← Quantery Blog

How to screen after a Fed rate hike

September 17, 2026 · 8 min readratesleveragethesis-craft

A Federal Reserve rate hike puts corporate financing claims under a harder test. Start with debt service: can current operating income cover current interest expense? Then test refinancing exposure, cash generation, and valuation separately. A useful thesis asks whether businesses with weak coverage and near-term refinancing needs fare differently from businesses whose cash flow can absorb dearer funding.

On 16 September, the Fed raised its target range by a quarter point to 3.75% to 4.00%. The vote was unanimous. The statement said inflation remained elevated while domestic spending had been resilient. Yahoo Finance's contemporaneous coverage called it the first increase since 2023 and reported that markets had largely priced it in. The announcement is the event. The investable question is what a higher refinancing rate does to company cash flows over time.

A policy rate isn't a company's borrowing cost

The federal funds rate is an overnight rate between depository institutions. A manufacturer doesn't roll its bonds overnight at the Fed. Its cost of debt depends on the Treasury curve and the credit spread investors demand. The maturity, fixed or floating terms, covenants and the date the company actually needs the money matter too.

That chain matters because the headline can move before the income statement does. A fixed-rate bond maturing years from now keeps its old coupon after a policy hike. A floating-rate loan can reset much sooner. A bond due next quarter has to be repaid or refinanced on whatever terms the market then offers. Two companies with identical debt balances can have very different exposure.

The market backdrop was already expensive. The Federal Reserve's H.15 release put the ten-year Treasury constant-maturity yield at 5.00% on 15 September. FRED's BBB corporate option-adjusted spread was 0.98 percentage points that day, while its US high-yield spread was 2.76 points. Those index spreads aren't quotes available to a particular borrower, and they aren't both additions to the ten-year yield. They show the extra compensation the market demanded over the relevant Treasury curve for broad credit baskets.

So don't build a thesis that treats the quarter-point move as a quarter-point jump in every company's interest bill. Build one that identifies where a higher cost could enter, then asks whether the business has room for it.

Interest coverage is the first filed test

Interest coverage is operating income divided by interest expense. It asks how many times the operating profit for a period could pay that period's financing cost.

interest coverage = trailing operating income / trailing interest expense

The numerator and denominator come from the income statement. Use trailing filed periods for both, so one seasonal quarter doesn't decide the result. Our article on trailing versus forward numbers explains why that window has to use filings available at the time of the test.

Coverage is useful because it measures the bill companies are paying now. It's also late. The ratio won't see a refinancing shock until the new interest expense reaches a filing, and management can have a problem months before that. A company with ample coverage today may have a large maturity coming due; another with thinner coverage may have locked its coupon for years.

The maturity schedule closes part of that gap. It sits in the debt footnote, commonly grouped by amounts due in future periods. Fixed-versus-floating terms sit there too. Structured fundamentals give a market-wide first pass, but they don't turn a footnote into a clean universal field. This is where the scan hands the job back to the reader: open the filings of the survivors and inspect the calendar.

The underlying leverage measure still matters. Net debt over EBITDA asks how large borrowings are relative to operating earnings before financing costs and non-cash asset charges. Coverage asks whether current operating income pays current interest. Gross debt, net debt, and coverage answer different questions; use both and preserve the disagreement between them.

The rate thesis needs separate failure modes

One giant "rate sensitivity" score hides too much. Break the idea into claims that can fail on their own.

Debt service. Low interest coverage means little room for operating weakness or a dearer refinancing. Negative operating income makes the ratio unusable in the flattering direction, so it should fail the test instead of sorting as an odd negative multiple.

Balance-sheet load. A borrower can cover a cheap legacy coupon today while carrying debt that becomes harder to refinance later. Net debt over EBITDA catches some of that stock of obligations. It still misses leases, pension claims, and restricted cash.

Cash conversion. Accounting earnings don't repay principal. Free cash flow shows what remained after operating cash movements and capital spending. A company can post acceptable coverage while working capital or required capex consumes the cash. Read operating cash flow against net income before treating the income-statement ratio as liquidity.

Price. A strong balance sheet can be priced as a strong balance sheet. Valuation belongs in its own criterion, where you can see whether the test rewards financial resilience or just expensive quality. Keep the numerator matched to the denominator; earnings yield and FCF yield walks through that pairing.

Keep these criteria separate. If the backtest works, remove each one in turn. An interest-rate story that survives without the coverage rule probably found a generic quality factor. That's still a finding, but it isn't evidence for the story you started with.

Write the rate-hike thesis so it can be wrong

The shape below is illustrative, and the bundled templates plus the Quantery DSL reference are the reference for exact supported fields. The thresholds are proposed research parameters. They're meant to move.

# "Cash flow with room for dearer debt" (illustrative)
params:
  coverage_floor: 4.0
  net_debt_max:   2.0
  fcf_margin_min: 0.05
  fcf_yield_min:  0.04

features:
  ebit_ttm:     ttm(operating_income)
  interest_ttm: ttm(interest_expense)
  ebitda_ttm:   ttm(ebitda)
  revenue_ttm:  ttm(revenue)
  fcf_ttm:      ttm(free_cash_flow)
  debt:         newest(total_debt)
  cash:         newest(cash)

  coverage: if(interest_ttm > 0, ebit_ttm / interest_ttm, null)
  net_debt: debt - cash
  leverage: if(ebitda_ttm > 0, net_debt / ebitda_ttm, null)
  fcf_margin: if(revenue_ttm > 0, fcf_ttm / revenue_ttm, null)
  fcf_yield: if(market_cap > 0, fcf_ttm / market_cap, null)

criteria:
  debt_service:
    rules:
      - { when: "interest_ttm <= 0 and debt <= 0", score: 2 }
      - { when: "is_null(coverage) or ebit_ttm <= 0", score: 0 }
      - { when: "coverage >= $coverage_floor", score: 2 }
      - { when: "coverage >= 2.0", score: 1 }
      - { else: 0 }
  balance_sheet:
    rules:
      - { when: "net_debt < 0", score: 2 }
      - { when: "is_null(leverage) or leverage > $net_debt_max", score: 0 }
      - { else: 1 }
  cash_buffer:
    rules:
      - { when: "fcf_margin >= $fcf_margin_min", score: 2 }
      - { when: "fcf_ttm > 0", score: 1 }
      - { else: 0 }
  valuation:
    rules:
      - { when: "fcf_yield >= $fcf_yield_min", score: 1 }
      - { else: 0 }

gate:
  mode: strict

A few choices are deliberate. newest() keeps debt and cash on the same filing date. The positive-EBITDA guard stops a loss-making borrower from masquerading as lightly levered through a negative ratio. A company with interest expense but missing coverage gets no benefit of the doubt. And valuation receives its own point instead of being folded into "quality."

The code still can't identify the portion of debt due soon. Treat that as a second-stage filing check, or split the study into groups after hand-labeling maturity exposure. Don't pretend the net-debt multiple carries a calendar when it doesn't.

Backtest how financing costs reach the filings

A same-day event study asks how prices reacted to a Fed announcement. This thesis asks something slower: whether resilient financing and cash-flow traits mattered while borrowing costs worked through filings. Those are different experiments.

Use rebalance dates after new filings become public. Run the rules on point-in-time fundamentals so the simulated decision never sees later restatements. Compare more than one holding window because refinancing costs arrive when debt resets or matures, while the market may anticipate them much earlier. Every result is hypothetical and excludes costs.

Then test the story against itself:

• Move the coverage floor and leverage ceiling across nearby round values. A result balanced on one cutoff came from the dial.
• Separate fixed-rate-heavy and floating-rate-heavy cases where the footnotes let you classify them.
• Split net-cash companies from indebted companies. If both behave the same, the leverage channel may not be doing the work.
• Compare sectors separately. Utilities and other capital-intensive businesses don't carry debt for the same reason a software company does.
• Start the clock from filing dates, then compare it with the policy-announcement date. That reveals whether you're measuring company fundamentals or one afternoon's market surprise.

What makes a backtest honest covers the mechanics behind that discipline. One extra trap applies here: choosing only famous rate-hike episodes after seeing which produced the cleanest chart. Define the episodes and the rate rule before looking at returns.

What a passing result would establish

Suppose the thesis holds across several tightening periods, nearby parameters, and sensible sector splits. You'd have evidence that the stated financial traits were associated with different hypothetical outcomes in those samples. You wouldn't have shown that the latest hike caused the difference, that every future cycle will transmit the same way, or that the maturity footnotes don't overturn an individual case.

That boundary is useful. This week's Fed decision is raw material for a stock thesis. Turn the observation into rules, make room for the details the dataset can't carry, and test which part of the story survives when you change the assumptions. The rules are yours, including the one that kills the idea.

Want to try this on your own rules? Quantery is free for 14 days: the full app, no card required.

← All articles