Paying For Results
or the past two years, federal employees have sent a clear message about the way agencies evaluate and reward their workers: The system isn't working.
In 1999, the Clinton administration's National Partnership for Reinventing Government (NPR) reported that in its annual survey of federal workers, "employees expressed the greatest dissatisfaction with how employee performance is handled. Two out of three employees believe rewards are based on something other than merit; many cited bias and favoritism. Still more employees say that no action is taken against poor performers; many pleaded strongly for something to be done about this problem."
In the 2000 survey, the results were only slightly better. Only 31 percent of employees said they understood what constituted "good performance" and only a little more than a third said "recognition and rewards are based on merit."
Surveys by the Merit Systems Protection Board have had similar results. The bottom line is that the traditional evaluation and rewards approach, based on subjective ratings by supervisors, is an utter failure. When rewards are based on subjective ratings, employees distrust and resent the process. Last year, when GovExec.com, the Web publication of Government Executive, invited e-mail messages on the ratings system, responses flooded in. A comment by Joe Chandler of the Marine Corps Logistics Base in Albany, Ga., was typical: "I can't believe that there can ever be a fair performance rating system. There are always going to be personalities involved, be they good, bad or indifferent, and there will always be a supervisor, and no matter how good or bad they are, or no matter how hard they try, there will always be favorites, and there will most certainly always be a whipping boy."
Managers find the process equally troubling. Few want to give employees poor ratings and take on the unwelcome job of documenting and justifying performance decisions. Nor do they want to anger or demoralize most of their employees by singling out a few for superior ratings. So most simply bless all but the absolutely worst employees. Hence, most federal managers and employees get positive ratings, and financial rewards are so diluted they are meaningless.
Exactly the same thing happens in other governments. When Australia tried a bonus pay program, for instance, 94.4 percent of those eligible got bonuses the first year.
Because it rewards marginal and superior employees equally, the traditional system promotes mediocrity. Superior performers become demoralized and cynical, either reducing their efforts or leaving their jobs.
Last year, the Clinton administration began to shift the focus from subjective evaluations to objective results. NPR officials had already worked with 32 agencies identified as having the greatest impact on the public to help them develop systems of "balanced measures" that include three basic elements: the operational results they measure under the Government Performance and Results Act; customer satisfaction; and employee satisfaction. Last year, the President's Management Council, an organization of top management officials at agencies, issued a policy requiring agencies to refocus their performance systems on those three performance factors. "Formal and informal recognition programs should be linked to desired performance outcomes," the council said.
The administration immediately required the new approach for political appointees, and last October the Office of Personnel Management released regulations putting it into effect for members of the Senior Executive Service.
"The regulations require agencies to evaluate senior executive performance using measures that balance organizational results with customer satisfaction, employee perspectives and any other measures agencies decide are appropriate," OPM said.
This practice, known as performance management, is the next logical step toward results-oriented government. Federal agencies that learned how to define and measure performance goals in the 1990s should now start learning how to reward people for achieving those goals.
A few agencies already have pioneered results-oriented performance management. The Veterans Benefits Administration (VBA), for example, uses a balanced scorecard for measuring agency performance. It includes five elements: timeliness, quality, customer satisfaction, unit cost, and employee development and satisfaction. VBA put together an executive appraisal system built around these and other factors to determine bonuses for senior executives. (By law, these bonuses range from 5 percent to 20 percent of salary.)
Now VBA is driving this approach down through the ranks. Last year, each of VBA's nine service delivery networks began receiving a quarterly pool of bonus money based largely on the performance of its "stations" (or regional offices). Each service delivery network awards bonus funds to its stations, which can decide how to distribute the money to employees. VBA officials are encouraging the stations to reward team performance, but aren't requiring any particular model. "We've encouraged folks to take the balanced scorecard down to the team level," says Mike Walcoff, associate deputy undersecretary of Veterans Affairs for operations. "For example, in Los Angeles, each team has a balanced scorecard. That is ideally what we want to see: that we're organized in teams, and that each team has a scorecard that measures their performance, and that they are rewarded based on their performancev as a team."
The IRS has also developed a set of balanced performance goals. This fiscal year, for the first time, SES members will be eligible for performance bonuses equal to 5 percent to 20 percent of salary based on a new appraisal system that combines objective results on performance goals with subjective ratings of their accomplishments. IRS Commissioner Charles Rossotti has also approved a broad pay band for senior managers below that level, where the agency will offer performance bonuses in the 5 percent range.
Managers' salaries also will rise based on performance. "Today, they advance based on time in grade," explains Chief Human Resources Officer Ron Sanders. The new system "will move them through based exclusively on performance evaluation. The principle is: the higher the pay, the higher the expectation. So as someone moves up in salary range, they have to perform better to get base pay increases, with a comparable formula for bonuses."
But the real leader is the U.S. Postal Service, which has far more freedom to innovate than other agencies because it is not subject to civil service regulations. In 1995, the Postal Service adopted a balanced measures approach and radically revamped its performance management system. Today, everyone and every team in the 800,000-member postal workforce has performance goals based on "the voice of the customer, the employee and the business."
All 84,000 managerial and executive staffers at the Postal Service have a smaller number of team goals that determine their lump-sum performance bonuses under the agency's Variable Pay Plan. The goals all focus on on-time deliveries, net income, productivity improvement, employee satisfaction (based on surveys), lost workdays due to injury and similar measures. Over the past five years, the average bonus for postal executives (the equivalent of SESers) has ranged from 9 percent of salary to 14 percent. Other managers, supervisors and professionals have averaged bonuses of 5 percent to 7 percent.
The Postal Service also did away with cost-of-living adjustments for managers, supervisors and other professionals, and created a merit pay program in their place. "The merit pay program is the way that people get [salary] increases, based on individual assessments of a person's performance," explains Paul Weatherhead, manager of pay programs. "We try to . . . tie the person's individual objectives back into the goals of the organization. For instance, human resource people have a balanced scorecard and could get [bonuses] based on three voices: the customer, the employee and the business. But as a function, HR probably has a greater impact on the voice of the employee. Finance people probably have a greater impact on the voice of the business. The merit pay program allows management to take it down and personalize it."
Finally, the Postal Service uses a separate recognition program to reward outstanding individual efforts. These rewards range from awards certificates to cash payments as high as 20 percent of salary for executives. (This maximum cash award is limited to 50 people a year). Postal workers' unions have resisted performance pay of any kind, but union members are also eligible for both financial and nonfinancial recognition awards for extraordinary efforts.
The results of the Postal Service's Variable Pay Plan have been dramatic. Since the program was begun:
- On-time scores for express mail have jumped from 82 percent to 94 percent.
- Lost workdays due to injuries have fallen from 3 percent to less than 2 percent.
- Personnel costs have fallen from 82 percent of operating expenses to 78 percent. Other strategies no doubt had an impact on these results. "But I think most executives and officers here will point to the Variable Pay Plan as a major contributor," says Weatherhead.
As the Postal Service's experience makes clear, performance management works. Success in the rest of the federal government will require two things: a focus on results that matter to citizens and customers, and a shift from subjective ratings to objective measures of performance. As OPM's Handbook for Managing Employee Performance says, "Too often, employee performance plans with their elements and standards measure behaviors, actions or processes without also measuring the results of [the] employee's work." And as correspondent after correspondent told GovExec.com last year, subjective ratings result in resentment, anger and anxiety.
The solution is to tie financial incentives to objective measures of performance, although there may need to be room for some subjectivity to adjust for realities outside the organization's control. A snow-plowing crew, for example, may do a heroic job in a bad winter and still end up with bad numbers. Conversely, if there is little snow, they'll almost inevitably look good.
As agencies make the shift to basing rewards on objective measures, here are 10 lessons they should keep in mind:
Money matters more than most people will admit, but so do psychological rewards, such as recognition and increased responsibility. There are dozens of ways leaders can publicly recognize employees' performance, from awards to newsletter stories to flowers, gift certificates, tickets, vouchers for a dinner out or even attendance at conferences or workshops. Some agencies, such as the Patent and Trademark Office, have created Employee Peer Satisfaction Awards, through which employees can recognize fellow workers.
2. Magnify the power of incentives by applying them to groups as well as individuals.
Most results are produced by groups of people working together, not by a single individual. In those cases, use collective rewards for performance. Unless individual work leads directly to outputs you can measure, you are better off rewarding teams. This binds employees to a collective purpose. At the Postal Service, says Weatherhead, "The key thing is we made it a group incentive program, so there were cooperative behaviors." At the Veterans Benefits Administration, if you are a regional service officer, 50 percent of your evaluation is based on results in your Service Delivery Network, 35 percent is based on results in your regional office, and 15 percent is based on overall results at VBA. The agency wants "people to not only think about how I'm doing in Phoenix, but . . . how I'm going to give resources to help Los Angeles," says Walcoff.
Team-based rewards are most effective if the teams have real control over their work. Can they get rid of bad apples? Can they change their work processes? Can they hold their internal suppliers accountable for their performance? Can they buy what they need elsewhere if support agencies let them down? If they can do these things, they can usually improve their performance dramatically.
3. Use lump-sum performance bonuses, not salary increases.
Salary increases are expensive, because they cost money not just in the year they are earned, but every year thereafter. Budget officials typically get nervous about escalating personnel costs and soon cap these costs, undermining the value of the tool.
Outstanding employees also hit the top of their salary range after several pay increases, at which point the tool becomes useless. When that happens, good employees often begin looking for jobs that pay more-an unintended consequence of the worst kind.
Salaries should be determined by what it takes in a particular labor market to attract and keep talented employees. To reward high performance, use bonuses.
4. Make performance bonuses big enough to get people's attention.
People with experience in performance management recommend bonuses that are at least double the size of a regular pay increase. The old Performance Management and Recognition System for federal managers failed in the 1980s in part because the bonuses and merit increases were simply too small. "Our managers told us, 'It's not worth all the time and effort it takes, for $1,000 or $1,500,' " says the IRS' Sanders. A good minimum threshold for bonuses is 5 percent of salary.
5. Avoid arbitrary targets.
One of the difficulties of any reward program is picking performance levels that will trigger rewards. W. Edwards Deming, the pioneer of total quality management, railed against using numerical targets. Among his famous 14 points, the 10th is: "Eliminate slogans, exhortations and numerical targets."
Deming argued that you only know how much better performance can be after you have improved your work processes. Hence, he reasoned, all targets are arbitrary and negative. Arbitrary targets create cynicism among the workforce. When workers fail to reach targets that are set too high, for example, they are blamed unfairly. If they exceed targets, they are praised and rewarded-regardless of how difficult or easy the improvements were.
Deming was right, but he went too far when he railed against all performance targets. The businesses with which he worked never eliminated all of them. They still measured profit and loss, market share, return on investment, net worth and customer satisfaction. And they faced real consequences if they succeeded or failed in hitting their targets. Deming may have been right to say that they did not need performance targets for individuals, but that is not the same as saying objective targets weren't needed for the entire organization, or for its divisions and work units. The best solution to this apparent conundrum is the one pioneered by the city of Sunnyvale, Calif., which in the 1980s was the first American government to create a successful performance management system. Sunnyvale simply measured current performance and created rewards for improvement. Managers in Sunnyvale whose units demonstrate significant improvement are eligible for raises and bonuses of up to 10 percent of their salaries. But the performance level they have achieved becomes their expected target.
6. Create winners, not losers.
Many programs limit the number of bonus winners. This creates a competition for bonuses, turning many employees into losers, even though they are improving their performance. Instead, bonuses should be tied to significant improvement, and those who improve more than others should get bigger bonuses. That way, people compete against themselves, rather than each other. This may be expensive, but the investment will be repaid in spades. At the Postal Service, "there has been some big money paid out," says Weatherhead. "But it has been worth it because of the organization's performance."
7. Involve employees, owners and customers in negotiating performance goals.
Setting performance goals should involve at least three parties: the organization, including both its leaders and employees; a neutral executive agency committed to performance improvement, such as the National Partnership for Reinventing Government; and something like a customer council. The nonprofit Fund for the City of New York, for example, has pulled together focus groups from neighborhoods to help several departments develop better performance measures. If possible, those involved in working out performance targets should have information about the performance of comparable organizations.
8. Don't make reward formulas too complex.
Many agencies at all levels of government have made this mistake. In an effort to create the fairest formula, they include so many goals and such complex weighting formulas that few employees can understand the process. And what people can't understand, they tend to ignore. Concentrate on the vital few results you want the organization or team to produce. You don't have to include everything you measure in the reward formula. You might use some data for feedback and learning, not for rewards. For example, a 360-degree evaluation, in which employees rate supervisors as well as the other way around, provides wonderful feedback. But if you attach financial consequences to it you may find that employees won't deal with it honestly and that it triggers serious trust and morale problems.
9. Develop a continuum of negative consequences.
Performance management giveth and, yes, it taketh away. Although you should rely primarily on positive incentives, you need clear signals when performance is not good enough. Without them, your system may lack credibility with both elected officials and employees. Start with training and help. If that doesn't work, make sure you have swift, defined penalties at the ready.
Stay away from financial penalties for poor performance, except at the very highest levels, such as agency directors and Cabinet secretaries. The threat of losing income fuels fierce resistance to performance management. And financial sanctions can curb innovation; if failure could mean losing pay, employees are less likely to try new things.
Organizations that use work teams find that peer pressure can be a powerful disciplining force. Several other types of negative consequences can be designed into performance management systems:
- Negative publicity. Most people don't want the world to know they are failing.
- Loss of privileges. If performance falls below certain levels, travel and other privileges can be revoked until it improves.
- Loss of autonomy. When organizations fail to perform, you can take away managers' control over their budgets or their flexibility in other areas. If you then provide the coaching and/or training they need, things sometimes turn around.
- Demotion. If poor performance continues despite training, help and sanctions, the employee is probably in the wrong job.
- Dismissal. This is the ultimate sanction. Surveys of federal employees show that it is rarely used, because most managers believe it is too difficult to fire someone for poor performance. The President's Management Council white paper on performance management urged that the laws covering termination be simplified, while preserving due-process protections.
If you push too hard on narrow results, people will concentrate on reaching their goals at all costs, regardless of whether that makes sense for the rest of the agency. To use a manufacturing metaphor, factories will meet their quotas, even if it means shipping junk.
The solution is to reward experimentation and learning as part of the organization's performance management system. Focus on rewards far more than penalties, and include rewards for innovation. Create a culture of trust, not fear.
Because performance incentives can be so powerful, federal managers must carefully select the results they intend to reward. If goals are all short-term, that's what people will work on-to the detriment of the long-term health of the organization. This is why the Postal Service puts two-thirds of its variable pay bonus each year into a reserve account, which is then paid out based on performance in future years. "You want to maximize performance over a three- or four-year period," says Weatherhead. If you emphasize productivity, but not effectiveness, that's what you will get. "Most program administrators will do whatever they can to generate good numbers," says Roger Vaughan, a veteran consultant to state and local agencies.
A striking example occurred at the IRS. Before Charles Rossotti took over and shifted to balanced measures, the agency measured employees' performance largely on how much money they collected. Managers' evaluations were tied to how well they achieved these goals, and the IRS' 33 district offices were ranked by how well they met collection targets. Critics said this approach placed taxpayer rights at risk. Members of Congress declared it a scandal in a series of highly publicized hearings. This is why a balanced approach, with a mix of business results, customer satisfaction goals and employee satisfaction goals, makes sense. As Vaughan says, "the wrong measure of performance can be-and has been-more harmful than no measure."
David Osborne, a partner in the Public Strategies Group, a consulting firm, is co-author of Reinventing Government (Addison-Wesley, 1992) and Banishing Bureaucracy (Addison-Wesley, 1997). This article is drawn from his latest book, The Reinventor's Fieldbook: Tools for Transforming Your Government (Jossey-Bass, 2000).