The South African government's objectives of job creation and poverty reduction depend on the success of policies that promote the employment of unskilled labour. Over the past decade, unskilled jobs in the formal sector have been lost, while the demand for skilled capital and labour has risen. In sectors that heavily employ less-educated workers, capital intensity has increased. Recent economic growth has mainly benefited sectors that rely more on relatively educated labour, and in these sectors capital intensity has not significantly increased. This paper explores how growth and job creation depend on private sector choices concerning how to produce, i.e. the relative proportions of capital and skilled or unskilled labour employed. \n\nSouth Africa's rising capital intensity and increased demand for skilled labour are part of a global phenomenon. Cross-country analysis, however, suggests that South Africa's rising capital intensity is greater than what would be expected given international experience. An analysis of the capital-to-labour ratio by major sectors of the economy documents substantial heterogeneity in sectoral trends. The evidence is circumstantial, but it is consistent with the hypothesis that education-intensive sectors of the economy are not flowing with the economy's trend towards capital intensity. The two education-intensive sectors of the economy - financial services and trade - are growing more rapidly but without significant increases in capital intensity, and they are creating jobs. The sectors that are not education-intensive - namely mining, construction and manufacturing - are growing more slowly or contracting even as their capital intensity increases, and they are shedding jobs. \n\nThe paper develops a theoretical model that offers one possible explanation for this phenomenon. In the model, job creation depends on: the rate of capital investment; the nature of productivity growth; and the degree of substitutability of capital and labour. Rising investment rates are not sufficient to generate job creation. If productivity growth is labour- augmenting and labour cannot be readily substituted for capital, more investment can lead to job losses, even as the economy grows. A symptom of this is a rising capital-to-labour ratio. \n\nThe evidence discussed in this paper suggests a number of important policy implications. Given the preliminary nature of this research, the paper aims to illuminate issues on the table rather than to pose solid policy recommendations. Further research, particularly with more disaggregated sectoral data and micro-economic analysis, is required to derive robust policy conclusions. \n\nThe cost of capital in South Africa does not reflect the true cost to society of diverting resources to employing machinery and equipment. Enterprises lay off workers and replace them with automated machinery based on expected cost-benefit analysis that assesses the relative costs to the firm. The additional costs to society of increasing unemployment do not enter the calculation--yet they are substantial, in terms of social safety net costs (both public and private), as well as increased crime and social unrest. In this sense, policy that affects the relative prices of capital and labour is important. Yet, any policy mechanism is likely to succeed only in the medium to long term. Transitional policies are therefore important. \n\nThe final section of this paper examines policy options, which can be categorised into two groups. The first takes the trend towards skills- and capital-intensity as a given, addressing the consequences by stimulating growth to offset the unemployment impact or by strengthening social security. The second set of options focuses on reversing the trend through policies that affect relative prices, productivity or specific industrial interventions. Relative prices can be shifted through labour policy, tax reform, industrial subsidies, or monetary policy. The policy options are not mutually exclusive. For instance, social security can be strengthened in a manner that emulates wage subsidies, improving society's ability to cope with high unemployment while potentially addressing the underlying problem. \n\nSince some of the causes of increased capital intensity are difficult to modify, there is a need for comprehensive social security reform that can cope with social dislocation while promoting job creation in a developmental manner. Other causal factors are amenable to policy intervention. The government's current and proposed industrial policies aim in part to address factor price distortions, skills shortages, and bottlenecks that stall labour-intensive small and medium enterprise promotion. This paper suggests that support for these policies, to the extent that they focus on increasing labour intensity, will yield returns not only in terms of economic growth, but also job creation.