CBO summarizes its main results in this handy chart:

Report author Justin Faulk summarizes the findings as follows:

Differences in total compensation—the sum of wages and benefits—between federal and private-sector employees also varied according to workers’ education level.

Federal civilian employees with no more than a high school education averaged 36 percent higher total compensation than similar private-sector employees.

Federal workers whose education culminated in a bachelor’s degree averaged 15 percent higher total compensation than their private-sector counterparts.

Federal employees with a professional degree or doctorate received 18 percent lower total compensation than their private-sector counterparts, on average.

Overall, the federal government paid 16 percent more in total compensation than it would have if average compensation had been comparable with that in the private sector, after accounting for certain observable characteristics of workers.

Of course, a lot is riding on the phrase “certain observable characteristics.” CBO did an extremely careful job of measuring total compensation and of controlling for observable factors such as education, age, and occupation. But many other factors are impossible to measure. CBO’s summary mentions effort and motivation. There are also issues such as job security and developing valuable skills and knowledge.

Joseph Kile, head of the microeconomics studies division at the Congressional Budget Office, discussed various policy options to deal with this funding gap in his testimony to the Senate Finance Committee on Tuesday. Most news coverage of Joe’s testimony emphasized his suggestion that taxes based on miles traveled, rather than gasoline consumption, might be a better way to finance America’s highways. After all, miles traveled is, along with weight, the primary driver of wear and tear on the roads. And it’s a decent proxy for the benefit that drivers get from having functioning roads.

That’s an interesting idea, but I’d like to highlight another important point that Joe made: the amount of infrastructure America should build depends very much on how we price it.

If a six-lane highway gets congested, that doesn’t necessarily mean that we need to build new lanes or lay out parallel roads.

We could charge congestion fees instead. That would discourage driving at peak times and thus speed traffic without new construction. That’s what London and Singapore famously do to limit traffic in their downtowns. And it’s something we should more here in the United States.

Joe reports estimates from the Federal Highway Administration (FHWA) that congestion pricing could decrease highway spending needs by 25 to 33 percent:

The federal government spent about $43 billion on highway investment in 2010. To maintain the same quality of highway performance would require an average of $57 billion in annual federal spending in coming years, according to the FHWA. That price tag drops to only $38 billion, however, if we make good use of congestion pricing. Congestion pricing would thus save federal taxpayers almost $20 billion per year; state and local governments would save even more, since they pay for more than half the costs of these projects.

Congestion pricing can make our roadways work better, save Americans precious time, and reduce federal, state, and local budget pressures. That a great combination in this time of growing infrastructure needs and tightening budgets.

Greg Mankiw set off a vigorous discussion in the blogosphere (see, e.g., Ezra Klein, Clive Crook, and the Austin Frakt) with a provocative analogy about health care reform:

I have a plan to reduce the budget deficit.  The essence of the plan is the federal government writing me a check for $1 billion.  The plan will be financed by $3 billion of tax increases.  According to my back-of-the envelope calculations, giving me that $1 billion will reduce the budget deficit by $2 billion.

Now, you may be tempted to say that giving me that $1 billion will not really reduce the budget deficit.  Rather, you might say, it is the tax increases, which have nothing to do with my handout, that are reducing the budget deficit.  But if you are tempted by that kind of sloppy thinking, you have not been following the debate over healthcare reform.

I read Greg as raising an important rhetorical / pedagogic question which, judging by some responses, may have been overshadowed by his satire.

That simple question is “what is health care reform?”

The policy community and commentariat often equate health care reform with the legislation (actually two pieces of legislation) that President Obama signed into law last year. As everyone knows, the Congressional Budget Office estimated that those two laws would, if fully implemented, reduce the federal budget deficit by $143 billion from 2010-2019. That’s the basis for the claim that “health care reform would reduce the deficit over the next ten years.” (CBO also discussed what would happen in later years, where the law, if allowed to execute fully, would have a bigger effect, but let’s leave that to the side right now.)

The complication, which Greg’s post partly addresses, is that the health care reform legislation included many provisions. Greg notes, for example, that some expanded health insurance, while others raised taxes. In his view, only the first part constitutes health care reform — an effort that by itself would widen the deficit — while the tax increases are what made the legislation deficit-reducing.

In fact, it’s more complicated than that. By my count, the two pieces of health care reform legislation combined seven different sets of provisions:

1. Expanding health insurance coverage (e.g., by creating exchanges and subsidies and expanding Medicaid)

2. Expanding federal payments for and provision of health care services (e.g., reducing the “doughnut hole” in the Medicare drug benefit)

3. Cuts to federal payments for and provision of health care services (e.g., cuts to Medicare Advantage and some Medicare payment rates)

4. Tax increases related to insurance coverage (e.g., the excise tax on “Cadillac” health plans)

5. Tax increases not related to insurance coverage (e.g., the new tax on investment income)

6. The CLASS Act, which created an insurance program for long-term care

7. Reform of federal subsidies for student loans

(The House Republicans’ effort to repeal health care reform would overturn 1-6, but leave the student loan changes in place.)

To capture these complexities, I occasionally refer to the legislation as the health care / tax / student loan / long-term care legislation. But whenever I write that for publication, my editors take it out. Although my lengthy description is accurate, it doesn’t work for friendly conversation. So the law (which again, was really two laws) gets called the health care reform law.

Greg’s point, I think, is that this rhetorical convention creates confusion when talking about the law’s budget impacts. To say “the health care reform law reduces the deficit over the next ten years according to CBO” is absolutely true. But it often gets elided to “health care reform reduces the deficit over the next ten years” which isn’t true if, like Greg, you think the revenue raisers, student loan changes, and CLASS Act aren’t really health care reform.

I think Greg is right to worry about this distinction. Because of the information loss as the details of CBO scores get transmitted through various layers of speakers and media (including this blog), some people are indeed under the mistaken impression that health care reform, by itself, reduces the budget deficit over the next ten years. It doesn’t.

However, Greg’s analogy has a flaw: it presumes that none of the tax increases count as health reform. I disagree.

Our current tax system provides enormous ($200 billion per year) subsidies for employer-provided health insurance. They should be viewed as part of the government’s existing intervention in the health marketplace. And rolling back those subsidies strikes me as essential to future health care reform. I would count any revenues raised from doing so as part of health care reform.

That didn’t happen, but the legislation did include a tax on “Cadillac” health plans as a partial substitute. That will clearly affect health insurance markets, and it offset a portion of existing tax subsidies. For both those reasons, it should be viewed as part of health care reform.

The key thing is not the difference between spending and revenues, but between provisions that fundamentally change the health care system and those that do not.

Happily, I am not alone in this view. Indeed, it has been endorsed by none other than the Congressional Budget Office. CBO grappled with this issue during the health care debate. And after much thought, it came up with a useful measure of the health care reform part of the legislation: the “Federal Government’s Budgetary Commitment to Health Care“. This measure combines the spending and tax subsidies that the government provides for health care.

Taking all the health care provisions into account, CBO concluded that the health care reform legislation would increase the federal government’s budgetary commitment to health care. But not as much as many critics suggest. Adding together items (1) through (4) on my list, CBO concluded that the health care reform parts of the legislation would increase the deficit by about $400 billion over ten years. That would then be more than offset by the other provisions — primarily taxes but also the student loan provisions and the CLASS Act. (In later years, by the way, CBO projects that the legislation would actually reduce the federal commitment to health care.)

Bottom line: Health care reform increases the federal deficit over the next ten years, but the health care reform legislation reduces the deficit. What could be simpler?

P.S. I hope it goes without saying–but will say it anyway–that one should not evaluate the health care reform legislation on its fiscal impacts alone … or even predominantly. The legislation has a wide range of benefits (e.g., 32 million more people with health insurance) and costs. The key question is how they net out.

What gives? Why would it cost $87 billion more to repeal the law than was saved by enacting it?

The main reason is that the 10-year budget window moved. The health debate started in 2009, so CBO used a 10-year window that ran from 2010 to 2019. It’s now 2011, so the repeal law will be judged against a 10-year window that runs from 2012 to 2021. The $230 billion figure reflects that longer window. Through 2019, the cost would be $145 billion.

The second reason is that the legislation President Obama signed last spring wasn’t the final word on health reform. In December, Congress was struggling to find a way to pay for the infamous Medicare “doc fix”, which now runs through the end of 2011. To do so, Congress decided to cut $15 billion from the subsidies created by the health legislation. Because those cuts reduced future subsidies, it is now $15 billion more expensive to repeal the overall health reform.

The third reason is that the original health legislation wasn’t just about health policy. It also included fundamental reforms to the way the government subsidizes college loans. The repeal bill wouldn’t undo those changes, which resulted in budget savings of $19 billion over 2010 to 2019.

Finally, the original health reform included about $7 billion in net budget costs during 2010 and 2011. It’s unlikely (to say the least) that the health repeal bill would be enacted in time to avoid those costs.

Bottom line: CBO estimated that the original legislation would reduce deficits by $143 billion over 2010-2019. CBO now estimates that repeal would increase deficits by $145 billion over the same period; the slight difference reflects the education provisions in the original legislation, the 2010 and 2011 costs that can’t be avoided, and the December 2010 changes to the law. The jump from $145 billion to $230 billion then reflects the addition of two years to the budget window.

P.S. The $230 billion figure is preliminary and subject to change once CBO has an opportunity to update its calculations to reflect the latest information about the economy, health care markets, etc.

P.P.S. Aficionados of the health debate will recall that many differences of interpretation surround CBO’s cost estimates for health reform. You can see some of my discussion here.

Readers may recall that last spring Phill Swagel and I proposed a reform in which Fannie and Freddie would be privatized, the government would sell guarantees on mortgage-backed securities composed of conforming loans, and that this guarantee would be available not only to Fannie and Freddie but also to qualified new entrants. (Here’s the blog version; here’s the full paper.)

CBO provides a thoughtful overview of such hybrid models:

A Hybrid Public/Private Model

Many proposals for the secondary mortgage market involve a hybrid approach with a combination of private for-profit or nonprofit entities and federal guarantees on qualifying MBSs. At its core, the hybrid public/private approach would preserve many features of the way in which Fannie Mae and Freddie Mac have operated, with federal guarantees (combined with private capital and private mortgage insurance) protecting investors against credit risk on qualifying mortgages. However, most hybrid proposals would differ from the precrisis operations of Fannie Mae and Freddie Mac in several important ways: A possibly different set of private intermediaries would participate in securitizing mortgages backed by federal credit guarantees, the guarantees would be explicit rather than implicit, and their subsidy cost would be recorded in the federal budget.14 As the public-utility and competitive market-maker models illustrate, a hybrid approach could be implemented in a way that involved more or less federal regulation of participants in the secondary market and a smaller or larger number of competitors in that market.

Advantages of a Hybrid Approach

Regardless of its exact design, a hybrid model with explicit federal backing for qualifying privately issued MBSs would have several advantages over the precrisis model, as well as over either a fully federal agency or complete privatization (approaches that are discussed below). An explicit federal guarantee would help maintain liquidity in the secondary mortgage market, in normal times and particularly in times of stress, and could retain the standardization of products offered to investors that Fannie Mae and Freddie Mac bring to that market. Compared with the precrisis model, imposing guarantee fees would ensure that taxpayers received some compensation for the risks they were assuming.

Compared with a fully federal agency, a hybrid approach would lessen the problem of putting a large portion of the capital market under government control, encourage the inflow of private capital to the secondary market, and limit the costs and risks to taxpayers by having private capital absorb some or most losses. Putting private capital at risk would also provide incentives for prudent management and pricing of risk.

Compared with a fully private market, hybrid proposals would give the government more ongoing influence over the secondary market and an explicit liability in the case of large mortgage losses that would be reflected in the budget. That arrangement might have the advantage of leading to a more orderly handling of crisis situations.

Disadvantages of a Hybrid Approach

Relative to other approaches, a public/private model has a number of potential drawbacks, the importance of which differs depending in part on the specific design chosen. Experience with other federal insurance and credit programs suggests that the government would have trouble setting risk-sensitive prices for guarantees and probably would shift some risks to taxpayers. A hybrid approach also might not eliminate the tensions that exist—with regard to risk management and pursuit of affordable housing goals—between serving private shareholders and carrying out public missions.

Another concern is that over time, the secondary-market entities might push for broader guarantees of their product lines and attempt to reestablish themselves as too-big-to-fail institutions backed by implicit federal guarantees. Consequently, regulators would need to be vigilant to control risks to the financial system and avoid regulatory capture, while also being open to market innovations.

To illustrate, suppose that Treasury used TARP to buy $10 of preferred stock in Bank X in 2008 and that a year later Treasury sold its position for $12, including accrued dividends. This investment would be recorded as achieving a $2 profit in TARP (subject to one technical caveat, see below).

That’s the normal way of calculating profit on an investment, and is what CBO was instructed to do for its part of TARP oversight. But as Jack and John point out, there’s an important complication here. During the year, the federal government undertook many other policy actions which may have boosted the value of Bank X (remember all the new acronyms?). From the perspective of policy evaluation, some or all of the $2 gain should be attributed to those other policies, not TARP.

It could be, for example, that absent further action, Bank X would have struggled, leaving Treasury with stock worth only $6. Other government actions, however, breathed enough life into the company (or, at least, boosted the value of its assets) that the stock ultimately became worth $12.

In that case, you could argue that TARP, by itself, resulted in a $4 loss, while the other government actions created a $6 gain. That puts the budgetary impacts of TARP in a different light: a 40% loss versus a 20% gain in this example.

Of course, you could also argue that the $6 gain was only possible because of the TARP ownership stake. There’s certainly an element of truth to that. But the basic concern still applies: the $2 gain in this example reflects both TARP and subsequent government actions, not just TARP alone. That’s an essential point when trying to evaluate these policies after the fact, and we commenters should keep that in mind when interpreting CBO’s findings.

And that’s not all. The other government actions may also have imposed additional direct or indirect costs on the federal budget. As a result, the $2 gain in this example may be offset (or more) by other costs that aren’t included in the calculation.

Bottom line: One reason that TARP appears much less expensive than originally predicted is that many of its investments benefitted from other government actions whose costs show up elsewhere in the budget.

Caveat: CBO’s methodology actually judges the profitability of investments relative to benchmark rates of return. The details are surprisingly complex, but just for purposes of illustration, suppose that the appropriate benchmark rate of return for investing in Bank X was 10%. If Treasury sold the stock for $11 after one year, CBO would deem that as breaking even. If it sold it for $12, that would be a $1 profit.

The good news, budget-wise, is that the government is on track to make about $22 billion on its assistance to banks.

However, CBO estimates that TARP’s other activities will cost $47 billion. This reflects aid to AIG ($14 billion), the auto industry ($19 billion), mortgage programs ($12), and a few smaller programs ($2 billion).

CBO now estimates that ARRA will cost $814 billion from 2009 through 2019. That’s up from the original $787 billion estimate, but down from the revised, $862 billion estimate released in January.

Spending exceeded original expectations because both unemployment and food prices rose more than anticipated, driving up the cost of extended unemployment benefits and expanded food stamp benefits. On the other hand, spending estimates have come down because “recently enacted legislation rescinded some of the funds appropriated in ARRA and limited the period in which higher payments under the Supplemental Nutrition Assistance Program [formerly known as food stamps] will be available.” (CBO did not update estimates for the tax provisions in ARRA.)

For a discussion of why the $814 billion figure (formerly known as the $862 billion figure or the $787 billion figure) is not really the right measure of stimulus, see this post.

On a related note: Earlier today, CBO released an updated analysis of the economic effects of ARRA. It estimates that ARRA reduced unemployment in the current quarter by 0.8 to 2.0 percentage points. In other words, without that stimulus CBO believes that the unemployment rate today would be between 10.3 percent and 11.5 percent, not the 9.5 percent reported in July.

However, that focus sometimes overshadows a key fact about our tax system: Revenues are already on track to rise substantially in coming years. And not just because of an economic rebound and expiring tax cuts. There are structural reasons why tax revenues will grow faster than the economy.

The Congressional Budget Office estimates that tax revenues will rise from 14.9% of GDP in 2010 to 20.7% in 2020 and 23.3% in 2035 if current law remains in place (the “extended baseline” scenario in pink):

To put those figures in context, note that federal revenues have averaged about 18.2% of GDP over the past forty years. Tax revenues today are thus remarkably low. Indeed, they are the lowest they’ve been since 1950. But that will quickly reverse under existing law. By 2020, revenues would near their all-time record (20.9% of GDP in 1944) and by 2035, revenues would be more than 25% higher than historical levels.

That rapid growth reflects six factors. First, the economy will recover, lifting revenues from currently depressed levels. Second, the 2001 and 2003 tax cuts will expire, as will tax cuts enacted in the 2009 stimulus. Third, the Alternative Minimum Tax, which is not indexed for inflation, will boost taxes for millions more taxpayers. Fourth, the new taxes that helped pay for the recent health legislation will go into effect. Fifth, retiring baby boomers will make more taxable withdrawals from tax-deferred retirement accounts. Finally, in a phenomenon known as bracket creep, growing incomes will push taxpayers into higher brackets and reduce their eligibility for various credits.

Together, those six factors will increase tax revenues by 8.4 percentage points of GDP over the next 25 years, according to CBO. About a third of that increase (2.7 percentage points) comes from expiring individual income tax provisions and the expansion of the AMT. Another third (2.6 percentage points) is due to real bracket creep and reduced credits. And about one-seventh (1.2 percentage points) results from the tax increases in the health legislation. The other factors account for the remainder.

We clearly have sizeable tax increases built into our revenue system. The trillion-dollar question, however, is whether policymakers will allow them to happen. That’s why CBO considers a second scenario in which Congress gives in to the temptation to cut taxes. Under that “alternative fiscal” scenario (blue), Congress would permanently extend most of the 2001 and 2003 tax cuts and would limit the growth of the AMT. That would slow the growth of tax revenues but they would still reach 19.3% of GDP by the end of the decade, well above the forty-year average. CBO assumes that policymakers would then enact a series of unspecified future tax cuts to hold revenues at that level rather than letting structural factors lift them higher.

CBO’s bottom line is thus simple: tax revenues will rise faster than the economy even if Congress does nothing new. Indeed, revenues may rise faster than the economy even if Congress enacts substantial tax cuts. Our long-run fiscal dilemma exists because the scheduled growth in future spending is even larger than the scheduled growth in future revenues.

This post first appeared on TaxVox, the blog of the Urban-Brookings Tax Policy Center. 

The first imagines a world in which lawmakers take pay-as-you-go budgeting really seriously. The budget baseline assumes that existing laws execute exactly as written: all the 2001 and 2003 tax cuts expire, the alternative minimum tax hits millions more families, real bracket creep drives taxes far above historical norms, Medicare payments to doctors are cut by more than 20%, discretionary spending grows only with inflation, and all the offsets in the recent health legislation – taxes on “Cadillac” health plans, cuts in provider payments, etc. – happen as scheduled. If Congress tries to avoid any of those changes, it would have to pay for them through offsetting spending cuts or tax increases.

In that strict—and unrealistic—PAYGO world, our debt would continue to increase faster than the economy, rising from about 60% of gross domestic product today to about 80% in 2035. That’s far above where we want to be. PAYGO policymaking cannot fix the budget pressures of an aging population and rising health costs. But you have to give PAYGO some credit. If Congress really acted that way on every future budget decision, it’s unlikely that we would have a fiscal crisis in the next decade or more.

Of course, no one believes that Congress will really be that disciplined. That’s why CBO offers a second vision, in which lawmakers give in to temptation. They extend most of the tax cuts, patch the AMT, limit bracket creep, increase payments to Medicare docs, allow discretionary spending to rise with GDP, and turn off some of the health legislation offsets after 2020.

If policymakers give in to all those temptations, the debt skyrockets, rising from about 60% of GDP today to 185% by 2035. And that’s assuming no negative effects on the economy. As my colleagues Len Burman, Jeff Rohaly, Joe Rosenberg, and Katie Lim have pointed out, out-of-control deficits would weaken the economy by crowding out investment and driving up interest rates, so the debt-to-GDP ratio would actually grow even faster.

CBO doesn’t include those economic effects in its official long-run projections. However, it does separately examine what would happen to the economy because of reduced investment. The results aren’t pretty. When CBO runs the giving-in-to-temptation world through its model, it discovers that the U.S. economy ceases to exist after 2027.

OK, maybe that’s a bit strong. What happens is that crowding out gets so severe that CBO’s model breaks down, overwhelmed by the debt explosion.

These findings provide ammunition to both sides of the great deficit debate. Budget hawks will point to the temptation scenario as further evidence that we are on a reckless fiscal path that threatens our economic well-being. Budget doves will emphasize the PAYGO scenario as evidence that we can muddle along for years without needing to fear an economic backlash.

I think the hawks have the better of this argument. Even under perfect PAYGO, our fiscal condition would deteriorate in coming years. Moreover, the odds that lawmakers will show that much discipline in the near term are nil. For example, the PAYGO law enacted earlier this year allows Congress to extend most of the 2001 and 2003 tax cuts and to avoid cuts to Medicare doctor payments without finding any offsets. Real life PAYGO doesn’t come close to the rigorous standards of CBO’s PAYGO scenario.

Nevertheless, I don’t really think that the U.S. economy will blink out of existence in 2027. We will find a way a muddle through. But to do so, we need to face up to CBO’s bleak vision of our fiscal future and the temptations that lawmakers will face. Now is not the time for sudden austerity—the G-20 bandwagon non-withstanding—but it is the time to develop a plausible plan to bring taxes and spending into long-run balance.

This post first appeared on TaxVox, the blog of the Urban-Brookings Tax Policy Center. 

Edited (7/6/10): Added “most of” in fourth paragraph (CBO’s alternative fiscal scenario extends most of the 2001 and 2003 tax cuts, but it allows the rates reductions for high-income taxpayers to expire).