PREMIUM ON NEW.

Editorial note: this piece was rewritten in August 2026. The original argued that the premium on new build had vanished, based on a comparison of transaction prices across the primary and secondary markets. I found three errors in it: I called averages medians; I wrote that Kraków showed “the same pattern” as Warsaw when in fact it is the reverse; and I drew a conclusion from the gap that does not follow from it. Below is the corrected version, with an account of each.

Where I started

The starting point was the NBP reading of average transaction prices for Q4 2025, in which the secondary market in Warsaw came in above the primary market: PLN 16,750 per square metre against PLN 16,583. The gap is PLN 167, or one percent. The observation is correct and can be pointed to in a specific table.

A second correction concerns the word. I previously wrote about medians. NBP reports averages in these tables, not medians — and the two are not the same, especially on the skewed distributions typical of housing.

And here is the thing I failed to check in the original version. I wrote then that Kraków showed “the same pattern”. It does not. In Kraków the primary market stands at PLN 15,588 per square metre against PLN 14,898 on the secondary — the primary market is PLN 690, or 4.6 percent, higher. That gap is more than four times the Warsaw one, in the opposite direction.

Two cities, the same quarter, the same calculation, opposite results. That alone should have stopped me before concluding that “the premium on new build has vanished”. If an indicator gives opposite answers in Warsaw and Kraków, it is not measuring a property of the product. It is measuring something that differs between those markets.

So the error was twofold: generalising an observation from one city to the whole country, and inferring from a gap in averages something that does not follow from it. The second error is the more serious one and deserves a separate explanation.

Why a gap between averages does not measure a premium

If in a given quarter the primary market sold more small flats on the outskirts while the secondary market sold more large ones closer to the centre, the gap between averages tells you something about the composition of what sold — not about what newness is worth. This is called composition bias, and it is one of the most common ways a correctly calculated number leads to a wrong conclusion.

This is not a theoretical objection. Reusens, Vastmans and Damen measured the effect on a decade of Belgian data and showed that the quality of the dwellings sold rose substantially — largely through improvements in energy performance — and that failing to account for this overstates measured price growth. Comparing averages without controlling for characteristics measures a change in the product, not a change in price.

The formal basis is old and uncontroversial: Rosen showed in 1974 that the price of a dwelling decomposes into the implicit prices of its characteristics. To value one characteristic — here, newness — you have to hold the others fixed: floor area, location, finish, floor level.

What the literature actually establishes

Coulson, Morris and Neill asked exactly this question in Real Estate Economics, and asked it properly: on a large set of transactions, testing several alternative ways of identifying a “new” home, split by period and location — including before and after the housing bust.

The result: a premium for new build appears primarily where the supply of new homes is relatively low, reaching over twenty percent in some cases. When new homes are plentiful, the premium disappears.

The premium on new build is not a property of the product. It is a state of the market.

This is a better answer than my previous one because you can act on it. “The premium has vanished” tells a developer nothing actionable. “The premium is a function of new supply in your submarket” tells them where to look.

The evidence that cuts the other way

I will not leave this in its convenient version. Nordahl and Sommervoll analysed roughly ninety thousand secondary-market transactions and nineteen thousand new-build transactions in Oslo between 2002 and 2015, published in Cities. The new-build premium turned out to be low, to vary significantly across project types, and to be negative in some projects.

Their conclusion should interest any board: developers do not always capture the value they create. Part of it leaks into the neighbouring existing stock, whose prices rise because of the development. This paper matters more to me than the American one — the data are European, the sample is large, and Norway is closer to Poland than Las Vegas during a boom and a crash.

What the secondary-market literature cannot see

All of the studies above compare new to used on markets dominated by resale, where a thousand independent sellers set prices. On the Polish primary market, prices are set by a dozen or so developers — each following its own price-list policy, release schedule and financing pressure.

Wu, Deng and Liu showed, on a market where new units make up the bulk of transactions, that simple averages fail to account for quality changes at the development level, while matching methods fail to control for developers’ own pricing behaviour — both biased downward. That is a variable the resale-based literature does not register at all: the price structure on a primary market is partly an artefact of developer decisions, not merely a resultant of demand.

What this means for positioning

The practical conclusion from the original piece survives the correction, though on different grounds. If an investment is positioned as “new, modern, energy-efficient” without answering why here and why at this price, it rests on a characteristic whose market value depends on the state of supply — and which, where new supply is plentiful, can fall to zero.

So the question to ask before setting a price list is: how many comparable new units are on the market at the same time in this submarket. Not in the city — in the submarket.

Sources and method

Method: a review of peer-reviewed literature on whether a new-build premium is documented and under what conditions, then a confrontation with my own earlier claim. I have not run my own hedonic estimate for Poland, and therefore give no percentage figure for the Polish market.

• Narodowy Bank Polski (2026). Informacja o cenach mieszkań i sytuacji na rynku nieruchomości mieszkaniowych i komercyjnych w Polsce w IV kwartale 2025 r. Financial Stability Department, March 2026. Data used here: average transaction prices for Q4 2025 — Warsaw, Table 22 (primary) and Table 23 (secondary), p. 41; Kraków, Table 16 (primary) and Table 17 (secondary), p. 38. Available in the NBP real-estate analytical materials section.
• Coulson N.E., Morris A.C., Neill H.R. (2019). Are new homes special? Real Estate Economics 47(3). doi.org/10.1111/1540-6229.12165 — direct test against comparable existing homes; Las Vegas, Nevada. Result is conditional on the state of supply.
• Nordahl B.I., Sommervoll D.E. (2023). Reaping the premium in urban redevelopment. Cities 138. doi.org/10.1016/j.cities.2023.104450 — Oslo 2002–2015, c. 90k resale and 19k new-build transactions. Premium low, negative in some projects. Contrary evidence.
• Wu J., Deng Y., Liu H. (2014). House price index construction in the nascent housing market: the case of China. The Journal of Real Estate Finance and Economics 48(3). doi.org/10.1007/s11146-013-9416-1 — bias in price measures in the newly built sector.
• Reusens P., Vastmans F., Damen S. (2023). A new framework to disentangle the impact of changes in dwelling characteristics on house price indices. Economic Modelling 123. doi.org/10.1016/j.econmod.2023.106252 — Belgium 2011–2021.
• Rosen S. (1974). Hedonic prices and implicit markets: product differentiation in pure competition. Journal of Political Economy 82(1). doi.org/10.1086/260169 — basis of the method.
• Kokot S. (2022). Identification of regularities in relation between prices on primary and secondary housing markets. Real Estate Management and Valuation 30(2). doi.org/10.2478/remav-2022-0020 — the only peer-reviewed Polish source I found on this relation. Low citation count, selected local markets. Open access.

What this text does not say: it gives no value for a Polish new-build premium, because I have not computed one; it does not claim the secondary market is the better investment; and it does not control for finish standard — which matters, because the primary market typically sells shell-and-core while the secondary sells completed flats, so part of the gap may simply be an underestimated fit-out cost. None of the cited papers controls for this. All were checked in the Crossref registry and in a citation-context registry for corrections, errata, editorial concerns and retractions — none carries such a notice. Source review date: August 29, 2026.

If you want to know how many comparable new units compete with your development in the same submarket, and what that means for your price list — let us compute it from transaction data. That is an absorption and price-position analysis, not a change of messaging.