Bank Credit Impulse

What the credit impulse measures

Total bank credit is the sum of loans and leases and securities held by all US commercial banks, reported every week in the Federal Reserve's H.8 release. It is one of the largest single numbers in the financial system: roughly twenty trillion dollars at the time of writing, against a nominal GDP of about thirty trillion. Almost every business that borrows from a bank, and most households that hold a mortgage or a card balance, appear somewhere in the total.

The level itself is close to useless as a signal. It has risen in 51 of the last 53 years, because nominal GDP rises and the banking system grows with it. What carries information is the rate at which the stock changes, and specifically whether that rate is accelerating or fading. That is the credit impulse: the flow of new credit, not the accumulated stock.

We report it as a thirteen-week change, annualised. Thirteen weeks is a quarter, which is long enough that one or two unusual weeks do not dominate, and short enough that a genuine turn shows up within about three months. The annualisation multiplies the quarterly rate by four so the number can be read on the same scale as a year-over-year figure, which makes it comparable to the growth rates quoted for GDP or for the money supply.

Why growth rather than the level

Credit growth is the financing side of nominal spending. A firm that expands payroll borrows to bridge the gap between wages paid and revenue collected. A household that buys a house borrows the purchase price. When the growth rate of credit falls, some of that financing is simply not there, and the spending it would have funded does not happen, or happens later and smaller.

This is why the growth rate leads the economy in a way the level cannot. In the four quarters before the 2008 recession, bank credit was still growing and the level was still setting records; the growth rate had already turned down. The same pattern appeared in 2000 and again in 2007. The stock is a lagging record of what has been financed. The flow is a leading indicator of what will be.

The asymmetry matters too. Credit can contract, and when it does the effect is not symmetric with expansion. A growth rate near zero means banks are roughly standing still, which in a growing economy is a contraction in real terms. The lowest reading in the record, minus ten percent annualised in October 2009, coincided with an unemployment rate heading toward ten percent.

What the record shows

The series starts in January 1973. Its median growth rate is a little under seven percent annualised, which is roughly the pace of nominal GDP growth over the same half century, as you would expect if credit and the economy grow together over long spans. Readings above twelve percent annualised are in the top tenth of the record; readings below two percent are in the bottom tenth.

The two extremes are instructive because they were made of different things. In May 2020 the impulse reached thirty-one percent annualised, its highest reading ever. That was not primarily a lending boom: banks added securities at an extraordinary pace as the Federal Reserve bought assets and deposits flooded in, and the H.8 total captures securities alongside loans. In October 2009 the impulse hit minus ten percent, driven by the opposite: loans written off, credit lines cut, and securities sold as banks shrank balance sheets to protect capital.

The 2022 and 2023 tightening cycle produced a milder version of the same shape. Credit growth fell from the post-pandemic peak toward zero without going negative, because large firms could still issue bonds when bank lending slowed. That substitution is the reason bank credit alone understates total credit conditions, and it is why we publish a separate reading on bank lending standards, which captures the supply side directly.

How it relates to M2 and to reserves

Bank credit and the money supply are two sides of the same balance sheet. When a bank makes a loan, it creates a deposit, and deposits are most of M2. The two series usually move together for that mechanical reason, which is why a divergence is worth noticing: credit growing while M2 is flat usually means deposits are leaving the banking system for money market funds, not that lending has stopped.

Reserves are a narrower thing still. Reserve balances at the Federal Reserve are the settlement medium between banks, not the funding of the real economy. In 2020 reserves and credit both exploded, which made them look like the same signal. Since 2022 they have diverged sharply as the Fed shrank its balance sheet while bank credit kept growing. Anyone who treated reserves as a proxy for lending got the last three years wrong.

How we calculate it

We take the weekly TOTBKCR series from FRED, which is the seasonally adjusted total of loans and leases plus securities held by all commercial banks, and compute the ratio of each week's level to the level thirteen weeks earlier. We raise that ratio to the fourth power, subtract one, and multiply by a hundred. No smoothing and no gap filling: if a week is missing, the calculation uses the closest prior observation, which is what the published series does anyway.

The percentile is the share of weekly observations since January 1973 that sit below the current value. Because the series is a growth rate with fat tails, a single extreme quarter can shift the percentile for years. Read the percentile alongside the raw number rather than instead of it.

Every input is a Federal Reserve release in the public domain. The growth rate, the annualisation and the percentile are computed here, so the series you see is ours and not a redistribution of anyone else's product.

Questions

Related

• Bank lending standards (SLOOS)
• M2 money supply growth
• Bank reserves at the Federal Reserve
• Credit market conditions
• Plans and full-data access