Mobile Edition
Print
Get the Mag
Weekly UpdatesThe Seventh Circuit's Guardian Pipeline decision encourages wider use of multiple regression analysis in valuation proceedings in place of the longstanding matched pair comparison canon. This article reviews applicable evidentiary rules, providing details on cases appraisers often hear cited but may not know well, such as Daubert and Kumho, and on Federal Rule of Evidence 702. A well-designed regression model would seem to meet all requirements for admissibility imposed by federal rules and case law, and by state rules favoring systematic analysis. Yet other techniques also might meet those requirements.
**********
On May 8, 2008 the U.S. Court of Appeals for the Seventh Circuit decided the easement condemnation case of Guardian Pipeline, L.L.C. v. 950.80 Acres of Land, et al. (1) In 2001 Guardian began construction of a 141-mile underground natural gas transmission pipeline, certificated by the Federal Energy Regulatory Commission, from Joliet, Illinois, to Ixonia, Wisconsin. The line's Illinois section traversed hundreds of tracts in four northern counties. Guardian negotiated for much of the needed right-of-way, a 50-foot wide permanent easement and adjacent temporary construction easements. Approximately eighty landowners rejected the offers, however, and Guardian initiated condemnation action under the Natural Gas Act in U.S. District Court in Chicago. In 2002 that federal trial court entered an order of condemnation, allowing pipeline construction to go forward, while leaving for later determination the just compensation that Guardian would have to pay for the easement interests.
As allowed by the Federal Rules of Civil Procedure, the court subsequently appointed a commission, consisting of three experienced Illinois lawyers, to conduct evidentiary hearings and recommend compensation for each landowner. Hearings were held, conducted in accordance with the Federal Rules of Evidence (FRE) per the court's instructions, over the course of seventy-six days between June 2003 and July 2005. Both Guardian and the landowners presented extensive appraisal testimony concerning the pipeline easement's impact on market values of the contested tracts. The evidentiary record ultimately consisted of 12,500 pages of testimony and hundreds of exhibits.
In July 2006, the commission submitted its 277-page report to the court, recommending aggregate awards of approximately $2.5 million, as contrasted with landowner demands totaling nearly $20 million. After a comprehensive review of the commission's report and the evidentiary record, the court entered a May 2007 judgment in the commission's recommended amounts. Disappointed at not having received more, some landowners appealed. In its May 2008 opinion, written by Chief Judge Frank Easterbrook, the federal appellate court rejected the landowners' various arguments and affirmed, in all respects, the district court's ruling.
An Unexpected Recommendation
Had it stopped there, the Seventh Circuit's opinion would be unremarkable, a routine affirmation of a district court judgment based on established legal principles. (2) But Judge Easterbrook went further and, in dicta, returned to a question he first had raised at oral argument two months earlier, wondering why the parties had expended extensive resources providing the commission with paired sales analyses by real estate appraisers to support their respective positions concerning the pipeline easement's impact on market value. As he put it in two paragraphs near the end of the opinion:
These comments surely came as a surprise to many. Throughout the hearings and during the extensive preceding research and preparations, the district court, the commission, the various parties, and their respective appraisal witnesses were guided by methodology spelled out in the Uniform Appraisal Standards for Federal Land Acquisitions, which explicitly embraces use of the sales comparison approach to value. (3) Judge Easterbrook, a disciple of the Chicago school of law and economics, suggests adopting a more efficient paradigm for valuation testimony in future cases of this nature. The two paragraphs quoted from the Guardian opinion thus contain potentially important implications for appraisers.
Judge Easterbrook's proposed new approach-questioning the use of typical adjustment factors derived from matched pair comparisons, and embracing multiple regression analysis as a faster, more accurate alternative--is so sensible, and long overdue, that it seems likely, eventually, to assume an important role in adjudicating such cases. (4) Appraisers, in turn, will need to become proficient in multiple regression methodology and, when testifying as expert witnesses in cases involving easement takings or other adjudicated valuation issues, be able to explain their analysis and data in a manner that is admissible in evidence.
The Essence of Multiple Regression Analysis
The term regression relates to the idea that when large amounts of data are examined, statistical measures of various characteristics should tend to regress to the true parameter values for the underlying populations. (5) The first uses of regression analysis in estimating real estate values involved academic studies of Midwestern agricultural land, conducted by C. G. Haas in 1922 and H. A. Wallace in 1926. (6) When applied to real estate valuation, regression analysis is often called hedonics or hedonic regression. Auto industry analyst A. T. Court coined the term hedonic in the late 1930s, borrowing from a psychological term suggesting pleasant states of mind, to convey the idea that an item's value is associated with features that give its users pleasure or utility. Outside the academic realm, pioneering users of regression analysis in real estate valuation over the ensuing decades frequently were local property tax assessment officials who had to conduct mass appraisals. Early users needed large-budget computer resources that were inaccessible to most private practitioners. But the needed software now is inexpensive (a serviceable version is provided as a Microsoft Excel feature), so cost should not impede appraisers' use of regression analysis today.
Indeed, regression analysis is hardly new to The Appraisal Journal's readers. In a 2005 article, Myllynen explains the use of simple regression analysis in estimating each added site's contribution to the value of a mobile home park in an illustrated market area. (7) For those unfamiliar with regression modeling, simple regression relates the magnitude of a dependent variable (e.g., property value) to just one independent or explanatory variable (e.g., square footage). Another 2005 article, written by Thomas O. Jackson, explains the use of multiple regression analysis in estimating environmental impacts on property value. (8) Multiple regression relates a dependent variable's magnitude to measures for a group of explanatory variables working together (e.g., zoning, size and age of improvements, proximity to schools or highway access).
In the most straightforward type of multiple regression model for real estate valuation,
Y = [alpha] + [[beta].sub.1] [X.sub.1] + [[beta].sub.2] [X.sub.2] + ... [[beta].sub.n] [X.sub.n] + [epsilon],
the measured property rights' value (Y) is theorized to bear a linear relationship to each feature (explanatory variables [X.sub.1] through [X.sub.n]) believed to be associated meaningfully with that value. (Value does not always increase in a straight-line manner with changes in possible explanatory features. However, sometimes simple adjustments allow the linearity assumption to hold; for example, the logarithm of value might change in a reasonably linear manner with a standard-depth parcel's frontage along a major road.) (9) The computational process generates (a lump-sum base or intercept magnitude) and coefficients [[beta].sub.1] through [[beta].sub.n], the identified explanatory variables' contributions to value, as derived from the data. It also measures the error term e that accounts for value impacts linked to conditions other than the included explanatory variables. A larger error term can indicate that important value determinants were not included as explanatory variables.
Real estate appraisers can productively apply regression analysis in estimating the total market value of specified property rights (a fee or easement value, for example). Appraisers also may apply regression analysis in estimating the magnitudes of adjustment factors for use in a traditional, grid-based sales comparison approach (corresponding to the [beta] values). (10) The Rubinfeld "Reference Guide," to which the Guardian ruling refers, (11) provides a concise summary of the theory underlying regression analysis and, in the appendix, a numerical example (involving gender discrimination in pay). But it is not likely to provide an appraiser new to the topic with sufficient grounding to apply multiple regression analysis in practice; so, other general references on statistics and econometrics, or other sources that focus specifically on regression analysis, may be needed. (12)
Current Standards for Admissibility of Expert Testimony
The appraisers involved in the Guardian case testified as expert witnesses. Thus, as the Seventh Circuit noted, their testimony's admissibility was governed by Federal Rule of Evidence 702. (13) Since that court found no abuse of discretion in the way the commission (and the district court judge) handled the admission of expert testimony, it rejected the landowners' challenges to various aspects of the Guardian appraisers' testimony. (14) It will be incumbent on appraisal experts who rely on multiple regression analysis instead of matched pairs to establish, in future cases, that their testimony satisfies Rule 702's requirements.
Federal Rule of Evidence 702 provides as follows:
FEED