France New Housing Loans: Monthly Mortgage Production Since 2003

This dataset tracks the monthly flow of new housing loans granted to French households, seasonally adjusted and expressed in billions of euros, as published by the Banque de France in its MIR1 statistics. It runs from January 2003 across 280 observations, from a low of EUR 4.2 billion in January 2009 to a record EUR 38.5 billion in January 2017. That record deserves an immediate caveat, because 61.8% of it was renegotiation of existing loans rather than new borrowing, a point the section on data quality develops.

Dataset: France New Housing Loans (2003–2026) · Updated 2026-07-01

Source: Banque de France, Webstat MIR1 · new housing loans to households, seasonally adjusted flows

Macro Takeaway

Dataset Overview

Dataset Variables

The CSV, Excel and JSON files contain the following columns.

Column names match the CSV headers exactly.

Download the Complete Dataset

The full France new housing loans dataset is available in CSV and Excel formats.

Direct CSV Access – Eco3min Structured Dataset

https://eco3min.fr/dataset/fr/fr-new-housing-loans.csv

This URL returns the complete dataset in CSV format. It can be used directly in pandas, R, curl, or any data tool. The upstream series is published by the Banque de France through Webstat, in the MIR1 dataset.

Using the Dataset in Python

import pandas as pd

url = "https://eco3min.fr/dataset/fr/fr-new-housing-loans.csv"
df = pd.read_csv(url, parse_dates=["date"])

print(df.tail())
print(df.set_index("date")["new_housing_loans"].resample("YE").mean())

Using the Dataset in R

library(readr)

url <- "https://eco3min.fr/dataset/fr/fr-new-housing-loans.csv"
df <- read_csv(url)

tail(df)
summary(df$new_housing_loans)

Both examples load the dataset directly from the URL, with no download or API key required.

Methodology

The Banque de France collects new-business volumes from a sample of French credit institutions in its MIR1 statistics, the national contribution to the harmonised euro-area survey of monetary financial institution interest rates and volumes. This line reports the amount of new housing loans agreed with households during the reference month, corrected for seasonal variation, which matters on a series where January and the summer months would otherwise be systematically distorted.

The Eco3min pipeline pulls the series through DBnomics, which mirrors Banque de France Webstat, and writes the CSV, Excel and JSON files, rebuilding the full history on each run.

Data Quality & Provider Notes

Gross production includes renegotiation, and the difference can dominate. This is the single most important caveat on the series. In January 2017 the published figure of EUR 38.5 billion contained roughly EUR 23.8 billion of renegotiated loans. Across 2015, 2016 and 2017 the renegotiation share averaged 50.0%, 42.6% and 36.3% respectively, which means the whole 2015 to 2017 plateau in this series overstates purchase activity by a wide and time-varying margin. Any reading of this dataset over that period that does not pair it with the renegotiation share will be wrong about the housing market.

Seasonal adjustment is applied at source. The published series is already corrected, so applying a further seasonal filter double-counts. The correction is also revised periodically as the seasonal factors are re-estimated, which changes history without any change in the underlying lending.

A flow, not a stock. Summing twelve months of this series does not give the change in outstanding loans, because the stock also falls through amortisation and early repayment. The two series answer different questions and are published separately for that reason.

Latency and revisions. MIR1 publishes with roughly a two-month lag. The pipeline rebuilds the whole file on each run, so revisions replace history rather than accumulating beside it.

Licence. Banque de France Webstat data is published under the Licence Ouverte 2.0 (Etalab), which permits reuse including commercial reuse, subject to attribution of the source.

Historical Regimes

2003–2008 – The pre-crisis build. Annual averages rise from EUR 6.4 billion to a 2006 peak of EUR 11.8 billion, then ease as the credit cycle turns. This is a market roughly half the size of the one that follows.

2009 – The floor. Production collapses to EUR 4.2 billion in January 2009, the lowest month in the entire file, with a 2009 average of EUR 6.8 billion. The financial crisis hit French mortgage production harder in a single month than anything since.

2015–2017 – The renegotiation plateau. Annual averages jump to EUR 17.8 billion, EUR 21.0 billion and EUR 22.4 billion, with the all-time high of EUR 38.5 billion in January 2017 and two further months above EUR 30 billion. As the data-quality section sets out, renegotiation accounts for most of the increase over these three years.

2019–2022 – Genuine expansion. Averages of EUR 20.6, 20.9, 23.0 and 21.6 billion, this time with renegotiation shares between 15% and 23%. The 2021 average of EUR 23.0 billion is the strongest year in the file on a purchase-adjusted basis.

2023–2026 – The rate shock and a partial recovery. Production halves to a 2023 average of EUR 12.8 billion and bottoms at EUR 10.1 billion in December of that year, then averages EUR 11.1 billion in 2024. Recovery follows, to EUR 14.3 billion in 2025 and EUR 13.8 billion so far in 2026, with the April 2026 reading at EUR 13.9 billion, close to the series median of EUR 12.1 billion.

Related Macroeconomic Datasets

A gross production series is only interpretable alongside the refinancing share and the stock it feeds.

Macroeconomic Dataset Hub

This dataset is part of the Eco3min macro-financial data repository.

Explore the Eco3min Dataset Hub

Sources

• Banque de France – Webstat, MIR1 statistics, new housing loans to households, seasonally adjusted flows, series M.FR.B.A22.A.5.A.2254U6.EUR.N
• Licence Ouverte 2.0 (Etalab), reuse permitted with attribution to the source
• Eco3min Research – structured dataset compilation

Dataset Reference

Eco3min Research. (2026). France New Housing Loans, Monthly Production (2003–2026).<br> https://eco3min.fr/en/france-new-housing-loans-dataset/