David R. Henderson  

The Coming Tax Increases

Outsourcing My Critique of Lef... Friedman on Immigration Histor...

Remember when critics of the Romney tax plan claimed that under his plan, the middle class would face tax increases of about 2,000 per family? [Here's just one of many examples.] Well, unless state and local government plans for their employees' pensions are reined in, households will face state and local tax increases of almost $1,400 per year forever. Here's what Robert Novy-Marx of the University of Rochester's Simon Graduate School of Business and Joshua D. Rauh of Stanford University's Graduate School of Business write in a recent NBER Working Paper:

We calculate increases in contributions required to achieve full funding of state and local pension systems in the U.S. over 30 years. Without policy changes, contributions would have to increase by 2.5 times, reaching 14.1% of the total own-revenue generated by state and local governments. This represents a tax increase of $1,385 per household per year, around half of which goes to pay down legacy liabilities while half funds the cost of new promises. We examine sensitivity to asset return assumptions, wage correlations, the treatment of workers not currently in Social Security, and endogenous geographical shifts in the tax base.

The paper is "The Revenue Demands of Public Employee Pension Promises, Working Paper #18489. [gated]

There is one error. The increase to 14.1% is not an increase of 2.5 times; it's an increase of 1.5 times. That's small comfort.

One other highlight:

In twelve states, the necessary immediate increase is more than $1,500 per household per year, and in five states it is at least $2,000 per household per year. A key feature of this analysis is that it accounts for the cost of new DB [defined benefit] accruals, for both current and future workers, not just the cost of unfunded legacy liabilities. Decomposing the results into these two components reveals that 49% of the increased contributions would be required to pay only the present value of new service accruals.

For California, the state I care most about (I live here), the increase would be $1,994 per household per year.

Here's an ungated version.

UPDATE: Russ Roberts interviewed Joshua Rauh on this issue on Econtalk.

Comments and Sharing

COMMENTS (15 to date)
Daniel Kuehn writes:

One of the worst things about the damage done to federalism in the U.S. over the last century is that people largely aren't aware of what is going on in their state capitols - and a lot still does go on there. Actually because of the issues you guys have had, Californians are probably more aware than most.

As badly informed as much of the public is about Social Security and Medicare, they're even worse on state pension funds and unemployment insurance trust funds. I'm quite guilty of that myself. I only have a sense of the UI trust fund situation in Virginia because I personally know an expert on the subject.

David R. Henderson writes:

@Daniel Kuehn,
Maybe you'll be happy to know, then, that the number for Virginia is a tax increase of "only" $1066 per household per year.

Daniel Kuehn writes:

Ah ha - chump change!

Ken B writes:


You're gonna need a bigger blog ...

Kevin Dick writes:

When I looked at the paper, it appeared they assume a real rate of return equal to the TIPS return. There was a dense paragraph explaining this assumption that I couldn't fully parse.

It seems like you want to run these calculations using the actual historical rate of return of the pension funds or some benchmark basket of assets. (Of course, you would never use the funds' rosy anticipated returns.)

Did I miss something?

David R. Henderson writes:

@Ken B,
Michigan: $1386.
@Kevin Dick,
Thanks for pointing that out. I should have. Here's their explanation in the NBER version:

These results are computed under an asset return assumption in which the pension fund assets earn a real risk-free rate of return, approximated by the yield on long-duration TIPS (1.7% as of December 2010) plus inflation. That is, we calculate required contributions under the risk-adjusted (or in finance terminology risk-neutral) mean real return. Considering the unadjusted return distribution would ignore the fact that the worst scenarios occur when marginal utility is highest. Under the risk-adjusted expected baseline, we also consider the possibility that public sector wages covary with priced risk factors (i.e., the market). When public sector salaries are exposed to priced risks, expected wages grow more slowly under the risk neutral measure than under the objective measure. Implementing wage risk in this fashion has a relatively modest effect on the results, reducing the required increases to $1,285 per household from $1,385.

I admit that I didn't totally get it either, especially the part about marginal utility.

Kevin Dick writes:

Thanks David. I'm glad it's not just me that couldn't make sense of it.

I'm in the VC business and so I talk to the guys who make investment decisions for pension funds. They clearly aren't targeting TIPS yields as their overall rate of return.

I remember seeing a paper a few months back on how much in the hole we are given various return assumptions, but I can't seem to find it.

David R. Henderson writes:

@Kevin Dick,
You're welcome. For some reason, my gut feel would be to use 3% real. But all it is is a gut feel. And this issue is within the authors' expertise. They're no slouches and so my other gut feel is to defer to their thinking on this. Still, I would like to understand their thinking better as, I gather, would you.

Tom West writes:

I'm a bit confused. I thought that people would be cheering tax increases on. After all, unless there are tax increases, it's very hard for the average person to understand that every extra government benefit comes at a *real* cost.

The *worst* case is making spending increases without any tax increase. It perpetuates the idea that there is a free lunch.

(Same with tax cuts. They should be accompanied by immediate program cuts so that people realize that decreasing taxes means you don't get something you had before...)

People need to understand that there are trade-offs before they can make informed decisions as to whether they tax cuts or more government programs. Program cuts/tax increases seem to be the right way to do it.

Kevin Dick writes:

FYI, I looked it up for California and the CalPERS Chief Actuary asserts that their 20-year annual return as of 2011 is 7.9% per year. Inflation during that period was 2.5%. So call it ~5% real. We should be conservative going forward, so 3-4% real seems defensible.


Chris H writes:

Haha! Georgia is only going up $803! Knew I picked a good state to live in (relatively).

Just glancing at the chart though it does look like there might be a correlation between states that voted for Obama and the size of the tax increase.

Methinks writes:

Very glad I escaped first New York and then Connecticut.

Same with tax cuts. They should be accompanied by immediate program cuts so that people realize that decreasing taxes means you don't get something you had before.

Doesn't quite work. People will just start screaming that "the rich" need to be robbed in order to provide everyone else with every desired comfort. The whole old fashioned notion of working to support the lifestyle to which you aspire is so passe in America. Theft laundered through government is the new black, dahhhling.

Jim Ancona writes:

@Kevin Dick, David,

You might want to listen to the Econtalk podcast linked in the post. In it, Joshua Rauh explains the argument against using historical returns to calculate pension liabilities. As I understood it, the premium over a risk-free return is due to the risk of principal loss. Because the taxpayers are taking that risk, not the pension recipients, you account for it by adjusting the return the same way the market does, i.e. the risk free return reflects the proper discount rate.

Eileen Norcross writes:

The Treasury rate is used because according to economic theory (Modigliani-Miller Theorem) how a liability is valued is independent from how it is financed. Current public sector accounting standards allow US state and local plans to calculate the present value of their liabilities using the rate of return the plan assumes it will earn on the assets when invested.

Most public plans are assuming they can earn between 7 and 8 percent on their asset portfolios annually.

But according to MMT, the performance of the assets is irrelevant to the value of the liability (*forget for a moment that it's probably unreasonable to assume those kind of returns. Critics often say that you have to look at historical returns over a long period and that it makes 7.5 percent very reasonable, but that's actually irrelevant to the core argument being made here.*)

Instead, the liability should be valued based on its risk characteristics. The discount rate should match the likelihood and timing of the payments of that liability to the workers.

A public sector pension is supposed to be risk-free for the worker. And often these plans are protected in state statute, constitution or enshrined in c.b. laws. The accrued liability is certain to be paid over a given period. Accrued pension benefits are like General Obligation debt for state governments and should be valued using a riskless rate of return such as the yield on Treasuries. Rauh and Novy-Marx cover this in their earlier papers. Also good discussions can be found in M. Barton Waring's book, "Pension Finance," Andrew Biggs uses an options-pricing model to address the same problem.

in effect, public plans use the higher (non-guaranteed) rate of return on risky assets to calculate the present value of (state-guaranteed) liabilities.

For now using these high discount rates is making life a little easier for policymakers since interest rates are so low. Raising the discount rate makes the present value of the liability (and current contributions) appear smaller. (Every one percent change in the discount rate changes the present value by 15-20 percent - a huge difference!)

But it's a deeply flawed approach. Using discount rates based on risky asset returns fails to capture the guarantee that these pension benefits are essentially a debt of the state. Undervaluing the liability means the plan is systematically underfunded. Valuing pension liabilities based on expected asset returns has other effects such as encouraging the plan to invest more heavily in riskier assets - And indeed, public plans have shifted more heavily to alternatives and equities and away from bonds and cash in recent decades (increasingly since 2002).

Hope this helps the discussion
Eileen Norcross
Mercatus Center

David R. Henderson writes:

Thanks, Eileen. It helps a lot.

Comments for this entry have been closed
Return to top