Once again, the bright and conscientious academics at Boston College's Center for Retirement Research (CRR) have issued a timely and insightful report on pension finances that should be studied closely by every professional and policy-maker in the public sector. This time, they have taken on the issue of pension spiking and the broader issue of how we calculate lifetime defined-benefit pensions in the first place.

Historically, pension plans have based their life annuity payments on the retiring employee's final salary. This "final average compensation" often includes the highest one to three years of the employees' earnings. The logic of these Final Pay formulas is that the pension benefit is designed to provide retirement replacement income at some fraction of the worker's final pay, so that they can afford to retire in dignity and with confidence that they can cover their expenses.

The problem with the formula is that employees, especially those in unions with seniority-based pay systems and practices, have figured out how to game the system with all kinds of maneuvers to inflate the final compensation of an employee just before they retire. Common abuses are the inclusion of overtime, accumulated sick leave, specialists' pay, bonuses and late-career promotions. This is not just a union problem, however, as the recent Bell, California, pension scandal so vividly reminds us. Management employees also game the system with late-life pay increases that were never funded in their earlier career years. In multi-employer pooled plans like Bell's, the former employers get stuck with the bills for outrageous pension costs for scandalous pay that they never approved.

This column focuses on one important feature of the CRR report: The mathematical defect of a defined benefit system that bases the pension on late-career earnings. We need to reform that calculation, and nowadays we have the computer systems to do it.

In the early days of government, the pension calculation had to be based on recent payroll records because many public employers had such primitive (often semi-manual) payroll systems and employment databases that it would have been virtually impossible to capture, retain and calculate the employee's lifetime earnings. That's no longer a problem, however, especially for new employees. One of the wonders of the modern digital age is that we can easily store all the employees' lifetime earnings in a database that can now be used to calculate the pension.

Pension actuaries calculate an assumed inflationary rate of salary growth, but pension plans always lose money when employees are promoted or otherwise earn significantly higher pay in their later years of employment. If you compare this math with the retirement benefits provided in a defined contribution system, the difference is obvious. Employees who hold a defined contribution account with previous contributions earned at earlier, lower pay rates, cannot possibly earn a retirement benefit as large as a pension, unless their investments outperform the pension fund average. Thus, the taxpayers bear all the excess cost for providing pensions to public employees whose earnings escalate rapidly over a career.

In statewide systems where employees enjoy portability from one employer to another, the problem is especially acute. Job-jumpers who pyramid their pay with promotions will receive benefits that nobody ever pre-funded. Police chiefs, fire chiefs, city managers, school superintendents and other similar highly compensated professionals who started out in entry-level jobs would be the most extreme cases of this inequity.

Don't get me wrong — I'm not opposed to promoting worthy public servants into higher-paying jobs. But the compensation reward for those promotions and upward-mobile job-jumps must be baked into the cash pay and deferred compensation plans in a visible, transparent way and not hidden from the public in the form of unfunded pension obligations that burden future taxpayers.

What the Boston College report suggests is that pensions should be calculated on a formula that awards lifetime retirement income as a multiple of lifetime earnings, not the Final Pay. I agree. Moreover, our governmental computers can handle this task now. And we also now have federal Medicare (tax) earnings reports that can be used to authenticate annual earnings if there is a dispute or missing data. At the very least, the calculation period should be extended to 15 or 20 years. This will require a refinement of the current pension multiplier, so that a worker can receive a reasonable replacement income at retirement, especially when inflation is a factor.

This approach also enables employers to transition fairly from a system where incumbent employees paid pension contributions for overtime and sick leave in previous years, which would then become part of the multi-year pension-base calculation. For new employees, such extraordinary compensation can then be excluded.

In states where courts and constitutions provide that pensions cannot be changed for incumbent employees, this change in the game rules would apply only to new employees. For states where incumbents' benefits can be adjusted, the best approach would be to install the new formula only for those under the age of 40.

Where pension funding has become an acute crisis for the employer, however, it may be necessary to invoke this strategy as a cost-cutting tool that applies also to senior incumbents if the law allows — unless they are willing to absorb a much larger share of the skyrocketing contributions required to pay their pensions under the current formulas. For example, workers under age 40 could be compensated using a 20-year final average, those now between 40 and 50 could use a 15-year final average and those now over 50 could use a 10-year final average. Another approach would place a ceiling or curb on pensions of older workers whose pay has increased by more than 6 percent compounded annually in the past decade and exclude excess compensation from the calculations.