How to compare policy rates across countries

Do it with the API

Policy rates are live on every plan, including Free. Each request returns one country's latest source observation at or before today UTC. The comparison uses USA, CAN, AUS, SWE, NOR, BRA, CZE, MYS and DEU. DEU represents the ECB deposit facility rate.

1. Set your API key Set the NARWHAL_API_KEY environment variable in your shell with export NARWHAL_API_KEY="your_api_key". Install the Python dependencies with python -m pip install requests pandas. The script reads the key from the environment and sends it in the Authorization header.
2. Request each country's latest rate Use one requests.Session and loop over the uppercase country codes. Call /v1/economics/{country}/central-bank/policy-rate once for each code. Keep the response fields together so each rate retains its currency, instrument and date.
3. Calculate one comparison value For a point response, read rate. For a range response, calculate (lower_bound + upper_bound) / 2 with decimal arithmetic. This midpoint is your own calculation: Narwhal API returns the published bounds and never computes a midpoint. Values are decimal strings in percent per annum.
4. Build and sort the table Create a pandas DataFrame with country, currency, instrument, rate and date. Take the date from effective_on or observed_on; each response has exactly one. Sort by the calculated rate. Keeping Decimal values preserves the precision of the supplied strings.
5. Run the complete script Save this as compare_rates.py and run python compare_rates.py after public access opens. It prints rates from lowest to highest. HTTP errors stop the script through raise_for_status(), so an unavailable country does not silently disappear from the comparison.
   Fetch and compare policy rates

   Copy

   import os
   from decimal import Decimal
   
   import pandas as pd
   import requests
   
   api_key = os.environ["NARWHAL_API_KEY"]
   base_url = "https://api.narwhalapi.com"
   countries = ["USA", "CAN", "AUS", "SWE", "NOR", "BRA", "CZE", "MYS", "DEU"]
   rows = []
   
   with requests.Session() as session:
       session.headers.update({"Authorization": f"Bearer {api_key}"})
   
       for country in countries:
           response = session.get(
               f"{base_url}/v1/economics/{country}/central-bank/policy-rate",
               timeout=10,
           )
           response.raise_for_status()
           observation = response.json()
   
           if observation["rate_type"] == "point":
               rate = Decimal(observation["rate"])
           elif observation["rate_type"] == "range":
               rate = (
                   Decimal(observation["lower_bound"])
                   + Decimal(observation["upper_bound"])
               ) / Decimal("2")
           else:
               raise ValueError(f"Unexpected rate_type for {country}")
   
           date = (
               observation["effective_on"]
               if "effective_on" in observation
               else observation["observed_on"]
           )
           rows.append({
               "country": observation["country"],
               "currency": observation["currency"],
               "instrument": observation["instrument"],
               "rate": rate,
               "date": date,
           })
   
   comparison = pd.DataFrame(
       rows,
       columns=["country", "currency", "instrument", "rate", "date"],
   ).sort_values("rate")
   
   print(comparison.to_string(index=False))
   

What does each country’s rate measure?

These rates are comparable only loosely. An overnight target, a repo rate and a deposit rate describe different instruments. Each also belongs to a currency with its own inflation rate. A numerical ranking alone does not establish which country has tighter monetary policy.

The saved US example has a federal funds target range of 3.50 to 3.75 percent, effective 11 December 2025. Its calculated midpoint is (3.50 + 3.75) / 2 = 3.625 percent. That figure is a comparison value derived from the example, not a published point target or a claim about today's rate.

Preserve the original bounds if you retain the comparison. A single midpoint removes the width of the target range from the printed table. For Brazil, use the Selic target returned for BRA.

The source notices for these observations are:

• Source: Board of Governors of the Federal Reserve System
• Source: Bank of Canada. This information is available free of charge on the Bank of Canada website: https://www.bankofcanada.ca. Narwhal normalised the format.
• Source: Reserve Bank of Australia 2026. This information is available free of charge on the RBA website: https://www.rba.gov.au. No endorsement is implied.
• Source: Sveriges Riksbank, with the date of the data.
• Source: Norges Bank
• Source: Banco Central do Brasil (SGS 432), Open Data Commons Open Database License (ODbL).
• Source: the CNB
• Source: Bank Negara Malaysia. Data and information are subject to Terms of Use for BNM Datasets. https://www.bnm.gov.my/terms-of-use-bnm-datasets
• Source: European Central Bank (ECB). This information is available free of charge on the ECB website: https://www.ecb.europa.eu. Narwhal's modifications are to the format only.

How to compare real rates

A simple approximate real rate subtracts inflation from the nominal policy rate: real rate ≈ policy rate − inflation. Use percentages on both sides of the subtraction. For a target range, identify the nominal input as your calculated midpoint.

For countries that have an inflation route, request /v1/economics/{country}/inflation and read year_over_year. This includes USA, CAN, AUS and DEU. A 404 means no data; leave that country's real-rate result absent rather than substituting zero inflation.

Keep the inflation period beside the policy-rate date, since the observations can refer to different times. The real interest rate calculation explains the subtraction. Inflation adjustment still leaves differences between the underlying policy instruments.

Why is the euro area one row?

Any supported euro member code returns the same ECB deposit facility rate. Using DEU once avoids repeating the same instrument across the comparison. For historical requests, coverage starts at each member's euro adoption date, so the available history can differ between member codes.

Why does the date differ between countries?

effective_on is the date the rate took effect. observed_on is the date the source recorded the value. Sources also publish different observation patterns, so latest responses need not share a date. Future effective dates never replace the current rate early.

For a historical comparison, add the same date=YYYY-MM-DD to every request. Each returns the latest source observation at or before that date. Days without source observations are not invented, so retain each returned date.