What Is the Credit-to-GDP Gap? Formula and Limits

Start with a ratio: credit outstanding for a defined sector divided by gross domestic product (GDP) over a defined period. Then compare that ratio with an estimated trend. The basic expression is:

Credit-to-GDP gap = credit-to-GDP ratio − estimated long-run trend

If both inputs are expressed as a percentage of GDP, the subtraction is reported in percentage points of GDP. The gap is not the growth rate of credit and is not the percentage by which debt exceeds a safe level. The “trend” is a statistical estimate, not an observed stock of sustainable lending. The BIS describes its published gap as the credit-to-GDP ratio less its long-run trend and presents the gap in percentage points of GDP. BIS credit-gap overview

This measure is different from the public debt-to-GDP ratio. The credit-gap series discussed here uses private non-financial-sector credit; government debt is outside that BIS input series. The similar names can obscure different borrowers, questions, and policy uses. For the separate government-debt comparison, see what the debt-to-GDP ratio measures.

The BIS credit-gap series begins with credit owed by the private non-financial sector. For its quarterly ratio, the outstanding debt at quarter-end is compared with the sum of nominal GDP over the latest four quarters. That pairs a point-in-time credit stock with a year of nominal economic output. The exact borrower boundary, credit instruments, timing, and GDP convention should always be read from the provider's metadata. BIS credit-gap overview

For a clearly hypothetical example, suppose a provider reports a credit-to-GDP ratio of 152% and an estimated trend of 143%. The gap is 152% − 143% = +9 percentage points of GDP. It is not “credit grew 9%,” and it does not mean debt is 9% above a safe or sustainable amount. Those inputs are invented to show the subtraction; they are not a current reading for a country.

The ratio and the gap answer related but different questions. The ratio gives the observed scale of credit relative to GDP under the chosen definitions. The gap describes how far that ratio sits above or below a chosen trend. A ratio can be high while the gap is modest if its estimated trend is also high; a lower ratio can have a positive gap if its trend is lower still.

The BIS comparison series aims for cross-country consistency. Its private non-financial sector includes households and non-financial corporations and covers borrowing from domestic and foreign, bank and non-bank sources. It excludes general government. The BIS notes that its underlying private-sector data are unconsolidated, so claims within that sector are not necessarily netted out. A national authority can use a different credit measure or data boundary and publish a different ratio or gap. BIS private-credit data BIS credit-gap overview

GDP is also a flow with a time window. In the BIS quarterly convention, the denominator is nominal GDP summed over the latest four quarters rather than one quarter's GDP alone. If a chart uses a different provider, check whether the credit measure is bank-only or broader, whether borrowers include government, and whether GDP is quarterly, annualized, or trailing four quarters. A comparison is meaningful only when the definitions are sufficiently aligned.

The stock-flow distinction helps prevent a common reading error. Credit outstanding is measured at a date, while GDP is produced during an interval. The ratio is a scaling device; it is not a repayment schedule and does not say that borrowers can use all of GDP to pay lenders.

The BIS estimates the long-run trend with a one-sided, backward-looking Hodrick–Prescott (HP) filter. “One-sided” means that the estimate for a given quarter uses information available up to that quarter, rather than future data. For quarterly credit-gap calculations, the BIS methodology uses a smoothing parameter of 400,000 and requires at least ten years of credit-to-GDP observations before publishing a gap. These choices make the trend a specific statistical construction, not a natural law or an equilibrium level of debt. BIS gap methodology

The filter separates the observed ratio into an estimated trend and a cyclical deviation by balancing fit to the data against smoothness of the trend. A smoother trend responds slowly to a rapid rise in the ratio, so the gap can widen during a credit boom. If a high ratio persists, the estimated trend may later catch up and the gap may narrow even if the ratio stays elevated. A narrowing gap therefore does not prove that financial vulnerabilities have disappeared.

The end of a time series is especially difficult to estimate because later observations are not yet available. As new quarters arrive, the estimated trend at recent dates can change. Revisions to credit or GDP, changes in source coverage, short histories, and structural breaks can also alter a gap. The BIS methodology explains its filter, its minimum-history rule, and the importance of the starting point. BIS gap methodology BIS review of credit-gap limitations

The ratio can rise because credit outstanding increases, because nominal GDP falls, or because the two change at different speeds. It can decline when GDP grows faster than credit even if the amount of debt has not fallen. A change in the ratio is therefore not identical to new lending, and it is not automatically evidence of deleveraging.

The gap has another moving part: the estimated trend. If the ratio is unchanged but the trend estimate moves, the gap changes. Conversely, the ratio may move while the gap changes less if the trend moves in the same direction. When describing a release, identify the observation date, the ratio, the trend, and the gap separately if the source provides them. This makes it easier to tell a change in borrowing from a change in the denominator or filter estimate.

For example, if nominal GDP contracts while outstanding credit is unchanged, the ratio can rise mechanically. That arithmetic does not establish that households and firms took out new loans. In the same way, a credit boom can raise the ratio faster than its slow-moving trend and widen the gap, but the gap alone does not reveal which borrowers, lenders, assets, or loan standards are driving the movement.

Basel III uses the credit-to-GDP gap as a common reference point for authorities considering the countercyclical capital buffer (CCyB). The buffer is intended to strengthen banks' resilience when system-wide vulnerabilities build up alongside credit growth. The common guide helps structure and communicate an assessment; it does not mechanically set a country's buffer. Authorities are expected to use judgment and may consider other quantitative and qualitative information. Basel Committee buffer guidance

This is a prudential policy connection, not a rule that central banks must raise interest rates whenever the gap is positive. The CCyB framework concerns bank resilience over the financial cycle. Basel guidance says the gap is a useful reference, that it may not work well in every jurisdiction or period, and that authorities should not rely on it mechanically. Basel Committee buffer guidance

If a national authority publishes a different gap from the BIS, the difference does not by itself show that one calculation is wrong. The data may use a different credit boundary or local input series. The BIS explicitly notes that its internationally comparable input can produce values different from those used in national CCyB decisions. BIS credit-gap overview

First check the units. A gap of +9 means nine percentage points of GDP above the trend when the series is expressed in percentage points of GDP; it is not a nine-percent rise in credit. Then check the reference period, release date, sector, credit sources, GDP denominator, trend method, and whether the chart shows a preliminary or revised series. A number without those labels can be easy to misread.

Next separate observation from interpretation. The measured ratio and estimated trend are data series; “excessive credit” or “systemic vulnerability” is an analytical interpretation. A gap can help policymakers examine a buildup of risk, but it is not a direct estimate of the amount of unsafe loans. The BIS presents the gap as a reduced-form proxy and notes that the definition and input data matter. BIS credit-gap overview BIS review of credit-gap limitations

Finally compare like with like. A BIS value may differ from a domestic authority's buffer guide because each uses its own input choices. Before comparing countries or vintages, verify that frequency, borrowers, lender coverage, consolidation, GDP window, and trend calculation match. For another supply-side macroeconomic estimate that depends on a constructed trend, see how economists estimate the output gap.

A positive gap does not identify the exact quarter a crisis will begin, the probability of failure for a particular bank, or whether a household's mortgage is affordable. Nor does a gap below zero prove the absence of risk. The measure is a broad indicator built from aggregate credit and GDP, not a loan-level assessment, a credit score, or a valuation threshold for property and other assets.

The trend is not an estimate of the “correct” amount of borrowing. A rapid financial deepening, a long credit boom, a break in the data, or a different starting point can affect what a statistical filter labels trend and cycle. Research and policy discussions examine the gap's warning properties and its limitations; the BIS review highlights endpoint, starting-point, and structural-break concerns and stresses that policymakers should use multiple indicators rather than a mechanical rule. BIS review of credit-gap limitations

Use the gap alongside measures that answer different questions: credit growth and borrower composition, debt-service ratios, lending standards, asset prices, bank capital, and the public debt ratio. For how household deposit pricing reflects policy-rate transmission rather than the volume of private credit, see why deposit rates do not move one-for-one with the policy rate. Each indicator has its own definition and limitations; combining them adds context but still does not produce certainty about future events.