Home | Faculty and Research | Academic Groups | Finance | Finance Matters

Author Archives: Jonathan Parker

The Financial Market Consequences of the Government Shutdown

In my previous blog post, I argued that the costs of the government shutdown on the real economy were, in the context of the media attention and the long-term fiscal issues, ‘small.’  And yet, the government shutdown seems to have hit consumer confidence hard, lowering it to levels seen early in the Great Recession in 2008.

In this post, I look at financial markets for answers.  I focus in particular on perceptions of the possibility of delayed debt payments, technical default and (gulp) actual default (emphasis: the effect of beliefs rather than actual default).  I make four points.  First, there were observable ripples in the market for government debt.  But only ripples.  Short rates were raised by the shutdown, but only short rates, not forward rates.  So markets expected rapid resolution.  Second, while rate increases raised borrowing costs, the additional cost for the government budget is far less than the costs discussed in the previous post.  Third, why did short rates move at all in zero interest rate environment?  Interestingly, I think they did because the market thought “payment-delayed” Treasury debt was not going to have the liquidity and safety features of the rest of the debt.  So the event provides a neat window into the liquidity benefits of Treasury debt.  Finally, we are still left with the question of why the shutdown was such a big deal.  I propose one answer: it raises the specter of default, or even, perhaps more accurately, the specter of others perceiving that default is a nontrivial possibility. And if markets perceive that a US default is possible, then even without actual default, lots of very bad scenarios are in play.  This answer is not the only answer.  Learning that your government is disfunctional is bad news on many fronts.  But market reactions to these beliefs can create economic disaster, with costs many times greater than those of the actual shutdown.

First, the shutdown did raise interest rates on US government debt.  Higher interest rates make new borrowing – to fund the deficit and to roll-over the existing debt that comes due – more expensive.  Maybe the budgetary effects of the shutdown through increased borrowing costs were substantial?  Turns out the answer is no.

To measure the effects of the shutdown on interest rates, I look at the movement of interest rates around the resolution of the shutdown.  Why the resolution and not the initiation of the shutdown?  Because market expectations of a shutdown slowly rose as we approached the debt limit, and during this time other factors might also be moving yields.  Because the news about the end of the shutdown was released relatively rapidly, the fall in government rates reveals the change in borrowing costs that the Federal government experienced as a result of the shutdown.

The following table from the Selected Interest Rates (H.15) release from the Board of Governors of the Federal Reserve System shows several things.  From October 15 to 17th (the deal to end the shutdown was announced October 16th 2013), the annualized yield on one-month nonfinancial commercial paper falls by 5 basis points (0.05 percent) while the yield on treasury bill coming due in a month fell by 31 basis points (a third of a percent).  (Note that these rates are estimates constructed from rates on securities of similar term.)  If markets were sure that the shutdown would not affect payments on debt due beyond one month, then there would be no effect on forward rates, and the movement in the three-year yield ought to be approximately 1/3 the movement in the one month yield.  And that is what we find: thee month annualized yields fell by only 9 basis points. (Note that all yields rise with the horizon – this is business as usual. Yields curves on government debt usually slope up.)


How much might this have raised the cost of borrowing for the U.S. government in this period?  In 2012, the Treasury issued about $650 billion in Treasury debt each month.  Take this as an estimate of monthly borrowing during our period of interest.  Further, following our above calculation, suppose that the borrowing rate for half of October was raised by a third of a percent at an annual rate. Finally, let’s assume that this rise was anticipated (fully and only) for three months, so that the government shutdown raised rates starting three months prior to the shutdown. Then the total costs on Federal borrowing are one third of a percent higher (forward) interest rate covering a two weeks affecting borrowing over three and a half months, or roughly: (.00333) * (2/52) * ($650 billion) * (3.5 months) < $1 billion.  Not only is this calculation rough, but it plays quick and dirty with compounding.  But it makes the point.  The costs on borrowing are smaller than estimates of the direct costs on the economy.  Increased borrowing costs cannot explain the nosedive in consumer confidence.

But this data raises an interesting question: why did the one-month yield move at all?  The only risk that I heard about in the markets was the risk that debt payments would be delayed for a short time.  If rates moved because of default fears, long rates would have moved more.  They did not.  So why would a Treasury bill that is yielding an almost zero rate of return suddenly have its yield rise by a third of a percent (annualized) at the threat of having its payment postponed?  The puzzling part is the “nearly zero interest” part.  If yields were 10 percent per year, then a postponed payment — effectively being forced to give the government an unexpected loan at a zero interest rate — would have real (opportunity) costs.  But in a zero interest rate environment, no.

The answer I think lies in the almost unique features of US debt: its perceived safety and liquidity.  These properties mean that Treasury debt serves as money for large US firms, foreign governments, and companies that do business all over the world.   As the budget fiasco emerged, markets perceived that payment might be delayed on Treasury debt due to be paid after the government hit the debt limit, and, more interestingly, that this delayed-payment Treasury debt would not have the liquidity and safety properties of not-yet-expired Treasury debt. Thus the yields on this debt rose.  But the yields on long-term debt did not, even though it was just as possible that coupon payments also might have been delayed.  So, and somewhat speculatively, it appears that the perceptions of the liquidity of long-term US government debt were largely unchanged. (I conclude this because we see similar movements in interest rates on long-term highly-rated corporate debt as for long-term Treasury debt. But this is a subtle quantitative question and I have not done the hard work — let me know if you have!)  My speculation is that what we saw in yields is that, for long-term debt, safety matters.  For short-term debt, liquidity is critical for value.

Why?  Businesses use the Treasury debt like you and I use money.  Businesses hold Treasury debt rather than money because Treasury debt pays interest and Treasury debt is safer than large accounts of cash which are not insured and more convenient than cash (which also can be stolen, lost or damaged).  Liquidity means that if a large corporation or other government wants to make a payment, it can easily sell the Treasury debt and get cash.  Or it can rehypothecate or ‘repo’ the Treasury debt, borrowing against this extremely safe collateral at a low interest rate (even in times of crisis). Or finally, it could simply make payment with Treasury bills – among large firms Treasury debt is often the means of payment.  US Treasury debt is the main international money supply of global businesses and governments.


The U.S. gets significant seignorage from this special feature of Treasury debt.  We get to borrow more cheaply.  This figure, (from this paper by Arvind Krishnamurthy and Annette Vissing Jorgensen), shows something like a demand curve for Treasury debt.  The vertical axis is the price – the lower yield that one accepts by holding Treasury debt instead of a diversified portfolio of highly rated (Aaa) corporate debt. When there is a lot of Treasury debt in the U.S. economy relative to the size of the economy (the horizontal axis, debt/GDP), the difference in yields is about 0.4 percent.  But when there is less debt outstanding, people don’t simply hold diversified highly rated corporate debt instead. They have a need for Treasury debt.  And this need is so strong that people, firms, and governments are willing to hold and use Treasury debt even when they are getting one and a half percent per year lower returns than the diversified pool of AAA corporate debt.  Krishnamurthy and Vissing Jorgensen estimate that this benefit of liquidity and safety have on average lowered the borrowing costs of the US by about 3/4 of a percent per year.  Obviously, this amount is higher when there is less debt, but this seignorage helps the US budget significantly (these amounts are much, much larger than any estimate of the effect of the shutdown to date).

Did the government shutdown change this liquidity premium?  Did it affect the liquidity and safety of Treasury debt or its widespread use as money?  The shutdown only caused tiny ripples in the overall market for US government debt. I saw some reports of a few financial businesses not accepting Treasury bills. And the shutdown seems to have affected the one-month yields only, as I showed above.  So, the basic answer is no.  The yield on very short debt rose, because it may have become illiquid and the people who hold it are those that value liquidity.  But the possibility of delayed payment did not bother investors with long-term government debt.

But it is here where the real danger of the government shutdown (and its deeper causes) lies. It raises the possibility that US Treasury debt might someday lose its special status as the money supply of global business.  And this would happen if markets more seriously thought the US might default.  And the shutdown raised this possibility.

If people thought that people thought that the US might default in some unlikely but not impossible state of the world, then Treasury debt would no longer have safety and its yields would become more volatile.  And it would no longer have liquidity and it would lose its status as the world’s money supply.  This would be bad for the US.  First, the US would lose the special terms at which it can borrow and lose the seignorage that come from liquidity and safety.  But one country’s loss would be another county’s gain.  Second, the US has very high debt levels.  A rise in borrowing costs pushes the long-term fiscal imbalance issues to the fore much faster.  And finally, as long as we are running significant deficits or rolling a significant amount of debt, a significant rise in rates raises the possibility of a run on the dollar.  The US could quickly be put in a position where creditor doubts could raise rates or severely limit US government access to credit markets and require immediate, and under current projects, massive cuts in spending and increases in taxes.  I won’t use the word ‘default’ here, but the recent experience of Greece and Italy come to mind, and I don’t see anyone big enough to play the role for the US that the ECB has played for parts of Europe. Gulp.

But that is not all.  Even without a run on the dollar, a switch from US Treasury debt as the global money supply to some other currency and debt instrument would not go smoothly.  This switch alone would cause a global recession.  Because the firms of the world currently use Treasury debt like money, a sudden collapse in the perception of the Treasury and the dollar as safe and liquid would be like a massive contraction of the money supply of global economy.  It could easily lead to a deep global recession.  It is almost impossible for such a switch to occur in an orderly way, because a switch to a new currency is only effective if fast – if people are using different moneys they find it hard to transact, so we need everyone to switch at once –  but coordination is difficult and preparation nearly impossible.  Markets are forward looking; they would start to react before plans could be set in motion.  Would you accept Francs as payment if you knew that next week there would be a different currency?  The transition to the Euro worked in Europe (but not the UK) because the central banks were willing to trade at a fixed price.  There is no World Central bank (and the IMF and World Bank have very few funds relative to US Treasury debt).  In this scenario, the perceptions of the perceptions of others – market psychology – could start to matter.  The global economy would be at mercy of whether markets believed that the US dollar and US debt would no longer be the world’s currency.  Increases in this belief would cause large disruptions in trade and economic growth.

What I have outlined  has ignored the massive disruptions that an actual default would have on the US and global economy, where Treasury debt is relied upon by banks, insurance companies, pension funds and so forth as a store of value.

Did the shutdown take us part of the way there already?  I don’t know. Just because long rates on Treasury debt did not decline much as the shutdown was resolved does not mean that move to and through this shutdown did not raise them.

I do not mean to be too pessimistic.  I expect the US to solve its long term fiscal imbalances and to rule out the possibility of default before it becomes a significant enough of a concern to cause some of the scenarios that I just sketched.  I wrote this post because I think these financial market concerns are at heart of the puzzlingly strong reactions of the media and consumer confidence to the recent Federal government shutdown.  The direct and indirect effects of the shutdown that we have observed are quite small in the scheme of things.  The danger lies not even in another shutdown—the US averaged a shutdown a year between 1977 and 1987 – but in the possibility of future shutdowns combined with high level of national debt and large deficits.  I am concerned about the extent of political conflict and of the open discussion of default as a possibility even among some of our political leaders.

 Jonathan A. Parker is the International Programs Professor in Management and a Professor of Finance at the MIT Sloan School of Management.

The Consequences of the Government Shutdown

Professor Jonathan Parker

While the press are still celebrating the end of the government shutdown and deconstructing the political fallout, I was on a lunch panel yesterday sponsored by the Center for Finance and Policy here at Sloan on “The U.S. Fiscal Crisis: Causes and Consequences.”  My esteemed colleagues talked about the long run fiscal imbalances, the rise of the dollar and U.S. Treasury debt as reserve assets, and the thorny legal issues facing the Administration as money ran out.   I talked about the consequences of the shutdown, with a focus first on the real economy and then on financial issues.

In this first post on the topic, I want to argue that the direct effects of the shutdown on U.S. economy were small, or at least small relative to the media storm the shutdown created and relative to the collapse in consumer confidence that has coincided with the shutdown.  A reasonable estimate is that the shutdown cost in the ballpark of 25 billion dollars.  I do not mean to imply that billions of dollars are unimportant.  Look after your billions and your trillions will take care of themselves.  And the process can be excoriated for burning resources while accomplishing very little.  But this amount is small relative to the long-term fiscal imbalances in the background of the shutdown and relative to the costs that the U.S. government flirted with.  In my next blog post I will talk about the financial issues and risks.

But for this post, how do we want to think about or measure the direct effects of the government shutdown on the real economy?  There are several sources of costs.  Most immediately, we measure national output, GDP, as the sum of all goods and services produced during a quarter or a year.  We measure the contribution of each good or service at its market value.  New cars, for example, each contribute to GDP at the price at which they are sold. And you can see issues with such a measure.  Cars of the same exact make, model, and features contribute different amounts to GDP if they are sold to different people at different prices. Another example, no matter how much you would have paid to read this blog post, it contributes nothing (directly) to measured GDP because it is free to view and read (my writing this does not change my salary, nor does MIT produce more in any measured way from any additional web hits).  These measurement issues are even harder for the government.  How should we value a road?  How about the administrative work done at the IRS?  There are no market prices to observe.  And people’s opinions differ on the value of these goods and services.  So, we tie our hands to a simple accounting rule.  When the government spends money, provides services, does administration, etc. the contribution to GDP is the cost to the government.

During the shutdown, the government stopped paying about 800,000 workers, as well as a smaller number of contract workers, and actual contracts to purchase goods and services.  Thus, there will be a reduction in measured GDP, not unlike that cause by labor strife.  The shutdown was like a lockout by the nation’s biggest employer.

But, this reduction in government spending is already being offset by back-payment for many government workers as well as higher payments in the near future due to needing to catch up on the work not done during the shutdown.  While creating variable rather than smooth production, and some inefficiency, the direct effect of the shutdown is small.

There is the possibility of secondary effects. As furloughed workers stop spending, business that they would have bought from stop hiring, investment on goods that they would have bought is postponed, and so forth.  Historically, we know there is a rapid and substantial pass through from take-home pay to consumer spending.  But in this case, these effects should be very small.  Not only is the income reduction for government workers temporary, full salary payments are being made after only a short delay.  And the fact that incomes for the year may even be higher due to the need to make up lost time implies that there may even be slightly higher demand from government workers.

The Federal government does not only buy goods and services, it makes monetary transfers through insurance programs for health, employment and disability, and pension programs and finally debt payments – we will return to debt payments, but for now I want to emphasize that had the government cut these programs, this would have led to more substantial declines in consumer demand because these programs are much bigger.  However, in this crisis these payments were almost entirely uninterrupted.

Finally, the government collects taxes and fees which raise revenue, it grants licenses and regulates activity, and the shutdown jeopardized a  lot of economic activity. But essential employees were not furloughed, licenses were only delayed, and the fees are not a major source of revenue for the Federal government.  In the future, the government will have to spend less or tax more both to cover the lost revenues during the shutdown and to cover the higher payments for workers needed to catch up on missed work.

Standard & Poor’s provided one estimate of the costs of the shutdown: $24 billion.  This estimate seems large to me.  But even 24 billion amounts to about one half of a percent of GDP in the fourth quarter of 2013, or about one eighth of a percent of GDP in 2013, which is within the range of statistical uncertainty with which we measure annual GDP.  That is, the effect of the government shutdown of 2013 is likely statistically un-measurable in annual data.

Now all these costs would have been larger if the shutdown had been more prolonged, and they may arise again, and this risk of another shutdown creates uncertainty for people and businesses, and all of this is bad for the economy.  And the shutdown that actually happened was a waste. But it just wasn’t that costly in the context of the issues at hand.  So this raises our first puzzle, what happened to consumer confidence?

The below Figure comes from Gallop, and shows their measure of consumer confidence plotted over time.  Confidence is lowest during the Great Recession, picking up just before the end of the recession and trending upward until the sudden recent decline.  There is also a decline and trough in the middle of 2011, which happens to coincide with the fight over the debt ceiling that temporarily closed the government in the summer of 2011.  So it appears that households are very concerned about government shutdown – showing almost Great Recession levels of concern.

Maybe this is because of what happens in financial markets?  This is the topic of my next blog post . . . .

 Jonathan A. Parker is the International Programs Professor in Management and a Professor of Finance at the MIT Sloan School of Management.