What is central bank credibility — and how do you actually measure it?

The definitional guide to sovereign credibility scoring, and the seven dimensions behind it.

TL;DR

Central bank credibility is the degree to which a central bank's stated commitments — above all its inflation target — are borne out by its actual policy, communication, and outcomes.
It is not a survey, a vibe, or a credit rating. It is a behavioral property you can measure from public data.
Quantamentry scores it as a 0–100 composite across seven dimensions, daily, point-in-time, for 169 countries.
The result often contradicts reputation: the Czech Republic (68.7) tops our gold tier and Brazil (59.8) ranks ahead of the United States and Germany, while Russia (30.8) and Turkey (36.4) sit near the bottom.
Credibility measures behavior and outcomes. A credit rating measures default risk. They are not the same thing, and they often disagree.

What is central bank credibility?

Central bank credibility is the gap — or absence of a gap — between what a central bank says it will do and what it actually does. A credible central bank has a stated objective (usually an inflation target), and its policy rate, its communication, and its realized inflation all line up behind that objective. An incredible one says one thing and delivers another: a 5% target with 30% inflation, or a "data-dependent" stance with a policy rate nowhere near where the data points.

Credibility matters because monetary policy works largely through expectations. If households, firms, and markets believe the central bank will return inflation to target, they price and bargain as if it already has — which makes the target self-fulfilling and cheap to hold. When belief breaks, every basis point of disinflation has to be paid for in lost output. Credibility is the cheapest asset a central bank owns, and the most expensive to rebuild.

How do you measure central bank credibility?

You measure it the way you'd audit any commitment: by checking the promise against the record, across several independent angles, so no single noisy signal dominates. Quantamentry decomposes credibility into seven dimensions, each scored 0–100, then combines them into a weighted composite (smoothed in-pipeline as 0.7 × raw + 0.3 × trailing-90-day average). Everything runs on free public data — World Bank, IMF, BIS, FRED, OECD, ILO, GDELT, and the European Central Bank reference feed.

Here are the seven, with a real example from our June 29, 2026 snapshot for each.

1. Credibility gap (weight 0.22)

The distance between actual CPI and the central bank's stated inflation target, scored with an asymmetric LINEX loss — overshoots and undershoots are penalized differently, with persistent overshooters hit harder. Countries with no formal target are scored against a normative 3.0% anchor. This is the single heaviest dimension because it is the most direct test of the promise. The Czech Republic, with CPI sitting on its 2% target, anchors the top of our gold tier at a composite of 68.7.

2. Behind-the-curve (weight 0.18)

A Taylor-rule deviation: given inflation and the output gap, is the policy rate where it should be, with r* anchored to trend growth? A bank that is structurally too loose for its own conditions is "behind the curve." On June 29, 2026, Russia scored just 10.0 here — the most behind-the-curve reading in the panel, despite a 21% policy rate, because ~8% inflation and the rule still call for more. (We break this dimension out in full in the most behind-the-curve post.)

3. Communication stance (weight 0.10)

Every English-language central bank statement is scored hawk/dove by FinBERT (CentralBankRoBERTa) and reconciled with an LLM read (60% model / 40% LLM). This catches the banks that talk tighter or looser than they act. Brazil's consistently hawkish statements are part of why it scores so well overall.

4. Geopolitical pressure (weight 0.13)

GDELT global event flow, scored for country-specific impact (mean of event scores through a sigmoid). Sanctions, conflict, and political ruptures all raise the pressure on a regime's ability to keep its commitments. This is where sustained external shock shows up before it reaches the inflation print.

5. Growth (weight 0.13)

Real-economy momentum: OECD CLI, business confidence, and the change in unemployment for gold-tier countries; GDP percentile and WEO deviation for the silver tier. A central bank fighting inflation into a collapsing economy faces a credibility test a booming one never sees.

6. Liquidity / financial vulnerability (weight 0.13)

Credit-to-GDP gap, debt-service ratio, and the real policy rate versus a per-country neutral rate (gold tier); REER, FX reserves, and external debt (silver tier); plus a sovereign-stress overlay combining fiscal vulnerability and the local-minus-US 10-year spread. This is the dimension that asks whether the financial plumbing can survive the policy the bank says it will run.

7. Governance (weight 0.11)

The World Bank's six Worldwide Governance Indicators — voice and accountability, political stability, government effectiveness, regulatory quality, rule of law, and control of corruption — each mapped from their −2.5..2.5 scale to 0–100 and averaged. Institutions are the substrate credibility grows in; a technically sound central bank inside a captured state is on borrowed time.

Add them up with the weights above and you get a single number per country per day. Brazil lands at 59.8 — ahead of the United States (58.7) and Germany (58.0), though still behind G10 names like Switzerland (65.3) and Japan (60.3) — because it scores on actions, communication, and outcomes rather than on the "EM political risk" discount most frameworks bake in. At the other end, Turkey sits at 36.4 and Russia at 30.8: high nominal rates that still fail the credibility-gap and behind-the-curve tests, dragged further by geopolitics and governance.

Credibility vs. reputation vs. credit ratings

These three get conflated constantly, and the differences are the whole point.

Reputation is the market's memory — the Bundesbank halo, the "EM is risky" reflex. It is sticky, lagging, and frequently wrong about the present. Our model is built to disagree with it when the data does.
Credit ratings measure probability of default — will the sovereign pay its bondholders? That is a fiscal-solvency question, not a monetary-commitment one. A country can be a reliable debtor with a chaotic central bank, or run a pristine inflation regime while its public finances deteriorate.
Credibility, as we score it, is behavioral: does the central bank do what it says, and does inflation behave as a result? It updates daily, off measurable inputs, with no committee vote and no rating-agency relationship to manage.

The cleanest tell is that the rankings diverge. Brazil ahead of the United States and Germany on credibility would be unthinkable on a ratings ladder. We've got a dedicated post coming that puts our scores next to the major agencies' ratings, country by country — it's the most direct way to see what each one is actually measuring.

What credibility scoring is not

Being precise about the boundaries is part of being a credibility product.

It is not a credit rating or a default forecast. We say nothing about whether a sovereign will repay its debt. If you need default risk, read the agencies.
It is not a buy/sell signal. A high score is not "go long the currency." It is one input — a measure of monetary commitment — that you combine with your own valuation and positioning view.
It is not a judgment of the people. A low score reflects the gap between stated policy and outcomes under the constraints a bank faces, not the competence of any individual. Germany's mid-pack reading, for instance, reflects inheriting a common ECB policy tighter than its own output gap warrants — a structural tension, not a central-bank error.
It is not equally deep everywhere. Every score carries a data-tier flag: gold (27 countries) with the full monthly stack, silver (53) on monthly CPI plus IMF WEO, and bronze (89) frontier markets on annual-only data. We never let "the model says X" hide how thin the inputs are.
It is not a black box. The dimensions, weights, data sources, and fallback logic are all documented. You can see exactly why a country scores what it scores.

What this is, in one line

A daily, point-in-time, 0–100 measure of how well a central bank's actions and outcomes match its stated commitments — built from public data, broken into seven auditable dimensions, and honest about its own coverage. If you want the full picture of how each dimension is computed, the methodology page has the equations; the coverage page shows exactly which countries sit in which tier.

What's coming next on Quantamentry

The ratings comparison: our credibility scores side by side with the major agencies, and where they disagree most.
Deep dives by dimension: starting with the most behind-the-curve central banks and what their Taylor-rule gaps mean for FX.
The full April rankings: we scored 169 central banks and the G7 didn't make the top 10.

If you want the daily snapshot or early access — [subscribe to Quantamentry].

— Quantamentry, June 30, 2026