This thesis examines self-employed individuals in the UK labour market, creating a new division that differentiates those who sustain by self-employment from those who move between self-employment and employee jobs. The thesis begins with an exploration of the characteristics of the different self-employed groups, compared to employees; then estimates earnings returns to human capital accumulation for the different groups; and scrutinises labour market transitions before and after the 2008 financial crisis, to see if the recession impacted behaviours. The analysis begins by establishing the criteria for partition between the self-employed, based on observation of the labour market behaviours of workers in our sample. We use an amalgamated dataset, the British Household Panel Survey, from the years 1991 till 2008, and its successor the United Kingdom Longitudinal Study, from years 2010 till 2014, following the same respondents in the UK for 23 years. We identify three categories of workers: those who we consider as ‘sustaining’ in self-employment; those who ‘dabble’ in self-employment [switching between self-employment and employee jobs] and those who are employees. The first empirical chapter sets out the criteria for the categorisation of individuals into sustained and dabbled self-employed and employees and describes the motivation for this additional distinction – not least the fact that most labour market studies consider ‘employees’ and ‘the self-employed’ as separate and distinct categories of worker. In this first chapter, we compare the characteristics of individuals in our two categories of self-employed with those of employees in our sample, using a Multinomial Logit Model. The analysis lends support to our categorisation of the two sets of self-employed workers. The sustained self-employed have characteristics that we more traditionally associate with the self-employed, for instance they are older, more male-oriented, non-white, non-UK born, and non-English native speakers. In contrast, those who we consider as dabbling in self-employment exhibit a unique set of attributes that place them in a distinct position when compared to both the sustaining self-employed and employees. They are found to be more advantaged than the wage earners, with respect to their observed socio-economic characteristics; not from ethnic minority groups, more male-oriented, UK nationals, native English speakers, middle-aged group of workers with good health status, reporting higher degrees of achievement and working in higher skilled industries. Also, they are more skilled than the sustained self-employed, with respect to their qualifications and skill levels, but are not well embedded in self-employment like the latter group. Sustainers are better off with respect to their home ownership, report higher work satisfaction, have fewer working partners, with a higher percentage of previously self-employed parents. What we expected to find was that dabblers were a disadvantaged group, pushed into self-employment because they cannot have access to paid employment, but what we found was that their movement between forms of self-employed and employee jobs seems to reflect a labour market ‘power’ of sorts and not a deficiency. They can move between forms of employment depending on the returns they perceive and are pulled rather than pushed. Hence, this does not fit with our original expectations and does not align with the disadvantage theory on which we base the formulation of our hypotheses. This implies that we have a sequential of highly professional and advantaged portfolio workers possibly making the best out of self-employment and paid employment jobs as they arise. The second empirical chapter estimates the returns to formal education for the ‘amalgamated’ group of self-employed and paid employees, without considering any heterogeneity within or among this group of workers in our data. We then estimate the returns for our categorisation and compare the results. We use the Log-Mincerian function to estimate the earnings returns for our workers, and the Ordinary Least Square model, Random Effect model, Fixed Effect model and Instrumental Variable techniques to interpret the results, and to deal with the potential endogeneity of education. We also employ the Heckman selection model to account for non-random selection into employment. The results from the [preferred] Fixed Effect model indicate that the returns for additional years of education are lower for the dabbled self-employed in comparison to paid employees. Similar findings are also shown when using credentials, although in both cases dabblers report a higher number of years of education and higher levels of educational attainments than paid workers. This validates with our own hypothesis, where we argue that dabblers are not able to enjoy higher returns to their human capital accumulation than paid employees, because they are unable to secure or ensure long term paid employment. Hence their lower returns to formal education might be explained by their oscillating pattern between paid employment and self-employment. Another possible explanation is that their agile way of working does not help us truly capture their earnings returns and for that reason the aggregate returns are shown to be lower. Moving towards the sustained self-employed, we could not find any robust evidence about higher returns in comparison to the dabbled self-employed and the always employed, although the estimates found lead towards this direction but do not hold any significant values. Hence, we could not validate our own hypothesis, based on the extension of the personal control theory, where we argue that sustainers should enjoy higher returns than dabblers because they have more planning advantages, continue longer in self-employment and are more established than dabblers. Also, sustainers should enjoy higher returns than paid employees because they are not bound by organisational rules since they are their own boss, they have more control over their own work and better use of their personal human capital. Furthermore, we could not detect any differences in the earnings returns between the general/amalgamated group of self-employed and paid employees, whereas our categorisation showed that the two subgroups of self-employed are different from each other and from paid workers, with respect to their observed socio-economic and demographic characteristics and their earnings returns. Although our results do not show any differences in the returns of the sustained self-employed, the returns for our dabblers seem to be lower, even though they are on average more advantaged. Hence, at the very least, we have found a group of workers not previously identified in studies that seem to suffer from some form of labour market disadvantage, when we consider their returns to education and account for the selectivity in occupational choice and endogenous problems of education. By doing so, we have contributed to new micro-econometric evidence on the heterogeneity of earnings returns to education for the self-employed in the UK labour market and offered a new comparable type of heterogeneity in the labour market that can be looked through in other studies and established in other countries. The third and last empirical chapter of this study looks at the transitions of workers in our sample, prior to and after the economic downturn and impact of the 2008 financial crisis on the behaviour of our division groups. This chapter explores the short and long-term trends in self-employment in the UK labour market, the changes in the nature of jobs and the demand and supply of workers in the market, the growth in self-employment in the UK labour market following the recession and the policies adapted as a response to the crisis. The aim is to find out if the transition behaviour of our workers helps us explain the overall changes in growth in self-employment that occurred in recent years within the UK labour market.