A statistical model that turns eight accounting and market inputs into an estimated probability that a company hits financial failure within the next 12 months.
Published by Quintessentia Network Inc. · Updated 17 July 2026 · Sources & disclosures
The CHS model blends accounting and market inputs into a 12-month failure probability.
▶ Watch: CHS Failure Hazard explained in 24 seconds
What it is
CHS Failure Hazard is a distress model from Campbell, Hilscher and Szilagyi (2008), published as "In Search of Distress Risk" in the Journal of Finance 63(6). It is a logit model: eight inputs are weighted, summed together with a constant, and passed through the logistic function to produce a probability of financial failure over a chosen horizon. The version Quintarthai reports is the 12-month specification. Unlike Altman's Z-Score, which uses accounting ratios alone, CHS deliberately blends accounting data (profitability, leverage, cash) with market data (excess returns, volatility, relative size, market-to-book, price level). It also weights the profitability and return inputs as geometric averages over trailing quarters rather than reading a single point in time.
Why it matters
CHS is useful because it mixes two kinds of evidence: the balance sheet's view of a company and the market's view, which often moves first. Its averaged inputs make the reading steadier than a model that reacts to one bad quarter, and because the output is a probability rather than a raw score, it is easier to compare across companies. Reading it beside Altman Z, the Ohlson O-Score and the Merton distance-to-default shows where different methodologies agree and where they diverge — divergence is itself informative and worth investigating. A pitfall is that the PRICE coefficient (-0.058) is statistically insignificant at the 12-month horizon (z = 1.40), so that term should not be read as carrying real explanatory weight, and more broadly a CHS output describes what the model estimates given its historical calibration — it is a research signal about model probability, not a prediction that a specific company will fail. The model also assumes a conventional industrial capital structure, which is why Quintarthai gates it to "n/m" for banks and insurers and whenever inputs are missing.
How it's calculated
Eight inputs are built for the company. NIMTAAVG is net income scaled by the market value of total assets (MTA = market equity + total liabilities), averaged over trailing quarters with geometric weights based on phi = 2^(-1/3), which halves each quarter's weight going back; EXRETAVG is the excess stock return averaged the same way. TLMTA is total liabilities over MTA, CASHMTA is cash over MTA, SIGMA is return volatility, RSIZE is relative size, MB is market-to-book, and PRICE is log price truncated above at $15. These eight are multiplied by the Campbell-Hilscher-Szilagyi 12-month coefficients (NIMTAAVG -20.264, TLMTA +1.416, EXRETAVG -7.129, SIGMA +1.411, RSIZE -0.045, CASHMTA -2.132, MB +0.075, PRICE -0.058) and summed together with the model constant of -9.164. The logistic function converts that sum into a probability between 0 and 1, reported as the 12-month failure hazard.
How Quintarthai uses it
The deep-analysis page on /app/ shows the CHS 12-month failure hazard in its risk block, alongside Altman Z, the Ohlson O-Score, the Merton distance-to-default, and the Economic-Profit spread, with each reading n/m-gated for banks and insurers and when inputs are missing. It is shown with methodology and caveats as an educational research signal — Quintarthai never gives buy or sell advice.
Cross-border note. The CHS model was estimated on US-listed equities, so its coefficients are US-calibrated; applying it to a TSX or TSX-V name uses those US-fitted weights on Canadian inputs. Quintarthai computes it for both US and Canadian tickers where the required market and balance-sheet inputs exist, and gates it to "n/m" when they don't.
FAQ
Does a high CHS hazard mean the company is going to fail?
No. It means the model, given this company's inputs and its historical calibration, assigns an elevated probability to a failure event within 12 months. It is a research signal describing model output, not a forecast of what will happen to a particular company. Many firms with elevated readings never fail, and the model can miss ones that do.
Why does CHS use averaged profitability and returns instead of the latest quarter?
The NIMTAAVG and EXRETAVG terms are geometrically weighted averages built on phi = 2^(-1/3). The exponents run in months, so consecutive quarterly terms carry weights of phi^0, phi^3, phi^6 and phi^9 — and because phi cubed equals one half, each quarter counts half as much as the quarter after it, with roughly a year of history still contributing. This lets recent quarters dominate while still carrying some history, so a single unusual quarter does not swing the score as violently as a spot reading would.
Check your understanding
What does the phi = 2^(-1/3) weighting in the CHS model do to the NIMTAAVG and EXRETAVG inputs?
phi = 2^(-1/3) is a geometric decay factor whose exponents run in months, so successive quarterly terms are weighted phi^0, phi^3, phi^6, phi^9. Since phi cubed equals one half, each quarter's weight is half the weight of the quarter after it — the halving happens every quarter, not every year. Recent profitability and excess returns dominate the average while older quarters still contribute, which keeps a single odd quarter from dominating the score.