The Credit Cycle and Bank Lending Standards

A mid-sized manufacturer walks into its bank in 2006 and asks to refinance a $300 million revolver. It gets the money in a week, at a thin spread, with a light covenant package. The same company, with the same machines, customers and order book, walks in again in early 2009 and is told the revolver is being cut to $180 million, the spread is doubling, and there is a new leverage test. Nothing about the borrower changed. Everything about the lender changed.

That is the credit cycle: not a cycle in the demand for money, since companies always want cheap money, but in the willingness to supply it. Because almost every decision that matters in the real economy — build a plant, hire a shift, buy a house — is financed, that willingness is one of the tightest constraints on growth there is.

The four phases

The cycle repeats with enough regularity to name its stages.

Repair. Losses taken, capital rebuilt, standards still tight and loan books shrinking. Spreads are wide and almost nobody is borrowing — the best time to lend and the moment lenders are least willing to.

Expansion. Losses fall, capital is ample, competition for good borrowers pushes spreads in and volumes grow faster than the economy. Standards ease gradually, one covenant at a time.

Euphoria. Credit growth outruns cash-flow growth and underwriting degrades invisibly, because a growing loan book always looks healthy — new loans have not had time to go bad. Minsky's taxonomy describes it exactly: borrowers migrate from hedge finance (cash flow covers interest and principal) to speculative finance (covers interest, must refinance principal) to Ponzi finance (covers neither, and needs the asset to appreciate).

Contraction. Something reprices — a default, a rate shock, a funding run. Losses hit capital, standards tighten across the system at once, and solvent borrowers are refused because their lender has no balance sheet, not because they are bad credits.

How you actually measure it

The single most useful instrument is the Federal Reserve's Senior Loan Officer Opinion Survey (SLOOS), run quarterly across roughly eighty large domestic banks and a couple of dozen US branches of foreign banks. Its headline is a diffusion reading: the net percentage of banks reporting tighter standards on commercial and industrial loans, meaning the share tightening minus the share easing.

The series is unusually well behaved. In calm expansions it sits slightly negative. It rose above 80 percent at the end of 2008, roughly 70 percent in the second quarter of 2020, and around 50 percent in mid-2023 after the regional bank failures. And it leads: Lown and Morgan showed that shocks to standards precede changes in loan volumes and output by several quarters.

Two market-priced gauges sit alongside it. High-yield credit spreads move far faster: the ICE BofA US high-yield option-adjusted spread has ranged from roughly 300 basis points in calm markets to about 1,100 in March 2020 and near 2,000 at the end of 2008. And the credit impulse — the change in the flow of new credit as a share of GDP — tracks the acceleration of lending rather than its level, which is what growth actually responds to.

Worked example: why a small loss removes a lot of lending

A bank holds $11 billion of common equity tier 1 capital against $100 billion of risk-weighted assets, a CET1 ratio of 11.0 percent, and it manages to that target because its regulator, its board and its bond investors all expect it.

1. Take a loss. Charge-offs of $1 billion come through. CET1 falls to $10 billion.
2. Restore the ratio. To get back to 11.0 percent on the new capital base, risk-weighted assets must fall to $10 / 0.11 = 90.9$ billion.
3. Read the consequence. That is a $9.1 billion contraction in risk-weighted lending caused by a $1 billion loss.

Leverage runs in reverse on the way down. The bank shrinks risk-weighted assets by refusing new loans, cutting undrawn commitments and shifting toward zero-risk-weight government bonds — none of which shows up in default statistics. Provisioning amplifies it: under the current expected credit loss standard, banks book lifetime expected losses when a loan is originated, so lending into a deteriorating outlook is immediately expensive.

Worked example: what tightening does to one borrower

The manufacturer from the opening has EBITDA of $100 million and $300 million of debt, so leverage is 3.0 times. It borrows at SOFR plus 200 with a maximum leverage covenant of 4.0 times.

The cycle turns. The bank reprices to SOFR plus 375 and tightens the covenant to 3.5 times. With SOFR at 4.30 percent, interest becomes $300 \times 0.0805 = 24.2$ million a year, up $5.3 million on the old spread.

Now the economy softens and EBITDA falls 15 percent to $85 million. Interest coverage is $85 / 24.2 = 3.5$ times, so the company can comfortably pay. But leverage is $300 / 85 = 3.53$ times and the covenant breaks. The default is triggered by a ratio, not by an inability to pay, and the remedy on offer is a smaller facility at a wider spread. The borrower cuts capex and headcount. That is the transmission channel, and it runs entirely through documentation.

Spreads are not standards

The two get substituted for one another constantly and should not be. Spreads are set by markets, reprice in minutes and mean-revert violently; they say more about liquidity and positioning than about credit availability. Standards are set by committees, move in one direction for years, and decide whether a borrower with no bond market access gets funded at all. Most of the economy — small firms, commercial real estate, middle-market borrowers — lives entirely on the second one.

Key terms

• Lending standards — the non-price terms of credit: covenants, collateral, maturity, facility size.
• SLOOS — the Fed's quarterly Senior Loan Officer Opinion Survey, reported as a net percentage tightening.
• Diffusion index — a measure of how many respondents moved in each direction, not by how much.
• Minsky's three postures — hedge, speculative and Ponzi finance, ordered by how much refinancing the borrower needs.
• Credit impulse — the change in new credit flow as a share of GDP; the acceleration, not the level.