This post is part of a series in which we discuss the principles of incentive design for blockchain platforms.

In a previous post, we discussed how incentive design is a much broader field than most of us probably think. The category of incentives that most often comes to mind when the topic arises is pay-for-performance incentives, which will be the focus of our next few posts in this series.

Many blockchain platforms attempt to crowdsource activities and resources by creating pay-for-performance systems that reward participants with tokens. One example of this is the block rewards used by Bitcoin. They have inspired a plethora of projects that aim to encourage contributions to decentralized platforms.

While the prospect of using tokens to incentivize countless activities may sound exciting, the benefits of pay for performance cannot be leveraged equally in all contexts. Understanding the tradeoffs involved in implementing this type of incentive is essential when designing a platform that achieves its vision.

What is pay for performance?

A pay-for-performance system uses incentives through financial compensation that is received by a worker or contributor. It is a function of a performance metric or metrics measuring the output of that agent’s efforts. For example, pay for performance is often used in sales, where a salesperson is paid a commission on the revenues generated or the number of units sold.

Pay for performance is only necessary in situations with a particular type of contractual incompleteness. These incentives are useful when the paying entity cannot specify key aspects of the desired outcome in advance. For example, because market conditions are variable, a business may want to avoid setting a precise sales target. In this instance, implementing pay for performance can be beneficial in sales. Similarly, the Bitcoin protocol does not specify the amount of computing power that each miner should contribute to secure its blockchain because in an open system it would be difficult, or impossible, to measure and enforce these contributions.

By using a pay-for-performance system, instead of directly compensating the agent for delivering a fully specified good or service, the payer can give variable rewards and let the agent determine the exact actions they will take in order to increase the probability of successful outcomes. In the case of Bitcoin, the miner who finds the correct hash first wins the block reward, while all other miners get nothing. Miners, therefore, are empowered to choose how much power to contribute.

Pay for performance incorporates two critical factors:

1. The agent is given discretion over a wide variety of choices that will impact the outcome. Bitcoin miners choose not only how much power to contribute but also what hardware to use and where to locate their rigs. Further, they are able to adjust their software to optimize their performance.
2. The monetary compensation received by the agent varies depending on what they deliver. In the case of Bitcoin, a miner can receive large rewards or potentially not get paid at all, depending on how often she can beat other miners at finding the correct hash.

Pay for performance involves known risks

Due to these features of pay for performance, such systems always involve risks to the parties on either side of a transaction, which can lead to two potentially detrimental effects:

1. When the desired performance has not been specified and agreed upon, the buyer or payer risks not getting what they want.
2. Sellers or service providers need to be paid more for the same level of performance so that they will accept the income risk that comes with participation.

In a pay-for-performance system, the ‘buyer’ cannot be certain that he will get what he wants because the ‘seller’ has the discretion to choose what to deliver. For example, while Bitcoin was intended to be very decentralized, unforeseen factors prevented that outcome. Over time, miners have become consolidated due to their choices regarding which equipment to use, how much power to supply, and how to organize themselves.

At the same time, a ‘seller’ in a pay-for-performance system cannot be sure if they will be paid at all, let alone how much they will be paid. In the case of Bitcoin, a miner only receives the reward when they win the race, even though they incur costs for each mined block. This has led smaller contributors to exit the market.

Pay for performance incentives are not always optimal

These two costs illustrate why using pay for performance can be counterproductive if you don’t need it — especially if you can specify what should be delivered. For Bitcoin, one can argue that the need for an open network outweighs the costs, but in many other contexts avoiding this type of cost is better. For these reasons, most consumer transactions do not involve pay for performance. Instead, a business typically offers a well-specified menu of goods or services. In those contexts, the contract, though implicit, is clear and well specified: the buyer pays the posted price, and the seller gives the buyer the item they paid for.

For both of these reasons, whenever the desired performance can be specified and compensated directly, using a specific execution contract is more efficient than using a pay-for-performance contract.

Designing incentives for impact

Once you have determined whether you will use pay for performance or a different structure, such as specific execution, you aren’t entirely out of the woods just yet. There is still a variety of economic design work that must be done to maximize value. For pay-for-performance systems, the performance metric(s) used and the size of the rewards delivered will both have an impact on what participants choose to do. For specific execution, the pricing mechanism will determine whether value producing trades occur.

Each platform design choice, starting with whether to use pay for performance, will have an impact on the economic functioning of the market and the outcomes it will produce. The effects include how decentralized the market is, how expensive transactions are, and who contributes resources and who gets served. It is essential that design choices are considered carefully with a profound understanding of the tradeoffs involved.