I’ve written over the past couple of weeks about public pensions in Illinois. Short version: they’re a possiblyunfixable mess. Since the state constitution forbids reducing promised benefits for current employees (or increasing contributions) and the state has failed to plan for their pension promises in a timely manner, the state is stuck between the proverbial rock and a hard place.
With this in mind, over the past few days I’ve been trying to think of unconventional ways in which the state can save money. This is a bit tricky, since in the case of public employees the state pays their salary when they’re working AND their pension when they retire. It’s the overall cost that matters. So, for example, when the University of Illinois had an early retirement program last year, the University stopped paying them and SURS, the state university retirement system, started paying them. But since both are ultimately using state dollars (but lessso in the case of the University, whose state appropriation as a fraction of overall cost has fallen drastically in recent years), this is really just a reshuffling of which pocket the money comes from. The state is still on the hook.
Thinking outside the box leads to some crazy ideas. And here’s one of them. I make no promises about whether it will work in practice. But it does point to some of the strange features of state finance.
Here’s the idea: to help the state’s pension system’s finances, the state should pay its workers more as they near retirement. That’s right. More.
As I started to play around with the idea, I had a dim recollection of reading something related. It turns out that a couple of weeks ago, Andrew Biggs wrote in the Wall Street Journal about how cutting the Social Security payroll tax for workers nearing retirement could actually help the system’s finances. The idea is simple: if older workers get to keep more of their wages, they’ll work longer. And, if they’re working, they’re not collecting Social Security. Lowering the payroll tax pushes back retirement, and this helps the system’s finances. The idea is also related to my post from two weeks ago, where I discussed research showing that retiree health benefits induce early retirements. If the state can’t pay retirees less and can’t ask them to contribute more, the only thing it can do to reduce pension costs is induce them to retire later, and it needs to do so in a way that costs less than the potential savings from delayed retirement.
So, how does it work? Consider a worker near retirement age who has been working for the state his whole career, or at least long enough to reach the earnings cap on the state’s retirement system. This worker, let’s call him Charlie, will earn 80% of his final salary after retirement. And, assuming this worker was actually fulfilling a necessary function (e.g., teaching students finance), that worker will have to be replaced after retirement by a new worker. Let’s call him David. New workers tend to earn less than senior workers, so David will earn less than Charlie did. Maybe David earns 80% of Charlie’s final salary. But, essentially, after Charlie retires the state will be paying both Charlie and David – two people – to do work that could be supplied by one person. While the state paid 100% of Charlie’s salary for that work before retirement, it pays 160% of Charlie’s salary after retirement!
So, the state has the potential to save a lot of money overall – 60% of Charlie’s salary per year – if it can induce Charlie to delay retirement. Due to the nonimpairment clause, a lot of the ordinary ways of doing this such as increasing the full retirement age are off the table. One thing the state can do is increase Charlie’s salary. This could be done through an actual wage increase or, as Biggs suggests, by reducing the 8% of wages that Charlie must pay into the retirement system as he nears retirement.
It is easy to see how it might be worth it to the state to spend more money on Charlie’s wages in order to delay his retirement. But, let’s make up some simple numbers. I’m going to ignore things like the fact that pension payments increase 3% per year and other details of the retirement system. They don’t change the basic insight, and the uncertainty involved with the other numbers that I’M JUST GOING TO MAKE UP is a much bigger deal than details like this. I’m illustrating – not proposing policy.
So, suppose that increasing Charlie’s wage by 10% per year leads him to delay retirement by 3 years. Suppose Charlie makes $50,000 per year and has maxed out his service so he’ll earn 80% of that ($40,000) after he retires. Assume that David will earn $40,000 after he’s hired.
There are two things that should be taken into account. If Charlie’s wage goes up, the basis for his pension will go up as well. Roughly speaking, pensions are based on average earnings over the last four years of work. Over these years, Charlie earns 50,000 for one year (the year before he gets the raise) and 55,000 for three years (after he gets the raise). His final pension is 80% of the average, or 0.8 * 53,750 = $43,000 per year. Again, there are subtleties to the formula, but too many details obscure the main idea. And, if Charlie works additional years, he will pay an additional 8% of salary into the pension system. This would seem to be money that the state gets back. But, as far as I can tell, these “excess contributions” are refunded to the employee at retirement. So, in the case of a worker who has maxed out his pension, there would be no additional benefit to the state. (For a worker who has not maxed out their pension, the state would receive additional contributions from the worker who delays retirement, but it would also have to pay an additional 2.2 percent of final earnings for each additional year of work, so it is unclear that this would benefit the state.)
Total 10 year cost if Charlie retires now:
Charlie’s pension payments: 
10*40,000 = 
400,000 
David’s wages: 
10*40,000 = 
400,000 
Total Cost: 

$800,000 
Total 10 year cost if Charlie retires in 3 years:
Charlie’s wages (years 1 – 3): 
3*55,000= 
165,000 
Charlie’s pension (yrs 4 – 10): 
7*43,000= 
301,000 
David’s wages (yrs 4 – 10): 
7*40,000= 
280,000 
Total Cost: 

$746,000 
So, the total savings over 10 years from my COMPLETELY MADE UP numbers is $54,000, or 6.75% of the cost under the current system. And, this savings occurs in the first three years from not having to pay David. Although I’ve ignored the timevalue of money to keep things simple, the fact that the savings come up front would favor giving Charlie the raise if there were a positive interest rate.
Whether a scheme like this could actually save money would depend on a lot of things. Among them are how much more nearretirees need to be paid to delay retirement, how long the delay retirement for, the relative cost of replacement workers, the length of time over which retirees draw pensions, and the timevalue of money. Again, for the purposes of illustration, I COMPLETELY MADE UP THE NUMBERS ABOVE. Economists invest a lot of time and energy in estimating quantities like these, though, and they’d need to do so before anything like this could go forward.
One crucial factor would be how well the state can target workers who are really on the margin of whether or not to retire. While a wage increase across the board would be extremely unlikely to save the state money, one that is targeted at workers who are thinking about retiring and induces them to delay retirement just might. One thing’s for sure: it wouldn’t run afoul of the nonimpairment clause!