Research reveals claw-back bonus schemes can lead to big losses

When sales bonuses are “loss framed”—that is, prepaid to salespeople then clawed back if they don’t meet their targets—the result is more sales. True or false?

Even though behavioral economics suggest that businesses could benefit by adopting the incentive, new research presents strong evidence against it.

The evidence comes from an unusually large field experiment featuring a car company, 294 car dealers and a whopping $66 million in bonus payments.

In all, 140 dealerships were assigned to the loss-framing/prepayment claw-back scheme. That led to an estimated 5% reduction in unit sales—resulting in more than $45 million of lost revenue over four months, said Lamar Pierce, WashU Olin’s Beverly and James Hance Professor of Strategy.

Pierce is one of three authors of “The Negative Consequences of Loss-Framed Performance Incentives,” forthcoming in the American Economic Journal: Economic Policy.  

The lost revenue could have been worse. If loss framing had been applied to the entire network of the car manufacturer’s dealers (as the manufacturer, in fact, originally wanted), the incentive scheme would have resulted in an annual revenue loss of more than $1 billion, the researchers found.

Pierce and coauthors Alex Rees-Jones at the University of Pennsylvania and Charlotte Blank, who was at Maritz during the research, cautioned the manufacturer against using the incentive companywide. Instead, their experiment randomized the payment of sales bonuses at 294 car dealerships, with the 140 in the prepayment group.

In addition to lost revenue, the study’s results indicate loss framing can increase in incentives for “gaming” behaviors at dealerships.

The findings are a cautionary tale for organizations designing incentive structures.

“They should reconsider the assumed benefits of loss framing in sales incentives,” Pierce said.

CarCo.

The research project began when a car manufacturer learned that behavioral economics literature encourages loss-framed incentives. The literature seemed relevant because the manufacturer provides monthly targets to dealers to sell cars—and it pays a big bonus for exceeding each target.

The car manufacturer in the study has models for sale at more than 1,000 dealerships across 50 states. The researchers’ access to its data was governed by a nondisclosure agreement, so they refer to the manufacturer as “CarCo.” Like other manufacturers, CarCo. uses incentive programs for certain vehicle models to motivate sales volume at its dealers.

CarCo. believed it could spur more effort among dealers by changing its existing post-payment scheme to a prepaid, loss-framed one. Toward that end, it recruited the authors of the paper to help it implement the new bonus scheme.

Pierce and the others recommended that they first test the policy in a randomized controlled trial before implementing it broadly. Many “nudge” interventions are low cost and may be effectively adopted even if their impact is low, but this one  differed: The costs of the prepayment of incentives would have been enormous given the scale of the bonus program.

To directly test if the benefits of prepayment merited the costs, the researchers set up 294 of the car dealerships with $66 million in bonuses randomized to pre- or post-payment.

Altogether, the 294 dealerships typically sold about 15,000 vehicles a month and pulled in monthly revenue of $600 million.

“The loss-framed prepayment scheme resulted in lower total sales,” Pierce said.

‘Gaming’ the incentive system

A decline in sales in one of two model groups drove the negative total sales of the bonus prepayment plan. This “small bonus” model group had substantially lower average sales volume, commensurate with a smaller bonus and thus a smaller potential loss.

The “large bonus” model group is estimated to have had a near-zero and statistically insignificant change in monthly sales.

The negative average effect of loss framing was influenced by multitasking incentives, the research found. When various dimensions are separately incentivized, as is the case for the two model groups, agents may benefit by “gaming” the incentive system: They neglect one dimension to attend to another with higher private returns.

The upshot: The overall effect of using loss framing was a reduction of average monthly sales. This negative effect was driven by a reduction of average sales for the small bonus model group. No effect was detected for the large bonus model group.

Under loss framing, the researchers found, dealers sacrificed low bonus group sales to ensure they hit targets in the high bonus group.

“In other words, loss framing might indeed have motivated them even more to hit the sales target, but in doing so induced them to sacrifice the smaller bonus for the larger one,” Pierce said.

These results suggest that one consequence of loss framing was a rebalancing of effort and resources across the two model groups. On average, the model group with the smaller potential loss was neglected, with this pattern reversing when the large potential loss was unavoidable.