A quarterly series of gig economy activity in the regions of the UK is constructed. Patterns of regional linkages are identified and the implications of spatial patterns for policymakers, businesses, workers and institutions are highlighted.
The Labour Force Survey data on main job self-employment in key gig economy sectors are used to construct the series. These are then analysed using vector autoregression techniques to identify patterns in the data and provide provisional forecasts.
The incidence of gig economy activity is greatest in the London region, characterised by high population density and a concentration of service sectors in which gig economy work, particularly of a highly skilled nature, takes place. Growth of gig economy activity in other regions has been more modest. In London, the percentage of workers in the gig economy is expected to rise to around 6.5 per cent over the next few years, while in other regions, the percentage is expected to settle at between 3 and 4.5 per cent.
These are the first regional estimates to be provided of the extent of gig economy activity. This is important in the context of discussions about the future of work, not least because regional disparities imply the need for policies addressing insecurity to have a spatial dimension.
Emerald Publishing Limited
Copyright © 2019, Emerald Publishing Limited
The term “gig economy” is of recent origin. It was first recorded in the wake of the great financial crisis as workers who had been made unemployed during the recession scrambled to assemble portfolios of jobs that were typically part-time and of short duration. The development of internet platforms to facilitate such engagements has meant that many workers now gain employment of this kind. Particularly well known platforms include Uber, which serves as a market bringing together those wanting to hire and those willing to supply car journeys. The online platform offers considerable flexibility to suppliers, who are essentially self-employed. Consequently, the supportive structure provided by a long-term association between employer and employee is absent; many of the characteristics that define the institution of a “job” – such as continuity, progression and training and development – are absent too. In some sectors, such as journalism, freelancing has long been common, but the development of platforms such as Uber – or, in other contexts Upwork (services including IT freelancing), Deliveroo (food delivery) and TaskRabbit (odd jobs) – has resulted in a rapid increase in such activity. These contexts have different implications in terms of their spatial dimension. Unlike taxi rides, IT work can be undertaken remotely. The spatial development of the gig economy is therefore unlikely to be even, and the emergence of new types of work may take different forms across space – be that travel-to-work areas or broader sub-national units such as regions. Yet, the spatial dimension of the gig economy has not been the subject of any analysis in the UK. This, specifically in the context of regions, is the subject of the present paper.
The remainder of this paper is structured as follows. The next section provides a brief review of relevant literature. This is followed by a brief consideration of salient economic characteristics of the regions of the UK. Next comes a section on the data used in the present analysis. The results of a forecasting exercise are then reported. The paper ends with a discussion and conclusion.
The development of cloud computing has been crucial in enabling the development of digital platforms capable of bringing together buyers and sellers of services in a virtual environment, hence enabling the growth of the gig economy (Kenney and Zysman, 2016). These platforms create value and distribute it across the buyers and sellers that they bring together. While the platforms themselves are small operations, they enable, catalyse and leverage a huge amount of value creation; they have the characteristic of natural monopolies, since sellers (buyers) have an incentive to use the platform that attracts most buyers (sellers). In contrast to a traditional model in which employees do most work, these platforms are characterised by a model where value is created primarily by independent agents. The implicit contract of a long-term engagement between employer and employee is absent, and new institutions may be required to ensure the continuation of training, development and progression. The legal definition of new categories of worker, aimed at ensuring that workers in the gig economy are not disadvantaged relative to employees, is a likely consequence (Stewart and Stanford, 2017; Taylor, 2017).
Over the 10 years to 2015, there was a sharp rise in the incidence of non-standard work in the US (Katz and Krueger, 2016). This is evidenced by data from the Contingent Work Surveys conducted first by the Bureau of Labor Statistics and subsequently at Princeton as part of the RAND American Life Panel. The definition of non-standard work used by Katz and Krueger is quite broad and includes agency workers, workers with no guaranteed minimum hours, contract company workers and freelance workers. On this definition, the proportion of workers employed (in their main job) under such arrangements amounted to almost 16 per cent in 2015. Ten years earlier, the corresponding proportion was just over 10 per cent. Not all of these may be considered gig economy workers – for instance, independent contractors are self-employed but may be working on relatively long contracts. However, the evidence is consistent with the view that, while only being a part of a bigger picture concerning insecure employment, gig economy activity has increased.
This increase has been in evidence around much of the globe (Kässi and Lehdonvirta, 2016), as shown by the Online Labour Index (http://ilabour.oii.ox.ac.uk/online-labour-index/) which shows especially stong growth in North America, Europe, Oceania and Asia. In the UK, for example, Kitching (2015) notes the growth in freelance employment from 4 per cent of total employment in 1992 to 6 per cent in 2015. It should however be noted that this sets an upper limit on the extent of platform working, since much freelance work is traditional in nature and is not mediated by cloud-based solutions such as those identified by Kenney and Zysman (2016).
Relatively little research has been undertaken on the differential impact of the gig economy on different regions, however. An exception, which makes use of data on major metropolitan areas in the USA, is the work of Hathaway and Muro (2016), who find considerable regional variation in the growth of own account working in the transportation and accommodation sectors (typified by Uber and AirBnB, respectively). In both cases, Austin TX and San Francisco and San Jose CA are among the cities where the growth is fastest. In the case of transportation, this is likely due in part at least to the time at which major services such as Uber launched in particular cities. More generally, differences in adoption of sharing models might vary across space as a result of differences in the nature of the platform. For example, some platforms require the provider to own capital assets (such as accommodation for AirBnB or a vehicle for Uber) in a particular location; others (such as Fiverr) allow professional labour services to be traded across geographies. For the former group, in particular, location in an area where there is a concentrated population is important; Davidson and Infranca (2016) investigate the importance of urbanicity in stimulating growth in the gig economy. An interesting application that highlights the importance of co-location is provided by Chan et al. (2018) who study the role played by Craigslist as a platform for gigs in the prostitution industry.
Several recent papers have focussed further on the role played by cities in promoting innovation. Tvetkova et al. (2017) show that self-employment growth has been fastest in small metropolitan areas and slowest in rural areas. Geissinger et al. (2018) highlight the importance of agglommeration in promoting the growth of digital industries characterised by high levels of entrepreneurship.
Regions of the UK
For statistical purposes, the UK is divided into 12 standard regions. Some – Wales, Scotland and Northern Ireland – are defined on the basis of their distinct histories, cultures and political arrangements. As the nation’s capital, London is characterised by high population density and an economy that is quantitatively and qualitatively different from other regions. The remaining regions within England divide the country into roughly equal areas.
Some salient features of these regions are reported in Table I. London has a considerably younger population than other parts of the country. Its rapidly expanding economy attracts young workers from elsewhere in the UK and beyond, and high real estate prices encourage many retirees to leave the capital once they no longer need to be there to work. Elsewhere, variation in the proportion of older people may be because of the high amenity value of areas to which people choose to retire, but also to the depopulation of regions that do not offer sufficient attractive employment opportunities for young people. The concentration of young people in London is relevant for our consideration, later in the paper, of gig economy activity in the capital.
Industrial structure also varies markedly by region. The strong financial sector in London (and correspondingly low proportion of workers in manufacturing) is symptomatic of a high-end service economy. This too is relevant for observations about the gig economy that will come later in the paper.
The data that form the basis of the analysis that follows are drawn from all sweeps of the quarterly Labour Force Survey since the first quarter of 2002. Data from earlier years are not considered owing to changes in the standard occupational classification. There is not direct way of identifying gig economy workers in the Labour Force Survey – and indeed there is no clear definition of such workers – and so, following the approach of Kitching (2015), we calculate the percentage of all working respondents who are self-employed in (three-digit) occupations in which gig economy work is believed to be common. This covers gig workers in construction, transport and delivery, and creative and professional sectors.
Figure 1 shows the percentage of all those in work that are self-employed in major gig-economy sectors, by region of the UK, for each quarter from 2002. Several features stand out from this graph. The gig economy (at least as represented by this measure) is neither huge nor new. Taking all regions as a whole, there has been only a small increase in the measure over the most recent 15-year period. One region – London – stands out, both in terms of level (by the end of the series) and the rate of increase. In Figure 2, we report the slope coefficients from a series of region-specific bivariate regressions of this measure against time (quarter). The slopes are significantly positive in all regions bar two (Northern Ireland, where the slope is insignificantly negative, and East Anglia, where the slope is significantly positive only at 10 per cent), but the coefficients are, for the most part, small. Other than London only in the rest of the South East, the South West and the West Midlands does the slope indicate an increase of more than 0.5 percentage points over a 10 year period, and the rate of change in London (0.037 per quarter) is twice as great as that in the South West (which is the region with the next highest slope). In this respect, the gig economy seems to be very much a London phenomenon. Elsewhere the preponderance of gig work has risen only very slowly over time.
Indeed the nature of the gig economy in London is very different from that observed in other regions. In particular, two groups of professionals who work in sectors where gig economy work is common are relatively heavily represented in the capital. Media professionals (SOC 247) account for 1.8 per cent of all workers in London in the most recent quarter for which data are available, but for only 0.2 per cent of all workers elsewhere in the country. Moreover, information technology (IT) professionals (SOC 213) account for 4.8 per cent of all workers in London, but just 2.0 per cent in the other regions. Of self-employed workers in gig economy sectors, in London, some 15.0 per cent have occupations in media or IT, but the corresponding figure outside London is just 8.5 per cent. It is therefore potentially highly misleading to extrapolate information about the gig economy in London to the rest of the UK.
The place at which a service is provided matters in some markets but not in others. Building work must be done in situ. Taxi rides must be provided between two points determined by the consumer’s needs. Information technology services can usually be provided remotely, with solutions being transferred around the globe electronically. In some contexts, such as media, agglomeration economies come to the fore. Therefore, some types of gig economy work are spread widely across space, while others are more concentrated. In the taxonomy provided by Sundararajan (2016, Table III.1), this is embodied by the criteria identifying whether the provider “has in-person customer contact” or “has virtual direct customer contact”.
The future evolution of the gig economy will depend on factors that are difficult to predict, including legal and technological developments. Nevertheless, it is instructive to consider the dynamics implicit in the data reported in Figure 1 – looking in particular at whether these suggest convergence to a new steady state, both across time and across regions.
Consider then the design of a forecasting model for twelve series reported in Figure 1, over the period from the first quarter of 2002 through the first quarter of 2017. The structure of the forecasting model is determined by first considering the appropriate number of lags to be accommodated. In view of the relatively short length of the panel, we choose one lag. We next consider the cointegrating rank using the Johansen (1991) test for cointegration. The alternative hypothesis of a rank of one is rejected in favour of the null that there are no cointegrating vectors. We therefore model the system as a straightforward vector autoregression (VAR) rather than as a vector error correction model.
The results of the VAR are reported in Tables II and III. As evidenced by the diagonal terms in Table III, there is strong autocorrelation in every region except the East Midlands. The off-diagonal terms suggest that spillover effects from one region to another are generally quite weak. There is, for example, a significant positive spillover from London to the West Midlands, but not from the capital to any other region.
The forecasts over the four years to the first quarter of 2021 are reported in Figure 3 and show continued growth to be likely over the short term in London, rapidly converging on a new steady state. There is short-term growth also in the North and in Wales. In many other regions, however, the forecasting model suggests that the steady state is somewhat lower than the current level. The confidence intervals, indicated by the shaded regions in the figure, are, however, wide.
The steady state in London appears to be around 6.5 per cent. Elsewhere, it ranges from about 3.5 per cent to 5.5 per cent. It is lowest in the North and in Scotland, and (outside London) highest in the Rest of the South East and the South West. This reflects a clear spatial pattern. That said, the confidence intervals within each region are wide. Notwithstanding this however, the forecasts do suggest that, in the absence of further technological shocks, the prospect for further growth in the gig economy is limited, and that the broader base of opportunities for (especially white collar) gig work in the concentrated market of the capital city renders greater scope in that region for platform-based employment. In regions such as the Midlands (both east and west) and Northern Ireland, the steady state appears to be around 4 per cent; current levels of gig employment in these regions may be slightly higher than this owing to unusually high levels of activity in the construction sector.
Discussion and conclusions
Growing insecurity in the labour market has been a recurrent theme amongst many commentators (TUC, 2017), and has resulted in widespread interest in the future of work (Eichhorst, 2017). It would, however, be premature to conclude that the typical worker’s experience is set to change in a way that renders the concept of a job outdated. While activity in the gig economy has increased in recent years, this activity has been confined to a few sectors and is concentrated in regions where these sectors are heavily represented (Ainsworth, 2017). Moreover, work that is mediated through online platforms takes a variety of forms, with a wide range of skill requirements. The nature of the gig economy varies markedly across regions, involving different constituencies of worker and implying a different pattern of evolution. While some challenges are common to users of Upwork and Uber, others are not; given the spatial distribution of different types of gig work, the spatial dimension is of clear policy relevance.
More generally, gig work presents challenges of relevance to businesses, workers and policy-makers. The creation of a spot market changes the nature of interaction between demand and supply sides of the labour market in fundamental ways. In particular, the long term relationship that has traditionally been characteristic of trade in this market – the institution that we call a “job” – is eroded (Lazear, 1995, ch.7; Drahokoupil and Piasna, 2017). Consequently, there is erosion also of the effectiveness of mechanisms encouraging firms to make long-term investments in developing workers’ human capital (Becker, 1962; Friedman, 2014). Long-term engagement has allowed firms to finance workers’ training with confidence that (on average) workers are likely to stay long enough for the firm to reap the benefits in terms of enhanced productivity. In the absence of such confidence, new mechanisms may be needed to finance training. Workers without the necessary resources to make this investment face a market failure – financial markets will not offer loans in the absence of collateral, and in a non-slavery society, workers are not able to offer themselves as security. Therefore, there may be scope for other players – including governments, trade unions and employer organisations – to provide mechanisms whereby training can be supported. Such mechanisms might involve a pooling of resource by such organisations to provide collateral and avoid the moral hazard issues that would arise from unsecured lending. This is likely to be of greater importance for occupations characterised by the acquisisition of human capital – often, but not always, the white-collar occupations that, as we have seen above, tend to be concentrated in the largest population centres. Thus, the appropriate policy response to challenges posed by the growing gig economy has clear spatial implications.
The media often present the gig economy as a simple phenomenon. The taxonomies provided by Kenney and Zysman (2016), Sundararajan (2016) and others make clear that it is, in fact, multifaceted. The present paper has demonstrated that the variety of forms that gig employment can take has important implications for the extent of gig activity in different regions. These, in turn, suggest that the response of policymakers, employers and institutions to the challenges presented by the growth of the gig economy should likewise have a spatial dimension, recognising that the impact of new forms of work differs for workers with different skill sets and resident in different labour markets.
Characteristics of standard regions of the UK
aged 65+ (%)
in manufacturing (%)
in finance (%)
in hospitality (%)
|Yorkshire and Humber (YH)||18.40||11.29||4.11||5.54|
|East Midlands (EM)||19.10||13.01||2.69||4.39|
|East Anglia (EA)||19.49||10.32||3.68||5.36|
|Rest of South East (ROSE)||19.11||7.9||4.3||4.49|
|South West (SW)||21.78||7.67||4.15||6.57|
|West Midlands (WM)||18.36||12.02||4.18||4.94|
|North West (NW)||18.43||9.41||4.67||4.58|
|Northern Ireland (NI)||16.19||8.94||3.40||5.83|
Data in column 1 are population estimates published by Nomis and refer to 2017; data in the remaining columns are author’s calculations from the q1 2017 Labour Force Survey
Vector autoregression results – goodness of fit
|Yorkshire and Humber (YH)||0.2008||0.7432||173.6481|
|East Midlands (EM)||0.2149||0.4509||49.2602|
|East Anglia (EA)||0.3064||0.5746||81.0476|
|Rest of South East (ROSE)||0.1921||0.7346||166.0349|
|South West (SW)||0.2158||0.8044||246.6928|
|West Midlands (WM)||0.1823||0.8090||254.1442|
|North West (NW)||0.2078||0.6411||107.1669|
|Northern Ireland (NI)||0.3443||0.8328||298.8516|
Vector autoregression results – parameter estimates
|N−1||0.594 (0.118)||0.144 (0.086)||0.049 (0.092)||−0.027 (0.131)||0.119 (0.111)||0.075 (0.082)||−0.174 (0.092)||−0.032 (0.078)||0.024 (0.089)||0.258 (0.145)||0.080 (0.090)||0.315 (0.147)|
|YH−1||−0.006 (0.135)||0.529 (0.098)||−0.094 (0.105)||0.215 (0.150)||0.048 (0.127)||−0.198 (0.094)||0.173 (0.105)||0.237 (0.089)||−0.154 (0.101)||−0.380 (0.165)||−0.125 (0.103)||−0.193 (0.168)|
|EM−1||−0.036 (0.171)||−0.196 (0.125)||0.259 (0.134)||0.021 (0.190)||−0.112 (0.162)||0.014 (0.119)||0.081 (0.134)||0.049 (0.113)||0.083 (0.129)||−0.255 (0.210)||0.234 (0.131)||0.295 (0.214)|
|EA−1||0.034 (0.095)||0.055 (0.069)||−0.083 (0.074)||0.526 (0.106)||−0.085 (0.090)||0.101 (0.067)||−0.208 (0.075)||−0.144 (0.063)||0.013 (0.072)||−0.012 (0.117)||0.233 (0.073)||−0.154 (0.119)|
|L−1||0.072 (0.110)||0.109 (0.080)||−0.045 (0.086)||0.193 (0.122)||0.670 (0.104)||0.090 (0.077)||0.170 (0.086)||0.246 (0.073)||0.153 (0.083)||0.174 (0.135)||−0.204 (0.084)||−0.130 (0.137)|
|ROSE−1||−0.056 (0.140)||0.248 (0.102)||0.223 (0.109)||−0.184 (0.155)||−0.082 (0.132)||0.771 (0.097)||0.118 (0.109)||−0.066 (0.092)||0.092 (0.105)||0.031 (0.171)||0.251 (0.106)||0.089 (0.174)|
|SW−1||−0.037 (0.165)||−0.147 (0.120)||−0.044 (0.128)||−0.170 (0.183)||0.349 (0.156)||0.050 (0.115)||0.497 (0.129)||0.024 (0.109)||−0.205 (0.124)||−0.174 (0.202)||0.050 (0.125)||0.042 (0.206)|
|WM−1||0.065 (0.177)||−0.061 (0.129)||0.306 (0.138)||0.049 (0.196)||−0.002 (0.167)||−0.156 (0.123)||0.118 (0.138)||0.409 (0.117)||0.247 (0.133)||0.301 (0.217)||0.232 (0.135)||−0.088 (0.221)|
|NW−1||−0.013 (0.161)||−0.001 (0.117)||0.016 (0.125)||0.200 (0.179)||0.224 (0.152)||0.010 (0.111)||0.044 (0.126)||0.189 (0.106)||0.318 (0.121)||−0.005 (0.197)||−0.196 (0.122)||0.298 (0.201)|
|W−1||−0.042 (0.088)||0.012 (0.064)||−0.024 (0.068)||−0.202 (0.098)||0.051 (0.083)||0.052 (0.061)||−0.126 (0.069)||−0.134 (0.058)||0.072 (0.066)||0.549 (0.108)||0.176 (0.067)||0.091 (0.110)|
|S−1||0.067 (0.128)||−0.049 (0.093)||0.144 (0.100)||−0.000 (0.142)||0.170 (0.121)||−0.152 (0.089)||0.116 (0.100)||0.018 (0.084)||−0.128 (0.096)||0.150 (0.157)||0.595 (0.097)||−0.634 (0.160)|
|NI−1||0.009 (0.057)||0.036 (0.041)||0.014 (0.044)||−0.039 (0.063)||−0.008 (0.053)||−0.022 (0.039)||−0.094 (0.044)||−0.075 (0.037)||0.079 (0.043)||−0.073 (0.070)||0.071 (0.043)||0.744 (0.071)|
|constant||1.110 (0.881)||0.814 (0.642)||0.993 (0.687)||1.938 (0.979)||−0.654 (0.832)||1.391 (0.614)||1.231 (0.690)||0.838 (0.582)||1.161 (0.664)||1.803 (1.081)||−1.898 (0.671)||1.330 (1.100)|
Author’s calculations. Standard errors in parentheses
Percentage of those in work that are self-employed in gig economy sectors
The operating models of these platforms are subtly different from one another, with implications for the degree to which they resemble a competitive market structure. Malone et al. (1987) provide useful insight into the impact that product complexity and the specificity of asset use have on the boundaries of the firm (Coase, 1937), and this has been extended by Sundararajan (2016) to define numerous dimensions across which platforms – and hence also their market characteristics – differ. For example, does the platform match providers with customers, or is this done by the users themselves?
Clearly, any empirical work on the gig economy requires an operational definition – and we provide this in the first paragraph of the ‘data’ section of the present paper.
This estimate is also referred to by Brinkley (2016). The 6-16 per cent figures obtained by Kitching and by Katz and Krueger are well below some other estimates (for example Manyika et al., 2016; Horowitz, 2017). The nature of activity in the gig economy renders problematic the construction of a meaningful definition of its extent – many people have occasionally sold something on eBay, but would not regard this as their primary economic activity. Implications of the advent of the gig economy for measurement in surveys are discussed by Abraham et al. (2017).
There is, however, a larger and older literature on regional aspects of more broadly defined casual and informal employment – see, for example, Button (1984).
A map is available at: https://bit.ly/2RejlRB
On the 2010 Standard Occupational Classification (SOC) these are 213, 222 (322 on the 2000 SOC), 247 (343), 341, 344, 531, 532 and 821.
This is supported by the Bayesian Information Criterion (Schwarz, 1978), which reaches its minimum value of 6.26 at one lag. However, other decision criteria (Akaike’s Information Criterion and the Hannan and Quinn Information Criterion) suggest three lags would be appropriate (with values of -4.00 and 2.18 respectively). As a robustness check, we have run the forecasting model with three lags, and the results are qualitatively similar to the model with one lag.
The trace statistic is 265.16, with a critical value at 5 per cent of 277.71.
Sims (1980) was an early advocate of the use of vector autoregression methods. In essence, the method involves regressing each variable of interest – in this case, the rate of gig employment in each region – on lagged values of both the dependent variable and of all the other variables of interest. In this way, in our context, we can evaluate the impact of gig employment in each region on subsequent levels of gig employment in each other region, thereby allowing analysis of regional spillovers.
Some evidence on this is provided by Deloitte’s crane surveys; see www2.deloitte.com/uk/en/pages/real-estate/articles/regional-crane-surveys.html
Abraham, K. Haltiwanger, J. Sandusky, K. and Spletzer, K. (2017), “Measuring the gig economy: current knowledge and open issues”, available at: https://aysps.gsu.edu/files/2016/09/Measuring-the-Gig-Economy-Current-Knowledge-and-Open-Issues.pdf
Ainsworth, J. (2017), “Gig economy: introduction”, House of Lords Library Briefing, available at: http://researchbriefings.files.parliament.uk/documents/LLN-2017-0086/LLN-2017-0086.pdf
Becker, G. (1962), “Investment in human Capital: a theoretical analysis”, Journal of Political Economy, Vol. 70 No. 5, pp. 9-49.
Brinkley, I. (2016), “In search of the gig economy”, The Work Foundation, available at: www.theworkfoundation.com/wp-content/uploads/2016/11/407_In-search-of-the-gig-economy_June2016.pdf.
Button, K. (1984), “Regional variations in the irregular economy: a study of possible trends”, Regional Studies, Vol. 18 No. 5, pp. 385-392.
Chan, J. Mojumder, P. and Ghose, A. (2018), “The digital Sin city: an empirical analysis of Craiglist’s impact on prostitution trends”, Informaiton Systems Research, available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3057722
Coase, R. (1937), “The nature of the firm”, Economica, Vol. 4 No. 16, pp. 386-405.
Davidson, N. and Infranca, J. (2016), “The sharing economy as an urban phenomenon”, Yale Law and Policy Review, Vol. 34, pp. 215-279.
Drahokoupil, J. and Piasna, A. (2017), “Work in the platform economy: beyond lower transaction costs”, Intereconomics, Vol. 52 No. 6, pp. 335-340.
Eichhorst, W. (2017), “Labor market institutions and the future of work: good jobs for all”, IZA Policy Paper 122, available at: www.econstor.eu/bitstream/10419/162533/1/87818922X.pdf
Friedman, G. (2014), “Workers without employers: shadow corporations and the rise of the gig economy”, Review of Keynesian Economics, Vol. 2 No. 2, pp. 171-188.
Geissinger, A. Laurel, C. Sandström, C. Eriksson, K. and Nykvist, K. (2018), “Digital entrepreneurship and field conditions for institutional change: investigating the role of cities”, Technological Forecasting and Social Change, available at: www.sciencedirect.com/science/article/pii/S0040162517316992
Hathaway, I. and Muro, M. (2016), Tracking the Gig Economy: New Numbers, Brookings Institution, Washington, DC, available at: www.brookings.edu/research/tracking-the-gig-economy-new-numbers/
Horowitz, S. (2017), “How to create a fairer gig economy for workers”, New York Times, 24 April.
Johansen, S. (1991), “Estimation and hypothesis testing of cointegration vectors in gaussian vector autoregressive models”, Econometrica, Vol. 59 No. 6, pp. 1551-1580.
Katz, L.F. and Krueger, A.B. (2016), “The rise and nature of alternative work arrangements in the United States, 1995-2015”, available at: https://krueger.princeton.edu/sites/default/files/akrueger/files/katz_krueger_cws_-_march_29_20165.pdf
Kenney, M. and Zysman, J. (2016), “The rise of the platform economy”, Issues in Science and Technology, Vol. 32, pp. 61-69.
Kitching, J. (2015), “Tracking UK freelance workforce trends, 1992-2015”, in Burke, A. (Ed.), Handbook of Research on Freelancing and Self-Employment, Senate Hall, Shankhill, available at: www.crse.co.uk/sites/default/files/Tracking%20UK%20Freelance%20Workforce.pdf
Kässi, O. and Lehdonvirta, V. (2016), “Online labour index: measuring the online gig economy for policy and research”, Oxford Internet Institute, available at: https://mpra.ub.uni-muenchen.de/74943/1/MPRA_paper_74943.pdf
Lazear, E. (1995), Personnel Economics, MIT Press, Cambridge, MA.
Manyika, J., Lund, S., Bughin, J., Robinson, K., Mischke, J. and Mahajan, D. (2016), Independent Work: Choice, Necessity and the Gig Economy, McKinsey Global Institute, San Francisco, CA, available at: www.mckinsey.com/∼/media/McKinsey/Featured%20Insights/Employment%20and%20Growth/Independent%20work%20Choice%20necessity%20and%20the%20gig%20economy/Independent-Work-Choice-necessity-and-the-gig-economy-Full-report.ashx
Malone, T., Yates, J. and Benjamin, R. (1987), “Electronic markets and electronic hierarchies”, Communications of the ACM, Vol. 30 No. 6, pp. 484-497, available at: http://is.esade.edu/faculty/wareham/Teaching/StratNetComp/Readings/Electronic%20Markets%20and%20electronic%20Hierarchies.pdf
Schwarz, G. (1978), “Estimating the dimension of a model”, The Annals of Statistics, Vol. 6 No. 2, pp. 461-464.
Stewart, A. and Stanford, J. (2017), “Regulating work in the gig economy: what are the options?”, Economic and Labour Relations Review, Vol. 28 No. 3, pp. 420-437.
Sundararajan, A. (2016), The Sharing Economy, MIT Press, Cambridge, MA.
Taylor, M. (2017), “Good work: the Taylor review of modern working practices”, available at: www.gov.uk/government/uploads/system/uploads/attachment_data/file/627671/good-work-taylor-review-modern-working-practices-rg.pdf
TUC (2017), “The Gig is up”, TUC, London, available at: www.tuc.org.uk/sites/default/files/the-gig-is-up.pdf
Tvetkova, A., Partridge, M. and Betz, M. (2017), “Entrepreneurial and employment responses to economic conditions across the rural-urban continuum”, Annals of the American Academy of Political and Social Science, Vol. 672 No. 1, pp. 83-102.