Why Does Uncle Sam Borrow?

By Dan Kervick

The Unites States government operates a fiat currency system.  The government is therefore the monopoly supplier of the final means of payment in our dollar-based economy, and is ultimately responsible, in one way or another, for any net increase in dollar-denominated financial assets in the private sector.

And yet, we continue to hear bipartisan expressions of fear and angst about the budget deficit and the national debt.  Both major parties seemingly agree that we are “out of money”.   They wrangle over various competing approaches to shrinking the gap between tax revenues and government spending.  They appoint commissions to study the government budget and recommend some combination of slashed spending and higher taxes in order to close that budgetary gap.   They warn us that we will transform ourselves into banana republic status if we do not urgently address our public debt problems.

This situation should be perplexing.  Why does a government that is the issuer of the national currency have to borrow that currency back from the public to which the currency is issued?   And how could such a government ever experience the kinds of budgetary squeezes and debt burdens that can pose severe problems for households and businesses?

I wish to make a radical suggestion:  Public borrowing is an outdated practice, and we could dispense with it entirely.   Borrowing by the public treasury and the accumulation of government debt obligations are legacies of the era that preceded the development of modern fiat currency, an era when governments were primarily users of traditional means of payment that lay outside their control, and not the producers and issuers of the primary means of payment.   That pre-fiat era is now dead in the US, and the chief remaining role of government borrowing in our time is to bamboozle the public, and to obscure the true nature and effects of government fiscal and monetary operations under a bewildering maze of bookkeeping ink and financial legerdemain.  Eliminating public borrowing, and replacing it with operations that are simpler, more direct and more transparent, would advance the cause of informed democratic debate over public spending and taxation.  Above all, the change would eliminate needless obscurity and confusion and help us all understand exactly whose bread is being buttered by the budgetary decisions made by politicians.

I will build up my argument against public borrowing in stages, through a series of thought experiments.  Let’s begin by imagining a country that operates differently from our own.  In this fictional country, the public treasury borrows directly from the central bank, but never borrows from any private lenders.  The central bank just deposits the borrowed amount into a special treasury account out of which the treasury spends, and also charges the treasury 5% interest on its borrowing.  Loans issued on any given date come due on the same date the following year, and must be repaid in full on that date, both principle and interest.

Imagine that the country’s GDP at some starting date is $2 trillion.  Over the next twenty years, the borrowing and repayment operations go like this:

On July 1st of Year One the treasury borrows $100 billion from the central bank.   It spends the entire $100 billion over the course of the year.

On July 1st of Year Two, the treasury borrows $205 billion from the central bank.   It then uses $105 billion of the borrowed funds to pay off the previous year’s debt entirely – $100 billion of principle and $5 billion in interest.  It then spends the remaining $100 billion over the course of the year.

On July 1st of Year Three, the treasury borrows $315.25 billion from the central bank.   It uses $215.25 billion of the borrowed funds to pay off the previous year’s debt entirely – $205 billion of principle and $10.25 billion in interest.   It then spends the remaining $100 billion over the course of the year.

This recurring process goes on for twenty years, so that finally …

On July 1st of Year Twenty, the Treasury borrows $3,306,000,000,000 from the central bank.   It uses $3,206,600,000,000 of the borrowed funds to pay off the previous year’s debt entirely- $3,053,900,000,000 in principle and $152.7 billion in interest.  It then spends the remaining $100 billion over the course of the year.

Let’s suppose that GDP grows by 5% per year, starting at the previously mentioned figure of $2 trillion in Year One, so that by Year 20, GDP is $5,053,900,000,000.   That means that in Year Twenty, the debt has grown to about 65.43% of GDP.

In this first example – call it “Example A” – government spending never increases.  It remains at $100 billion per year.  Thus it declines from 5% of GDP to just below 2% of GDP.   Let’s consider, then, a more realistic and slightly modified example – call it “Example B” – where annual government spending grows by 5% per year and so keeps space with GDP.  In this revised example, spending in Year Twenty is $253 billion, and the public debt in Year Twenty is 100% of GDP.

We can make a few immediate observations about these imaginary scenarios in our fictional society.  First, the growing debt numbers in these examples might seem very scary, but they are effectively meaningless.  The debt consists entirely of money owed by one part of the government to the other.  The central bank produces increasingly large amounts of money each year to extend to the treasury, and the next year almost that entire amount is delivered back to the central bank, financed entirely by more borrowing from the central bank.  But year after year, only modest amounts are actually spent into the economy: in Example A, $100 billion is spent into the economy each year, and in Example B amounts ranging from $100 billion to $253 billion are spent.  There is never any government solvency issue, because the debt is simply owed from one part of the government to another, and the money generated to fund the borrowing and spending is directly created by the central bank, a creature of the government itself.

Second, notice that there is something absurdly convoluted about the whole procedure.  The exact same functional operation could be accomplished by the central bank if it simply credited the desired amounts to the treasury each year, with no borrowing and no debt repayment.   In Example B, it would credit $100 billion in Year One and $253 billion in Year 20.  No debts would ever be booked, and no additional intra-governmental payments would ever be made.

Third, would the spending be inflationary?  It hardly seems so.  Even in Example B, spending remains steady at 5% of GDP, and the government is only injecting a fixed proportional level of new monetary assets into the economy each year, no more nor less than the overall rate of economic growth.

Suppose the central bank, in a fit of whimsy, increased the interest rate on its loans to the treasury to 25%.  In Year Twenty, then, assuming the treasury’s spending is no different than it was in Example B, the amount borrowed would be 832% of GDP!   But again that number has little inherent significance.  The effect is still exactly the same as if the central bank had just credited an amount equal to 5% of GDP to the treasury.  Each year the central bank lends the treasury an increasingly massive amount of money, but each year the treasury sends almost the entire amount back to pay off the debt of the previous year.  It is a little game they play.  The large sums are never spent into the economy to bid for goods and services, in a manner that might generate inflation.  They are just passed back and forth as ledger entries between two government agencies.

Now let’s modify the thought experiment by changing this first scenario slightly.   Assume that there is a law in our fictional country prohibiting the treasury from borrowing directly from the central bank.   The treasury can only borrow from private sector lenders.  However, the central bank can purchase any financial assets it likes from private sector owners of those assets, including debt assets issued by the treasury.   And the treasury and central bank can therefore work around the legal restriction by relying on those private sector intermediaries.

To simplify, suppose they rely on a single, large private sector dealer.  Each year, the dealer loans the treasury the entire amount the treasury desires to borrow.  The central bank then purchases the entire debt from the dealer at a price somewhat greater than the amount loaned, but less than the dealer would receive if they were paid the next year at the official interest rate.   For the sake of simplicity, let’s assume that the central bank pays the dealer exactly half the rate of interest the dealer would have received by holding the debt to maturity.   For example, if the dealer loans the treasury $1,000 at 5% interest, the central bank then purchases that debt for $1,025.  The dealer thus makes a profit of $25, and the treasury now owes the central bank $1,050.

So now let’s take another look at Example B above, but this time we will modify the numbers to take account of these new procedures:

On July 1st of Year One the treasury borrows $100 billion from the dealer at 5% interest.   The central bank then purchases that debt for $102.5 billion from the dealer.  The dealer has made a profit of $2.5 billion on the total transaction.  The treasury spends the entire $100 billion that it borrowed over the course of the year.  It owes the central bank $105 billion as a result.

On July 1st of Year Two the treasury borrows $210 billion from the dealer at 5% interest.   The central bank then purchases that debt for $215.25 billion from the dealer.  The dealer has made a profit of $5.25 billion on the total transaction.  The treasury spends $105 billion of the borrowed amount over the course of the year, and uses the remaining $105 billion to pay off the previous year’s debt now owed to the central bank.  Due to the new borrowing, it now owes the central bank $220.5 billion.

On July 1st of Year Three the treasury borrows $330.75 billion from the dealer at 5% interest.   The central bank then purchases that debt for $339.02 billion from the dealer.  The dealer has made a profit of $8.27 billion on the total transaction.  The treasury spends $110.25 billion of the borrowed amount over the course of the year, and uses the remaining $220.5 billion to pay off the previous year’s debt now owed to the central bank.  Due to the new borrowing, it now owes the central bank $347.29 billion.

This recurring procedure again goes on for twenty years, so that finally …

On July 1st of Year Twenty the treasury borrows $5,053,900,000,000 from the dealer at 5% interest.   The central bank then purchases that debt for $5,180,250,000,000 billion from the dealer.  The dealer has made a profit of $126.35 billion on the total transaction.  The treasury spends $252.7 billion of the borrowed amount over the course of the year, and uses the remaining $4,801,200,000,000 to pay off the previous year’s debt to the central bank.  It still owes the central bank $5,306,600,000,000 as a result of the new borrowing.

Now how have these new procedures changed the fundamental economic outcomes in our fictional society?  As far as the treasury goes, nothing has changed.  The treasury still spends the same amount each year, and it still lends the same amount each year.  The treasury debt year-over-year is unchanged from Example B.   But notice that the central bank is injecting more money overall into the system.  Some of that money still goes to the treasury and is then spent into the economy just as before.  But some now goes to the dealer.   And notice also that the quantity of central bank monetary injections, as a percentage of GDP, has changed.  Because the central bank is injecting this money on the dealer side of the operation in line with a fixed percentage applied to an ever-increasing government debt total, the dealer’s annual take rises at a pace that exceeds the annual increase in GDP.  The debt the treasury owes the central bank is still meaningless, but total central bank injections as a percentage of GDP climb from 5.13% of GDP in Year One to 7.50% of GDP in Year Twenty.  Treasury spending stays at 5%, but the dealer share of GDP goes from .125% to 2.5% over the twenty year period, and the dealer share of raw injected dollars rises from $2.5 billion to $126.35 billion.

Note also that, as before, the entire functional effect of the operation could be accomplished by directly crediting the treasury account by the annual treasury share of the injection, and directly crediting the dealer account by the annual dealer share of the injection.   The borrowing pays no essential role.

We have considered two main scenarios so far: one in which the treasury borrows everything it spends directly from the central bank, and one in which it borrows from private sector dealers who then sell the debt to the central bank at a profit.   Now let’s consider a third scenario.   Suppose that just as in the second scenario the treasury sells debt to the private sector to finance its spending and to repay all of its outstanding debt.   But in this new scenario, the central bank does not buy all of the debt from the private sector dealers, but instead buys just a portion of it.   For simplicity, we will consider a situation in which the central bank always buys exactly half of the treasury debt purchased in that year.  The ultimate effect is that annual size of total government money injections into the private sector goes down.  Let’s see why:

Each year the dealer sends a payment to the treasury to purchase the entire debt issue for that year.   And each year the government sends three types of payment to the private sector: (i) the annual treasury spending amounting to 5% of GDP as before, (ii) the central bank purchase from the dealer of half of the debt purchased by the dealer that year, and (iii) the treasury debt repayment to the dealer for the portion of the previous year’s debt that was not purchased by the central bank.

The result is that in Year Two, for example, the treasury borrows $210 billion from the dealer at 5%.  The treasury spends $105 billion which is 5% of GDP.  The central bank purchases half of the $210 billion debt from the dealer for $107.63 billion, giving a 2.5% profit to the dealer on that portion, and the treasury sends the dealer $52.50 billion to repay, with 5% interest, the portion of the previous year’s debt retained by the dealer.  The net government injection into the private sector, which includes the three government payments to the private sector minus the dealer payment to the treasury to purchase debt, is $51.25 billion – which is 2.56% of GDP.  And in Year Twenty, the net government injection into the private sector turns out to be $189.52 billion, which is 3.75% of GDP.

For each year after the first year, we can calculate the total profit the dealer has booked on the previous year’s lending.  Some of that profit was gained immediately after the debt purchase, when the central bank buys half of the current year’s debt, and some of the profit is reaped in the next year, when the treasury repays the portion of the debt that was not purchased by the central bank.  It turns out that the dealer profit continues to rise, going from .18% of GDP in Year Two to 3.39% of GDP in Year Twenty.   And the total government monetary injection into the economy goes from 2.56% of GDP in Year One to 3.75% of GDP in Year Twenty.

But how can this be?  How can the total government monetary injection be so low in each year when the treasury continues to spend, as before, 5% of GDP?  One way of looking at what is happening this is that an increasingly large proportion of the ordinary treasury spending, which is always 5% of GDP, is absorbed each year by the dealer to purchase additional debt from the government, and increasingly significant portions of that migration of funds are converted into dealer profit.

It’s clear that we are beginning to touch on complex distributional issues here, but that the distributional questions are obscured behind lending operations that, even in our extremely simplified fictional scenario, are not easy to track.  If these were real-world operations, we would have some important questions to ask: Who is the dealer?  Who does the dealer represent?  Who loses and who gains if we employ the operations described in the third scenario rather than the second one or the first one?

And further distributional issues are raised when we move to a final fictional scenario, one that begins to resemble the real world US system.   Suppose the government in our fictional example now voluntarily decides to collect taxes, in order to reduce the amount of debt it issues each year.   The taxes supplement the amount that is borrowed, and then the total revenues from taxes and borrowing are used by the treasury to fund both its spending and debt repayment.  Clearly the addition of taxation introduces other distributional factors into these operations.  Adding taxation to the mix affects who is receiving payments from the government, since fewer payments will go to the dealer, and it also affect who is sending payments to the government, since some payments are shifted from the dealer to the taxpayers.

But again, it should be clear that in both scenarios three and four, the government could achieve the same functional effects by changing the operations, eliminating the borrowing altogether, and instituting some combination of direct central bank credits and debits – both crediting and debiting dealer accounts and crediting and debiting other private sector accounts.

So what lessons can be drawn from these fictional examples for the real world of government financing operations under a fiat currency regime, a regime such as the one that operates in the United States?

I propose we view these operations in this way:  Our government is the monopoly supplier of net financial assets to the private sector economy.  But it supplies and extracts these monetary assets through two different channels: one is the US Treasury channel though which most of what we consider government spending occurs.  The Treasury channel is also the channel through which money is extracted from the private economy in the form of taxes.   The other channel is the banking channel operated by the Federal Reserve System.  The banking channel generates IOUs for dollars of its own accord, in response to the demand for credit in the marketplace for loans.  The Fed continually accommodates this expansion of bank credit by supplying additional bank reserves to the system.  These reserves provide banks and their customers with the final means of payment to which the bearers of bank IOUs are entitled.  The chief mechanism through which the Fed provides these reserves is via the purchase of US Treasury debt.

So the issuance of US Treasury debt plays a dual role.  On the one hand, as debt is issued, rolled over and re-issued over time, the accumulated effect is to convey Fed-issued money into the public treasury.  On the other hand, the debt is purchased by dealers representing financial institutions.  When the debt is then subsequently purchased by the Fed, the Fed is in effect creating additional money and supplying to the financial system as in the form of bank reserves.   This is free money, because it comes in the form of profit on the purchase and re-sale of government debt.  The government sells debt to dealers, and then pays the dealers a premium to sell the debt back to them.  The dealers profit from serving as middle-men between two branches of the same government!

But as we have seen from considerations of the thought experiments above, the entire functional effect of these operations could be accomplished in a precisely equivalent way without the issue of a single Treasury bond, note or bill.  The mechanisms of borrowing and repayment now serve mainly to bamboozle the public and empower unscrupulous opponents of public spending and progressive government.  These borrowing and lending operations create large, frightening debt numbers on government account books.   These numbers are exploited by demagogues to create insolvency fears among the public, and these fears are in turn exploited to pressure citizens into shrinking the active role of government.

People who have not thought through all of the byzantine bookkeeping of government finance and Fed-Treasury interactions – and thinking through these operations is hard! – naturally seek to apply the best analogy they can find in order to understand them.  Unfortunately, the analogy they typically employ is the analogy with household debt.   But that analogy is completely inappropriate.  The issuance of government debt under a fiat currency regime by a government that is the monopoly supplier of the public currency is nothing like household debt.   The government issues this debt solely as a mechanism for injecting additional currency into the private sector economy, not because it needs to “raise” the additional funds.  Households, on the other hand, are users of the public currency, not issuers of it.  They have a finite budget constraint, and issue debt in order to acquire funding they would not otherwise have.  And when they can’t pay their debts, they face bankruptcy.  The government under a fiat currency regime, in strong contrast, can never be in this situation.

The sovereign national government in a fiat currency system has an option available to it that no currency user in the system possesses: it can run a pure deficit.  That is, it can simply spend more than it takes in, without borrowing to cover the gap.   The Federal Reserve System in the US is a creature of the US Congress.  It was created by an act of Congress and represents a delegation of the constitutionally-provided monetary authority that belongs by right to Congress.    Congress could at any time choose to direct the Fed to credit the account of the US Treasury by any amount Congress desires.  It could expand the deficit, and help counteract demand shortfalls in a stagnating economy, without issuing a single penny of additional debt.

The only limit on the process is the policy goal of price stability.  A significant net expansion of dollar-denominated financial assets could be inflationary in some circumstances.  However, in circumstances of high unemployment and underutilized manufacturing capacity, we are probably far away from that price stability danger horizon.    Clearly, the quantity of net financial assets in the private sector always needs to increase continually along with the growth in the economy, that is, along with increases in the total volume of goods and services available for sale.   The spending channel is one way in which the consolidated government – Fed and Treasury combined – accomplish that task.   The bank credit channel is another way.  But sometimes the effectiveness of the credit channel breaks down when interest rates are near zero, credit demand remains low, household incomes are stagnant and declining, and the desire to pay down debt is strong.  The obvious solution in such circumstances is to directly inject monetary income into household and business bank accounts, either in exchange for goods and services, or with no strings attached.

The American people are by right the sovereign masters of their currency system, and should demand that the US Congress take charge of that system and end the illusory game of public debt and public debt fear-mongering.  If we need to extract monetary assets from an overheated economy to reduce demand, we can do it by taxing.   If we need to inject monetary assets into an underperforming economy to increase demand, we can inject those assets directly.   If we need to redistribute wealth and purchasing power to restore democratic balance in our society, we can both tax away monetary assets and inject monetary assets at the same time.   A system relying solely on taxation and direct monetary crediting, and that eschews public borrowing altogether, would be much more transparent.  The effects of its operations would be more open and obvious to the public, and that would improve intelligibility and clarity, and re-empower democratic decision-making.

The borrowing game is pointless.   Let’s end it.