What’s the Best Way to Pay Employees?

Companies have long struggled with ways in which to motivate employees. To that end, many organizations offer a mix of base salary, bonuses, stock options and a variety of benefits, including a retirement plan, health-care program and so on. But what’s the most cost-effective package for getting employees to deliver their best efforts? What, for instance, provides more bang for the buck — merit raises or bonuses?

To investigate the effects of pay, Michael C. Sturman, an associate professor with the School of Hotel Administration at Cornell University in Ithaca, New York, studied the employees of a large diversified corporation with a variety of service-related businesses. Although the organization has operations in 35 countries, Sturman conducted his research in just the United States because those employees were paid under the same system. He also wanted to avoid any potential effects that different cultures might have on the data. Altogether, he studied almost 700 employees who worked in the U.S. operations of the company from 2001 to 2004.

Sturman focused his research on three major components of employee compensation: base pay, merit raises and bonuses. All the individuals in his study were paid a base salary and were eligible for raises and lump-sum bonuses depending on their performance, as assessed by their supervisors. At the company, supervisors use a four-point scale to assess an employee’s performance: (1) fails to meet expectations, (2) meets expectations, (3) exceeds expectations and (4) significantly exceeds expectations. Specifically, Sturman was interested in determining the difference between “how much” employees got paid (the total monetary amount) versus “how” they got paid (raises versus bonuses, and a person’s total pay compared to the average for that position, based on compensation surveys).

Not surprisingly, the best predictor of future performance is the past. That is, an individual who has received exceptional ratings in the past will tend to receive high ones in the future. But that knowledge doesn’t provide guidance about how to pay those top employees to help maintain their performance, nor does it offer advice about how to get other workers up to speed. In his research, Sturman was particularly interested in how pay could be used to motivate employees so that, for instance, a mediocre worker would become a stellar performer. Thus, in his study, he controlled the data for past performance.

Sturman also considered an employee’s manager. Specifically, some supervisors strongly tied merit raises and bonuses to an employee’s performance. Other managers were far less diligent. By investigating those differences, Sturman could determine the effects of any pay-for-performance link. The goal was to understand how employees responded when their pay was strongly tied to their performance versus when it was only weakly connected.

The Dos and Don’ts of Pay

In general, Sturman found that “how” employees were paid was often just as important (if not more so) than “how much” they were paid. For instance, people’s salaries were unrelated to their future performance except with respect to the market going rate. The bottom line was that absolute pay wasn’t as important as relative pay: Employees who were being paid above the market average were more likely to improve their future performance. One explanation is that these individuals might have realized how good they had it so they pushed themselves to go the extra distance to ensure that they’d keep their jobs.

Conversely, people who were paid below the market rate tended not to perform as well in the future. One interpretation is that those individuals, feeling that they weren’t being treated fairly by the company, became demotivated. Apparently, employees know what they could be making elsewhere, and they adjust their behaviors accordingly. That finding was hardly startling; past studies in various fields have repeatedly shown that people’s happiness often depends on how they’re doing in comparison with their neighbors.

With respect to merit increases, Sturman found that employees receiving a large raise were more likely to improve their performance than those receiving a bonus of equal size. In fact, the effect of a 1% raise was about equal to that of a 3% bonus. Again, this result was not surprising because the enduring nature of raises makes them, in the long run, more valuable.

But then Sturman discovered something unexpected: Tying pay to performance for merit raises seemed to have no effect. In other words, making raises more contingent on performance did not appear to spur employees to do better at their jobs. A possible explanation is that, because base salary is very important to people (many, for example, set their household budgets by it), employees will be hesitant to take any risks that might endanger that income

(or potential increases to it). But the willingness to take risks is often what enables employees to excel.

When it came to bonuses, though, Sturman found that tying pay to performance had a strong effect. That is, making bonuses more contingent on performance had a substantial impact in motivating employees. Perhaps people perceive bonuses as “extra” pay; thus they are more willing to take risks with it, and that then helps elicit their top performance.

From a company standpoint, bonuses are more economical because they are onetime events, whereas raises carry over into successive years. (Because of that, raises become a fixed cost whereas bonuses do not.) But the effects associated with raises are larger than those associated with bonuses. So, then, what’s the most cost-effective system for awarding raises and bonuses?

To answer that, Sturman conducted various “what if” experiments, and the results are provocative. By performing a regression analysis to predict future performance based on how bonuses had affected performance in the past, he could investigate the potential effects of different bonus systems. Through this work, he discovered that if all managers were to tie bonuses strongly to an individual’s performance, the company would theoretically see an overall 16% increase in employee performance. The important thing here is that substantial improvement could be obtained without any increase to the payroll budget; instead, the company would merely be maximizing the allocation of that resource.

In another model, Sturman analyzed what would happen if the company were to change its annual average raise from 2% to 3%. Even though that total expense would be considerable, the company would see only a 2.2% increase in performance. But if that increase in average raise were implemented along with changes to how bonuses were allocated, the theoretical predicted increase in performance would rise to 19%.

Using Pay As a Strategic Mechanism

Sturman is the first to admit the limitations of his research. For one thing, he conducted his study at just one company (albeit a large, diversified one). Also, he did not study how pay might be related to employee retention. Could a bonus system tied strongly to performance lead to higher turnover, and if so would the company be losing just the C performers or would many A and B individuals leave, too? Moreover, Sturman investigated just three components of compensation (base salary, raises and bonuses). As many organizations scale back on health and retirement benefits, broader research to investigate the effects of such cutbacks will be sorely needed. Could increases in raises and bonuses, for example, be used to offset any reduction in employee motivation that might occur when other benefits are trimmed?

Those limitations aside, Sturman’s study provides valuable insights into ways in which an organization can provide pay-based incentives for motivating its workers. At the very least, he has shown that companies might do well to consider employee compensation as not just an unavoidable cost but as a mechanism for improving worker performance in strategic ways. Moreover, Sturman has shown that there is no single, ideal compensation system. In each case, the desired results must first be identified, and potential benefits need to be weighed against their expected costs.

Details of Sturman’s research are contained in his October 2006 paper “Using Your Pay System to Improve Employees’ Performance: How You Pay Makes a Difference,” which is currently available for download. For additional information, Sturman can be contacted at michael.sturman@cornell.edu.