Odds have risen that non-essential functions of the federal government will shut down temporarily when the current stop-gap funding measure expires on Monday, September 30. This would reduce GDP during the fourth quarter of 2013, but output would rebound in first quarter of next year.
- The direct effect on fourth-quarter real GDP growth of a 2-week shutdown starting on October 1 would be 0.3 percentage point.
- All of that would be reflected in GDP produced by the government sector.
- Because we expect any shutdown to be brief, induced effects on private production and repercussions in financial markets would be modest.
- There is no risk to Social Security or Medicare benefits.
- A protracted shutdown would disrupt private production and financial markets.
- However, the larger fiscal risk is the possibility of hitting the debt ceiling this fall.
A stop-gap measure for funding the federal government expires September 30. Unless Congress and the Obama Administration reach an agreement beforehand, parts of the federal government will shut down with the new fiscal year. <1>. House Republicans have tied to a “continuing resolution” that would fund the government for 11 weeks (through mid-December) a provision to defund the Affordable Care Act. The Senate will strip the defunding provision from the resolution and send it back to the House where it likely will pass, but perhaps not in time avoid a shutdown—the odds of which have risen sharply in recent weeks. However, we believe that because many Republicans fear the potential political fallout, any shutdown will be relatively brief. We see the possibility of hitting the debt ceiling later this fall as a far larger threat to the economic outlook, and frankly are incredulous that some Republicans seem more concerned about a shutdown than a sovereign debt crisis.
History Repeats Itself?
The last shutdown occurred in 1995-96, when Congress, then led by Speaker of the House Newt Gingrich, played a game of budgetary chicken with President Clinton that ultimately closed the federal government for the better part of a month. <2>
During the 1995-96 episode there were actually two separate shutdowns. The first lasted five days, from November 14 to November 19. It affected roughly 258,000 civilian defense employees and 489,000 nondefense workers. The number of civilian workers affected (747,000) was 36% of the roughly 2.1 million federal civilian employees (excluding postal workers). The second shutdown lasted 21 days, from December 16 to January 6. It affected roughly 284,000 nondefense employees, or 14% of the total.<3>
Together, then, these shutdowns covered almost 4 weeks. What is known about the resulting impact on GDP, and how does that experience inform our estimates of the impact of a shutdown today?
In the National Accounts, the main direct or accounting effect on GDP of a shutdown arises because compensation of federal employees is treated as GDP produced by the federal government. However, the distinction between real and nominal compensation is important here. Nominal compensation reflects pay accruing to workers. Real compensation is based on hours actually worked. Following the shutdowns of 1995-96, workers were eventually paid for their time not at work. Hence, the GDP accounts now show no direct effect of those furloughs on nominal GDP produced by the government. However, because the employees were not working during the furloughs, there is a recorded decline in real GDP produced by the government. <4>
In particular, the BEA estimates that the Clinton-era shutdowns reduced real government output $4.6 billion (at an annualized rate) during the fourth quarter of 1995 and by $1.0 billion during the first quarter of 1996.<5>
This subtracted roughly 0.25 percentage point from real GDP in the fourth quarter of 1995 but added back roughly the same amount to growth over the first and second quarters of 1996. <6>
Suppose that today, as in November of 1995, 36% of all 2.1 million federal civilian employees (other than postal workers) were furloughed for two weeks starting October 1.<7>
The direct impact would be to reduce real GDP growth by roughly 0.3 percentage point during the fourth quarter. <8>
A three-week furlough translates to 0.5 percentage point of growth. If the shutdown lasted all of October—that is, 4½ weeks—the direct effect would be to reduce GDP growth for the quarter by roughly 0.7 percentage point. If, as happened in 1995-96, federal employees expected eventually to be paid for the time not at work, there would be little if any induced effect on personal consumption expenditures (PCE). Even if workers did not expect to be paid, the transitory nature of the interruption in their income suggests that PCE would be little affected.
Timing and Paybacks
The GDP lost when government workers are furloughed is not made up except in the very unlikely case that federal employees subsequently work that much extra time. Hence, following a shutdown, real compensation would simply return to its previous level, temporarily boosting GDP growth by roughly the same amount that the decline in real compensation reduced it. Another implication is that the effect of furloughing workers on GDP growth during a given quarter does not depend on when during the quarter the shutdown occurs.
This is not the case for other kinds of federal spending which, for a given level of appropriations, are only delayed by a shutdown. <9>
If that delay is intra-quarter, there is no impact on the quarterly growth rate of such spending. If the delay is inter-quarter, there is an effect. Hence, a shutdown that occurs relatively late in the quarter is more likely to have an effect on GDP growth than one that occurs early enough in the quarter that any lost production is made up by the end of that quarter. Furthermore, for spending that is merely delayed, the pattern of the “payback” is different because, after the shutdown, the level of such spending must temporarily move above the previous level, creating a saw-tooth pattern of growth rates: first down, then up, then down and finally back up. Since the looming shutdown would start at the beginning of the quarter, inter-quarter considerations would become important only if the shutdown proved protracted.
Other Non-defense Consumption & Gross Investment (C&GI)
Not all spending that is delayed from one quarter to the next translates into a temporary reduction in reported GDP during the quarter of the shutdown. For example, in the National Accounts, federal non-defense spending other than compensation is currently about $212 billion per year. However, the Bureau of Economic Analysis (BEA), lacking information on the quarterly pattern of most of these outlays, smoothes its estimates over the four quarters of the year. A shutdown would have a major impact on other non-defense C&GI only if: (a) the delay was inter-annual; or (2) the BEA made a special adjustment to the quarterly smoothing. No such adjustment was made in 1995-96. Hence, we do not expect a brief shutdown to depress the numbers for non-defense spending (other than compensation) reported in the National Accounts.
Defense Gross Investment
Funding for defense gross investment is usually set contractually about a year before expected delivery. So, most equipment scheduled for delivery late in 2013 was financed out of appropriations authorized in 2012. These monies are not threatened by a shutdown. In any event, equipment is recorded in defense gross investment when it is delivered. Furthermore, when such equipment is delivered there is no impact on GDP because the value of the delivery is subtracted from private inventories where it accumulated as the equipment was fabricated. Similarly, ships and structures are recorded in defense gross investment as the work progresses and also are paid for with funds obligated out of prior appropriations. So, unless a shutdown actually delays private production or construction of defense-related capital from one quarter to the next, there would be no impact on reported GDP.<10>
Since contractors know the funding is secure and will be forthcoming eventually, the best assumption is that a brief shutdown would have no impact on defense gross investment.
Defense Consumption (Other than Compensation)
These items are recorded in the National Accounts as final sales upon payment. Most of the data on these outlays come from the Monthly Treasury Statement (MTS), so there is information about the quarterly pattern for these expenditures that gets reflected in the National Accounts. If such outlays are delayed from one quarter to the next, then final sales are likewise delayed.
Sales, Production, Goods, Services
Note, however, that such a wiggle in final sales does not necessarily imply a like wiggle in GDP. This is because some portion of defense consumption is for intermediate goods rather than services. A temporary reduction in defense spending for goods translates into a temporary reduction in GDP only if private production of goods temporarily dips. Otherwise, private inventories of goods will simply rise only to be drawn down after the shutdown. For a brief shutdown, it seems unlikely that contracted production of defense-related goods (mostly weapons systems, munitions, and petroleum) would be even temporarily curtailed.
Services, however, cannot be inventoried. So, if defense spending on services is delayed from one quarter to the next, there is a corresponding effect on GDP. At current rates, the delay from one quarter to the next of one week’s worth of defense spending on services (other than compensation) would knock about 0.13 percentage point off GDP growth during that quarter, then raise it by roughly 0.26 percentage points the next quarter, reduce it by roughly 0.13 percentage point the quarter after that, and then finally raise.
The rhetoric here is sharp but does not match the reality. Social Security checks were mailed during the 1996-96 shutdowns. The prevailing legal view is that since Social Security benefits are paid from a trust fund rather than discretionary funds, these benefits are not affected the shutdown. Because of that, the Social Security Administration has implicit authority to keep on workers involved in processing benefit checks. Indeed, during the last shutdown, eventually workers were brought back to process new claims for benefits. So, it seems highly unlikely that an interruption in Social Security benefits will occur. In any event, an interruption would likely be viewed as temporary and so have a limited effect on PCE.
Medicare payments are also paid from a trust fund, so all the arguments surrounding Social Security apply here, too. Furthermore, in the National Accounts, PCE on healthcare services funded by Medicare are counted as they accrue, not as the Medicare payments are made. That is, a delay in payments would not reduce measured consumer spending on healthcare services.
The Shutdown Calculator
The nearby table shows our “shutdown calculator” into which one can enter assumptions for the length and the extent of a shutdown beginning October 1.
The calculator then returns the approximate direct impact on fourth quarter real GDP growth. As shown here, it is populated with our assumptions for a 2-week shutdown affecting 36% of civilian compensation with no intra-quarter delays. This reduces GDP growth by 0.3 percentage point during the fourth quarter.
The direct impact on fourth-quarter GDP of a brief shutdown in the federal government in early October would be modest. By our measure, a two-week furlough of 36% of civilian workers would reduce GDP growth during the fourth quarter by 0.3 percentage point. Growth would be boosted by 0.3 percentage point in the first quarter of 2014. A protracted shutdown would cause larger, escalating disruptions, including induced declines in private production and rising risk premiums in financial markets. However, we see the possibility of hitting the debt ceiling later this fall as a far larger threat to the economic outlook.
1 Additional information on federal shutdowns is found in Clinton T. Brass, “Shutdown of the Federal Government: Causes, Processes, and Effects”, Congressional Research Service (September 27, 2010).
2 In August of 1996 the CBO estimated that the shutdowns in 1995-96, which covered nearly three weeks in November and December of 1995, subtracted only 0.5 percentage point from GDP growth during the fourth quarter of 1995. See the Congressional Budget Office, The Economic and Budget Outlook: An Update (August 1996), Chapter 1, page 2.
3 Between the first and second shutdowns, an appropriations bill covering the Department of Defense was passed.
4 Since nominal government GDP was not affected by the shutdowns but real government GDP did decline, there was a corresponding temporary increase in the price index for government compensation during this episode.
5 Survey of Current Business, June 1996, footnote on page 6. These figures are in 1992 chain-type dollars.
6 This is MA’s calculation made using the data as reported in June of 1996.
7 Military personnel are assumed not to be affected.
8 We made separate calculations for the contributions to real GDP growth of furloughing non-defense and civilian defense workers and then summed the resulting calculations.
9 Indeed, most of this spending is contractually guaranteed.
10 Normally payments are used to gauge the progress (e.g., production) on these projects. However, unusual variations in the timing of payments — such as those related to a shutdown — are adjusted for by the BEA, leaving no impact on the estimate of GDP.