sustainable development there. Specifically, the factors found to be contributing to crashes in Namibia are presented in terms of how they relate to important currents of transportation-related development such as reliance on transport services for basic mobility, increased pressures from goods movements, and a rapidly growing fleet of privately owned cars and trucks. Where appropriate, findings are extended to the broader sustainable development context.

5.1. Transport services

While much of the transport planning and development literature emphasise the importance of transport services in providing basic mobility and accessibility, they are too often a dangerous mode of travel. The Namibian crash data indicate that crashes involving transport services vehicles are more likely to be serious than crashes involving the other two modes. The finding that younger drivers have such an impact on the safety of transport service users is quite interesting. For one, it underscores issues related to licensing, training, and overall behaviour of these drivers and the apparent importance of transport services as a source of employment for these individuals.

The relative riskiness of young drivers worldwide is well documented (e.g. Jonah, 1997; Møller & Haustein, 2014) and it is no different throughout SSA (e.g. Azestop, 2010; Porter et al., 2010; Sinclair, 2013; Amweelo, 2016). This is especially true in the instance of drivers participating in informal transport sectors and some researchers have even documented specific psychological conditions within these driver populations (Møller & Gregersen, 2008; Ulzen et al., 2018). In many cases, transport service vehicles across Africa are poorly maintained and ill-equipped (van Garenne et al., 2001; Akloweg et al., 2011).

Transport services, however, are essential to sustainable development (Turner & Kwakye, 1996; Bryceson et al., 2003; Too & Earl, 2010; Porter, 2014). They are also often the primary form of mobility (other than walking) for many rural women and children who typically do not have access to private vehicles (Porter, 2002, 2014; UN Women, 2015). As such, improvements in rural transport services have the potential to not only address targets 3.6 and 11.2 of the SDGs but also those that explicitly address the advancement of women and children. Goal 5, and the need for safe public spaces set out in target 11.7 (Nhamo et al., 2018). The essential mobility afforded by transport services also supports inclusion of elderly populations as contained in target 10.2 (Oppong, 2006; Porter, 2014; UNDP, 2016).

There are enormous possibilities for job creation and innovation (Goals 8 and 9) in the provision of safe transport such as the safety-promoting, mobile phone-based ridesourcing app SafeMotos recently launched in Rwanda (Ismail, 2016; Roberts & Lall, 2019). Also, such formal promotion of a positive safety culture would specifically contribute to Target 8.8 to ‘promote safe and secure working environments … ’ and support the use of sustainable infrastructure for ‘economic development and human well-being’ as set out in Target 9.1.

5.2. Goods movement

There were two significant findings related to goods vehicle crashes in Namibia. The first is the involvement of younger drivers. Many of the issues elucidated above apply equally to the behavioural issues compounded by lack of education, training, and enforcement of these drivers when transporting various commodities within and across Namibia, often in large trucks – again the importance of participation in the transport workforce for these younger people is underscored. Of particular interest, however, is the role non-Namibians play in goods vehicles crashes. As a country with a robust mining industry itself and one on the coast serving oil-rich Angola, and its landlocked mineral producing neighbours, Botswana and Zambia, as well as economically active South Africa, Namibia accommodates substantial amounts of regional freight traffic (Savage et al., 2013; Roberts, 2014). The fact that this freight traffic evidently has an adverse impact on road safety in the country is alarming and poses a development challenge.

The challenges and opportunities of providing sustainable transport services (e.g. driver training, technological innovation, and corruption mitigation) relate equally to the movement of goods across Namibia as well and other developing economies (e.g. Maldonado et al., 2002; Kircher & Andersson, 2013). As such, goods movements place pressures on and can be improved by the same SDGs as the transport service sector.

5.3. Private motorisation

Finally, although not a source of the most serious crashes in Namibia, private vehicles accounted for almost half of the overall crashes recorded during the 2010–12 study period. The number of motor vehicles is rapidly increasing in Namibia and elsewhere in the developing world. The very driving behaviours and unsafe driving cultures that plague heavily motorised countries will no doubt continue to rise in Namibia and other developing countries if left unchecked as new private vehicles are constantly added to the roads. And while the private vehicle crashes were not the primary source of serious injury crashes, even non-serious crashes negatively impact development in terms of direct and indirect costs to society through insurance payouts, time spent resolving crash-related issues, time spent in crash-induced traffic congestion, lost productivity, etc.

Again, the role of behaviours and safety culture are evident. As the prevalence and use of private vehicles grows in Namibia, the exposure of drivers and passengers to potential crashes will also increase. While driving behaviours and related safety culture issues (e.g. corruption) must be addressed, the increased use of private vehicles will require community outreach and education to encourage passenger safety – whether through programmes to enhance seat belt usage and child restraint devices or, in countries where motorcycles are prevalent, the availability and use of helmets. The responsibility for such efforts will fall far beyond that of transport professionals and law enforcement and require a holistic approach involving educators, social workers, policy makers, public relations and marketing businesses, and government agencies at all levels.

5.4. General

The provision of safe transport is essential to meeting the SDGs. In addition to the targets where safe transport is explicitly addressed, the role it plays in creating and supporting social and economic opportunities for all are clear. In order to successfully realise these, however, the importance of transport safety and the issues elucidated above (e.g. driver training, vehicle maintenance and repair) need to be explicitly addressed in Target 4.4 of the SDGs wherein the need for relevant technical and vocational skills is asserted. It could even be said that the call in Target 4.7 to ensure ‘ … all learners acquire the knowledge and skills needed to promote sustainable development … ’ should promote a culture of safety inasmuch as it does ‘peace and non-violence’. Furthermore, Target 4A implores a safe learning environment that must include sensitivity to the well-documented road safety hazards facing school children worldwide (e.g. Hyder et al., 2008; Ackaah, 2010; Alonge & Hyder, 2013).

The widespread and nefarious impacts of corruption also links Targets 16.5 and 16.6 to the provision of safe transport (Lagarde, 2007; Hua et al., 2010; Jones et al., 2016; Jones et al., 2016; Peiffer & Rose, 2018). Whereas, safe and inclusive transport is an essential component in mitigating fragility and promoting resilient, sustainable communities in Africa as broadly set out in SDG 16 (Jones et al., 2019). And finally, the ability to conduct the current study in Namibia and present the findings within a development context was dependent on the availability of quality data – the importance of which is noted in Target 17.18 of the SDGs,

6. Conclusion

Based on the data for the period 2010–12, a latent class multinomial regression model was estimated for three categories of transport modes: Private Vehicles, Goods Vehicle, and Transport Service Vehicles. The purpose of the study was to identify and perhaps better understand factors contributing to crashes in Namibia and how they relate to sustainable development.

The findings of this study reveal that in spite the significant role of transport service vehicles in the developmental of Namibia and other developing countries, the costs (i.e. loss of life and attending socioeconomic impacts) of traffic crashes involving these vehicles can potentially offset development gains. The observation that younger drivers have such an impact on the safety of transport service users and goods movement is quite interesting. This underscores issues related to licensing, training, and overall behaviour of these drivers and the apparent importance of the transport sector as a source of employment for these individuals. Also, the number of motor vehicles is rapidly increasing in Namibia and elsewhere in the developing world. The very driving behaviours and unsafe driving cultures that plague heavily motorised countries are likely to rise in Namibia and other developing countries if left unchecked as new private vehicles are constantly added to the roads. This study has ultimately revealed that the underlying factors contributing to crashes in Namibia are largely attributable to driving behaviours and safety cultures. The study also documented the importance of safe transport to achieving sustainable development in Namibia, especially as expressed in the SDGs.

Disclosure statement

No potential conflict of interest was reported by the authors.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

1 Unobserved heterogeneity implies some level endogeneity in the data (i.e., explanatory variables correlated with regression error terms) that can be attributed to (a) missing or omitted parameters or (b) errors or simultaneity at the time of measurement (Mannering et al, 2016).