The study counts articles from Slovenian web media from December 2023 to September 2026, 1,373,091 articles from the 170 outlets that published every month. An article is "about real estate" if its title says so. No text is read: everything below is counts of titles and first lines.
There are two questions. Does coverage follow the market? And do the media only relay official sources, or do they look into the market themselves?
How much they write
On average 4.1 in 1,000 articles are about real estate, 5,617 in all. August is the lowest month every year (2.4 against 4.2 in other months): the country is on holiday, and the media and the market with it.
A real-estate article: the title mentions real estate, flat prices, buying or renting a flat, housing policy, housing construction, loans or the tax (fires, break-ins and accidents excluded). A share within the 170 outlets that published in every month. No text is read.
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
What sets coverage off
The four highest weekly peaks are Croatia's property tax, GURS's trial value calculation, the government's tax proposal and GURS's new generalised values. None of them is a market event.
Peaks are labelled with the event that dominated that week's titles. The week of 13–21 September 2025 is missing: the collector was down.
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
What they write about
Most about the market and prices and about renting, least about loans: in the last twelve months 580 articles on the market and 433 on renting against 50 on loans. Pick a topic.
Topics overlap: 'affordable housing construction' counts under construction and under policy.
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
The market grows, coverage falls
The last twelve months against the first twelve: the share of real-estate articles −17%, the median flat price per m² +13%, sales +14%, new mortgages +67%.
The rate on new mortgages fell from 3.96% to 2.80%, yet loans are the subject of only 2.5% of real-estate articles. HICP rents rose 4% to March 2026 and then 11% in one month (the non-profit rent decree); there were 433 articles on renting in the last year and 453 in the first. Pick a series.
Price: the median €/m² of clean flat sales. Sales to June 2026, when the register is complete. Loans: ECB MIR, Slovenia. Below: the articles that concern the series (on loans for loans, on renting for rents).
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
Coverage does not follow the market
With the season removed and monthly changes compared, no market series tracks coverage. The closest are consumer confidence (r = 0.42) and the mortgage rate (0.26), both weak; with nine tests one or two such are expected by chance. Prices (−0.17) and sales (0.11) show nothing.
Month-on-month changes with the season removed (divided by the mean of the same calendar month). 95% interval, bootstrap with four-month blocks. Nine tests: one or two weak 'hits' are expected by chance.
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
Who writes
Finance and Cekin write about real estate 7.6 times as often as the average. Both specialise in business and personal finance.
All outlets with at least 3,000 articles, not only the steady panel. Finance and Cekin specialise in business and personal finance.
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
The property tax
The tax set off the most coverage. In December 2024 and January 2025 there were 228 articles about it, 44% of all real-estate articles in those two months.
A tax article: the title mentions a tax or taxation together with real estate or a flat.
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
Part two: relay or investigate?
Who the source is
Real-estate articles were classed by whom the title or lead names. 40% name the state (the government, a ministry, parliament, a law, a municipality, a fund), 7.7% the market (agencies, banks, investors), 6.2% official statistics. 1.8% carry a marker of the outlet's own work ("we checked", "we calculated", "we reveal"), 102 of 5,617 articles. For each of those there are 25 relaying the state or the statistics. Almost half (49%) name none of these.
The source is the one the title or lead names. Own work: the title or lead says 'checked', 'calculated', 'we reveal' and the like. It is a floor: an investigation that does not announce itself in its first lines is not counted. Classes overlap.
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
The source is the one the title or lead names. Own work: the title or lead says 'checked', 'calculated', 'we reveal' and the like. It is a floor: an investigation that does not announce itself in its first lines is not counted. Classes overlap.
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
On the day a statistic is released
On a day SURS or GURS publishes price data (18 such days), there are 5.5 times as many price articles as on an average day. Even without the articles that name the statistic, other price articles run 2.8 times the average. By the second day coverage is back to average (1.0 times), with no second wave in the two weeks after.
A release day: at least three price articles name SURS, GURS, BS or Eurostat in the title or lead. Dark: articles naming the statistic; light: other price articles.
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
Own work by outlet
No outlet with 80 or more real-estate articles marks its own work in more than 6.2% of them (zurnal24.si). STA 0.6%, Delo 1.3%, Finance 3.6%. The own work that exists is often a data piece on public statistics, not field reporting.
Outlets with at least 80 real-estate articles. Own work is a floor; paywalled and print investigations are under-represented in the feed.
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
The same title in several outlets
12% of articles share an identical title with another outlet's article, and 61% of those titles were also on STA. A rewritten wire story with a new title is not counted here, so this is a floor.
Only an identical title counts: a rewritten wire story with a new title counts as its own, so the share is a floor.
Source: a feed of Slovenian web media, ETN, ECB, Eurostat · bloki.si
The answer: mostly relaying
Real-estate coverage is set off by the state and GURS, not the market, and it arrives on the day a statistic is released and is gone two days later. Own work that announces itself in the first lines is rare. This is a floor, not a census: an investigation that does not say so in its title or lead is invisible to a count, so the claim is "rarely", not "never".
§ How it is calculated
A real-estate article is one whose title matches one of six topics (market and prices, renting, housing policy, the tax, housing construction, loans) or contains "nepremičnin" or "stanovanjsk". Fires, break-ins, accidents and floods at a family house and war news are excluded. "Stanovanj" alone is not enough: on its own it is mostly crime news.
A share, not a count. The feed loses outlets, so the share is taken within the 170 outlets that published in every month. The outlet charts use all outlets.
The market. Prices and sales from the real-estate transaction register (sales to June 2026, when the register is complete), loans and rates from the ECB, rents from the HICP. ETN rents are not used, because from 2025 only a sixth of new leases are filed.
Comparisons. Monthly changes divided by the mean of the same calendar month, with intervals from a bootstrap with four-month blocks.
Sources and own work are read from the title and lead. Own work is a floor. A release day is a day with at least three price articles naming official statistics.
What this cannot say. No text is read. The feed is a list of web sources, mostly front pages: paywalled and print investigations are under-represented. 34 months describe and do not prove. One week of September 2025 is missing because the collector was down; shares are unaffected. The market series are national, local stories local.
Can the Bloki server answer questions like these?
The server does not read the media. It answers about the market they write about, from the registers, the same day and for every municipality.
- Yes
What is happening on the market in my town right now?
Asking prices and rents today, how much is sold and let, and the price index, all with their n.
asking_pricesmarket_activityprice_index - Yes
How far are the new GURS values from prices?
The ratio of price to generalised value by year and municipality, with its dispersion, and one property's value.
valuation_baseget_property - Partly
Is the price in the news a 'record'?
It shows the distribution of registered sales in the area; it finds a single sale only if you know the property.
market_statsproperty_history - No
What do the media write about my neighbourhood?
The server neither holds nor reads the media. This article comes from a separate feed, counted without reading the text.