Nobel prize winners’ research worked out a theory on worker productivity – then Amazon and Deliveroo proved it wrong
Financial incentives are important. We all know that’s true. If you were offered a job that paid £10 an hour and then someone else came up offering to pay you £11 an hour for identical work, which one would you choose?
Most of us would also accept that well-designed employment contracts can get more out of us. If we could take home more money for working harder (or more effectively), most of us would.
Bengt Holmstrom won the Nobel economics prize this week for his theoretical research on the optimum design for a worker’s contract to encourage the individual to work as productively as possible.
The work of Holmstrom and his fellow Nobel laureate, Oliver Hart, is subtle, recognising that the complexity of the world can cause simplistic piece-rate contracts or bonus systems to yield undesirable results.
For instance, if you pay teachers more based on exam results, you will find they “teach to the test” and neglect other important aspects of children’s education. If you reward CEOs primarily based on the firm’s share price performance you will find that they focus on boosting the short-term share price, rather than investing for the long-term health of the company.
Holmstrom and Hart also grappled with the problem of imperfect information. It is hard to measure an individual worker’s productivity, particularly when they are engaged in complex tasks.
So how can you design a contract based on individual performance? Holmstrom’s answer was that where measurement is impossible, or very difficult, pay contracts should be biased towards a fixed salary rather than variable payment for performance.
Yet when information on an employee’s performance is close to perfect, there can also be problems.
The information problem seems to be on the way to resolution in parts of the low-skill economy. Digital technology allows much closer monitoring of workers’ performance than in the past. Pickers at Amazon’s Swansea warehouse are issued with personal satnav computers which direct them around the giant warehouse on the most efficient routes, telling them which goods to collect and place in their trolleys. The devices also monitor the workers’ productivity in real time – and those that don’t make the required output targets are “released” by the management.
The so-called “gig economy” is at the forefront of what some are labelling “management by algorithm”. The London-founded cycling food delivery service app Deliveroo recently tried to implement a new pay scale for riders. The company’s London boss said this new system based on fees per delivery would increase pay for the most efficient riders. UberEats – Uber's own meal delivery service – attempted something similar.
Yet the digital productivity revolution is encountering some resistance. The proposed changes by UberEats and Deliveroo provoked strikes from their workers. And there is a backlash against Amazon’s treatment of warehouse workers.
It is possible that some of this friction is as much about employment status as contract design and pay rates. One of the complaints of the UberEats and Deliveroo couriers is that they are not treated like employees at all.
It may also reflect the current state of the labour market. If people don’t want to work in inhuman warehouses or for demanding technology companies, why don’t they take a job somewhere else? But if there are not enough jobs in a particular region, people may have no choice. The employment rate is at an all-time high, but there’s still statistical evidence that many workers would like more hours if they could get them.
Yet the new technology does pose tough questions about worker treatment. And there is no reason why these techniques of digital monitoring of employees should be confined to the gig economy or low-skill warehouse jobs.
One US tech firm called Percolata installs sensors in shops that measure the volume of customers and then compare that with the sales per employee. This allows managements to make a statistical adjustment for the fact that different shops have different customer footfall rates – it fills in the old information blanks. The result is a closer reading of an individual shop worker’s productivity.
Workers who do better can be awarded with more hours. “It creates this competitive spirit – if I want more hours, I need to step it up a bit,” Percolata’s boss told the Financial Times.
It’s possible to envisage these kinds of digital monitoring techniques and calculations being rolled out in a host of jobs and bosses making pay decisions on the basis of detailed productivity data. But one doesn’t have to be a neo-Luddite to feel uncomfortable with these trends. It’s not simply the potential for tracking mistakes by the computers and flawed statistical adjustments that is problematic, but the issue of how this could transform the nature of the workspace.
Financial incentives matter, yet there is rather more to the relationship between a worker and employer than a pay cheque. Factors such as trust, respect and a sense of common endeavour matter too – and can be important motivators of effort.
If technology meant we could design employment contracts whereby every single worker was paid exactly according to his or her individual productivity, it would not follow that we necessarily should.