The upsides and downsides of online labour

Labour costs are one of the biggest expenses for many organisations. With global competitive pressures, companies need to be able to scale up and downsize labour just as they do other resources. Having a significant proportion of full-time employees can limit flexibility to adapt to changing circumstances. So how can companies build in flexibility? One solution is to look for gig economy workers like those in online labour markets.

What are these markets? What are the benefits and risks? Our research, “Scoundrels or Stars? Theory and Evidence on the Quality of Workers in Online Labor Markets”, by Anne Farrell, Jonathan H. Grenier, and Justin Leiby, The Accounting Review, January 2017, Vol. 92, No. 1, pp. 93–114, found that online labour markets offer large numbers of workers who complete tasks quickly and often to a high standard.

The internet has given rise to hundreds of online labour markets that connect workers and employers — for example, Upwork, Guru, and CrowdSource. Perhaps the most famous is Amazon Mechanical Turk (MTurk). MTurk is a crowdsourcing labour market that allows employers to hire temporary workers for short, simple tasks that are difficult to automate, such as identifying music clips or the content of photos.

The relatively low pay, anonymity, and unsupervised work environments prompt some to dismiss this as a labour market with unskilled and unmotivated workers, but research suggests otherwise. MTurk — and similar platforms — may represent a significant untapped labour resource for firms, even for tasks that are more demanding than what could be considered typical MTurk tasks.

How they work

Here’s how many online labour markets work. An employer looking to hire someone for a one-time task sets up an account with an online marketplace. After some level of vetting by the platform, the employer can post a job to be completed by one or more contractors. The posting includes the deadline for completion, compensation, required skills, and any other terms and conditions. The employer can even require potential workers to pass skills or knowledge quizzes.

Potential workers browse the platforms to find and complete work that interests them. Once a worker accepts the job, the employer receives the end product within a very short time. The employer confirms the quality of the work and then uses the platform’s portal to pay the worker. Task completion speeds depend on the complexity of the task but can typically range from just minutes to a few days.

Suitable tasks

• Purchasing. To inform sourcing decisions on materials or products (from places including, but not limited to, Amazon), companies can set up an online labour task to find the “lowest price” or “highest quality given a price point” item. The business’s purchasing team would still place the order and sign the contract, but the time-consuming searches can be outsourced. This is likely to save money both in labour and cost of items purchased.
• Market research. Watching the market and catching trends are difficult and time-consuming activities. Hiring many online workers to search articles and websites can provide a solution. For example, restaurant food supplier US Foods uses workers on MTurk to gather information from sources such as menus and websites, and then uses the data collected in customer and market insight analyses.
• Training of machine learning. Artificial intelligence and machines are trained with cases of human experience. Online employment marketplaces can be cost-effective ways to generate human decision examples. For example, The Allen Institute for Artificial Intelligence in the US uses MTurk workers to build common-knowledge datasets that are then used to “train” machine-learning models.
• Translation. Machines cannot yet fully translate between languages. Monolingual online labour can be used to describe videos in the worker’s language, and then the computer can learn how to move between languages from these examples. This method, and others, can be cost-effective for translation.
• Labelling and categorisation. Organisations can use crowdsourcing labour to properly identify or categorise images or text.

In addition to the examples above, there are many other time-intensive tasks that can be done via an online labour market. These include transcription services, audio editing, and information gathering tasks (surveys).

Potential downsides

While online labour markets offer companies convenient and fast task completion, there are potential downsides.

Workers are rightly concerned that they have no labour protections and that employers routinely pay below-market wages for high-quality work.

Employers worry that unsupervised, unknown, remote employees will not work hard or that the quality of the work will be below expectations, so many companies are sceptical.

However, research suggests these are not unskilled workers who are suitable only for simple tasks. For example, our study found that MTurk workers in particular have the following characteristics:

• They are often qualified to do more complex tasks than those offered to them, so they may be a scalable workforce your company can use without having to significantly increase headcount or reduce output quality.
• They are comparably honest, provide quality output, and are as motivated as participants in other research studies who completed the same tasks although from different participant groups (MBA students and undergraduate accounting students).
• Many are intrinsically motivated to perform well on engaging computer-based tasks for a flat rate. For less-engaging tasks, a flat wage with small performance-based incentives is as effective as a much higher flat wage for motivating high performance.
• They score better than the general population, although not quite as well as business students, on financial literacy measures.

The gig economy is here. We use gig workers for transportation, food delivery, business tasks, and even healthcare. Understanding the skills of workers in and the mechanics of different online labour markets can increase a business’s flexibility to stay competitive. As a starting point it’s worth considering the merits and costs of hiring workers in online labour markets and, perhaps, running a low-risk trial.

Online labour marketplaces

Some of the many online labour marketplaces are:

• Amazon Mechanical Turk: Transcription, identifying audio or video clips, surveys, and online research (mturk.com).
• Clickworker: AI development, ecommerce, and retail marketing solutions (clickworker.com).
• CrowdSource: Copywriting, transcription, and other data management tasks (crowdsource.com).
• Figure Eight: Transcribing texts and annotating images to train machine-learning algorithms (figure-eight.com).
• Guru: Professional programming, design, and business services (guru.com).
• Microworkers: Visual and classification tasks (microworkers.com).
• Samasource: Computer vision and natural language processing tasks (samasource.org).
• Upwork: Data analysis, web development, writing, administrative support, and other business services (upwork.com).

—Anne Farrell, CPA (inactive), CGMA, Ph.D., is a professor of accountancy; Jonathan H. Grenier, CPA (inactive), Ph.D., is an associate professor of accountancy, both at the Miami University Farmer School of Business in Oxford, Ohio, in the US. Justin Leiby, Ph.D., is an associate professor of accountancy at the University of Illinois at Urbana-Champaign Gies College of Business in Champaign, Illinois. Ethan Retcher is an investment banking analyst at Lincoln International in Chicago. Margaret Shackell, CPA (Canada), Ph.D., is an assistant professor of accounting at Ithaca College in Ithaca, New York. To comment on this article or to suggest an idea for another article, contact Oliver Rowe, an FM magazine senior editor, at Oliver.Rowe@aicpa-cima.com.