"""fed-funds-vs-cpi-real-rates.py — the effective fed funds rate minus CPI inflation, 1955 to the latest month.

Input : datasets/fed-funds-vs-cpi-real-rates.csv (FRED FEDFUNDS monthly %, CPIAUCSL index 1982-84=100),
        retrieved 2026-09-05 via the public fredgraph CSV. Re-download with --download.
Method: cpi_yoy = 100 * (CPI_t / CPI_{t-12} - 1); real = FEDFUNDS - cpi_yoy. This is an ex-post real
        rate (realized inflation), not a rate against expected inflation. Tables by decade, the
        longest negative stretches, and the latest month.
Run   : python code/fed-funds-vs-cpi-real-rates.py [--download]
Needs : Python 3.13, pandas 3.0.2.
"""
import io
import sys
import urllib.request

import pandas as pd

CSV = "datasets/fed-funds-vs-cpi-real-rates.csv"


def download():
    frames = []
    for sid in ("FEDFUNDS", "CPIAUCSL"):
        req = urllib.request.Request(f"https://fred.stlouisfed.org/graph/fredgraph.csv?id={sid}",
                                     headers={"User-Agent": "prism-data-lab/1.0"})
        txt = urllib.request.urlopen(req, timeout=60).read().decode()
        frames.append(pd.read_csv(io.StringIO(txt), na_values=".", index_col=0).iloc[:, 0].rename(sid))
    df = pd.concat(frames, axis=1)
    df.index.name = "date"
    df[df.index >= "1954-07-01"].to_csv(CSV)


def main():
    if "--download" in sys.argv:
        download()
    df = pd.read_csv(CSV, parse_dates=["date"], index_col="date")
    df["cpi_yoy"] = 100 * (df["CPIAUCSL"] / df["CPIAUCSL"].shift(12) - 1)
    df["real"] = df["FEDFUNDS"] - df["cpi_yoy"]
    d = df.dropna(subset=["real"]).copy()
    d["decade"] = (d.index.year // 10) * 10
    print("| decade | months | mean real rate (pp) | min (month) | max (month) | months below zero |")
    print("|---|---|---|---|---|---|")
    for dec, s in d.groupby("decade")["real"]:
        print(f"| {dec}s | {len(s)} | {s.mean():.2f} | {s.min():.2f} ({s.idxmin():%Y-%m}) | "
              f"{s.max():.2f} ({s.idxmax():%Y-%m}) | {int((s < 0).sum())} |")
    neg = (d["real"] < 0)
    grp = (neg != neg.shift()).cumsum()
    runs = []
    for _, x in d[neg].groupby(grp[neg]):
        runs.append((x.index[0], x.index[-1], len(x)))
    runs = sorted(runs, key=lambda r: -r[2])[:5]
    print("\nlongest stretches of a negative real rate:")
    for s, e, n in runs:
        print(f"  {s:%Y-%m} to {e:%Y-%m}: {n} months")
    last = d.iloc[-1]
    print(f"\nlatest month with both series: {d.index[-1]:%Y-%m}: FEDFUNDS {last['FEDFUNDS']:.2f}, "
          f"CPI y/y {last['cpi_yoy']:.2f}, real {last['real']:.2f}")
    print(f"whole sample {d.index[0]:%Y-%m} to {d.index[-1]:%Y-%m}: mean real {d['real'].mean():.2f}, "
          f"median {d['real'].median():.2f}, share of months below zero {100 * (d['real'] < 0).mean():.1f}%")
    peak = d["cpi_yoy"].idxmax()
    print(f"peak CPI y/y in sample: {d.loc[peak, 'cpi_yoy']:.2f} in {peak:%Y-%m}; "
          f"2020s peak: {d.loc['2020':, 'cpi_yoy'].max():.2f} in {d.loc['2020':, 'cpi_yoy'].idxmax():%Y-%m}")
    lo = d.loc["2020":, "real"]
    print(f"2020s real-rate low: {lo.min():.2f} in {lo.idxmin():%Y-%m}; first positive month after it: "
          f"{lo[lo.index > lo.idxmin()][lo[lo.index > lo.idxmin()] > 0].index[0]:%Y-%m}")


if __name__ == "__main__":
    main()