Errata and Corrections

Errata and Corrections

The Valuation Engineer Journal corrects the record in public. The version of record is the PDF carried by each article’s DOI.

Where an error is found after publication, the Journal corrects the article’s file and issues it as a new version, numbered on the first page, in all twelve language editions; the article’s DOI is unchanged, its record notes the version, and the earlier version remains archived at Zenodo under the same concept DOI. The correction is recorded on this page with the date and what was changed. A correction that cannot be made in this way is published as a corrigendum with its own DOI. The published pagination is kept where the correction allows and otherwise restated here.

Reporting an error. Write to editor@valuation-engineer.com, naming the article, the page, and what is wrong. Corrections from practitioners working in the jurisdictions the Journal surveys remotely are particularly welcome; the Recent Court Decisions department reports instruments as written and cannot warrant how they are applied in practice.

Corrections issued to date

Foundations, Vol. 1 No. 2 — the Overview and entries 007 to 010, version 1.1.0 (released 5 October 2026). The four statistical entries and the department’s Overview share one constructed market, and its description of condition was wrong: the entries presented a three-level coding (1–3) as a condition rating, which the Journal does not use, and treated a remodel as a feature with a market price of its own, independent of condition. Condition is a state of the house, estimated by the Journal’s condition score (0.00 to 10.00, the share of houses in the market area in worse condition); a remodel is an event whose value is the condition and quality it produced. Every price, every regression result and every verification is unchanged; the entries are restated in the Journal’s units (average 5.00, good 7.00, excellent 9.00; $20,000 a point) and the remodel indicator’s $60,000 is the price difference the constructed data attach to a remodel, at the same GLA, lot, view and condition score; it stands for whatever the remodel brought that the other features do not record. (An earlier statement on this page divided it into $20,000 of condition and $40,000 of quality; that division has been withdrawn.) In detail:

  • 000 Overview (10.67330/h5s4c242): the generating function restated; the Comp H sentence corrected; the Prolog companion omitted entry 014 from the issue’s contents and is corrected; earth prunes by generalized cross-validation, not fold-based cross-validation.
  • 007 Conditional Expectation (10.67330/99hhbc69): the equation and the Issue 1 regression restated in the Journal’s units; the statement that the identity E[ε | Z] = 0 “either does or does not hold” was wrong — it always holds, and what can fail is the same property for the error of the equation actually fitted; a paragraph on the Journal’s own method (Residual Constraint Approach) added.
  • 008 Omitted Variable Bias (10.67330/460smg10): condition, the condition score and a remodel distinguished; the claim that renovation “is not shown in the condition assessment” was wrong under UAD and is removed; the leave-one-out ranking corrected (E is third).
  • 009 Endogeneity (10.67330/z9feps66): “renovation is a choice, not a form of treatment” corrected (a remodel is a treatment that is not randomly assigned); the statement that appraisers rate condition knowing the price withdrawn and restated as a measurement risk for comparables; measurement error is attenuating only in the classical case; “impossible to detect endogeneity” softened; “the subject is a randomly chosen property” corrected; the effect to be estimated named before the variables are chosen (confounder, mediator).
  • 010 Parameter Identification (10.67330/xjszh949): the subsample regression restated in the Journal’s units with both normalizations shown; the two cuts that make the subsample distinguished; bracketing separated into contrast and support; the two time-adjustment failures separated; the formal definition corrected.

Pagination: entries 009, 000 and 007 each grew by one page; page numbers from p. 14 onward in Vol. 1 No. 2 are those of version 1.1.0 of the issue.

Foundations, Vol. 1 No. 2 — entries 012 and 013, with the Overview and entry 014, released 5 October 2026. The Purposes of Valuation (012, version 1.1.0; 10.67330/g9knsr34) presented the levels beneath the purpose as a fixed chain (basis of value, then property rights, then legal context, then methodology) and said the purpose determines them. It now presents the property interest, the governing law and the basis of value as settled together, in no fixed order, and the purpose as constraining them without fixing a single answer to each; purpose 5 is renamed damages and compensation and limited to valuations that measure a loss; the Ukrainian war-damage compensation law is recorded as a compensation regime beside the eight purposes, not under insurance; Finland’s compensation (market value plus twenty-five percent) is distinguished from its basis of value; purpose, basis of value and property interest are defined by contrast with their everyday appraisal meanings. Terminological Divergence (013, version 1.1.0; 10.67330/symwrt78): a missing equivalent is stated not to mean a missing concept; the alignment rule is stated; the negative findings are qualified as findings in the materials surveyed; “a date is also a word” withdrawn. The Overview (version 1.2.0) and The Basis of Value and the Approach (014, version 1.1.0; 10.67330/jp7pr043) restate the relation of purpose to basis accordingly. No page number changes.

Vol. 1 No. 2 — Foundations version 1.2 and revised articles, released 7 October 2026. Every Foundations entry, every article, Comparative Practice M1 and the Editors’ Message were reviewed for readers who are appraisers rather than statisticians or lawyers, and then reviewed again independently. Each item keeps its DOI; the earlier versions remain archived at Zenodo. Each item’s version number counts its own revisions: the articles, Comparative Practice and the Editors’ Message are in their first revision (1.1.0), most Foundations entries in their second (1.2.0). The English edition is now 746 pages, and the page numbers of every item from page 13 onward have changed. The translated editions of these versions will follow. Only errors are itemized below; clarifications for the same reader are summarized in one line per item.

Foundations, Vol. 1 No. 2: Conditional Expectation, version 1.2.0 (released 7 October 2026) (10.67330/99hhbc69)

  • Formal definition. “Reproduces the function exactly when it is linear” could be read as the regression being linear. It now says least squares reproduces the CEF exactly when the CEF itself is linear.
  • Appraiser encounters. A conditional expectation was described as “the middle value of the range” of prices. It is the average, not the median or the midpoint.
  • Intuitive framing, the causes of a misleading regression. “Because its coefficients are read as what a buyer pays” was given as a cause of the error moving with the features. Reading a coefficient does not change the error. The causes are now a left-out feature, price acting back on a feature (days on market), and measurement error.
  • Structural error. u was described as the error of the fitted equation. It is the structural error, as defined in Endogeneity.
  • “Any difference … is bias and nothing else” contradicted Parameter Identification. It now reads: bias or a failure of identification, never noise.
  • The constructed data were said to remove sampling variability altogether. They remove random noise from the prices: every price lies exactly on the formula.
  • Equal weighting. Averaging three indications was said to assume only that each is an unbiased draw from the same conditional mean. Equal weight also assumes that all three are about equally precise.
  • Conditioning set.
    • The choice of conditioning set was called “not a constraint imposed by the data”. It defines what is estimated; whether the data can estimate it is a separate question, the subject of Parameter Identification.
    • The conditional expectations with and without condition were said to have different values and slopes. They may.
  • “Holding other things equal” was said to be “used in almost all appraisal reports”. That claim was removed. The phrase describes the appraiser’s reasoning, and the analysis behind it usually goes into the workfile that USPAP’s Record Keeping Rule requires, not into the report.
  • Defensibility. “E[P | GLA, view, conditioning]” now reads “condition”.
  • Quotation marks printed reversed in two places.
  • Remodel. The $20,000 + $40,000 division of the remodel indicator was removed.
  • Issue 1’s regression. It is now stated as restated on the Journal’s condition score (Issue 1 printed it on its 1–3 coding).
  • Clarifications for the same reader:
    • ambiguous pronouns resolved
    • terms defined at first use, among them “nonparametric estimator”
    • a notation key and a figure of the conditional expectation function added
    • prediction intervals stated in MARS (earth varmod) terms

Foundations, Vol. 1 No. 2: Omitted Variable Bias, version 1.2.0 (released 7 October 2026) (10.67330/460smg10)

  • Comp H’s gap. It was given as “about $60,000 more than its four recorded features explained”. Issue 1’s regression gap was $46,455; the $60,000 is what Issue 1 says the kitchen remodel adds.
  • Theil’s formula was called “not an approximation”. It holds exactly with the estimated coefficient, and on average with the true one.
  • Features left out. They were called features “regularly” left out, each affecting price and most going with something the grid records. Now: features commonly left out, each of which can affect price, many going with something the grid records.
  • The remodel line. The grid was said to need no remodel line because the remodel’s value is “already in” the condition and quality lines. That holds only when those lines capture what the remodel brought.
  • More data. A larger sample with the same feature left out was said to give “the same wrong figure”. It gives an estimate centered on the same wrong figure.
  • Cross-validation was said to be unable to distinguish bias from noise. It does not rule bias out: good cross-validation results do not show that the coefficients are free of bias, because a misspecified model can predict well.
  • Prediction.
    • The “0.14 percent” example is Comp N, a sale in the fit.
    • The typical error is about $21,000.
    • “Fails only at the properties that break the pattern” contradicted the entry’s own held-out errors.
    • Each held-out error was called bias. These are prediction errors; each comes from the left-out remodel indicator, none from noise.
  • View coefficient. It was said to “give up” $20,822 to condition. The two biases are separate effects of the same omission, not a transfer.
  • Least squares was said to “reproduce the sample’s prices”. It comes closest to them.
  • Remodel. The $20,000 + $40,000 division and “a 9.00 is in truth a 10.00” were removed.
  • Prolog addendum. A paired-sales adjustment far larger than cost was said to “often reflect” leakage from a left-out feature. It can.
  • Issue 1 correction. Issue 1’s Latent Variable said the bias “would shrink further with more comps” and biased the other coefficients upward. Bias does not shrink with more sales, and in Issue 1’s own eight sales the view, GLA and lot coefficients are biased downward; only condition is biased upward. The entry now says so.
  • Clarifications for the same reader:
    • “traditional appraisal” and “RCA-based appraisal” named where meant
    • the condition score defined in the editor’s wording
    • terms defined at first use

Foundations, Vol. 1 No. 2: Endogeneity, version 1.2.0 (released 7 October 2026) (10.67330/z9feps66)

  • Formal definition. The regressor was said to move with “the part of price the model does not explain”, which is the residual, and a residual cannot move with the regressors. The definition now names the structural error: the price minus what the causal relationship says it should be.
  • Measurement error.
    • A condition score’s error was said to depend on how good the house really is. It may.
    • A second, independent measurement was offered as the remedy by itself. It needs a method designed for measurement error to correct the estimate.
  • Two-way causation. School quality was said to raise prices while prices raise school quality, and crime, retail amenity and tree canopy were said to be both cause and effect of prices. These are now stated as possibilities (text and Prolog addendum).
  • Selection of sold properties. It was said to make a sold property’s price error non-random in every case. That holds when the properties that sell differ from those that do not in ways the model does not see.
  • Unsold properties. The selection assumptions were said to be ones “reports almost never state”, which is an unsupported count. It now reads: which a report should state.
  • Price acting on the score. The direction of the resulting bias was given as known in general. It is known for a nudge that carries part of the price’s deviation into the score (text and Prolog addendum).
  • The coefficient was described as “the price difference that goes with the feature”. It is the difference in fitted price per unit of the feature.
  • Natural experiments. The designs were presented as more defensible without qualification. Each rests on assumptions that must be defended, and only a design whose assumptions are defended is more defensible.
  • “The specification is now correct” was said of the model once the remodel indicator is included, while the same passage leaves the structural error u out. It now says the indicator is included and that u, which no dataset records, is still left out.
  • Worked example. The $60,000 was called “what a remodel does to a price”. It is what a remodel adds with the recorded condition score held fixed; a remodel that also raised the score would add more.
  • Renovation payback. The second appraiser’s model, “with condition and quality held fixed”, is now stated as one that also measured them.
  • Practice claims.
    • Comparables are no longer said to be “commonly” scored after their prices are known. That is now a hypothetical.
    • Homeowners are no longer said to renovate “because” they expect a return.
    • The basement and larger-lot examples were removed. The lot example did not show a feature caused by the price.
  • Remodel. The $20,000 + $40,000 division was removed from the text, the formula label and the Prolog addendum.
  • Prolog addendum. It gave the selection of sold properties a single upward direction. It now gives both directions, as the text does.
  • Clarifications for the same reader:
    • the causal relationship written out before the structural error
    • terms defined at first use
    • “traditional appraisal” and “RCA-based appraisal” named where meant
    • the Durbin and Wu names dropped from the test, which is cited as Hausman’s

Foundations, Vol. 1 No. 2: Parameter Identification, version 1.2.0 (released 7 October 2026) (10.67330/xjszh949)

  • Parameter. An adjustment was listed as a parameter. The parameter is a rate or coefficient, from which an adjustment is computed.
  • Opening example. “If every remodeled house is also in excellent condition” was given as enough for the two to be inseparable. Both directions are needed: every remodeled house excellent and every excellent house remodeled.
  • Sample size. “A greater sample size will be of no help” contradicted the worked example, which adds sales. It now reads: a greater sample of the same kind of sales.
  • A more sophisticated model was said not to help, which the entry’s own causal case contradicts. It does not help unless it adds an assumption the data cannot test.
  • Near collinearity. It was called “in practice” unidentified. It is identified but imprecise, and the text now keeps the two ideas apart.
  • Contrast. Any contrast in a characteristic was said to secure identification. The characteristic must vary apart from the others (text and Prolog addendum).
  • The time adjustment was called “identified but composite”, which can be read as identifying the effect of market conditions. The time coefficient is identified, but it is composite and is not the effect of market conditions alone (text and Prolog addendum).
  • Causal parameters. They were said to be unidentified “whenever” the data are observational. The text now adds: unless an assumption the data cannot test is added.
  • Reporting an unidentified split. The advice to give an honest account of the split was stated for every unidentified parameter. It applies when the data determine a combined effect but not its division.
  • USPAP and the Fannie Mae Selling Guide were said to make the identification question obligatory. That is now stated as the Journal’s reading of them (text and Prolog addendum).
  • Worked example.
    • What the ten sales determine was called “the sum” of the two effects. It is two points of condition plus the remodel, $100,000.
    • Each pair on the line fits only with its own intercept.
    • The intercepts were misdescribed for the condition-dropped fit.
    • The direct contrasts identify the split “given only the model’s own form”, not “with no further assumption”.
    • The ten-sale subsample of a constructed dataset was said to be “not artificial”. It now says the subsample leaves a realistic pattern.
  • References to other entries. The composite time adjustment was credited to Omitted Variable Bias, which has no time content.
  • Remodel. The “condition the record does not show and quality the dataset does not carry” was removed.
  • Clarifications for the same reader:
    • the condition score in the editor’s wording
    • traditional appraisal and RCA-based appraisal named
    • terms defined at first use
    • “rates” (adjustments) kept apart from interest rates

Foundations, Vol. 1 No. 2: Overview, version 1.3.0 (released 7 October 2026) (10.67330/h5s4c242)

  • Conditional Expectation was said to name “what a regression estimates”. It names what a regression is trying to estimate; a linear regression recovers it exactly only when it is linear, and otherwise recovers its best straight-line approximation.
  • Omitted Variable Bias was described as the case where the missing information could have been obtained. That is the Pacifica case; a missing feature no one could measure biases the coefficients the same way.
  • Bracketing. The bracketing requirement was called “an identification requirement”. Parameter Identification says the opposite: bracketing secures support, and only contrast in the sales, a feature varying apart from the others, secures identification. The Overview now agrees.
  • “Every gap … is bias” contradicted Parameter Identification. It now reads: bias or a failure of identification, never noise.
  • MARS.
    • Statistical statements were said to hold “term by term”. That holds for the algebraic results, not for standard errors and tests.
    • GCV was called “a penalty”.
  • The chain. It was called a chain of four failures. The first entry names the target, and the other three are failures.
  • Terminological Divergence. Its finding was summarized as “no statute carries an abstract noun for a value effect”, the idea appearing only as a directional pair or a verb. That left out the neutral nouns the entry itself reports. It now gives all three forms (a directional pair, a neutral noun for a change in value, a verb) and says that no statute uses a general noun for the effect as a term of art (text, abstract and Prolog addendum).
  • The Purposes of Valuation.
    • Its classification was described in the old fixed hierarchy (abstract and Prolog addendum).
    • The war-damage regime was called “a purpose of its own kind”. It is a regime of its own kind, not a ninth purpose.
    • The empty purposes were counted as two.
  • Remodel. The $20,000 + $40,000 division was removed (text and Prolog addendum).
  • Prolog addendum. The diagnostic ladder was said to be listed “in order of asking”, which contradicts the text. It is listed in printed order; in practice it is asked from the last step back to the first.
  • Other corrections.
    • “The eight entries” listed nine items.
    • “Three points of posture carry over” from Issue 1, when two do.
    • Parameter Identification was said to refer back to Issue 1’s regression.
    • “Real data never behave this way” now reads “real sales data”, since the constructed data in the entries do.
  • Clarifications:
    • one name for the remodel indicator
    • the condition score defined as a percentage divided by ten
    • traditional appraisal and RCA-based appraisal named
    • “basis of value” given its USPAP equivalent
    • “tribunal” given its US equivalent
    • entries named by title
    • the structural error named in the four questions

Foundations, Vol. 1 No. 2: The Barrier to Human Understanding of MARS (supplement), version 1.2.0 (released 7 October 2026) (10.67330/qmnctv78)

  • Reading the slopes. The rule that a segment’s slope is “the running total” of the coefficients to its left (text and Figure 3) leaves out the first segment’s rate, which reflected hinges set. The running total starts from that rate.
  • Contributions and fitted values. The adjustment was computed from figures called “contributions” that are fitted values. Both are now shown, and they give the same $500,000.
  • Where the rate changes. MARS was said to let the rate change “where demand changes”. It lets the rate change where the sales show it changing.
  • Below the smallest house. The variable graph was said to have “no value” there. The formula still returns a number, but it is an extrapolation, not a value the sales support.
  • The 800-square-foot floor was called “standard”. The figures use it, but the floor is a characteristic of the market area.
  • Paired knots. In the seven-knot example the knot at 1,200 also carries a pair, not only 2,400.
  • Interactions.
    • Corollary 4 overstated when partners give different rates.
    • Two one-feature-at-a-time grids were said to agree on the total. They do so only in opposite orders.
    • The partner in the sale-age example was described as the age of the house.
    • Figure 7 matches Figure 2’s rates, not its level.
    • The plot of a variable in a model with interactions was described as a slice at one setting of the partners. earthUI’s plot draws the variable’s terms alone, which equals the slice where every interaction term is zero (here, a new sale); plotmo’s degree-one plot is a slice with the other variables at their medians. Both programs are now named.
  • The first segment was said to be “typically the steepest”, which the supplement’s own example contradicts.
  • Friedman (1991) was cited as pages 1–141. The article is pages 1–67; the discussion runs to page 141.
  • Smaller corrections.
    • The method is now said to take the difference of contributions and explain it by the rates.
    • “Every figure” became “every graph of a fitted model”.
    • The builder history is marked as a constructed illustration.
    • “Most appraisers’ regressions” became “regressions in traditional appraisal”.
    • “Premium” was removed.
    • The rate is defined as a change in fitted value.
    • The summary of the four statistical entries was corrected.
    • American spelling is used throughout.

Foundations, Vol. 1 No. 2: The Scheme, and Effects on Value, version 1.1.0 (released 7 October 2026) (10.67330/30ncvp43)

  • England and Wales. The disregard was given as the Pointe Gourde rule. Since the Neighbourhood Planning Act 2017 it is the statutory no-scheme principle (Land Compensation Act 1961, s. 6A). Pointe Gourde remains the rule elsewhere in the Commonwealth.
  • Severance damages. “Severance damages” was given as the US equivalent of injurious affection alone. It covers severance and injurious affection together.
  • France, art. L. 322-2. The three-year public-works clause applies to works in the built-up area where the property lies (text and Prolog addendum).
  • Goldmate.
    • The original determination’s wider frame was presented as a hypothetical. That determination did take the frame, and valued the land as rural.
    • The amounts are now marked as Australian dollars.
    • The Court of Appeal decision is cited, [2024] NSWCA 292.
  • Market movements. Excluding them was called “expropriation without compensation”, which is true only of a rise.
  • The remainder. None of the instruments surveyed was said to “answer” whether the disregard applies to land the owner keeps. None says so expressly.
  • Other corrections.
    • The “never existed” rule was attributed to “most jurisdictions”; the entry’s own survey shows exceptions.
    • The extent of the scheme was said to be fixed by a tribunal. What counts as the scheme is a question of law; how far it extends is a question of fact, which the appraiser must take a position on and a tribunal decides when it is disputed.
    • The digest regions listed did not match the issue’s installments.
    • A doubled period printed after the Recent Court Decisions line.
  • Clarifications:
    • US equivalents given for compulsory acquisition, motorway, tribunal and remitter
    • entries named by title

Foundations, Vol. 1 No. 2: The Purposes of Valuation, version 1.2.0 (released 7 October 2026) (10.67330/g9knsr34)

  • Finland.
    • The Lunastuslaki was cited by its amending act (520/2025). It is Act 603/1977, as amended by 520/2025.
    • Its selection rule names “the reliable” methods, of which the highest-value one is used.
  • France, art. L. 322-2. The table gave one of the three causes of value change the article disregards. All three are now given (text, footnote and Prolog addendum).
  • Italy, DPR 327/2001, art. 37. The article applies to building land only.
  • Statutory purchases between private parties. These were described as “at a formula price” and as “compulsory purchase”. Japan’s is at market value. The sub-branch is now “statutory forced sale between private parties at a statutory price”.
  • Mexico. Jurisprudencia binds lower courts; it is not “authoritative guidance”.
  • Ukraine’s war-damage compensation. It was recorded “outside the eight” purposes, as “a different purpose” and in the same paragraph “not a ninth purpose”. Its objective is to measure the owner’s loss, so it is now placed under purpose 5, damages and compensation (also in the Overview).
  • Qatar. The statute was said to let “the stated purpose” determine the basis. The statute names the basis, just compensation (art. 2), and fixes who values (committees, art. 5), the method (official price lists) and the date (art. 11).
  • The basis of value was said to be chosen by the appraiser where the law prescribes none. It is set for the assignment, by the client’s instructions or the appraiser’s choice, from those the purpose admits.
  • Disputes. Purchase and sale was said to “rarely” produce reported valuation disputes. The claim was removed and replaced with an example: misrepresentation suits, where the valuation measures the buyer’s loss.
  • Prolog addendum.
    • Its own consistency check failed in the published version: a hierarchy level renamed in v1.1 was left under its old name in three facts. Fixed.
    • Its header spoke of a “five-level hierarchy” while the addendum defines three levels. It is a hierarchy of five elements in three levels.
  • Other corrections.
    • “No financial-reporting instrument” exists, was the claim. None entered the quarter’s material.
    • The basis of value for co-ownership matters was said to be usually set by family or succession law. Corporate and partnership law, or the parties’ agreement, also set it.
    • “Three cases” listed items that were not all cases.
    • The United States AVM rule was called “most often cited”.
    • A doubled period printed after the Recent Court Decisions line.
  • Clarifications:
    • American terms (assignment, engagement, appraiser, fee simple, government bonds)
    • mass appraisal named for the tax purpose
    • purpose names made consistent
    • instruments named in full
    • the digest installments named

Foundations, Vol. 1 No. 2: Terminological Divergence, version 1.2.0 (released 7 October 2026) (10.67330/symwrt78)

  • Directional speech. “Every checked statute speaks directionally” was contradicted by the entry’s own quotations of neutral statutory nouns (German Wertänderungen, French changements de valeur, Korean 가격의 변동). The text now says what the statutes do.
  • The general noun. No general abstract noun for the value effect was said to have been found, beside the neutral nouns the entry names (German Wertänderung). No statute uses one as a term of art, and the text, the Prolog addendum and the Overview now say so.
  • France. It was said to have plus-value “alone”, beside a neutral statutory noun. Of the directional pair the Code uses only plus-value. Its disregard uses the neutral changements de valeur and reaches both directions (text, table, Prolog addendum).
  • The scheme. “No surveyed system outside the Commonwealth” was said to have a single word for it. The United States has “the project”, and Japan and Korea have statutory terms (text and abstract).
  • England and Wales. The disregard is now the statutory no-scheme principle (Land Compensation Act 1961, s. 6A), not the Pointe Gourde rule.
  • Hungary. The törvényszék is the regional court, not “the trial-level court”.
  • Kuwait. It is Decree-Law No. 33 of 1964.
  • Smaller corrections.
    • The Russian quotation printed with stray marks.
    • An internal reference pointed to the wrong section.
    • The East Asian statutes’ place relative to the “axis” was misstated.
    • “Eminent domain” was called “United States vocabulary only”.
    • A doubled period printed after the Recent Court Decisions line.
  • Clarifications:
    • US equivalents for betterment, worsenment, hope value, freehold and severance losses
    • official sources cited for the statements on Mexico’s Supreme Court and Hungary’s courts, and § 87(1) of the Flurbereinigungsgesetz quoted with its source
    • entries named by title
    • the digest installments named

Foundations, Vol. 1 No. 2: The Basis of Value and the Approach, version 1.2.0 (released 7 October 2026) (10.67330/jp7pr043)

  • Formal definition. The term “indication of value” did not print (“each approach gives an , and …”).
  • Poland.
    • The replacement-value rule was cited as art. 134(1); it is art. 135(1).
    • The statute was said not to select a method. Art. 152(3) ties market value to the comparative or income approach and replacement value to the cost approach.
    • The Supreme Administrative Court decision III FSK 560/24 is a tax case.
  • Hedonic regression. Refitting on other sales was said to leave the fit statistics “unchanged”. The form of the equation stays the same; the statistics change but may look just as good (text and abstract).
  • The sales comparison approach.
    • It was said to take into account “only market activity”. It draws its inputs almost wholly from actual transactions, with none specific to one party.
    • It was said to “nearly always” indicate market value. It most often does.
  • The worked example.
    • Its two assignments both “asked for market value”, yet differed in basis.
    • The text said the basis separates the two assignments, while the Prolog addendum gave both the same basis. Both now say that what separates them is the definition of the value sought, which the shared name “market value” hides.
  • Other corrections.
    • A sentence implied the basis can be read from the approach when the approach is clear.
    • Subdivision analysis was placed both inside and outside the approaches.
    • The Austrian LBG was said to establish “the basis and no more”.
    • The US fair-market-value definition was attributed to a taxing statute; it is in the Treasury regulations.
    • The court summaries were said to apply the entry’s principle.
    • A doubled period printed after the Recent Court Decisions line.
  • Clarifications:
    • “premium” removed
    • American terminology and units
    • entries named by title
    • the cross-references corrected

Articles, Vol. 1 No. 2: The Languages of Valuation Law, version 1.1.0 (released 7 October 2026) (10.67330/ejz4v620)

  • Abstract. It said that an apparent convergence between distant legal systems turned out to be partly common ancestry. The article finds the reverse: an agreement the department had attributed to common ancestry (Türkiye) turns out to be independent convergence.
  • Source languages. The abstract gave thirty-five. The survey used thirty-six.
  • Türkiye was placed among the jurisdictions that arrived independently at the same rule on delay. The decision cited concerns land taken without proceedings, which is valued at the date the owner sues, not the date of the taking. Türkiye now reaches a related result by a different route.
  • Nigeria and the Democratic Republic of the Congo.
    • They were said to disagree on whether “the holder’s interest” is compensable. Nigeria pays for improvements and a year’s rent; the two disagree on whether the value of the land itself is compensable.
    • The example was offered as showing that shared ancestry does not entail shared doctrine. The two belong to different legal families, so it shows only that shared premises do not entail shared results. The ancestry clause was dropped.
  • The European Union was said to operate the same mechanism as OHADA, one language carrying one body of law. EU law is equally authentic in twenty-four languages.
  • The European Court of Human Rights. Its expropriation case law was said to bind Türkiye. Only its judgments in cases against Türkiye bind it (Article 46).
  • OHADA. Valuation for secured lending, enforcement and realization was said to run on one body of law. It runs largely on common OHADA law (the Uniform Acts on secured transactions and on enforcement), and registration of a mortgage follows each state’s land-registry rules.
  • Japan. The roadside price was said to “yield to appraisal” under provision 6 of the National Tax Agency’s circular. Provision 6 sets the price aside when it is markedly inappropriate, and the value is then fixed on the Commissioner’s instruction, which may rest on an appraisal.
  • Korea. The supplementary method was cited to the Enforcement Decree, Article 49(1) proviso, which concerns market value from sales outside the valuation period. The method is in the Inheritance and Gift Tax Act, Articles 60(3) and 61.
  • Indonesia.
    • The link between hak guna bangunan and the Dutch opstal right was said to be disputed among scholars. The 1960 Basic Agrarian Law’s own conversion provisions turned opstal rights into hak guna bangunan, so the link is statutory; the content of the right is fixed by the 1960 Law.
    • The Law was said to have revoked the colonial code’s land-law provisions, from a text the footnote said was not consulted. It revoked them except those on mortgages (hypotheek). The footnote now cites the official text.
  • Ukrainian and Turkish material. The Ukrainian framework was said to be in English “precisely because” of international attention, and the Turkish rent-determination decisions to be “of far greater methodological interest”. Both claims went beyond the record; the Turkish decisions were identified but not confirmed, and the text now says so.
  • Smaller corrections.
    • Footnote 1 named an “Australia and New Zealand” installment. The installment is Oceania.
    • India and Pakistan were said to have no ancestry “to share”, two sentences after naming the 1894 Act they share. They share none with the civil-law jurisdictions.
    • The OHADA treaty’s French title was misspelled (harmonization for harmonisation).
  • Clarifications for the same reader:
    • the Court of Cassation, the valuation date, de facto expropriation, immovables, formulary tax systems, rosenka, the supplementary method and “jurisprudence” in the civil-law sense explained
    • “fixed-term right to build” for “term building right”
    • a legend for the footnote tags
    • American terms (“engaged”, “state”, “colonial power”)
    • sources added: Zweigert and Kötz; official texts for Indonesia, Egypt and Morocco; the Yargıtay decision’s record

Articles, Vol. 1 No. 2: Comparison of International Valuation Standards, version 1.1.0 (released 7 October 2026) (10.67330/y3ejb015)

  • The roster. It was said to list every jurisdiction above US$100 billion, but it left out China, Russia, Saudi Arabia, the United Arab Emirates, Vietnam, Iran, Egypt and others, listed Hong Kong apart from China, and missed Algeria, Angola and the Democratic Republic of the Congo, which the text names as surveyed. Every jurisdiction above the threshold is now listed, those not surveyed are marked, and China (with Hong Kong) is one entry.
  • GDP figures. They were cited to the IMF but taken from a Wikipedia tabulation. They are now read from the IMF World Economic Outlook, April 2026.
  • Figure 1. Layer 3 (the Red Book and EVS) was unshaded, so the caption classed it as doctrine, regulation and case law. It is now shaded as a standards layer.
  • The Red Book.
    • Its application guidance (the VPGAs) was described as mandatory. It is advisory; only the professional and valuation practice standards are mandatory. The contrast with the United States now rests on RICS publishing both in one document.
    • Its topics were said to include comparable evidence and development property. Comparable evidence is a separate RICS standard, and no VPGA covers development property.
    • It was said to reproduce IVS in full. It incorporates IVS.
  • USPAP was called principally a professional practice standard, which contradicted Figure 1 and the article’s thesis that the United States lacks one. It is a standard of professional conduct and minimum performance.
  • EVS was described as a “regulatory framework”. It is a set of professional standards, drafted with EU lending regulation in view.
  • Australia and New Zealand. The professional bodies were said to require IVS “with effect from 31 January 2025”. That is the date the current edition, IVS 2025, took effect, not the date of adoption.
  • IVS by law. Several jurisdictions were said to bind their valuers to IVS by law. They require IVS, or a national standard built on it, within the domains the tables give.
  • The bank-supervisor pattern was said to hold in “the Caribbean”. The appendix’s own Caribbean table shows the Red Book as the de facto standard there; the pattern holds in Central America and the Dominican Republic.
  • United States, layer by layer. Federal acquisition was called “statutory”. The binding rules are the Uniform Act regulations; supervised lenders are governed by regulation and supervisory guidance, and sales of loans to Fannie Mae and Freddie Mac by contract.
  • Business valuation. The ASA and AICPA standards were said to cover machinery and equipment. Neither does.
  • Level 3 in the United States. The United States was said to lack any universally applicable Level 3 standard. That limit is now stated as this article’s, for the appraisal of an individual property, and the IAAO standards for mass appraisal are listed.
  • EU mortgage directive. It was cited through a UK reference copy labelled “Retained EU Law”. The EUR-Lex text is cited.
  • Clarifications for the same reader:
    • acronyms expanded at first use
    • “basis of value”, “exposure time” and “classification” explained
    • Levels and layers related to each other
    • the German passage reworded
    • UASFLA’s use of USPAP’s Jurisdictional Exception stated
    • American spelling (“License”, “authorized”, “harmonization”)
    • “valuation in divorce”
    • the Yellow Book’s “unverified” mark removed once verified

Articles, Vol. 1 No. 2: Toward Valuation Engineering Protocols, version 1.1.0 (released 7 October 2026) (10.67330/7at33590)

  • Latent characteristics. The appraiser’s model was said to estimate the contributions of latent characteristics, such as condition and quality, as well as measured ones (text and abstract). The model estimates the measured characteristics. The appraiser then assigns the contributions of the latent ones.
  • Scope.
    • The limit to the appraisal of an individual property was presented as the Journal’s. It is this article’s.
    • Mass appraisal and AVMs were excluded together as “these forms of valuation”, and mass appraisal was called “automated or tax valuation”. Mass appraisal is a process within the Journal’s scope; an AVM is a software tool outside it.
    • The rules for both were said to “often contain the most detailed numerical requirements”. The unsupported claim was removed.
  • Protocols were defined differently from the Journal’s Mission. They are now defined in the Mission’s words: specific instructions for executing valuation tasks.
  • The three levels. Minimum sample sizes were placed at the third level, testable requirements, although the minimum numbers of comparables in Spain and Poland belong to the second, documentation. The third level is now testable statistical requirements.
  • Taiwan.
    • The joint test was stated as “the probability that all coefficients are jointly zero may not exceed 5%”. That misreads a significance test. The F test must reject that hypothesis at the 5% level.
    • The caps were said to apply to the adjustments “a model may produce”. They apply to the adjustments made to any comparable.
    • The 2006 revision was said to make the rules “more exacting, not looser”. It raised the required fit but cut the required number of sales and added a waiver of the caps.
  • Germany. Practitioners were said to derive the market data. The expert committees (Gutachterausschüsse) derive them; the appraiser must apply them consistently with the model used.
  • Spain and Brazil. Spain’s rules were presented as applying to all valuations. They govern valuations for mortgage lending and other regulated financial purposes. Brazil’s “federal regulation” is an instruction for valuations of federal property.
  • Modern methods. No system was said to specify a regression technique or to anticipate modern methods, which Brazil (linear regression) and Colombia (machine-learning models) contradict. Now: none mentions MARS, and those that set statistical tests (Brazil and Taiwan) wrote them for linear regression (text and abstract).
  • Fannie Mae was called “the largest secondary-market purchaser”. It is a principal purchaser of US residential mortgage loans.
  • Brazil’s precision grades. The limits were listed in the opposite order to the other lists, so 30 percent read as the lowest grade. They are 50, 40 and 30 percent for Grades I, II and III.
  • Citations.
    • The editor’s works were cited under two name forms, so “Craytor 2026b” pointed to two different works, and Issue 1’s Foundations entries were cited by numbers they do not carry.
    • IVS and the Red Book were cited as 2024 beside the table’s 2025 editions.
    • A quotation from Issue 1’s RCA paper attached “stringent enough” to “a strict protocol”; in the source it qualifies standards and protocols.
    • A section number printed as a page (“p. 9.1.2”).
  • Clarifications for the same reader:
    • statistical terms glossed (regressor, F test, extrapolation, adjusted R², coefficient of variation, hinge functions, generalized cross-validation)
    • Portuguese and Spanish terms explained
    • Poland’s “pairwise method” kept apart from American paired-sales analysis
    • “latent” explained
    • acronyms expanded
    • VE-P1 named
    • the Colombian resolution cited by article and paragraph

Articles, Vol. 1 No. 2: Architectural Style as Evidence, version 1.1.0 (released 7 October 2026) (10.67330/s6sq5e55)

Published with the subtitle “California Bias”, which reads as statistical bias; the subtitle is now “The California Styles”.

  • The central example argued the wrong way round. Encoding “every Craftsman is a bungalow” would make every Craftsman, two-story ones included, assert bungalow. It is the reverse rule that would make every bungalow assert Craftsman and Arts and Crafts.
  • Bungalows. A Greene and Greene “ultimate bungalow” was listed among houses of which “none is a bungalow”. None is the compact bungalow the word usually names, as Figure 4 shows.
  • Printed Prolog output. The printed query output now matches the program, as the article’s statement that every output was checked requires:
    • the two evidence_of results were printed in alphabetical order; the program returns them in the order the facts are listed
    • confusable(tudor, X) was printed with two answers; the program gives four, each pair twice. The query is now printed with setof, which gives the two.
  • Gambrel. A gambrel was said to settle the style “on its own”. It is claimed by Dutch Colonial alone in this ontology; the Shingle Style, whose sketch in the article has a gambrel, also uses it.
  • “No detail sets a style flag”. The grammar’s bare word “gambrel” does set Dutch Colonial. The exception is now stated.
  • Hipped roof. The caption said it identifies Prairie and the Foursquare. It is typical of Prairie and French Normandy (and of the Foursquare form).
  • Hedgerow Houses. The sketch showed a gable. The houses have single shed roofs; the sketch was replaced.
  • Dates. The Larkin House was built 1835–37, not in 1834; the Carson Mansion 1884–86, not in 1884.
  • “All four shapes” referred to figures that show six roof shapes.
  • “The two commonest styles” in a caption. The program rates seven styles as very common; these are two of them.
  • Craftsman and Prairie. Four details were said to “genuinely” not separate the two. Two are shared, and the two that are not point one each way.
  • Ranch table. The row “Cliff May design” read as a claim that Cliff May houses lack the ranch form. It is replaced with the grammar’s own example, a ranch-influenced split-level.
  • Market claims.
    • Eichlers were said to carry a “measurable premium”, with no measurement shown. Now: the trade attaches a higher price to them.
    • The Tuscan idiom was said to meet “now often negative” buyer reaction, with no source. Now: a distinct market period, which a model can test separately.
    • Prairie was said by historians to parallel the Arts and Crafts movement rather than belong to it. That is contested, and the text now says so.
  • Rights in the sketches. They were said to “carry no third-party rights”, an unsourced legal claim. The sentence was removed.
  • Dependencies. The ontology was said to need only the standard list library. Under Scryer it also needs two further standard libraries.
  • Appendix A. Two comments in the frozen code speak of “Arts and Crafts beyond Craftsman” and a “sibling style”, which the article’s own output contradicts. A line in the appendix’s preamble now says so; the code is unchanged.
  • The Tuscan sketch was said to show more than the text describes. It omits the front-facing garage of the tract idiom the text defines.
  • References.
    • Blumenson is the second edition, revised and enlarged, 1981, with the title ending “1600–1945”; it was given as 1990.
    • Hess, The Ranch House, is 2004, not 2005.
    • Sunset Western Ranch Houses (1946) was published in San Francisco, not Menlo Park, by the editorial staff of Sunset in collaboration with Cliff May.
    • The Library of Congress address printed broken; the list now points to the references.
  • Clarifications for the same reader:
    • the subtitle
    • ontology, entailment and massing explained
    • acronyms expanded
    • a key to the relations arrows
    • the weight points explained
    • wording on the Bay tradition, the Monterey Revival and the date “current then”
    • American spelling
    • the “Prolog code (zip)” download named

Articles, Vol. 1 No. 2: The Curious Case of Brazilian Valuation Engineers, version 1.1.0 (released 7 October 2026) (10.67330/410wq635)

  • Brokers and valuation. The abstract said brokers are “not allowed to carry out valuations”, the table recorded them as “excluded”, and the text called the matter “ongoing litigation”. Whether a broker may value was litigated between the two councils; CONFEA’s appeal to the Supreme Federal Court failed in 2012. The table now reads “contested”.
  • The CNAI document. The text said it would not describe the document CNAI brokers issue, then called it “less costly” and “more limited” than a laudo, with no source. Those sentences were removed.
  • The CNAI certificate was presented as what lets brokers give opinions on commercialization. Every broker may already do so under Lei 6.530/1978; the certificate adds no statutory power.
  • Where the reservation comes from. The article said in one place that the reservation of valuation is statutory and in another that the council’s resolution makes it. Lei 5.194/1966 grants the competence and makes it illegal for an unregistered person to perform reserved acts, without listing them; CONFEA Resolution 345/1990 states that valuation is among them and declares valuations by others void. The Prolog program now records the statute’s grant as “permitted” and the reservation as “exclusive”, from the resolution only.
  • Architects were placed under the engineering council on identical terms. Since Lei 12.378/2010 they keep valuation, but under their own council, the CAU.
  • NBR 14653-2, clause 7.3.1, was cited for a rule excluding outside market data. The rule is in clause 8.2.1.4.2; it covers adjustment factors, not market data, and it lets the engineer derive his own factors. The conclusions drawn from it were corrected.
  • NBR 14653-1:2001 was cited as current. It has been superseded by the 2019 edition, and the article now says its citations are to the 2001 edition.
  • Lei 6.530/1978.
    • Article 3 was quoted as “opinar sobre”. It reads “opinar quanto à”.
    • The law was said to create the broker councils. It governs them.
    • Brokerage (intermediação) was recorded as “exclusive” (table and Prolog program). Article 3 grants it (“Compete”) without reserving it; it is now “permitted”.
  • Separate statutes. The terms were said to be “defined in separate statutes”. Neither statute defines them; the statutes assign the acts to different professions.
  • Definitions. Vistoria, perícia and arbitramento were glossed differently from Resolution 345/1990, art. 1, and the abstract described arbitramento as acting as an arbitrator on value. They now follow the resolution: arbitramento is deciding contested technical questions (text, abstract and Prolog program).
  • “One competence.” Brazilian law was said to treat the reserved acts as one competence. That is the Journal’s reading, and is now stated as such.
  • Unsupported claims removed or recast.
    • That rural work goes to agronomists and urban work to engineers “in practice”.
    • That there are not enough engineers, that valuation therefore costs more and takes longer, and that local authorities must wait. These are now stated as COFECI’s position, and no count of engineers was found to test it.
    • That the credential is most common in states with limited engineering capacity.
    • A general statement about why the Supreme Federal Court refuses leave to appeal.
    • That Brazil “has not adopted” a second, durable basis of value. None was found in the Brazilian sources examined.
  • The report requirement was stated as Brazilian law. It comes from the engineering standard and CONFEA’s rules.
  • The Supreme Federal Court case.
    • The case was said to have been “first heard” by the regional federal court under the number given. It came to the Supreme Federal Court on appeal from that court; the number is the first-instance number.
    • It was said to have “now come before” the Court. It did so in 2012.
    • IBAPE was called an intervening party. The record lists it as an interested party.
    • 4292636 was given as the case number. It is the Court’s internal record number.
  • Counts. Engenheiro de avaliações appears twenty times in NBR 14653-2, not eighteen. The count of twenty-seven for fundamentação was removed; the total depends on what is counted.
  • Mortgage lending value was said to rest on criteria required by the EU Capital Requirements Regulation. It is set by Germany’s Mortgage Lending Value Regulation (BelWertV); the CRR (art. 229, as amended in 2024) likewise requires prudently conservative criteria.
  • Glossary.
    • Emphyteusis was called a long-lease institution. It is a perpetual right to use another’s land against a fixed annual ground rent, and none may be created since the 2002 Civil Code.
    • Padrão construtivo was given as field usage. NBR 14653-2 defines it (clause 3.49).
  • Sources. CONFEA Resolutions 218/1973 and 345/1990 were cited only as quoted in a standard, while Appendix B called them verified. They are now cited to their official texts.
  • Smaller corrections.
    • The United States was described as supervising appraisers federally. The states license them under national criteria, with federal oversight of the state programs; brokers are licensed separately and are typically paid through the transaction.
    • A stray word (“Restricted.”) printed in the text.
    • A citation printed after the sentence’s period, and a clause number printed as a page (“pp. cl. 7.3.1”).
  • Clarifications for the same reader:
    • acronyms and Brazilian court terms explained
    • “valuation” for “assessment”
    • “durable” for “sustainable”
    • the 2001 standard named by year
    • the Supreme Federal Court’s judgment cited in full
    • the other article named by title
    • American spelling

Comparative Practice, Vol. 1 No. 2: Who May Sign a Value Opinion: An Ontology of Apex Signing Authority in 134 Jurisdictions, version 1.1.0 (released 7 October 2026) (10.67330/grbrbw81)

  • Tables 8 and 9 in the first printing. Table 8 lost all nineteen of its unverified-fact marks (daggers), which printed as “[native script]”. Table 9 lost native-script words in four rows (China, Thailand twice, and Iran). The marks are restored and the words romanized. The Taiwan entry kaigyo, the Japanese reading, is now kaiye.
  • Counts.
    • “The 115 regimes that impose an education requirement” are the 115 whose requirement is known; two of them have none (text and abstract).
    • The American type was called the second rarest. With two regimes against five for “state valuers only”, it is the rarest.
    • The list of fifteen other split regimes named the United Kingdom, which is coded market-only. The fifteenth is Puerto Rico.
    • Vietnam was missing from the regimes at 7.
    • The opening paragraph named six arrangements but spoke of “five ontologies”.
  • Puerto Rico. The United States was said to be alone in its combination. Puerto Rico shares its license and examination (text and abstract).
  • The United States’ rigorous examination was said to be unique as a private credential. Canada’s and the United Kingdom’s top examinations are also private; what is unique is a low-rigor state license examination beneath a rigorous private one (abstract).
  • Statutory footing. The managing designations (MAI, SRA) were said to have no statutory recognition. A federal tax statute recognizes appraisal designations generically (26 U.S.C. § 170(f)(11)(E)); they have no footing in licensing law, and courts weigh them as credentials.
  • Type G was said to be constituted by the client. The license is issued by the states under Title XI; Figure 4 is redrawn.
  • The Certified General was said to presuppose the Certified Residential (text and ontology data). The criteria allow direct entry from Trainee.
  • Clustering.
    • The United States was said to share a cluster with Canada, the United Kingdom and Ireland at every cut from 2 to 12. It does so with all three from 2 to 10, and with Canada alone through 12.
    • “In no run” does more than half of the regimes fall in one cluster. That holds for the three primary runs; one run in Table 7 has 85 percent.
    • A pair was said to be split “in every run under the primary encoding”. That holds in every run in which both appear.
    • Two clusters were said to “both require” a specific degree. Both are centered on one.
    • Collateral demand and the data regime were called “the strongest correlates” of a profession’s shape. Only collateral demand was analyzed.
  • Lending.
    • The mortgage-market family was described as one in which lending demands an appraisal for every loan. It requires a per-loan appraisal as a matter of routine; the United States, for one, has a $400,000 exemption.
    • Where it does, an L1 tier was said to exist. It tends to exist.
    • Germany, coded as moderate collateral demand, was grouped with regimes where it is absent.
    • German lenders were said to regard the HypZert certificate as mandatory under the Mortgage Lending Value Regulation. It is the usual way they meet it; § 6 presumes the competence of state-appointed or accredited valuers.
  • Jurisdictions.
    • The standards article was placed in the previous issue. It is in this one.
    • The Netherlands’ specialty was called a legal axis. It is the axis of a private professional register.
    • The German publicly appointed expert was described as holding a chamber board membership. Appointment follows an examination before the chamber’s expert committee.
    • Jordan was listed among the Gulf regulators.
    • The Cayman Islands and Bermuda were said to share one regulatory regime. They share a pattern; the eastern Caribbean states share a currency union and a bank supervisor.
    • Every jurisdiction was said to have its own ontology of appraisers. Every jurisdiction that regulates who may value has one (abstract).
  • RICS membership was said to have no legal significance yet to be the gateway to nearly every institutional assignment. A few statutes name it (Charities Act 2011, s. 119(3)), and it is widely treated as the gateway to institutional assignments.
  • Brazil. The engineering degree was presented as the only gate. Registered real estate brokers issue market-value opinions under a COFECI resolution (Res. 1.066/2007).
  • Smaller corrections.
    • The de jure annotation was defined in circular terms. It records the level at which the law permits the holder to sign.
    • “Lending is underwritten on” now reads “largely on”.
    • The software versions printed did not match the environment that reproduces the results.
  • Clarifications for the same reader:
    • acronyms and designations expanded
    • statistical terms glossed (silhouette width, cophenetic correlation, Adjusted Rand Index, Stress-1, Gower distance, multidimensional scaling)
    • “jurisdictions” for “countries”
    • the L3/L4 labels explained
    • the Yellow Book named
    • a note added on the Appraisal Qualifications Board’s Third Exposure Draft of 24 Sep 2026

Editors’ Message, Vol. 1 No. 2, version 1.1.0 (released 7 October 2026) (10.67330/krpyy115)

The Editors’ Message is the editor’s own text. Only its factual corrections are listed.

  • The fifth article. The message announced five articles but described four. The Curious Case of Brazilian Valuation Engineers is now described.
  • New departments. The message spoke of one new department, Protocols. There are two: Comparative Practice and Protocols (text and abstracts).
  • Foundations. The abstracts counted four new Foundations entries. There are eight: four on legal concepts and four on the statistics behind appraisal models, which the text now names.
  • Eighty-four jurisdictions. The count holds only with the jurisdictions of the Caribbean installment counted together as one, which the message did not say.
  • Languages. The survey was said to use thirty-five source languages and machine translation for thirty-four. The figures are thirty-six and thirty-five.
  • Article 3 was summarized as saying that standards “specify the results of a valuation”. It says they set out ethics, scope of work, the recognized approaches and reporting.
  • India. “No matter what adjustments are made” was attributed to the Supreme Court. Those words were the survey’s gloss. The Court reached a similar result under the statute, which requires sales of a similar type of land. The judgment is now cited as reproduced by LiveLaw; the Court’s own copy was not found.
  • Parameter Identification was summarized in its earlier form. A relevant characteristic must vary apart from the others, and the remedy is different data or a narrower claim, not simply better data.
  • Poland. Replacement value was said to apply only where property of the kind does not trade. It applies where property of that kind, use or designation does not trade, or where another statute so requires (art. 150(3)).
  • Nigeria. Replacement cost less depreciation was presented as the rule for the whole taking. It is the rule for buildings and other improvements.
  • Sweden and Finland were said to require market value plus twenty-five percent. That is the general rule.
  • Reconciliation by statute was said to be “now increasingly” the case. It is the case in a number of jurisdictions.
  • Sources. The survey was said to reference each jurisdiction’s primary sources and to “obtain and check the main law”. Some footnotes rest on secondary sources. It now says primary sources where they could be reached, and the main law in the official text or a reliable reproduction.
  • “Two cases” in the Americas. Quebec’s change is a statute. They are now two developments (text and abstract).
  • Central Asia. “The Central Asian states” included Turkmenistan, which the same sentence had already listed for another reason. Now: the other Central Asian states.
  • Clarifications: wording made clearer in several places at the editor’s direction; sources added or marked (South Africa’s Expropriation Act, Ontario’s Expropriations Act s. 13(2), the Yargıtay decision as reported in the survey); “license” spelled in the American way.