How free is the market

ETN tags every sale with a transaction type: open market, voluntary auction, forced sale (enforcement, bankruptcy), sale between family members or related parties, and leasing. The study reads those tags instead of guessing non-market deals from the price.

93.5% of contracts are open-market sales; forced sales are 2.3%. ETN holds not one sale in lieu of expropriation or expropriation (0): the state does not file those purchases as sales. Farmland is the most open-market segment at 96.4%; its pre-emption order does not show, because a pre-empted sale is still a sale. Commercial space is the exception at 81.8% open market: leasing carries 6.7% and auctions 4.9%. Pick a segment in the chart.

Forced sales were 0.4% of contracts in 2009, peaked in 2015 at 4.3%, and have held at about 2% since 2019.

Two jumps in the chart are not the market. Related-party sales fall from 1.0% in 2023 to 0.1% in 2024, in every segment at once. Leasing is 4.9% and 7.7% of contracts in 2013 and 2014 and almost nil before and after. Both are changes in how deals are filed.

If a tag is right, a forced or family sale should fetch less than the open market. The price is set against the GURS generalised value, which holds size and location roughly constant, and each type's median against the open-market median in the same segment.

A forced house sale fetches 47% of the open-market price, a voluntary auction 42% and a family sale 33%. For flats: 65%, 68% and 52%. The tags separate deals that really are not market deals.

From 2013 to 2024, 6.4% of filed dwelling leases carry a rent set by law, at a median €3.30/m² a month against €4.98 on the open market. All non-market types together are 10.8%.

That is a share of filings, not of rented dwellings. If private landlords file less often than public ones, it overstates. After 2024 filings fall fivefold, so the chart stops there.

A gain on selling real estate is taxed at 25% in the first five years, 20% to the tenth, 15% to the fifteenth and 0 after fifteen years (since December 2022; before that after twenty). If sellers wait for the lower rate, more should sell in the anniversary month than in the months around it.

They do. Among 40,897 consecutive open-market sales of the same flat or house, the month of the tenth anniversary holds 191 resales and the month before it 88. The excess is 1.16× at five years, 1.43× at ten and 2.13× at fifteen, against about 1.04× at other anniversaries. The tax decides when sales happen. The excess at four years (1.33×) is unexplained.

The other tax mark, developers designing new flats just under the reduced-VAT threshold, is shown in Who the market suits.

What is it worth to be allowed to build on a plot? For every farmland sale the comparison is the median price of building land sold within 1 km and within two years either side.

Building land sells for 7.5 times the farmland beside it (n = 25,103, interquartile range 3.0–16.1×), a median gap of €16.8/m². Against forest the ratio is 18.1× (n = 4,711). It is flat over time: 6.6× in 2007–10, 7.0× in 2023–26.

The ratio is a lower bound: farmland next to a settlement already carries the hope of being built on one day, and it is dearer where dwellings are dear. So it peaks in Ljubljana's hinterland, not in the cities: Trzin 25×, Log-Dragomer 22×, Brezovica 20×, Vrhnika 19×, against Ljubljana 7×, Koper 7× and Kranj 7×.

In the 101 municipalities with enough sales in both windows, the median price over the same generalised value rose 86% from 2013–15 to 2023–25. Dwellings built 2015–2025 are only 5% of the 2014 stock.

Across municipalities the two are unrelated (correlation −0.14). Where prices rose most (Mislinja, Bovec, Litija, Jesenice, Kamnik: by 112%–121%), the stock grew a few per cent. Where most was built (Rače-Fram 18.5%, Benedikt, Hrpelje-Kozina), prices rose less than the median. The market sets prices freely and taxes leave visible marks, but quantity barely answers price.

Part one compares municipalities over a decade. Part two asks quarter by quarter: when the cycle turns, do permits, construction, completions and the number of sales move? And if the response stops somewhere, is it at the permit or after it?

The chart shows levels, 2007 = 100. From 2015 to 2025 prices rose 115%, permits 65% and completions 75%. In 2025 completions are 53% of 2007, the peak of the previous boom. Pick a series.

In the bust, from 2007 to 2014, prices fell 18%, permits 74% and completions 67%. Supply fell four times as hard as price. For each 1% fall in price completions fell 5.6%; for each 1% rise in the boom they rose only 0.7%.

The study does not separate the causes of the collapse: prices fell, builders went bankrupt and credit dried up at once.

Permits and construction move with price growth in the same quarter (r = 0.40, interval 0.24–0.57, and 0.47), but swing far harder: annual permit growth has a standard deviation of 22%, price 6.7%. Permits peak two quarters before price and construction two quarters after it: permit, price, build.

Completions follow permits (their annual growth agrees at r = 0.58 within the same year), so nothing suggests a bottleneck in construction after the permit. The number of flat and house sales does not move with price in the same quarter (r = −0.03 and −0.18). Pick a quantity.

When the mortgage rate rises, fewer houses and flats sell a quarter or two later: r = −0.72 for houses and −0.59 for flats. None of 50 random shifts of the rate series by three years or more gives a link that strong (p = 0).

The link is real, but leans heavily on 2009–2010, when the ECB's cuts and the rebound in sales coincided. Without those years it is −0.43 for houses and −0.28 for flats; in 2017–2026 −0.51 and −0.24. Prices do not fall with rates; they rise with them (What moves house prices). Higher rates cut the number of deals and leave the price alone.

In the same 101 municipalities, dwellings completed per 1,000 existing ones (2015–2025) are unrelated to price growth (Spearman −0.05). Building permits per 1,000 dwellings are even negatively related to it (−0.38), because they count every building, farm and ancillary ones included, and so read high in the countryside. Permits do agree with completions (0.51). Pick a measure.

Who buys. GURS collects the parties' legal form, position and citizenship on every contract, but the public ETN extract carries no party file. In the ownership register every private owner hides behind the same marker. Reciprocity for non-EU buyers under Article 68 of the Constitution is therefore law here, not a measurement.

Pre-emption. Whether it was exercised is not a field in ETN. 27 legal notes across 622,630 contracts mention a pre-emption right: an anecdote, not a rate.

Credit. The Bank of Slovenia has capped loan-to-value and debt-service-to-income since November 2019. There is no mortgage data, and the start coincides with covid.

Fees. A broker's fee is capped at 4% of the price. ETN records only whether an agency took part (26% of flat sales where the field is filled), not the fee.

Price is free: more than nine in ten contracts are open-market sales, and the non-market tags mark deals that really fetch less. Taxes leave marks on what is built and when it is sold. Quantity is constrained, but in place, not in time: supply answers the cycle, harder than price does, but it does not move to where prices rose. The bottleneck is probably land and location, as the 7.5-fold premium for buildable land suggests, rather than the construction pipeline. Rates change the number of deals, not prices. The wait for a building permit by administrative unit is in Who the market suits; there too, a slow office does not build fewer homes.