Mass Appraisal Across Six Jurisdictions: A Synthesis of Thin Case Studies

Summary

Lars Doucet's Does Georgism Work? Part 3 closes its methods survey with a run of six single-country citations that the wiki's Doucet de-referencing wave flagged as a synthesis candidate rather than six separate stub pages: Kilić, Rogulj & Jajac (2019, Croatia), Yalpır & Ünel (2017, Turkey), Raslanas, Zavadskas, Kaklauskas & Zabulėnas (2010, Lithuania), Aragonés-Beltrán, Aznar, Ferrís-Oñate & García-Melón (2008, Spain), Xue, Hu, Yang & Chen (2008, China), and Kettani & Khelifi (2001, Quebec, Canada).

Read individually, none of these is a landmark. Each is a single case study — one region, one dataset, sometimes one parcel — reported in a specialist operations-research or land-management journal, not the top-tier econometrics literature the wiki's Kolbe or Berry citations draw on. This page's contribution is not to inflate any one of them — it is to make their collective pattern visible and to grade each one's actual accessibility honestly, rather than let a single retelling (Doucet's) stand in for six. Read together, they show statistical and expert-weighted land/property valuation being attempted, independently, across six unrelated legal and institutional systems spanning Southern and Eastern Europe, Anatolia, East Asia, and North America — evidence of breadth, not depth.

The Six Cases, Graded by Access Level

1. Croatia — Kilić, Rogulj & Jajac (2019)

Fuzzy Expert System for Land Valuation in Land Consolidation Processes, Croatian Operational Research Review 10(1), pp. 89–103.[1] Case: agricultural land parcels on the island of Hvar, Split-Dalmatia County. Method: a fuzzy-logic expert system that classifies cadastral parcels into value ("bonitet") categories to support land consolidation — the process of reallocating fragmented smallholdings into more efficient parcels, where fair relative valuation of what each landowner gives up and receives back is the central problem. This is a narrower task than tax assessment: the system need only rank and classify parcels for equitable exchange, not produce absolute market values.

Access level: open-access journal (hrcak.srce.hr, Croatia's national open-access repository), abstract-level read. Direct PDF fetch was blocked by the host during this wave; the description above is drawn from the published abstract and secondary indexing (Google Scholar record, related-paper descriptions), not a full-text read. The paper is nonetheless freely available in principle — a future pass should fetch it directly rather than through automated tooling.

2. Turkey — Yalpır & Ünel (2017)

Use of Spatial Analysis Methods in Land Appraisal; Konya Example, presented at ISITES2017 (5th International Symposium on Innovative Technologies in Engineering and Science), Baku, Azerbaijan.[2] Case: plots in the settlement districts of Konya province, Turkey. Method: a GIS-built spatial index combining neighborhood and locational factors (fed partly by a public questionnaire on which criteria buyers weight) with multiple regression analysis (MRA) on market land prices.

Access level: full text read (freely downloadable conference PDF). Reported results, verified against the paper: the model, built from 41 criteria (reduced by indexing from an initial 64), reached R² = 0.85 between modeled and market value, with a mean absolute percentage error (MAPE) of 0.287 (≈29%), RMSE of 0.013, and MAE of 0.008.[2] The paper's own reading of this: "the model obtained with 41 criteria... seems to explain 85% of the plot value... The part of unexplained 15%... has been determined to be due to variable economics, rent [and unearned income]."[2] An R²=0.85 with a ~29% mean percentage error is a real but moderate result — closer to Bencure et al.'s Philippines case (Adj. R²≈0.67, already on the wiki) than to Kolbe et al.'s Berlin correlation of 0.845, and a useful reminder that headline "explains X% of value" figures can coexist with substantial per-parcel percentage error.

3. Lithuania — Raslanas, Zavadskas, Kaklauskas & Zabulėnas (2010)

Land Value Tax in the Context of Sustainable Urban Development and Assessment. Part II: Analysis of Land Valuation Techniques, the Case of Vilnius, International Journal of Strategic Property Management 14(2), pp. 173–190, CC BY 4.0.[3] Case: a single 6.89-are site on Švitrigailos Street, Vilnius, appraised on 12 March 2008 using three different methods: the sales comparison approach, MAMVA (Multiple Attribute Market Value Assessment — a multi-criteria weighting method), and a mass-valuation (value-zone-average) approach.

Access level: full text read (freely downloadable, Creative Commons). This is the most rigorously reported and most honestly complicating case in the set. The three methods diverged sharply on the same site: sales comparison → LTL 1,500,000; MAMVA → LTL 1,730,000; mass valuation (the zone-average method actually used for Lithuania's tax base) → LTL 511,370 — 65.9% lower than the sales-comparison figure and 70.4% lower than MAMVA's.[3] The paper's own conclusion is blunt: "fairness of the land tax is not ensured, because the tax value makes up only about 30 percent of the market value... Highly inaccurate valuation prevents municipal budgets from considerable tax revenue and the tax efficiency is low."[3] Applied to the 1.5% Lithuanian rate then in force, the resulting land-tax bill would have been LTL 25,950 (MAMVA), LTL 22,500 (sales comparison), or LTL 7,670.55 (the mass-valuation method actually used) for the identical parcel[3] — a roughly 3.4× spread depending on which recognized method is applied.

This case is a genuine caveat, not just a confirmation, and the wiki should not launder it into the "CAMA works" pile without saying so. It shows that (a) different valuation methods can disagree substantially on a single site even when each is methodologically defensible, and (b) at least one operational mass-valuation system (Lithuania's zone-based approach, at the time of the study) tracked market value poorly enough to matter for tax fairness and revenue. This cuts in the opposite direction from the "spatially smooth, therefore easy" framing elsewhere on the wiki and belongs in this page's Limits section as much as its case tally.

4. Spain — Aragonés-Beltrán, Aznar, Ferrís-Oñate & García-Melón (2008)

Valuation of Urban Industrial Land: An Analytic Network Process Approach, European Journal of Operational Research 185(1), pp. 322–339.[4] Case: an industrial park in Valencia, Spain. Method: the Analytic Network Process (ANP), a multi-criteria decision-analysis technique that (unlike simple weighted-sum methods) explicitly models interdependencies between valuation criteria, intended to handle situations with partially available data and qualitative variables — an expert-weighted alternative to regression-based methods, in the same family as the AHP method Bencure et al. use for the Philippines.

Access level: paywalled (ScienceDirect); no abstract available on either the publisher page or its RePEc/IDEAS mirror. Both were checked directly during this wave and both show no abstract text. The description above is reconstructed from secondary characterizations (indexed search summaries of the publisher's own framing, and the discussion of this paper in Raslanas et al. 2010's literature review, which cites it as addressing "some of the drawbacks found in classical property valuation methods"[3]) — not from a first-hand read of the paper's results or figures. No quantitative accuracy result from this paper is reported here because none could be verified.

5. China — Xue, Hu, Yang & Chen (2008)

Land Evaluation Based on Boosting Decision Tree Ensembles [Chinese original], Transactions of the Chinese Society of Agricultural Engineering 24(7), pp. 78–81.[5] The full four-author byline (Yaolin Xue, Yuting Hu, Jingyao Yang, Qiu Chen) was previously unconfirmed on the wiki (flagged for verification in the Doucet de-referencing wave); it is now confirmed via Raslanas et al. (2010)'s own reference list, which cites the paper in full.[3] Doucet's essay glosses the title slightly differently ("Land valuation using C5.0 with a Boosting decision tree"), which is either a paraphrase or a translation variant of the Chinese title — not independently resolved here.

Access level: bibliographic record only. This is the thinnest entry in the set. Despite searches in both English and Chinese (including the Chinese-language journal's own site, which returned a server error), no abstract, full text, or even a secondary description of the paper's method or findings beyond its title could be located this wave. What is confirmed — author names, journal, volume/issue/pages, and that the paper concerns applying boosting-ensemble decision trees to land evaluation — comes entirely from other papers' citations of it, not from the source itself. No claim about this paper's findings should be made beyond "it exists, is by these authors, in this journal, and appears to apply a boosting-tree ensemble method to land evaluation."

6. Quebec, Canada — Kettani & Khelifi (2001)

PariTOP: A Goal Programming-Based Software for Real Estate Assessment, European Journal of Operational Research 133(2), pp. 362–376.[6] Case: municipal ad valorem property assessment for the Communauté urbaine de Québec (CUQ), an administrative grouping of 12 Quebec-area municipalities. Method: PariTOP, a decision-support system built on goal programming (a mathematical-optimization technique for satisfying multiple, sometimes conflicting, target constraints simultaneously) to estimate market values across the tens of thousands of properties in CUQ's assessment roll, developed as a partnership between CUQ and the software firm Modellium.

Access level: paywalled (ScienceDirect states "No abstract is available for this item"). The description above is reconstructed from secondary sources — publisher/indexing-site summaries and the existence of a 2015 follow-up paper by the same lead author, Designing and Implementing a Real Estate Appraisal System: The Case of Québec Province, Canada (not itself read this wave) — not from the 2001 paper's own text. No accuracy metric (R², COD, or similar) from the original paper is reported here because none could be located.

What This Collectively Shows

Individually, four of the six cases are read only at second or third hand (Croatia's abstract; Spain's and Quebec's paywalled records reconstructed from indexing; China's title-only citation). Only two — Turkey and Lithuania — were read in full, and one of those two (Lithuania) is a cautionary case, not a confirming one. No single one of these six papers should be cited on the wiki as strong evidence that land/property valuation "works" in its country.

What they support collectively is narrower and more defensible: the attempt to formalize land and property valuation — via regression, fuzzy logic, multi-criteria expert weighting, or machine learning — is not a peculiarity of a handful of rich, data-abundant, English-language jurisdictions. It shows up independently in a Croatian operations-research journal, a Turkish geomatics-engineering conference, a Lithuanian property-management journal, a Spanish-authored operations-research paper, a Chinese agricultural-engineering journal, and a Canadian public-private assessment partnership — six unrelated academic and administrative traditions converging on the same underlying problem. That convergence is itself a (weak, indirect) data point for the mass-appraisal-methods page's claim that assessment technique, not fundamental infeasibility, is the live constraint on land valuation. It is a breadth argument, not a depth argument, and the wiki should keep it labeled that way rather than let six thin citations accumulate the rhetorical weight of six confirmed results.

Limits and Caveats

• Access asymmetry. Four of six sources are graded at abstract-level or bibliographic-record-only access in this pass; only the two least prominent venues (a regional conference, an open-access CC-BY journal) were fully readable. This is itself worth noting: some of the most-cited operations-research venues (Elsevier's EJOR, twice in this set) are the least accessible to a reader trying to verify the claims Doucet's essay repeats.
• The Vilnius case argues against, not for, easy assessment. Raslanas et al.'s finding that three defensible methods produced land values 3.4× apart for the same parcel, and that Lithuania's actual tax base captured only ~30% of market value, is real counter-evidence to any claim that land valuation is a solved problem — it belongs on the objection's steelman side as much as this page's tally belongs on the response side.
• None of these six is peer-reviewed at the level of the wiki's strongest assessment evidence (Kolbe et al. in a dedicated German land-value working-paper series with a verified 0.845 correlation; Berry (2021)'s 2,628-county US study). They are supplementary-tier by the wiki's own evidence-ordering rule and should never be cited ahead of that stronger evidence.
• Method diversity is not method validation. That six different statistical/expert-weighting techniques have been tried across six countries says less about whether any of them reliably matches market value than a single well-validated study would. Breadth of attempt is not the same claim as breadth of confirmed accuracy.

See Also

• Mass Appraisal Methods — the concept page this synthesis supports
• Objection: Land value can't be assessed accurately
• IAAO Standards: Ratio Studies & Property Tax Policy — the professional-standard metrics (COD/PRD/PRB) these case studies are informally approximating
• Bencure et al.: iLVM, Baybay City, Philippines — the wiki's other thin-data / non-Anglophone developing-country case, method-comparable to the Spain and Croatia entries here
• Kolbe et al.: Berlin land-value appraisal — the wiki's strongest single-country accuracy evidence, for contrast
• Doucet, Does Georgism Work? — the essay whose citation tail this page de-references
• Gloudemans & Almy, Fundamentals of Mass Appraisal — the IAAO textbook backbone behind the standardized technique family (sales comparison, regression, allocation) these six countries' methods approximate

Sources

1. Jelena Kilić, Katarina Rogulj & Nikša Jajac (2019), "Fuzzy Expert System for Land Valuation in Land Consolidation Processes," Croatian Operational Research Review 10(1), pp. 89–103, DOI 10.17535/crorr.2019.0009. hrcak.srce.hr record — open-access journal; used at abstract level (direct PDF fetch blocked this wave) for the Hvar, Croatia land-consolidation case and the fuzzy-expert-system method.
2. Mürşide Şükran Yalpır & Fatma Bünyan Ünel (2017), "Use of Spatial Analysis Methods in Land Appraisal; Konya Example," ISITES2017 (5th International Symposium on Innovative Technologies in Engineering and Science), Baku, Azerbaijan, pp. 1573–1582. PDF (read in full) — used for the Konya, Turkey case, the GIS+MRA method, and the R²=0.85 / MAPE≈29% results (verified verbatim against the PDF, 2026-07-18).
3. Saulius Raslanas, Edmundas Kazimieras Zavadskas, Artūras Kaklauskas & Arūnas Remigijus Zabulėnas (2010), "Land Value Tax in the Context of Sustainable Urban Development and Assessment. Part II: Analysis of Land Valuation Techniques, the Case of Vilnius," International Journal of Strategic Property Management 14(2), pp. 173–190, DOI 10.3846/ijspm.2010.13, CC BY 4.0. PDF (read in full) — used for the Švitrigailos Street, Vilnius site, the sales-comparison/MAMVA/mass-valuation value comparison (LTL 1,500,000 / 1,730,000 / 511,370), the resulting land-tax figures, the paper's own conclusion on tax-base fairness, and the confirmed full byline for source [5] (verified verbatim against the PDF, 2026-07-18).
4. Pablo Aragonés-Beltrán, Jerónimo Aznar, Jesús Ferrís-Oñate & Mónica García-Melón (2008), "Valuation of Urban Industrial Land: An Analytic Network Process Approach," European Journal of Operational Research 185(1), pp. 322–339. ScienceDirect (paywalled; no abstract available on this page or its RePEc mirror, both checked directly) — used for the Valencia, Spain industrial-park case identification only and the ANP-method description, both reconstructed from secondary indexing rather than a first-hand read; no findings from this paper are reported.
5. Yaolin Xue, Yuting Hu, Jingyao Yang & Qiu Chen (2008), "Land Evaluation Based on Boosting Decision Tree Ensembles," Transactions of the Chinese Society of Agricultural Engineering 24(7), pp. 78–81. ResearchGate record — bibliographic record only; full byline and journal/volume/pages confirmed via citation [3] (Raslanas et al. 2010's reference list) rather than the source itself; no abstract, full text, or findings could be located in English or Chinese this wave.
6. Ossama Kettani & Karim Khelifi (2001), "PariTOP: A Goal Programming-Based Software for Real Estate Assessment," European Journal of Operational Research 133(2), pp. 362–376. ScienceDirect (paywalled; "No abstract is available for this item") / RePEc mirror — used for the Communauté urbaine de Québec case identification only and the goal-programming/PariTOP method description, reconstructed from secondary indexing of the paper and its 2015 follow-up by the same lead author; no findings from the 2001 paper are reported.