The impact of a labour-intensive road construction programme in the Vhembe District, Limpopo Province

Abstract

This study investigates the distributional impacts of labour-intensive projects under the Gundo Lashu programme in the Limpopo province. The aim was to evaluate infrastructure effectiveness of the Extended Public Works Programme, focusing on the interface between road investment and economic development as the central premise. The objective was to evaluate the impact of the programme on the participants and their communities. Specific areas of impacts investigated are mobility, income, work opportunities, poverty reduction, sustainable livelihoods and opening of access to market opportunities. A matched case–control study design was adopted. The study found that the programme had achieved its objectives related to the total number of jobs created and the total road length constructed. However, the programme showed mixed results on communities' socio-economic outcomes and the impacts of the programme on poverty and sustainable livelihoods.

Keywords:

• road construction
• labour-intensive methods
• transport infrastructure
• impact evaluation
• Extended Public Works Programme
• economic development

1. Introduction

The South African economy is characterised by mass unemployment within its working-age population, with a large proportion of the population lacking skills to enable them to participate actively in the economy. After several years of decline, official Labour Force Survey estimates for the first quarter of 2006 reported the unemployment rate to be 25.6% (Statistics South Africa, 2007). A comparison with the Quarterly Labour Force Survey for Quarter 1 of 2012 shows a slight decline in unemployment rate by 0.4% (Statistics South Africa, 2012). The estimates for the first quarter of 2012 reported the unemployment rate to be 25.2% (Statistics South Africa, 2012). While this is lower than some past statistics, the rate is still of great concern.

The Extended Public Works Programme (EPWP) was initiated with the purpose to draw a significant number of the unemployed into productive work (Philips, 2004b; Department of Public Works, 2007). The key rationale for the establishment of the EPWP was not to create a separate entity to address the unemployment problem; its framework was built on existing job creation programmes focusing on introducing labour-intensive methods. Based on the high unemployment rate amongst the youth, the EPWP aims to provide additional and mostly temporary work opportunities that are combined with training, in all spheres of government and state-owned enterprises (Philips, 2004a; Department of Public Works, 2007).

Labour-intensive construction programmes, however, are not without criticism. Most of the programmes are stalled by various barriers related to ineffective management systems and structures as well as operational barriers associated with project planning and implementation. Amongst the key challenges are the following:

•	Inappropriate budget allocation and poor workmanship.
•	Although employment opportunities are created in large numbers, the work generated remains temporary, associated with minimal long-term empowerment of the people.
•	Development of skills during the construction phases does not result in sustainable entrepreneurial or small contractor abilities that could result in small business development (McCutcheon, 1990).
•	Furthermore, the majority of the labour-intensive programmes, especially within the road sector, are initiated as pilot programmes, and since there is no proper planning for sustainability most projects fail to transform from pilot phase into a full-scale implementation phase (Pama, 1992).

In considering ways to provide transport infrastructure that benefits the local communities through job creation and skills transfer, government policies and strategies, specifically within the transport sector, focus more on the application of labour-intensive methods as opposed to the traditional capital-intensive methods. In his explanation, McCutcheon (2008:9) reflects that labour-intensive methods are suitable for a wide range of construction activities and can generally produce the same quality of product as capital-intensive methods. These methods may also be applied on the construction and maintenance of high-volume and high-standard roads and not just low-volume, low-standard rural roads (McCutcheon, 2002).

Labour-intensive work, specifically within the road sector, is not a phenomenon exclusive to developing countries, it is also found in the local traditions of the developed countries. In the past, communities would build paths and trails to satisfy their need for access. These initiatives have a long history of altering both social and economic structures of local and surrounding communities in various ways, including positive trickle-down effects on local economic and community development, and a potential contribution to fighting poverty. Dawson & Barwell (1993) explored general impacts of labour-intensive programmes, especially within the road sector, and classify them as follows:

•	Generation of employment, both temporary and long term.
•	Contribution towards poverty alleviation through income generation.
•	Provision of necessary infrastructure to improve mobility and access.
•	Increase in gross domestic product through improved trade facilities.
•	Transfer of knowledge and skills to local communities.

Community-based labour-intensive work has a long history in Africa. In Morocco, a large-scale national promotion programme was launched in 1961. The programme aimed to enhance opportunities for rural unemployed in productive works and to address the problem of rural migration. Although started initially as an emergency relief works programme for rural areas, the programme gradually acquired a development orientation (Thwala, 2001).

The Tunisian and Algerian works programme known as Worksites was developed to combat underdevelopment through the creation of job opportunities. In Algeria, this programme began operating in 1962 as a relief operation. However, the programme emerged as a development orientated to maximise employment in various land development programmes and agrarian reforms. In Kenya the programme played a key role in road infrastructure development, where over 12 000 km of rural access roads were constructed and more than 80 000 man-years of employment were created (Thwala, 2001).

Major lessons from labour-intensive works programmes, within the Southern African Development Communities, can be drawn from the labour-intensive works programme in Lesotho and Botswana. The main objectives of this programme were articulated in these countries' National Development Plans focusing on rural development and creation of employment.

The major component of the programme was gravelling, maintenance and reconstruction of roads. The analysis of the physical results of the pilot project and financial costs showed that labour-intensive works resulted in the gravelling of between 1000 and 1400 km of roads and a total of 404 roads were reconstructed, at the relatively high wages prevailing at that time (McCutcheon, 1992). The building of high-standard gravel roads even in mountainous terrain was achieved and a sense of ownership of the roads was generated in the communities (Pama, 1992). In Botswana, an independent evaluation concluded that the programme was a success in relation to most of its objectives. Almost 2000 km of roads were upgraded and 3000 people per year were employed (McCutcheon, 1992).

Community-based labour-intensive works have also been used in South Africa for development purposes. The major turning point was the launch of the National Public Works Programme after the 1994 elections. The emergence of Community Based Public Works further advanced these goals by focusing on the development of smaller companies and institutionalising regulatory bodies (Thwala, 2001). The EPWP later became one of the measures agreed upon, to address the problem of unemployment in situations of high unemployment or underemployment. This was done through road infrastructure development and maintenance programmes such as the Gundo Lashu programme in Limpopo province and the Zibambele programme in KwaZulu-Natal province (Philips, 2004b).

The aim of this paper is to report on a study that was carried out in Limpopo province to investigate the distributional impacts of the labour-intensive Gundo Lashu programme. The infrastructure effectiveness of the programme was evaluated in terms of the interface between road investment and economic development. The objective of the study was to describe the nature and delivery mechanisms of the projects, to evaluate the impact thereof on the participants and their communities and to explore the constraints and challenges experienced by these initiatives. After describing the methodology used in the research, the setting of the Gundo Lashu programme is discussed and a full discussion of the key findings of the research is then provided.

2. Overview of methods

According to Harmse (2004:28): ‘methods used in both economic and transport geography have changed over time, partly due to the social nature of science and the struggle between competing philosophical positions’. For this study, the collection of the primary data was based on ethnographic strategies, which are not generally used in economic and transport geography as a methodology. An empirical exploratory research design, guided by critical realism, was used in order to focus on people's sense of, and relationship with, places. The main instrument used for data collection was a comprehensive questionnaire that collected an array of information on respondents' individual and household socio-economic circumstances. This was complemented by observations and structured interviews with the two sampled groups. Furthermore, unstructured interviews were also conducted with the project manager for the Gundo Lashu programme. Project administrative data, programme management data, and the existing records within Limpopo Road Agency (RAL) were used as secondary data sources.

At the time of data collection, the Gundo Lashu programme had employed a total of 1339 people in labour-intensive jobs in the six selected projects (RAL, 2009). The study sample covered a total of 128 respondents from the treatment group, which was drawn from project communes where the Gundo Lashu programme was implemented, with a specific focus on projects executed through labour-intensive construction methods. The control group was drawn from non-project communes where labour-intensive road construction and maintenance projects were prioritised but had not yet taken place.

The study covered six projects which were constructed using labour-intensive methods to provide access to communities and villages within the Makhado and Thulamela local municipalities. Table 1 provides a list of the sampled road projects.

Table 1: List of sampled projects

Project area	Road number	Length (km)
1 Tshifulanani–Tshivhumbe	D3724	3.7
2 Tshifulanani–Tshivhumbe	D3724	6.8
Muraleni–Madabani	D4016	1.6
Maebane–Makhitha	D3676, D5003	3.3
Schoemansdal–Makhitha	D3700	5.2
Muraleni–Madabani	D5003	3.5

The analysis of the data collected for the study was performed using STATA version 10. Comparisons were made between the two groups (control and treatment) based on odds ratios estimated from conditional logistic regression analysis.

3. The setting

Limpopo province, where the programme was implemented, is situated at the north-eastern corner of South Africa, sharing international borders with Botswana, Zimbabwe and Mozambique. The province is divided into five district municipalities, namely Capricorn, Mopani, Greater Sekhukhune, Vhembe and Waterberg (Statistics South Africa, 2009). Transport infrastructure in the Vhembe district, specifically in Makhado and Thulamela municipalities, are composed of provincial road networks, local roads and rural access roads. Table 2 shows the status of roads in various municipalities within the Vhembe district. Most of the roads in the Makhado municipality (1675 roads) are unpaved while there are almost 696 unpaved roads in the Thulamela municipality (RAL, 2009).

Table 2: Road network in Vhembe district 2005

Municipality	Local municipality	Paved roads (km)	Unpaved roads (km)	Total roads (km)
Vhembe district	Makhado	608	1 067	1 675
	Thulamela	258	696	954
	Musina	353	543	896
	Mutale	88	424	512
	Total	1 307	2 730	4 037
Source: RAL (2005).

While there are no formal statistics on those who walk for various purposes such as leisure, tourism and social activities, the province has the highest number of households who walk to work. Official statistics from the National Household Travel Survey show that more than 45% of the total workforce within the Vhembe district walk to work, as compared with those who use private cars and public transport such as buses, trains and taxis (Department of Transport, 2005).

Two broad labour-intensive models were used in the Gundo Lashu projects in order to capacitate local communities to become independent emerging contractors. Within the first model a total of 32 emerging contractors were trained in labour-intensive methods. These contractors were then provided with small trial contracts for execution. Those who demonstrated their ability by the successful completion of these small contracts were then allowed to tender for larger contracts. The programme also adopted a conventional contract model with labour-intensive prerequisites. This model had a very basic contractual arrangement from the client's point of view, as the only contracts involved are entered into with a consulting engineer to provide design and supervisory services and the selected contractor to execute the project. In this case, contractors were fully responsible to find and manage the labour, materials and plans required to execute the labour-intensive work and roads were constructed using labour-intensive practices in order to maximise the number of job opportunities per unit of expenditure.

During the construction stage, local community members used light equipment such as wheelbarrows for the transportation of the aggregate from the stockpile to the location of the mixing plant. This was followed by the construction of a base layer by mixing of the stabilising agent using a shutter system and screening off. Once the base is completed, the base surface is primed and fine, coarse concrete is sprayed.

The maintenance of the road infrastructure using labour-intensive methods is a key focus area within the Gundo Lashu programme. Figure 1 depicts some of the road maintenance activities within the Gundo Lashu programme. Projects of this nature were also sampled for the study. Administrative data from the RAL reflect that, since inception in 2004, the programme has been funded by the Department of Public Works, and in 2003/04 the budget was R50 million. The budget increased to R103 million in the 2004/05 financial year. As reflected in the annual report, the total Gundo Lashu programme budget for the period 2002–09 was R22 790 728.31. The programme had constructed 222.8 km of roads by labour-intensive methods using cement and slabs, natural pavement, and emulsion-treated and composite emulsion treatment base. During the training phase a total length of 110.5 km roads were constructed, and an additional 112.3 km were constructed during the post-training phase (RAL, 2009).

Figure 1: Labour-intensive road construction and maintenance activities

4. Key findings

4.1 Impact on travel behaviour and access

Basic to the relationship between transport and development are the associated concepts of mobility and accessibility. According to Hilling (1996:34), accessibility may be interpreted in two ways. At the one level, it is related to connectivity and may be defined as the number of links that have to be transverse, or sub-journeys made, from a particular node. At another level, accessibility may refer to joins in a network (Hilling, 1996:36). In this case, the frequency at which community groups, from both the control and treatment groups, visit the local towns using public transport was investigated.

The data obtained from the questionnaires reveal that the travel patterns of the respondents from the treatment group had increased as a result of the new road infrastructure. The majority of the respondents (60%) indicated a significant improvement in their travel behaviour. They were able to visit their nearest town at least once in a week, more often than before the construction of the road. Only 36% of the respondents indicated that their travel behaviour had changed only slightly since public transport, to visit their nearest communities, became available. Only 4% indicated that their travel pattern had not changed at all over time and they were not affected by the development of the road.

The majority of the respondents from the control group (89%), on the other hand, indicated that there was no tangible contribution of road infrastructure to their personal and community well-being. This was substantiated by their verbal responses indicating that the government's investment in the Gundo Lashu programme is just a waste of resources. The majority of these respondents within the control group (65%) also indicated they would not recommend or give their approval to any labour-intensive road construction in their communities.

Most of the issues raised did not dispute the significant contribution of the projects towards improved access, but they are specifically linked to the quality of roads constructed as wastage of government resources on projects that are not preferential to most rural communities. The fact that there was a significant improvement in the travel behaviour of the treatment group, while there was little improvement in that of the control groups, indicates that the people participating in the Gundo Lashu programme benefited from the programme and its projects.

In the study, access to educational facilities, in terms of the time used to reach these facilities, was also used to measure the differences between the two groups in the study. However, before comparing the means of the two sampled groups, the variance of the two independent sample groups were compared using the variance ratio test (sdtest). The resulting p value was 0.2439, which is >0.05. It was concluded that there is similar access to educational facilities for both the treatment and the control communes. There is equal variance and a comparison of mean access to educational facilities can be performed.

The mean access to educational facilities was compared by making use of a two-sampled t-test with equal variance. The p value was 0.1407 (>0.05), which implies that the mean access to educational facilities was equal at the 0.05% level of significance for both the treatment and the control groups. The study found that the total mean access to educational facilities for the treatment group was 32.34 minutes with travel times ranging from 5 to 90 minutes. The control group, on the other hand, had a mean access to educational facilities of almost 37 minutes with travel times ranging between 15 and 80 minutes.

The conclusion that can be drawn from this analysis is that although the mean travel time to educational facilities was less for the treatment group (32.34 minutes) than the control group (37 minutes), the accessibility to roads did not have a significant impact on the travelling time to educational facilities. These results challenge the notion that rural transportation and access to roads in rural areas significantly increases the access of local communities to services. Ellis & Hine (1998) concluded that villages in Zimbabwe with an average of one motorised transport per 300 people were worse off due to lack of transport infrastructure compared with Sri Lanka, where the level is five times higher. They further concluded that the travel cost for journeys of up to 30 km was more than twice expensive in Zimbabwe than in Sri Lanka (Porter, 2002:286).

The time it took participants in the study area to travel to health facilities was used to measure the differences between the two groups with regard to access to health facilities. The means of the times it took to access health facilities for the two sampled groups were compared. Before comparing the means of the two sampled groups, the variance of two independent sample groups were compared using the variance ratio test (sdtest); since the p value was 0.3156, which is >0.05, it was concluded that there is similar access to health facilities for both treatment and control communes and a comparison of mean access to health facilities can be performed. The analysis of mean access to health facilities found that the p value was 0.2606 (>0.05), which implies that the mean access to health facilities of both the treatment and the control groups are equal at the 0.05% level of significance. The empirical data showed that the total mean access to health facilities for the treatment group was 36.01 minutes with travel time ranging from 5 to 80 minutes. The control group, on the other hand, had a mean access to educational facilities of 39 minutes with travel times ranging between 15 and 80 minutes. Although the mean travel time to health facilities was less for the treatment group (36.01 minutes) than for the control group (39 minutes), the availability of roads did not have a significant impact on the travelling time to health facilities.

4.2 Impact on income (personal and household) and expenditure patterns

Wage levels in the Gundo Lashu programme are determined and governed by the terms of the Code of Good Practice for Special Public Works Programmes (Department of Public Works, 2007). Contractors are contractually required to employ local workers on the basis of gender, age, disability and quotas. This was because of difficulties in setting up labour-intensive wages in line with the legislated minimum wages. Within the sector there was fear of cases where the rate could be set too high if compared with prevailing unskilled market wages for similar work in the areas concerned and this would result in the project not being economically viable. In other cases there was a fear that the minimum rate could be too low to attract and motivate workers.

In general, the administrative data obtained from the RAL show that the total overall wages for the Gundo Lashu programme were R40 981 655 (RAL, 2009). This amount is divided into R32 933 255 for wages paid during the training phase and R8 048 400 paid to project participants in the form of wages during the post training phase (RAL, 2009). The study found that a total amount of R21 795 000 was earned by beneficiaries for road construction in the sampled projects in the Thulamela local municipality and R1 871 236 was paid in the form of wages to the participants in the Makhado local municipality.

Besides the wage level, various concerns were raised by project participants such as the inconsistent implementation of wages, the application of wage rates and confusion and uncertainty over the period of participation and the duration of project implementation. While the limitations of investment in roads as an instrument of social and economic development have been recognised, the most direct benefit of road construction work is the income wages and salaries for the workers. Other benefits are indirect and dependent on various other factors (Howe, 2003).

The information on the respondents' household income obtained though the questionnaire was used for household income analysis. The total household income was calculated by aggregating income from all of the sources provided by the respondents. A comparison of mean household income of the two sampled groups was made. Before comparing the means of the two sampled groups, the variance of the two independent sample groups was compared using the variance ration test (sdtest). Since the p value was 0.3131, which is >0.05, it was concluded that household incomes for both the treatment and the control communes are comparable and a comparison of mean household income was performed. The p value was 0.1117 (>0.05), and therefore the mean household income of both the treatment and the control groups was significant at the 0.05% level.

The mean monthly value of household income for the treatment group was R1669.37, with a minimum monthly household income of R180.00 and a maximum of R3090.00. For the control group, the mean monthly household income was R1324.844 with household income ranging between R400.00 and 4000.00. In both cases the data were skewed by high transfers from private maintenance, government social security transfers, pensions, and other transfers.

Respondents were asked to indicate the amounts of income they receive from various sources. Most respondents received income from more than one source. About 60% of the treatment group (households where a member is involved in the Gundo Lashu programme) and about 80% of the control group received government support in the form of grants and other social transfers. Among the treatment group, almost 20% would not disclose their major source of financial support. A mere 12% of the respondents reported that they receive financial support from family members while about 40% had their own business. In the control group, a large percentage of respondents (58%) received financial support from their family members while only 42% were involved in their own business. Almost 19% reported other sources of income as they did not want to disclose their major sources of financial support.

In contrast to the notion that the Gundo Lashu programme would have a direct impact on the respondents' household income, the study found that the impact was statistically insignificant. It appeared that the development and maintenance of local roads through areas in which the Gundo Lashu programme was implemented did not improve household incomes in the affected communes, since there was no evidence that the wages of the project participants increased the total household income significantly.

When considering wages earned within the Gundo Lashu programme, the average duration of employment, where programme participants are exposed to one episode of employment, was four months. These projects are therefore not a sustained source of income and are not a remedy for chronic poverty. This conclusion is based on the fact that the stabilisation effect of income is dependent on the duration of the employment, and it is achieved through sustained employment. Therefore, the short-term nature of their participation did not provide the people in the treatment group with an opportunity to accumulate assets that would then enable participants to move out of poverty.

One can therefore conclude that while the Gundo Lashu programme led to large gains in terms of wages during the course of the project, there was no specific long-term effect on the treatment group's household income. Even though the treatment group members had received significant gains through personal income from the project, this was offset by the communities over-reliance on substantial government social assistance programmes, such as child support grants, leading to virtually no increase in total household income.

In addition to the household income patterns, household income expenditure was also examined in order to gain further insight into the effect that project wages had in the households of the treatment group (project participants). Respondents were asked to estimate the monthly amounts spent on a range of expenditure categories in the household. The data reveal that more than one-half (56%) of the treatment groups' income was spent on food. This is much the same as the almost 57% spent on food by the control group. For both groups, expenditure on food was the largest expenditure in the household budget. The statistical test of significance shows that there were no significant differences between the treatment and control groups in terms of the proportion of household expenditure per defined expenditure class. The income derived from participation in the projects therefore appears to be used primarily to buy essential food. It is therefore unlikely that the income could be used to cover a wide range of household expenditures.

Although salaries from the projects are used to purchase various consumable products for the families, the discussions thus far suggest that in families where a member is a programme participant, their project salaries were pooled with other income sources such as government welfare grants to meet the broader needs of vulnerable families. Respondents were further asked to indicate how they spend their project income on a monthly basis, and their responses are summarised in Figure 2. This was a huge challenge since most of the respondents indicated that they do not keep transactional records of all their monthly purchases.

Figure 2: Classification of expenditure for the treatment group

The study revealed that about 67% of households that received wages and salaries from the Gundo Lashu programme pooled the money with other income to cover general household expenses. However, almost 16% of the respondents reported that they spend their salaries and wages exclusively on their children. This must, however, be seen within a context where pooling of project income is combined with child grants and other income that generally contributes to the improved functioning of a poor household.

Besides evidence that the Gundo Lashu programme is making a significant contribution to households with poor incomes, there is also evidence that the participation of poor rural households enabled them to open bank accounts and thus promote some form of savings. Project participants are also involved in financial decision-making in their households regardless of gender, and generally had some say in how the project incomes were spent.

4.3 Impact on employment and work opportunities

Construction and maintenance of the road infrastructure generates labour demand, especially for low-skilled labour. Employment is generated during the construction of the infrastructure as well as during the maintenance and operation (McCutcheon, 2008). Various research studies within the field of geography, transport, economics and other related fields (Adato et al., 1999; Thwala, 2001; Adato & Haddad, 2002; Agingu, 2004; Philips, 2004a; McCord & Van Seventer, 2004) provide evidence of employment capacity for labour-intensive programmes, both within the road and other infrastructure programmes.

In this instance, the Gundo Lashu programme created a total of 6827 jobs and 1 285 189 person-days of work from 2004 to 2009. During the training phase a total of 4311 jobs (and 1 092 741 person-days of work) were created. Only 192 448 person-days of work and 2516 jobs were created in the post training phase. These jobs were, however, short term and unsustainable. Table 3 shows that the programme created a total of 3206 jobs for women (comprised of females of all age categories) during the project lifecycle. Almost 1884 jobs for women were created during the training stage. The number of jobs decreased to 1322 during the post training phase. A total number of 2996 jobs were created for youth (comprised of males and females below the age of 35 years); this included 1838 jobs that were created during the training phase and 1158 jobs that were created during the post training phase.

Table 3: Employment disaggregation by participant groups (2004–09)

Project phase	Length (km)	Youth	Women	Disabled
Training	110.5	1 838	1 884	13
Post training	112.3	1 158	1 322	2
Total	222.8	2 996	3 206	15
Source: RAL (2009).

The analysis shows that the total number of jobs created for both the youth and women decreased by almost 50% in the post-training stage despite the fact that the total road length constructed increased from 110.5 to 112.3 km (RAL, 2009). The sharp decline in employment output creation between the training phase and the post-training phase for the period 2008/09 may be due to limited reporting and unavailability of consolidated programme performance data.

The temporary nature of employment in the Gundo Lashu programme is dependent on the programme's objective of providing spill-over benefits instead of contributing significantly to aggregate employment. The purpose of the temporary jobs created and training through the Gundo Lashu programme was thus to improve the labour-market performance of the participants once they have exited from the programme.

Data analysed in the study showed that the programme managed to improve the quality of the labour supply based on the training and work experience provided to the participants during the duration of the project. Practical work experience and training have the potential to enable those who participated in the programme, to be employed in suitable jobs in the future. From a theoretical perspective it is therefore reasonable to conclude that the implementation of public works programmes, such as the Gundo Lashu programme, had a positive impact on South Africa's unemployment challenges and growth path.

Since the Gundo Lashu programme only provided employment and earnings for a short period, the key question is what happened to participants' household income after their participation in the Gundo Lashu programme. One expectation is that a reduction in income due to the loss of the Gundo Lashu programme earnings would be partly offset by increased new employment opportunities. Participating members were encouraged to seek employment in anticipation of lost of the Gundo Lashu programme earnings, and participants were well aware of the duration of their participation in the project.

Although the Gundo Lashu programme is not entirely a training programme, working in the project has provided participants with valuable experience and improved their employability. The training also provided participants with relevant skills, which improves their effectiveness. The overall positive effects of the Gundo Lashu programme on major working skills provides hope that a community-based employment programme can improve participants' opportunity for future employment. Although formal skills development training was not the key objective of the Gundo Lashu programme, participants' skills were thus enhanced through the provision of basic job training, combined with work experience in the different areas of construction.

The offering of skills-training and capacity-building programmes to participants, in exchange for their participation in the Gundo Lashu programme, can lead to changes not only in participants' skills but also in their attitudes towards work. Participants in the Gundo Lashu programme typically did not have a strong attachment to the labour market before participation. The participation of the treatment group in the Gundo Lashu programme is expected to offer them a marked change in their employment situation in future. This change is reflected in their responses in terms of the programme's effects on their attitudes towards work, thereby excluding them from the ‘discouraged work-seekers’ category as described by Statistic South Africa (2007, 2012).

In line with the quarterly labour force guide, discouraged work seekers are categorised on the basis that they wanted to work, but there were no available jobs for them. This category also include those who were unable to find work requiring their skills and those who lost hope of finding any kind of work (Statistics South Africa, 2007). The responses obtained from both the treatment and the control groups on the question regarding the possibility of finding a job were subjected to a conditional logistic regression analysis. A conditional (fixed-effects) logistic regression (clogit) was calculated for jobs, age, gender, marital status and levels of education and the results are presented in Table 4.

Table 4: Possibilities of finding work

Job	Coefficient	Standard error	z	p >|z|	95% confidence interval
Age	–0.03963	0.04573	–0.87	0.386	–0.12926 to 0.05001
Gender	–0.33914	0.78789	–0.43	0.667	–1.88338 to 1.20510
Marital	0.04385	0.27899	0.16	0.875	–0.50297 to 0.59068
Education	–0.37624	0.53256	–0.71	0.480	–1.42005 to 0.66757

The p values obtained for the jobs, age, gender, marital status and levels of education variables were all larger than 0.05. At the 0.05 level of significance, there is no statistically significant difference between the control group and the treatment group when it comes to the possibility of finding work in future. It seems as if the intervention in terms of participation in the programme did not achieve the intended aim of assisting participants in obtaining jobs after their involvement in the programme ended.

The responses to the question of whether they worked for pay during the week preceding the survey indicated that many of the treatment group did not have a job and that they have returned to the unemployed labour pool after their term in the short-term public works programmes. Most of them were not absorbed into other labour-intensive programmes, such as the programmes within the social or environmental sectors, after completing their term with the Gundo Lashu programme. These multiple short-term public works initiatives only serve to churn the unemployed, replacing one cohort of unemployed with another, instead of addressing the underlying unemployment problem.

Within the existing literature on the development of the informal sector, some researchers have noticed the existence of a sub-category of businesses that face growth barriers in the economy regardless of their significant contribution to the gross domestic product. This subcategory is called survival(ist) enterprises (South Africa, 1995). According to the White Paper on National Strategy for the development and promotion of small business in South Africa (South Africa, 1995), the subcategory is characterised by set of activities conducted by the unemployed to generate income. Most of the incomes generated from these businesses, usually fall short of a minimum standard of income. According to Wood (2003), these enterprises survive by ‘destructive uncertainty’. The entrepreneurs have to cope not just with short-term shocks but also with unpredictable hazards, and the avoidance of unnecessary risks is absolutely imperative.

The study found evidence of the development of these businesses by both the project participants and other local community members. These businesses, however, were not initiated by choice; they existed out of desperation from people who could not find waged employment. The opportunity for business start-up was investigated and the study found that about 17% of the respondents in the treatment group started businesses since the roads were constructed, with almost 69% of them indicating that the road was a deciding factor for business start-up. The majority of these businesses were small informal markets along the roads, supplying consumable and agricultural products to the local community.

Within the project communities, the study found that there has been a better availability and mix of transport services, and an increased flow of goods and services. These areas are primarily agricultural communities with higher vulnerability to natural shocks or fluctuations in commodity prices. Their ability to diversify to other sources of income was therefore important to reduce poverty. In practice, those community members who are most secure and have savings are the ones who are able to make the best use of the opportunities from better road infrastructure. However, the outcomes of the Gundo Lashu programme in the sampled study areas suggest that better rural roads have allowed those with some savings to diversify into activities with substantial better returns. Project communities have demonstrated the ability to generate surplus funds to invest in trading, purchase products to sell, or create a network of connections and relationships outside the community enabling them to take advantage of trading or working opportunities.

Involvement in different work activities also provide an indication of the extent to which project participants are able to make a living or earn an income after the project has ended and is a good example of the programme's contribution to the gross domestic product. The respondents were asked to indicate whether they were involved in any or all of a list of seven different activities during the week preceding the survey. The results from the survey show that larger percentages of people in the treatment group were engaged in specified activities than the people in the control group. The activities provided covered both paid and unpaid jobs. The people in the control group tended to be mainly involved in running a business, helping unpaid in a household business or working for a wage or salary. While these were also the most popular activities among the control group, the number of people involved in each activity was much smaller than in the treatment group.

The results from the survey showed that about 59% of the respondents in the treatment group were involved in running a business or doing any kind of business for themselves or with one or more partners (including family members). Only about 41% of the respondents in the control group were involved in running a business or doing any kind of business. Not surprisingly, 52% were involved in construction or major repair work on their homes while only 23% of the control group were busy with such activities.

These findings are consistent with a study by Binswanger et al. (1993), who used data from 85 randomly selected districts of India to examine the role of rural roads, among other factors, in agricultural investment and output. The study found that accessible roads contributed directly to the growth of agricultural output, to increased use of fertiliser, and to commercial bank expansion. Some of the extended benefits included higher agricultural outputs and increased use of modern inputs. Improved access to education and health facilities also had a positive impact on school enrolment and access to healthcare services (Wei-Bin, 2007).

5. Conclusion

The Gundo Lashu programme had a mixed effect. Most notably was that there was little evidence generated from the study to demonstrate that the road construction projects within the Gundo Lashu programme have had a direct impact in terms of reducing the poverty of the people living in the sampled study areas and project participants. The conclusion drawn from the analyses indicates that the programme has in fact achieved its economic development objectives regarding, among others, the number of jobs created, the gender breakdown of these employments, the total road length constructed and maintained, inter-community access and mobility. Furthermore, the programme was successful in improving the chances for those who participated to acquire employment in the long term. However, once the contract implementation and handover is completed, it is unlikely that the skills left behind would be sufficient to maintain the infrastructure. One can therefore conclude that the development of rural road infrastructure alone is not sufficient in tackling poverty; other mechanisms must be combined with road construction programmes.

A further conclusion of the study was that other the social dimensions related to poverty reduction, sustainable livelihoods and opening of access to market opportunities were not sufficiently achieved. The programme did not have a significant impact on the communities who participated in the projects compared with those communities and individuals who did not participate in the projects. The project participants' socio-economic status did not change significantly from those communities without projects.

The results from the study have made a significant contribution to the in-depth understanding of the role of labour-intensive road projects to socio-economic development. Examining the distributional consequences and effects of rural road investments on self-employment, community empowerment and sustainable livelihoods have been addressed extensively within various literatures. However, they provide limited critical appraisal on evidence from a labour-intensive road development programme and the changes taking place over the project lifecycle. The results from the study have the potential to disrupt the existing literature as it revisits the longstanding research problem on the contribution of transport to socio-economic development in general.

Notes