April 6, 2017
Although the Recent Weakness in Bank Credit Growth May Not Be a Concern to Others, It Is to Me
Starting around this past December, growth in commercial bank credit (loans and securities) slowed precipitously (see Chart 1). Annualized 13-week growth in bank credit of late is the slowest since the summer of 2013. This weakening in bank credit growth has been noticed and commented on by at least two economic analysts besides me – University of Oregon economics professor Tim Duy and Goldman Sachs economist Spencer Hill. These two analysts have concentrated on the weakness in one component of bank credit – commercial and industrial(C&I) loans – and concluded that there is nothing to get excited about with respect to the pace in U.S economic activity. I do not share their sanguine view. Notice that the data in Chart 2 show that the growth in bank credit excluding C&I loans also has slowed precipitously since this past December. If history is any guide, this weakening in bank credit growth excluding C&I loans is cause for concern with regard to the pace of economic activity.
Professor Duy, employing sophisticated econometrics techniques, elegantly corroborated what the Conference Board told us decades ago – the behavior of commercial and industrial bank loans is a lagging indicator of economic activity. What Professor Duy did not do was explain in Dick and Jane (Mr. Pero, that was for you) or otherwise why this is the case. As the data in Chart 3 show, percentage changes in business inventories have a relatively high contemporaneous correlation (0.57) with percentage changes in bank C&I loans. So, businesses rely heavily on bank loans to finance their inventories. To understand why bank C&I loans are a lagging indicator of economic activity, we need to understand the behavior of business inventories relative to business sales in the business cycle. As business sales start to slow, inventory growth tends to picks up. This is shown in Chart 4. This increase in inventory growth relative to sales growth typically is involuntary. With slower growth in revenues, businesses rely even more heavily on their banks to finance their higher involuntary inventory builds.
So, a surge in inventories and C&I loans often is associated with a slowdown in the growth of final demand for goods and services. Hence, bank C&I loan growth often tens to lag growth in final demand for goods and services. That is, the behavior of bank C&I loan growth provides more information as to where the overall economy has been rather than where it is headed.
Goldman’s Mr. Hill hypothesized that the recent weakness in bank C&I loan growth was due to the re-opening of the bond markets to oil and gas industry borrowers. According to Mr. Hill, when energy prices fell in 2015 and 2016 (obviously due to the anticipation of a Trump administration that would promote oil/gas exploration and the elimination of environmental regulations), oil and gas producers had to tap their bank lines of credit because bond-market lenders became more wary. According to Mr. Hill, then, the recent weakness in bank C&I loan growth is largely attributable to a more receptive bond market toward oil and gas industry borrowers and does not signal an imminent slowdown in U.S real GDP growth from its blistering Q4:2016 annualized pace of 2.1%.
And I agree with Messrs. Duy and Hill that the recent slowdown in bank C&I growth is not alarming with regard to the future course of U.S. economic activity. But I believe bank C&I loan growth is a red herring with regard to the future course of the economy. Instead, I focus on the growth in bank credit excluding the C&I loan component. And as I mentioned at the outset of this commentary, growth in bank credit ex C&I loans also has weakened precipitously starting in December 2016.
I am arguing that thin-air credit growth (you knew it was coming) excluding C&I loans is a better leading indicator of the pace of domestic demand than is thin-air credit growth including C&I loans. I will demonstrate this to you by comparing changes in correlation coefficients when thin-air credit leads and lags growth in domestic demand. Plotted in Chart 5 are year-over-year percent changes in quarterly observations of the sum of depository institution credit (including C&I loans) and the monetary base (the sum of depository institution reserves at the Fed and currency) along with year-over-year percentage changes in quarterly observations of Gross Domestic Purchases. The contemporaneous correlation coefficient between these two series is relatively high 0.61. (Remember, a perfect correlation is 1.00). Plotted in Chart 6 is the same thing except that C&I loans are excluded from the thin-air credit growth aggregate. The contemporaneous correlation coefficient between growth in thin-air credit growth excluding C&I loans and growth in Gross Domestic Purchases is 0.55 – not bad for private-sector analysis, but less than the 0.61 correlation coefficient when C&I loans are included in thin-air credit growth.
Remember, though, I am trying to discern which measure of thin-air credit growth is a better leading indicator of economic activity – thin-air credit growth with C&I loans or excluding C&I loans. So, we need to see what happens to correlation coefficients when thin-air credit growth leads and lags Gross Domestic Purchases growth. Contemporaneous correlation coefficients tell us nothing about leading or lagging characteristics. Does thin-air credit growth “cause” Gross Domestic Purchase growth or vice versa? When thin-air credit growth including C&I loans is advanced by one quarter, implying that today’s thin-air credit growth “causes” tomorrow’s Gross Domestic Purchases growth, the correlation coefficient falls from 0.61 to 0.59. When thin-air credit growth excluding C&I loans is advanced by one quarter, the correlation coefficient rises from 0.55 to 0.57. Now let’s retard thin-air credit growth by one quarter, implying that Gross Domestic Purchase growth “causes” thin-air credit growth. When we do this, we find the correlation coefficient for thin-air credit growth including C&I loans is 0.61, the same as its contemporaneous correlation coefficient and higher than 0.59, its correlation coefficient when thin-air credit growth including C&I loans is advanced by one quarter. These changes in the correlation coefficient suggest that thin-air credit growth including C&I loans is a lagging indicator of economic activity. When thin-air credit growth excluding C&I loans is retarded by one quarter, the correlation coefficient falls to 0.51 compared to its contemporaneous level of 0.55 and its one-quarter-advanced level of 0.57. These changes in correlation coefficients suggest that thin-air credit growth excluding C&I loans is a leading indicator of economic activity.
By the way, in case you think that there might not be much left of depository institution credit once C&I loans are excluded, take a look at Chart 7. From 1952 through 2016, the median percentage of nonfinancial business loans from depository institutions as a percent of total depository institution credit was 14.7. In 2016, C&I loans accounted for 13.9% of total depository institution credit. So, C&I loans, although a significant portion of total depository institution credit, are far from the whole ball of wax.
Okay, now that I have established (beyond a shadow of doubt?) that thin-air credit excluding C&I loans is the better leading indicator of the two, let’s see how it has been behaving on a year-over-year basis in recent weeks and months. This is shown in Chart 8. Although year-over-year growth in weekly observations of commercial bank credit excluding C&I loans has been slowing since October 2016, there has been some acceleration in the growth of combined bank credit ex C&I loans and the monetary base in recent weeks. Mind you, at 3.4% in the 52 weeks ended March 22, growth in this version of thin-air credit still is weak in an historical context. If commercial bank credit is not boosting modestly the growth rate of thin-air credit of late, it must be the monetary base. As can be seen in Chart 9, one important factor that has been increasing the monetary base since January is the decline in Treasury deposits at the Fed. All else the same, when these deposits decline, depository institution reserves increase. But as the April 15 tax payment date approaches, Treasury revenues will spike up. To the degree that these revenues are transferred to the Fed, all else the same, the monetary base will decline. In addition, when the Fed raises the federal funds rate, it has to reduce the supply of reserves in order to push up the interest rate. In sum, I would expect that in coming weeks the monetary base will be contracting. Unless there is a resurgence in bank credit growth, total thin-air credit growth will slow from an already tepid pace.
On January 17, I published a commentary entitled “2017 – Shades of 1937”. In the commentary, I wrote: “Based on the recent slowdown in thin-air credit growth, I believe that a significant slowdown in the growth of nominal and real U.S. domestic demand will commence in the first quarter of 2017.” Perhaps “significant” was too strong an adjective, but I hold by my prediction of a slowdown in the growth of real domestic demand. Despite relatively strong employment gains in January and February and hinted at by the March ADP employment guesstimate, real GDP growth in Q1:2017 appears to have come in at an even weaker pace than that of the paltry 2.1% annualized in Q4:2016. (Perhaps the depths of the productivity labor pool are being plumbed, requiring a larger quantity of workers to get a given amount of output produced.) The Federal Reserve Bank of Atlanta’s GDPNow Q1:2017 real GDP annualized growth estimate as of April 4 is 1.2%. Of course, this does not yet incorporate March data. Real personal consumption, which has accounted for about 68% of total real GDP in recent years, is coming in weak based on January and February readings. If the March level of real personal consumption were to be unchanged from the February level, Q1:2017 real personal consumption will have grown at an annualized pace of 0.3% -- not 3.0%, but 0.3%. In order for Q1:2017 real personal consumption expenditures to grow at the 3.5% annualized pace of Q4:2016, March real personal consumption would have to grow at annualized rate of 9.75% vs. February. How likely is this given that from January 2010 through February 2017 there have been only two month when real consumption expenditures grew by at least 9% annualized month-to-month? The median month-to-month annualized growth in real personal consumption from January 2010 through February 2017 has been 2.3%. If March real personal consumption were to grow at an annualized 2.3%, this would imply Q1:2017 real personal consumption growth of only 1.1%. What is arguing against a strong reading of Q1:2017 real personal consumption growth is the annualized 17.1% Q1:2017 contraction in unit sales of light motor vehicles, the largest quarterly contraction since the 30.2% contraction in Q4:2009, the quarter after the federal “cash-for-clunkers” program that boosted motor vehicle sales.
In conclusion, with or without C&I loans, thin-air credit growth remains weak. Weak thin-air credit growth implies weak growth in domestic demand. If the Fed does raise its federal funds rate target a couple of more times this year as it has indicated it might, this would weaken thin-air credit growth further and would likely bring on a recession. My bet is after seeing the weakness in Q1:2017 real GDP growth and the lack of rebound in early Q2:2017, the Fed will hold its fire.
Paul L. Kasriel
Founder, Econtrarian, LLC
Senior Economic and Investment Advisor
“For most of human history, it made good adaptive sense to be fearful and emphasize the negative; any mistake could be fatal”, Joost Swarte