BoE UK rates — morning report

Report date: 2026-09-15
Parquet fetched: 2026-09-15T06:21:31.405712+00:00
OIS curve as of: 2026-09-11
Data cleaned: 2026-09-15T06:59:50.901512+00:00

Data pipeline workflow

End-to-end path from Bank of England original sources to this morning report.

Source → storage → report use

Extract step: generate_boe_report.py loads the four parquet files (no live BoE call), builds matplotlib charts, and writes Markdown/HTML. Use --full-refresh to re-fetch BoE data first.

What we built this session

This folder now holds a small Bank of England data pipeline for UK mortgage / OIS analysis:

Parquet store (data/boe/)

Data quality work

Early IADB values were stored with a flawed to_decimal() rule (values ≤ 1.5 not divided by 100).
clean_boe_parquet.py truncated bad history per series; audit now passes 0 fail / 0 warn.

Cutoffs applied: SONIA/Bank Rate from 2022-08-04; mortgages from 2021-11-30 (var from 2017-11-30); OIS 10y+ nulled before 2021-12-01.

Latest rates (one-page table)

Policy & mortgage rates

OIS par yields (BoE curve)

Retail margin (advertised − OIS)

BoE mortgage rates — what a householder pays

The Bank of England does not publish “what John Smith pays on his mortgage.” It publishes quoted (advertised) household interest rates — monthly, weighted averages of headline rates that UK lenders advertise for standard products.

Official methodology: Further details about quoted household interest rates. Key points:

• Product data is sourced from Moneyfacts (the same product universe used by many comparison sites).
• 75% LTV, owner-occupier, loans roughly £25,000–£500,000.
• The lowest qualifying headline rate for each product type (fees/cashback help select the product but are not included in the rate).
• Weighted using FCA product sales data — reflects market mix, not necessarily the cheapest deal on a comparison site.
• This is not APR (fees are excluded from the quoted rate).

Advertised rates in our parquet (latest)

10-year fixed — the BoE benchmark rate

Series: IUMBV45 → mtg_10y in mortgage_quotes.parquet
Latest: 5.250% as of 2026-08-31

This is the official BoE answer for: “typical advertised 10-year fixed at 75% LTV.” A householder who matches that profile (75% LTV, standard criteria, near the weighted-average product) might expect a headline rate around this level — but individual quotes vary.

Decomposition (month-end, from mortgage_spreads.parquet):

Latest OIS curve (2026-09-11): 10y par = 4.936% — useful for repricing if mortgage quote is stale.

Illustrative payment: £300,000 loan, 25-year repayment, 5.250% fixed → ≈ £1,798/month (excludes fees, ERC, insurance; not APR).

Recent 10y advertised trend:

BoE database: IUMBV45 — 10y fixed 75% LTV

15-year fixed — not published by BoE

The BoE does not publish a 15-year mortgage quoted rate. Fixed quoted series in IADB run 2y, 3y, 5y, and 10y only (we fetch 2y/5y/10y; 3y is IUMBV37 and could be added). See BoE mortgage series search.

15-year fixes exist in the market but are uncommon. We can only estimate from swap curve + margin proxy:

These ~5.6–5.7% figures are indicative only — for a real 15y quote use a broker or comparison site.

BoE average vs what you might get elsewhere

Compare like with like: same LTV, fixation period, and whether you want average (BoE) or best-buy (brokers).

Typical pattern:

• BoE 10y (5.250%, 2026-08-31) — market average at 75% LTV.
• Comparison site “from” rates — often lower for strong profiles on promotional fixes (check same LTV and term).
• 90% / 95% LTV — typically +30–80 bp above the 75% benchmark.

Use BoE for: historical trends, macro analysis, and retail-margin vs OIS (our spread charts).
Use brokers / comparison sites for: “what could I get today?” before applying.

Charts (Python / matplotlib)

Reading the spread chart

• Solid lines: month-end retail margin (advertised fixed rate minus matching-tenor OIS par). Labels at right show latest level in bp.
• Dashed lines: 36-month rolling median for each tenor.
• Lower panel: 12-month change — red = margin widening (lenders charging more over swap), green = tightening.

Regenerating with fresh data

Run from the project root (finance/):

# 1. Fetch latest BoE curves and IADB quotes into parquet
python3 fetch_boe_yield_curves.py

# 2. (If audit shows IADB errors) clean bad history and rebuild spreads
python3 clean_boe_parquet.py

# 3. Validate
python3 audit_boe_parquet.py

# 4. One-page rate snapshot (HTML + MD)
python3 export_boe_latest.py

# 5. This morning report (MD + charts)
python3 generate_boe_report.py

# Optional: push into Excel workbook / YTD export
python3 refresh_uk_rates.py
python3 export_boe_ytd.py

One-liner (fetch + audit + snapshot + report):

python3 generate_boe_report.py --full-refresh

Outputs from this script:

Audit history is archived separately under data/boe/audit_history/<timestamp>/.

Notes

• BoE yN columns are spot-starting par OIS from the published spot curve — not forward-curve points. When comparing to BlueGamma or Chatham, match spot par swap rows and the same as-of date.
• Mortgage quotes are monthly; OIS and policy rates can be daily.
• Retail margin uses BoE advertised 75% LTV series (IUMBV34/42/45), not live lender menus.

Generated by generate_boe_report.py at 2026-09-15 08:24 UTC.