The Yield Curve, Thin-Air Credit and Real Aggregate Demand

In this commentary I will explain the relationship between the shape of the yield curve, the behavior of nominal thin-air credit and the behavior of real aggregate demand for goods and services. Specifically, I will argue that there is a positive relationship between the “slope” of the yield curve and the percent change in thin-air credit. Further, I will argue that there is positive relationship between percent changes in real thin-air credit and percent change in real domestic aggregate demand for goods and services. For example, the steeper the slope in the yield curve, the faster will nominal thin-air credit grow. In turn, the faster will real domestic aggregate demand grow. The yield curve concept, I will be referring to is the spread between the yield on the Treasury 10-year security vs. the overnight federal funds rate (hereafter referred to as the yield spread). The importance of the federal funds rate in this yield spread will be explained later in this commentary. Thin-air credit for the purposes of this commentary will be defined as the sum of the asset categories of securities and loans on the books of depository institutions (currently, primarily commercial banks). “Thin-air” refers to the fact that the system of depository institutions has the unique ability among various other private lending entities to create credit figuratively out of thin air. (The central bank also has the ability to create credit out of thin-air. In fact, the central bank creates the “seed money” that enables the depository institution system to create thin-air credit.) The important implication of the ability to create credit out of thin-air is that the borrower can spend borrowed funds without necessitating any other entity to reduce its current spending. Thus, when new thin-air credit is created, we can assume that nominal current spending will increase, all else the same. (We cannot determine, a priori, what will be purchased by the borrowers of this new thin-air credit. It could be newly-produced goods and services or it could be existing assets, physical or financial.) We cannot make this same assumption when new credit is extended by non-depository entities. For example, if an individual extends credit, in order to fund that loan, she either has to cut back on her current spending or run down her current deposit holdings. If she reduces her current spending, i.e., increases her saving, no new net spending will occur from this loan because she will merely be transferring her spending ability to the borrower. If she funds the loan by running down her deposits, then there will be some net new spending in the economy. In economist jargon, this would be an example of an increase in the velocity of money. An increase in the velocity of money is another way of stating that there has been a decrease in the public’s demand for money balances. I am not aware of any statistic that measures total transactions in the US economy, transactions that include not only expenditures for newly-produced goods and services but also purchases of existing assets. So, I must rely on expenditures recorded in the National Income and Product Accounts (where the GDP data are reported). Because the bulk of thin-air credit created by US depository institutions goes to domestic borrowers, I have chosen Gross Domestic Purchases as my measure of aggregate domestic demand.

If you intend to read further, I would advise you have some Red Bulls on ice. It’s going to be a long one. Plotted in Chart 1 are quarterly averages of the level of federal funds rate (the blue line) and the yield spread between the Treasury 10-year security and the federal funds rate (the red bars). The gray shaded areas demarcate periods of recession. Notice as the federal funds rate declines, the spread tends to widen. Similarly, as the federal funds rate rises, the spread tends to narrow, sometimes going into negative territory. Why might this be, especially, why might the spread narrow when the federal funds rate rises? Remember, the Federal Reserve sets the level of federal funds rate. It does so by regulating the supply of cash reserves it creates out of thin-air in relation to the amount of these reserves demanded collectively by depository institutions (hereafter, banks). Prior to March 26, 2020, the Fed imposed reserve requirements on banks. That is, banks were required to hold as cash reserves (either on deposit at the Fed or as coin/currency in their vaults) in an amount equal to a specified percentage of their deposits. The Fed varied this percentage from time to time for reasons known only to the Fed. Prior to October 9, 2008, when the Fed began paying interest on reserves held by banks, these required reserves largely defined banks’ demand for reserves. If the Fed wanted to increase the level of the federal funds rate, it would reduce the amount of reserves it supplied relative to banks’ demand for reserves. Even if the Fed had not imposed reserve requirements on banks, they still would have had a demand for reserves. Banks would want to maintain a certain level of reserves to cover their clearings with other banks. As mentioned above, on October 9, 2008, the Fed began paying interest to banks on their reserves holdings. The reason it did so was to increase banks’ demand for reserves. In November 2008, the Fed began its first round of quantitative easing (QE), which added reserves to the banking system, but had not yet lowered its federal funds target to zero. In order to prevent the federal funds rate from falling to zero, the Fed induced the banking system to hold these extra reserves by paying them interest on their reserves holdings. Eventually, the Fed lowered its target for the federal funds rate to zero, yet the Fed continued to pay interest on reserves held by banks’ as it still does to this day. Why?

But I digress. The federal funds rate represents the marginal cost of funds for a bank and it is pegged by the Fed. The farther out you move along the maturity curve, yield movements are more independent of the influence of the Fed. Yes, the yield on the 10-year Treasury security is affected by the current level and expected future levels of the federal funds rate. But the effect on the 10-year yield of expectations of future levels of the federal funds rate is de minimis compared to the effect on the 2-year Treasury. The behavior of the 10-year yield is a better representation of the true forces of the supply and demand for credit. Let us assume that the demand for credit declines. All else the same, the whole maturity structure of interest rates would “want” to fall, reflecting the decline in credit demand at the current level of interest rates relative to credit supply. But the federal funds rate cannot fall on its own volition because it is pegged by the Fed. However, the yield on the 10-year can fall. In order to maintain its growth in earning assets, banks would be inclined to lower their loan rates given the fall in the demand for credit. But because the pegged federal funds rate is banks’ marginal cost of funds, banks cut back on their acquisitions of earning assets in light of compressed interest-income margins. So, a reduction in the yield spread between the Treasury 10-year security and federal funds would result in a slowdown in the growth of thin-air credit. Plotted in Chart 2 are quarterly observations of the year-over-year percent changes in thin-air credit, i.e., the sum of loans and securities on the books of depository institutions, (blue line) and the four-quarter moving average of the yield spread (red bars). Although the fit is far from perfect, in general, when the yield spread widens, growth in thin-air credit increases and vice versa.

To recap, when the yield spread widens, this is an indication that the nominal demand for credit is increasing. When the yield spread narrows, this is an indication that the nominal demand for credit is decreasing. With the federal funds rate pegged by the Fed, when the demand for credit increases, banks are willing to accommodate the increased demand for credit because of widening net-interest margins. Thus, bank credit expands. When the demand for credit decreases, banks net-interest margins narrow because of the pegged federal funds rate. Thus, bank credit growth decreases.

Changes in real bank credit are positively correlated with changes in real aggregate domestic demand. This can be seen in Chart 3 in which are plotted year-over-year percent changes in the sum of depository institution loans and securities deflated by the chain-price index for Gross Domestic Purchases (blue line) and the year-over-year percent changes in real Gross Domestic Purchases, i.e., the sum of real expenditures of households, businesses and government entities (red bars). From 1955 through 2018, the correlation between these two series is 0.62 out of a possible 1.00.

Now, let us bring things up to date. The Fed has not yet released its Q2:2023 Financial Accounts report, so Q2 data for loans and securities at depository institutions are not available. But because loans and securities held by commercial banks now account for most of these data at all depository institutions, I am using bank credit as a proxy. Plotted in Chart 4 are the monthly observations of the year-over-year percent changes in nominal bank credit (blue line), bank credit deflated by the All-Items CPI (red line) and the 12-month moving average of the yield spread (green bars). The 12-month moving average of the spread peaked in mid 2022 at 1.67 percentage points and has trended lower to minus 0.50 percentage points in July 2023. Similarly, the year-over-year percentage changes in nominal and real bank credit have trended lower since mid 2022. In the 12 months ended July 2023 real bank credit contracted by 2.78%.

So, a narrowing in the yield spread is associated with a slowdown in the growth of both nominal and real thin-air credit growth, which, in turn is associated with a slowdown in real Gross Domestic Purchases growth. Recently, there has been a lot of “it is different this time” talk with respect to the behavior of the spread between the Treasury 10-year security yield vs. federal funds rate and the prospect of a near-term recession. There has been no talk about the contraction in real thin-air credit growth and its relationship to the prospect of a near-term recession because only my few readers and I are familiar with the concept of thin-air credit. As Chart 5 shows, a negative yield spread has usually, but not always has presaged a recession. For example, the yield spread did not turn negative prior to the recessions commencing in 1957 and 1960. But the yield spread did narrow prior to these recessions. Sometimes a negative spread has not presaged a recession. For example, the yield spread did go negative in late 1966 and early 1967. There was no official recession called in 1967, but growth in real Gross Domestic Purchases slowed significantly. This period has come to be referred to as the mini-recession of 1967. There is nothing magic in the yield spread going from +0.01 percentage points to -0.01 percentage points. What is more important is the trend of the yield spread. Also shown in Chart 5, recessions differ in their severity. The recessions commencing in 1960 and 1969 were relatively mild. The recession commencing in 2001 was so mild that the year-over-year percent change in real Gross Domestic Purchases stayed in positive territory.

The point is that significant narrowing in the yield spread is a leading indicator of a significant slowing in the growth of real Gross Domestic Purchases. I looked at monthly data of the behavior of the yield spread in relation to the seven recessions from 1969 through 2007, specifically from the month that that the yield spread turned negative through the peak month of the particular expansion. The range was nine months for the November 1973 peak to 18 months for the December 2007 peak. The median number of months from when the yield spread turned negative through the peak month of the business expansion was 12 months. Through July 2023, the yield spread has been negative for eight consecutive months. Again, there is more to it than just the fact that the yield spread is negative. The magnitude matters too. At minus 1.22 percentage points, the July 2023 yield spread is three basis points above the July 1989 nadir, six basis points below the December 2020 nadir and 52 basis points below the March 2007 nadir. So, both in terms of when the yield spread turned negative and the magnitude of the negative spread value, the current situation looks ripe for a recession to commence in 2023. Remember, jumping off the 15th floor of a building is a leading indicator of death. But death is not evident until the jumper reaches the ground! But maybe it is different this time.

So, there you have it. The “grand theory” linking the behavior of the yield spread to the behavior of thin-air credit and the behavior of aggregate domestic demand. The behavior of the yield spread with respect to recessions is not just some technical analysis. It has a theoretical basis. When the yield spread turns negative on a sustained basis, expect slower growth in thin-air credit and aggregate domestic demand growth to follow. Whether this process results in an “official” recession is uncertain. But something close to real economic stagnation is certain.