The property market, in the institutions' own figures

What it costs

How much transaction prices moved in a year.

Eurostat · Q2 2026 this exact figure ↗ · NSI · Q2 2026

The chart below measures something else: how many times higher prices are than one year taken as the yardstick. That year is 2015, and on the chart it is the ×1 line. It is the yardstick rather than the start of the record, which is why there are points below ×1 to the left of it: homes cost less then.

The two lines measure the same thing two ways. The solid one counts money: how many times more of it changes hands for a home. But money buys less of everything than it did; the dashed one takes exactly that out and answers the other question: have homes got dearer than everything else we buy.

The money paid for a home today is ×2.8 what it was in 2015. But everything else got dearer too, and measured against that a home is ×1.6 what it was. On that measure today's level is 3.4% below the highest Eurostat have recorded, in Q3 2008.

Eurostat · Q1 2005 – Q2 2026 this exact figure ↗ · Eurostat · Q1 2005 – Q2 2026 this exact figure ↗

Eurostat · Q1 2005 – Q2 2026 this exact figure ↗

A shorter column above the zero line is a smaller rise, not a fall: prices fell only in the quarters whose column is below it.

Eurostat · Q1 2006 – Q2 2026 this exact figure ↗

The average deal

"Total paid" is the price of the dwellings themselves together with the land under them, VAT included on new builds; the notary, the agency commission and the bank fees sit outside it, so the average deal is what a home costs rather than what buying one costs.

Eurostat · Q2 2026 this exact figure ↗

A dwelling changed hands for 84,139 € on average in the quarter: 110,306 € for a new build and 72,366 € for an existing dwelling. The two lines below are drawn on one scale.

Eurostat · Q1 2017 – Q2 2026 this exact figure ↗

The average deal is our arithmetic: the total paid divided by the number of deals, €1,732,668,652 over 20,593 dwellings. It is a mean amount paid for a dwelling, not a price per square metre and not the middle price of the quarter's deals; the mix of flats and houses sold moves it. Eurostat are not responsible for the division or for conclusions drawn from it. check 1 check 2  How, and why →

How much changes hands

Eurostat count the dwellings households bought during the quarter, flats and houses, bought at a market price. Gifts, inheritances, discounted sales between relatives and anything built for oneself stay outside the count.

Eurostat · Q2 2026 this exact figure ↗

The columns alternate high and low every year because winter and summer are not traded alike.

Eurostat · Q1 2017 – Q2 2026 this exact figure ↗

The year-on-year change is our arithmetic: this quarter's count against the same quarter a year earlier. Like quarters are compared rather than neighbouring ones: a summer-to-winter fall measures the calendar rather than the market. check  How, and why →

The series' first point is the start of a record rather than a floor in the market: there were sales before it, just not in this table. Nor is this a count of all property sales: the land register also records land, garages, shops and offices, and so counts considerably more for the same quarter. The two measure different things and neither is wrong.

The two charts below are for the same quarters, and the bottom one is the same series drawn in full under what it costs, here only over the quarters that also have a count. Each keeps its own scale, so the two move by very different amounts.

Eurostat · Q1 2018 – Q2 2026 this exact figure ↗ · Eurostat · Q1 2018 – Q2 2026 this exact figure ↗

The top series is our arithmetic from the counts: each quarter against the same quarter a year earlier. The bottom one is the figure Eurostat publish and is not worked out here. The two measure different things and are drawn together so they can be seen together; nothing on this page claims that either one moves the other. check  How, and why →

Prices and sales by city

NSI publish the same movement for the six cities over 120,000 people, and beside it how much the number of sales there changed.

The price changes are NSI's, from the house price index by city; the sales counts come from dwelling sales by city. Every value is a cell NSI published; nothing in this table is computed by us.

An asking price and a paid price are different things. Asking prices per m² by city come from imot.bg and are in the calculator and in the section below; the figures here are transaction figures and come from Eurostat. Nobody publishes a transaction price per square metre for an individual Bulgarian city, which is why there is no such table here.

How many years of pay a home costs, city by city

The longest is not Sofia but Varna: 11.0 years of pay against 9.7 in the capital. Another 5 cities take longer than Sofia does.

"11.0 years" means: that many years of the whole wage, down to the last euro, going to the home. We take what a 70 m² home costs at the city's average asking price and divide by the average wage for the oblast, after contributions and tax. The figure is not a budget, it is a yardstick for comparing cities.

Since 2020 a home has grown dearer against pay in 21 of 25 cities. Most of all: Gabrovo +45.7%, Vratsa +43.1%, Kyustendil +40.4%. In Sofia the change is +7.7%, and in half of them it is above +9.4%.

Our arithmetic over two published figures. The prices are the average asking prices per m² by district from imot.bg, and the wages are the average gross wage by oblast as NSI publish it for Q2 2026 and for the same quarter of every earlier year. Neither of the two published figures is altered.

An asking price and a paid price are different things: these are the asking ones. The wage is the average for the whole oblast rather than for the city itself. Between the two years the figures do not move in a straight line: every year is in the table above.

Silistra, Sofia oblast, and Smolyan are not in the table: imot.bg publish no price for 2020 to compare today's with.

Who borrows

The figures are on one and the same base: owners and renters make a hundred.

Eurostat · 2025 this exact figure ↗

That row looks impossibly small and it is not wrong. It counts people rather than purchases: how many of those living in the country have a loan on their home, children and pensioners included. New lending is a different thing, a flow of how many contracts are signed in a year, and the two can move in opposite directions for years.

For the people in the last three rows the price of housing is the rent. It is measured every month rather than every quarter: for August 2026 rents are +7.1% against the same month a year earlier. That is the price of renting rather than the price of a home, and it covers every rent actually paid in the country.

Eurostat · August 2026 this exact figure ↗

How much of the money is borrowed

It is measured two ways and the two do not give one figure. One line is the growth in the banks' housing loans: what was lent less what was repaid. The other is only what was lent on new home loans, with the repriced old ones taken out. The true figure is between the two.

Below zero the banks' housing loans shrank over the year: more was repaid than was lent.

what part of the money paid for dwellings comes from a bank loan

Eurostat · 2015 – 2025 this exact figure ↗ · BNB · 2015 – 2025 · ECB MIR · 2020 – 2025

Both lines are our arithmetic: the lending is the BNB's and the ECB's, and what was paid for dwellings, which we divide it by, is Eurostat's. The ECB amounts before January 2026 are in leva and are converted at the official rate before the division. Eurostat are not responsible for the division or for conclusions drawn from it. check 1 check 2 check 3 check 4  How, and why →

A housing loan is also taken out for a renovation, for a plot or to build a house of one's own, and Eurostat count only the dwellings that were bought.

What part of those loans carry a floating rate is on the borrowing page.

What the census counted

The other figures on this page are measured every quarter or every year and are replaced by the next edition. This is a count: taken once, and not updated until the next census, the last of which was 2021.

"Unoccupied" means the dwelling was nobody's permanent home at the time of the census. Who slept there is not the test, so holiday homes, second homes and the homes of people abroad are all inside it, not only the genuinely empty ones.

Eurostat · 2021 this exact figure ↗

The share is our arithmetic: unoccupied dwellings over all the dwellings from the same census: 1,657,674 over 4,258,585. Both counts are in the table above. check  How, and why →

Expensive against incomes?

The figure below does not measure what one home costs: it counts the people who spend more than 40% of their household's disposable income on housing.

The figure divides one thing by another: what housing costs, and the household's disposable income. A bar falls when housing takes a smaller share of that income, and that happens while the bills themselves are going up, so long as income goes up faster than they do.

"Housing costs" here is everything around the home: electricity, heating, water, maintenance, insurance and tax; rent for tenants, and for owners with a loan the interest alone, never the capital. The percentage is a share of people rather than of households: it counts everyone living in a household over the line, out of the people in private households, and for 2025 that is 6.9%. Most people here live in a home they own outright, so this series moves with those bills rather than with transaction prices.

share of people spending over 40% of their disposable income on housing

Eurostat · 2006 – 2025 this exact figure ↗