Free Response to Motion - District Court of Federal Claims - federal

Case 1:01-cv-00201-VJW

Document 219-3

Filed 10/05/2006

Page 1 of 28

IN THE UNITED STATES COURT OF FEDERAL CLAIMS

CAROL AND ROBERT TESTWUIDE, et al., ) ) Plaintiffs, ) ) v. ) ) THE UNITED STATES OF AMERICA, ) ) Defendant. ) _______________________________________

No.: 01-201L

Judge Victor J. Wolski

PLAINTIFFS' REBUTTAL TO THE EXPERT REPORT OF DR. DAVID DALE-JOHNSON (Analyses of the Impact of the Realignment on House Values)

JON P. NELSON, Ph.D.

December 12, 2005

Case 1:01-cv-00201-VJW

Document 219-3

Filed 10/05/2006

Page 2 of 28

TABLE OF CONTENTS

Page I. Introduction ............................................................................................. 1

II.

Repeat Sales Methodology ...................................................................... 2 A B. C. The Repeat Sales Model ............................................................... 2 Aggregation Bias and Sample Selection Bias............................... 4 Sampling Methods Used by Dale-Johnson .................................. 5

III.

Noise Measurements and Alternative Repeat Sale Estimates ........... 7 A. B. C. Noise Data Used by Dale-Johnson ............................................... 7 Differences Between BASE 2000 DNL and ARS2 DNL ............ 8 Repeat Sale Regression Results for ARS2 DNL ........................ 9

IV.

Event Study Method and Results ........................................................... 11 A. B. Methodology Used by Dale-Johnson ........................................... 11 Additional Regression Results for Price Indices .......................... 13

V.

Effects (if any) of the Realignment on Housing Demand .................... 13

VI. .

Conclusions .............................................................................................. 16

Tables 1 - 10 ............................................................................................. 17

Case 1:01-cv-00201-VJW

Document 219-3

Filed 10/05/2006

Page 3 of 28

I. Introduction 1. This document presents my evaluation of the expert report by Dr. David Dale-Johnson on the impact on property values of the realignment of 10 squadrons of F-18 C/D aircraft to NAS Oceana and NALF Fentress (hereafter Dale-Johnson report). Dr. Dale-Johnson presents two complementary analyses of the realignment. First, a "repeat-sales analysis" of a sample of residential properties that were sold before-and-after July of 1999. This analysis uses a "refined" sample of 6029 properties that sold twice during the period 1995-2003, and subsamples of these properties are used as the basis for sensitivity analyses. Second, an "event study analysis" of housing price indices constructed for the 36 quarters during 1995-2003 and various subsamples of this longer time period. Empirical results from a modified repeat sales analysis are used to construct the event study price indices. Hence, the event study results hinge on the validity of the repeat sales model, sample, data, and econometric analysis. 2. Based on these analyses, Dr. Dale-Johnson concludes that the realignment of 156 F-18 aircraft did not have a (net) negative effect on residential property values in the vicinity of NAS Oceana. With respect to the results of the repeat sales analysis he states that: The statistical model demonstrates that there is no statistically significant relationship between the changes in noise for houses due to the realignment and changes in property values from before to after the realignment. In fact, the sign of the noise variable is positive, which is counter to the plaintiffs' hypothesis. The model demonstrates that houses that experienced large increases in noise due to the realignment did not appreciate at a different rate than houses that experienced slight increases in noise (Dale-Johnson report, p. 16). and with respect to the event study analysis he states that: The regression results show a positive, not statistically significant coefficient on the event dummy variable. Therefore, this is evidence that the realignment did not have a negative impact on house values around NAS Oceana (Dale-Johnson report, p. 19). 3. These results are contrary to a large body of accumulated evidence on the empirical relationship between aircraft noise exposure in the vicinity of major airports and residential property values. This evidence was summarized in the form of a meta-analysis in my Expert Report of September 2005 (hereafter Nelson report). Thus, in terms of the econometric evidence, Dr. Dale-Johnson's results are clearly an outlier. His explanation for his null findings rests on the possible positive impact on housing demand from the transfer of U.S. Navy personnel to Virginia Beach: The realignment resulted in the movement of an estimated 8,300 people from Florida to the Virginia Beach area, including approximately 3,700 military and civilian personnel. In addition, the realignment was expected to add roughly $308 million annually to the regional economy and increase taxes by $7.8 million annually. Such increases in population and spending would be expected to result in an increase in the demand for housing which in turn would be expected to cause house values to increase ­ an increase as a result of the realignment (Dale-Johnson report, p. 4).

Case 1:01-cv-00201-VJW

Document 219-3 2

Filed 10/05/2006

Page 4 of 28

4. In order to evaluate these claims, I will first develop the methodology used to conduct a repeat sales analysis, including an examination of the sampling and measurement procedures used by Dr. DaleJohnson. The methodology of a repeat sales analysis is subject to a number of well-known problems, all of which are largely ignored in his report. Second, building on my criticisms of his work, I present alternative regression estimates for the repeat sales model that demonstrate that the "Change in Noise" coefficient is negative and statistically significant. Furthermore, these negative effects agree with the noise damage estimates presented in my Expert Report (Nelson report, p. 29). Thus, using the same basic methodology and data available to Dr. Dale-Johnson, I demonstrate that plaintiffs' damage formula is statistically sound and correct. Third, I examine selected aspects of Dr. Dale-Johnson's event study analysis. Fourth, I evaluate his hypothesis for his null findings, i.e., the transfer of U.S. Navy personnel to the Virginia Beach area and the possible effects (if any) of this transfer on housing demand and values.

II. Repeat Sales Methodology The repeat sales model is commonly used to construct real estate price indices. It is subject to several methodological and econometric problems that require caution in its application. Two widely discussed problems are aggregation bias and sample selection bias. A. The Repeat Sales Model 5. As an econometric technique, the repeat sales method was first developed in 1963, and it has been widely used in recent years to develop housing price indices. Consequently, the econometric problems or difficulties associated with this technique are well known. The repeat-sales model as applied to environmental effects or damages was pioneered in 1982 by Dr. Raymond Palmquist.1 Assume a significant environmental "event" that occurs at a specific date in time and which has different measurable physical effects on residential properties. In the case of NAS Oceana, the transfer of 10 squadrons of F/A-18 aircraft occurred during December 1998 to September 1999, with 2 squadrons arriving in December 1998, 6 squadrons in July 1999, and 2 squadrons in late September 1999. Thus, in Dr. Dale-Johnson's analysis, the "event date" or "base quarter" is the second quarter of 1999 (April-June 1999), which is the quarter just before the increase in noise levels. The physical effects of this event refer to different changes in noise exposure, e.g., +0, 5, 10, or 15 dB changes measured relative to the

R.B. Palmquist, "Measuring environmental effects on property values without hedonic regressions," Journal of Urban Economics, 11 (1982): 333-47. I am aware of only one other study that uses the repeat sales method for aircraft noise; see N.N. Knickerbocker, "Aircraft Noise and Property Values," Unpub. Ph.D. dissertation, University of Maryland, 1991. This study of National Airport finds negative effects that agree generally with other studies for National Airport and other airports, i.e, the noise discount is -1.29% per dB (p. 149).

1

Case 1:01-cv-00201-VJW

Document 219-3 3

Filed 10/05/2006

Page 5 of 28

pre-event exposure level. The important reasons for using a sample of repeat sales of residential properties are: (1) it might eliminate the necessity to collect numerous housing features required for the hedonic price model; (2) it focuses the analysis on the change in environmental quality and damages; and (3) it can be used to construct housing price indices for additional analysis of the event. 6. The repeat sales model is based on the notion that there exists a set of residential properties that were sold before and after the event, and which did not change except for the differences in noise exposure. Consider two residential properties, both of which were sold for $100,000 in 1998 and which sold again in 2000. Assume that neither property underwent any structural, environmental, or neighborhood quality changes during 1998-2000, except that Property A experienced a 10 dB increase in noise exposure in July 1999 and Property B did not. If Property B was sold for $120,000 in 2000, the appreciation was $20,000 or 20%. According to plaintiffs' damage formula, Property A should appreciate less in value. Suppose that Property A sold for $110,000 in 2000 or a 10% appreciation. Hence, the damage due to the increase in noise was $10,000 or -1% per dB change in noise exposure. 7. A more technical description of the repeat sales model helps reveal some of the assumptions that underlie the model and analysis. The starting point is the standard semi-log hedonic price function ln(Pit) = "t + $'Zi + (Nit + *Ait + uit

(1)

where ln(Pit) is the log of the sales price of house i at time t; Zi is a vector of housing characteristics that do not vary with time; Nit is the noise exposure level at time t; Ait is the age of the house; and uit is the error term (aka the "residual"). The error term captures the effects of omitted variables, measurement errors, and purely random or stochastic aspects of the data. The depreciation rate due to age is given by

*. The noise damage index or hedonic price of noise in percentage terms is given by ( × 100. The
expected sign of ( is negative. 8. For the repeat sales model to be appropriate, there must be a sale of each property before and after the event date. Suppose the earlier sale of the property occurred at time t and the second (or subsequent) sale occurred at time s. The first sale occurs before the event and the second sale occurs after the event. For these houses, the price-relative is given by (Pis / Pit). Using the log transformation, the rate of appreciation function for the repeat sales model is given by ln Pis - ln Pit = "s - "t + $'(Zi - Zi) + ((Nis - Nit) + *(Ais - Ait) + (uis - uit)

(2)

Case 1:01-cv-00201-VJW

Document 219-3 4

Filed 10/05/2006

Page 6 of 28

Using an assumption due to Dr. Palmquist, houses are assumed to depreciate geometrically with age. The age variable is eliminated by a uniform adjustment of the price-relative using an independent estimate of the depreciation rate. The values of the constant terms, "s and "t , form a house price index. These values are the estimated coefficients for dummy variables indicating the quarter(s) of sale.2

B. Aggregation Bias and Sample Selection Bias 9. In the standard hedonic model, $ is a vector of parameters to be estimated. In the repeat sales model, this is not necessary because the Z variables are constant over time. Hence, the repeat sales model requires a sample of properties for which the structural, environmental, and neighborhood features of the houses are unchanged between time t and time s. This condition is referred to as the constant quality assumption and violation of this assumption is called aggregation bias. 10. While the repeat sales model can eliminate the necessity to collect data on housing characteristics required for the hedonic price model, it does this under stringent conditions. It is clear from equation (2) that the sample of repeat sales must not undergo any changes whatsoever ­ no housing improvements or disrepair other than the uniform effects of age ­ except for the change in noise level. There also must be no changes in environmental or neighborhood quality variables that vary across properties in the sample, i.e., any important changes in neighborhood quality must be area-wide changes that affect all properties equally. Further, there must be no change in the underlying mathematical relationship for the hedonic price function, i.e., the $ parameters must not change over time. 11. The constancy assumption is very difficult to meet in practice, and also leads to features of the repeat sales sample that impose subtle constraints on the data. It is very likely that the sample of repeat sales is not a random draw from all properties or a random draw of those properties that were sold on the market, including properties sold once during a given time period. Hence, past studies using repeat sales have found that properties that sell twice are likely to be different from the rest of the housing stock.3 This is referred to as sample selection bias. In fact, there are two possible selection biases in repeat sale studies. First, the repeat sale sample is not a random draw from the existing stock of properties (i.e., not all properties were sold in the time period). Second, all sales are not repeat sales, and

Because the repeat sales model is derived from the hedonic price model, it does not avoid various econometric problems that have been the subject of recent discussion for the latter model, including spatial autocorrelation and spatial autoregression. Housing is a unique commodity because it trades infrequently. This is in contrast to other commodities, such as stocks and bonds, that might trade each business day. As a result, transactions are sparse relative to the outstanding stock of houses. As a consequence, external information from other markets, such as that from a metaanalysis, are informative about the effects of environmental disruptions, including the realignment at NAS Oceana.
3

2

Case 1:01-cv-00201-VJW

Document 219-3 5

Filed 10/05/2006

Page 7 of 28

a repeat sale sample is biased toward those properties that were re-sold during a specific time period. Hence, properties that sold more than once may not be representative of all residential properties in the market area. Increasing the length of the sample period might reduce selection bias but this aggravates the problem of heteroskedasticity ­ the price change variability is greater the longer the period between transactions. Further, increasing the length of the sample period increases the likelihood of aggregation bias due to changes in the underlying hedonic price function or unobserved changes in housing quality. 12. The effects of aggregation bias and sample selection bias have been widely discussed in the real estate literature, including a special issue of the Journal of Real Estate Finance and Economics (January 1997). Because Dr. Dale-Johnson was an Editor for this journal and has published several papers in the journal, he was obviously in a good position to take account of the special issue papers. However, these well-recognized econometric problems are not discussed explicitly in his report.

C. Sampling Methods Used by Dale-Johnson 13. Repeat-sale aggregation bias results from omission of characteristics that contribute to value (including unobserved characteristics) and parameter instability over time. The effect of this bias on the "Change in Noise" coefficient is difficult to establish without greater knowledge of the omitted characteristics and parameter instability. Only the application of other methodologies or more complete data would resolve this issue. Previous studies of this problem have established that the bias can be substantial for the derived price indices.4 Hence, there is reason to suppose that the event study conducted by Dr. Dale-Johnson is biased statistically. 14. Dr. Dale-Johnson does attempt to ensure that the houses in his sample have not been "improved" between the time of first sale and the second sale. Specifically, he starts with samples of 154,628 "housing units," 141,642 residential properties, and 70,520 properties that were sold at least once between 1/1/1995 and 12/31/2003 (Dale-Johnson report, Exhibit 1). From the latter sample, he omits 4,533 properties that were missing basic information on the living area and 57,085 properties that sold only once or were non-arms-length sales. This leaves him with a raw sample of repeat-sale properties with a sample size of 8,902 (or about 6.3% of the residential housing stock). This raw sample is further reduced by 2,420 properties with recorded improvements and 453 properties with other nonarms-length sales (sales to relatives, banks, etc.). This leaves a final repeat-sale sample of 6,029 properties. The final sample is only 4.3% of the residential stock and 8.5% of all property sales between

See J. Dombrow, et al., "Aggregation bias in repeat-sales indices," Journal of Real Estate Finance and Economics, 14 (1997): 75-88; and J.E. Zabel, "Controlling for quality in house price indices," Journal of Real Estate Finance and Economics, 19 (1999): 223-41;

4

Case 1:01-cv-00201-VJW

Document 219-3 6

Filed 10/05/2006

Page 8 of 28

1995 and 2003. While this screening procedure is necessary to ensure a sample of constant-quality repeat sales, it produces a non-random or selected sample of the Virginia Beach housing stock and housing sales. Further, the screening procedures do not ensure parameter stability or ensure that neighborhood and environmental features are constant, other than the measured change of noise levels. 15. The effect of sample selection bias has been investigated in a number of repeat sale studies. These studies indicate that repeat-sales data are biased toward older and smaller properties. Furthermore, repeat-sale properties tend to appreciate more than properties that sold only once during a specific time period. This difference has been attributed to unobserved or unrecorded improvements to the repeat sale properties between the dates of sale:

(1) Clapp, Nanda, and Ross (1991) found that houses selling repeatedly were older and smaller ("starter homes") than houses that sold only once during a ten to twenty year period.5 (2) Meese and Wallace (1997) found that repeat sale houses were smaller and in worse condition ("fixer-uppers") than the average for single-sale properties.6 (3) Goetzmann and Spiegel (1997) found that repeat sale homes in San Francisco tended to be sold more often by higher-income families living in racially-mixed neighborhoods.7 (4) Case, Pollakowski, and Wachter (1997) found that repeat sale homes were smaller, but also tended to appreciate at a higher rate.8 They believe that this is due to property improvements that are not adequately reflected in the available data. Further, homes that have depreciated in value are less likely to be sold. Both conditions lead to sample selection bias. (5) Gatzlaff and Haurin (1997) found that repeat-sale price indexes tended to be biased upward during periods of economic growth and biased downward during periods of economic weakness.9 This bias is with respect to never-sold homes or less-frequently sold homes.
5

J.M. Clapp, C. Giaccotto, and D. Tirtirglu, "Housing price indices based on all transactions compared to repeat subsamples," AREUEA Journal, 19 (1991): 270-85. R.A. Meese and N.E. Wallace, "The construction of residential housing price indices: A comparison of repeat-sales, hedonic regression, and hybrid approaches," Journal of Real Estate Finance and Economics, 14 (1997): 51-73. W.N. Goetzmann and M. Spiegel, "A spatial model of housing returns and neighborhood substitutability," Journal of Real Estate Finance and Economics, 14 (1997): 11-31. B. Case, H.O. Pollakowski, and S.M. Wachter, "Frequency of transaction and house price modelling," Journal of Real Estate Finance and Economics,14 (1997): 173-87. D.H. Gatzlaff and D.R. Haurin, "Sample selection bias and repeat-sales index estimates," Journal of Real Estate Finance and Economics, 14 (1997): 33-50.
9 8 7 6

Case 1:01-cv-00201-VJW

Document 219-3 7

Filed 10/05/2006

Page 9 of 28

(6) McMillen and Thorsnes (2005) found that the average price of houses in Chicago rose an average of 7.5% per year during 1993-2002, but repeat sale houses increased by 8.5% annually.10 They attribute this difference to unobserved remodeling and renovation. 16. Summarizing the discussion of the model, a repeat sales study requires a sample of properties that were sold at least twice during a given time period and which, aside from normal "wearand-tear," have not changed in quality. Any unobserved improvements (or major disrepairs) will bias the results, such as improvements done by the first owner or those that go unrecorded at an assessor's office, such as landscaping. The constant quality assumption also applies to other environmental features of the residential area. For example, if traffic noise or air pollution changed in a non-uniform manner across the study area, this will bias the results. There also must be no non-uniform neighborhood quality changes. For example, if one neighborhood had an increase in crime rates, this also can bias the results. Finally, the hedonic price function must not change over time, so that the underlying marginal prices are constant over time. The screening procedures used by Dr. Dale-Johnson are insufficient to ensure that these conditions are met. 17. Using repeat sales and screening the data to delete "improved" properties results in sample selection bias. A sample selected according to various screening procedures is not equivalent to random sampling and a regression analysis based on a non-random sample does not as a general rule produce a true description of the population regardless of the sample size. Previous studies demonstrate that repeat sale samples are biased toward smaller and older properties, and these properties tend to appreciate more than properties that sold only once during a given time period. Due to these problems, the two regression studies conducted by Dr. Dale-Johnson should not be taken as an accurate description of the impact of the realignment on all residential properties in the vicinity of NAS Oceana.

III. Noise Measurements and Alternative Repeat Sale Estimates The post-realignment noise measurements used by Dr. Dale-Johnson ­ BASE 2000 DNL levels ­ have not been universally adopted for analyses of the realignment. Due to errors of measurement, this feature of his study alone could bias his regression results toward a null outcome. Regressions using ARS2 noise exposure data yield significantly negative noise coefficients that agree with my meta-analysis and plaintiffs' damage formula. A. Noise Data Used by Dale-Johnson 18. The repeat sales analysis requires data on sales prices before and after 1999Q2; dates of the

D.P. McMillen and P. Thorsnes, "Housing renovations and the quantile repeat sales price index," Unpub. paper, University of Illinois at Chicago, August 2005.

10

Case 1:01-cv-00201-VJW

Document 219-3 8

Filed 10/05/2006

Page 10 of 28

two sales; and location of the property. The location information allows the measurement of noise levels on a before-after basis for each repeat sale property. For the pre-realignment sales (i.e., 1995Q1 to 1999Q2), Dr. Dale-Johnson uses the BASE 1997 DNL data produced by Wyle Labs (Dale-Johnson report, p. 15). I also employ these data. For the post-realignment sales (i.e., 1999Q3 to 2003Q4), two noise exposure measurements are available, ARS2 DNL or BASE 2000 DNL. Dr. Dale-Johnson states that it is his understanding that the BASE 2000 data are ". . . the most accurate reflection of the noise environment after the realignment" (Dale-Johnson report, p. 15). 19. This statement ignores the fact that several post-realignment studies continue to rely on the ARS2 data for noise exposure levels after July 1999:

(1) As part of the U.S. Navy's AICUZ program, the web-site for NAS Oceana uses ARS2 noise contours for purposes of depicting noise exposure levels after the realignment: http://www.nasoceana.navy.mil/ http://www.nasoceana.navy.mil/graphics/NOISE%20CONTOURS%20PAGE%201.pdf (2) The NAS Oceana web-site is linked to a GIS eMapping facility operated by the City of Virginia Beach. This site maps property locations by address (or GPIN) according to ARS2 noise contours and contains links to the ARS2 maps used in the Joint Land Use Study: http://www.vbgov.com/e-gov/emapping/ http://www.vbgov.com/e-gov/emapping/disclaimer.asp?returnURL=/e-gov/emapping/access/ (3) The Hampton Roads Joint Land Use Study (Hampton Roads PDC, April 2005) continued to use the 1999 AICUZ maps, which depict ARS2 noise contours. These maps also are linked to Virginia Beach's GIS eMapping site: http://www.vbgov.com/dept/planning/ http://www.vbgov.com/dept/planning/plans/corridor/0,,16452,00.html (4) The study of The Economic Impacts of the NAS Oceana BRAC Decision (City of Virginia Beach, November 2005) also used the 1999 AICUZ maps and ARS2 noise contours as a baseline: http://www.vbgov.com/dept/brac/ http://www.vbgov.com/dept/brac/vgn_files/economic_impacts_of_BRAC_decision_draft.pdf

B. Differences Between BASE 2000 DNL and ARS2 DNL 20. The differences between the ARS2 and BASE 2000 data are substantial, and these differences are not revealed by simple correlations. Define the affected geographic area as DNL 65+ (i.e., DNL equal to or greater than 65 dB using either BASE 2000 or ARS2 data). Using the BASE 2000 and ARS2 data, the distribution of post-realignment noise levels for the repeat sales data are as follows (no. of repeat sales by noise zone for the period 1995-2003):

Case 1:01-cv-00201-VJW

Document 219-3 9

Filed 10/05/2006

Page 11 of 28

DNL Zone 65-70 dB 70-75 75-80 80-85 85-90 90-95 Total Sales (no.)

Base 2000 619 321 259 25 6 0 1230

ARS2 771 662 353 203 49 12 2050

Absolute Difference 152 341 94 178 43 12 820

The comparisons indicate that Dr. Dale-Johnson's study may have misclassified as many as 820 properties in the DNL > 65 zone. This is 40% of the ARS2 sample size of 2050 sales. 21. Using the BASE 1997 data, the distribution of changes in noise levels for the repeat sales data are as follows (no. of repeat sales by noise change for 1995-2003):

Change in DNL 0 - 4.99 dB 5 - 9.99 dB 10 - 14.99 dB 15 dB + Total Sales (no.)

Base 2000 126 701 396 7 1230

ARS2 0 123 1343 584 2050

Absolute Difference 126 578 947 577 2228

The comparison suggests that the method used by Dr. Dale-Johnson produces measurement errors in the "Change in Noise" variable (in his data files this is the DNOISE variable). It is well known that errors of measurement will bias an estimated coefficient toward zero. This is one plausible reason for his null results, which are contrary to results contained in numerous hedonic price studies.

C. Repeat Sale Regression Results for ARS2 DNL 22. The continued use of ARS2 data by several agencies, including the U.S. Navy and BRAC, casts doubt on regression results using BASE 2000 data. As a result, plaintiffs' attorneys asked that I examine the repeat sales data using ARS2 noise exposure measurements. Although Dr. Dale-Johnson did employ ARS2 data for some of his sensitivity analysis, he also changed the event date from July 1999 to a modified date of July 1997 to June 1998 (Dale-Johnson report, p. 20, Appendix F and Appendix G). In the analysis that follows, I have defined the event date or "date of taking" as July 1999, which is

Case 1:01-cv-00201-VJW

Document 219-3 10

Filed 10/05/2006

Page 12 of 28

consistent with Dale-Johnson's Exhibits and plaintiffs' position. In order to reduce aggregation bias, I focus on two time periods: 1996Q1 to 2002Q4 and 1997Q1 to 2001Q4. Results are reported for ARS2 DNL > 65 and ARS2 DNL > 75. For my regressions, I constructed a new "Change in Noise" variable for each property, DNOISE2, which is defined as ARS2 DNL minus BASE 1997 DNL.

For the properties with DNL > 65, the regression results using ARS2 indicate that the noise discount lies between -0.61% to -1.01% per dB change in noise exposure. These values are consistent with my meta-analysis and plaintiffs' damage formula for properties located in the noise zone 65 < DNL < 80. For properties with DNL > 75, the regression results for ARS2 indicate that the noise discount lies between -0.83% to -1.62% per dB. These values are consistent with my meta-analysis and plaintiffs' damage formula for properties located in the noise zone DNL > 80. These results refute Dr. Dale-Johnson's null findings for the realignment. 23. Tables 1 - 3 show the results for the sample time period 1996-2002. The event date is 1999Q2. The "Change in Noise" variable is denoted by DNOISE2. For sensitivity analysis, I report regression results for two geographic areas: ARS2 > 65 and BASE 2000 > 65. The sample sizes are 1304 and 760 repeat sales, respectively. The noise discounts in Tables 1 and 2 are -0.61% per dB and -0.65% per dB, respectively. Both regression coefficients are statistically significant at the 95% level or better. Table 3 restricts the noise levels to ARS2 > 75. The sample size is 386 repeat sales. The noise discount is -0.83% per dB. The change in noise has a negative impact on property value appreciation. 24. Tables 4 - 6 show the results for the sample time period 1997-2001. The event date is 1999Q2. The "Change in Noise" variable is denoted by DNOISE2. For sensitivity analysis, I report regression results for two geographic areas: ARS2 > 65 and BASE 2000 > 65. The sample sizes are 607 and 343 repeat sales, respectively. The noise discounts in Tables 4 and 5 are -0.85% per dB and -1.01% per dB, respectively. Table 6 restricts the noise levels to ARS2 > 75. The sample size is 172 repeat sales. The noise discount is -1.62% per dB. The change in noise has a negative impact. Expanding the sample to 1995-2003 produces a smaller, but significantly negative coefficient. It may be that either aggregation bias and/or sample selection bias are worsened by the longer time period. Regardless of the time period, the results refute the claim made by Dr. Dale-Johnson that the "Change in Noise" coefficient is insignificant (or positive). 25. Tables 7 - 9 show the results for the sample time period 1995-2003. The event date is 1999Q2. The "Change in Noise" variable is denoted by DNOISE2. For sensitivity analysis, I report regression results for two geographic areas: ARS2 > 65 and BASE 2000 > 65. The sample sizes are 2049 and 1216 repeat sales, respectively. The noise discounts in Tables 7 and 8 are -0.28% per dB and -0.38% per dB, respectively. Table 9 restricts the noise levels to ARS2 > 75. The sample size is 617 repeat sales. The noise discount is -0.50% per dB. The change in noise has a negative impact.

Case 1:01-cv-00201-VJW

Document 219-3 11

Filed 10/05/2006

Page 13 of 28

26. My results refute the findings and conclusions produced by Dr. Dale-Johnson. The ARS2 noise exposure contours have continued to be used by the U.S. Navy, BRAC, and the City of Virginia Beach. Using a "Change in Noise" variable defined as ARS2 DNL minus BASE 1997 DNL, the noise coefficient is negative and statistically significant. For DNL > 65, the damage estimates are in the range from -0.61% per dB to -1.01% per dB. The latter estimate matches the value used in plaintiffs' damage formula (Nelson report, p. 29). For DNL > 75, the damage estimates are in the range from -0.83% per dB to -1.61% per dB. The latter estimate agrees with the value used in plaintiffs' damage formula (Nelson report, p. 29). These results also are consistent with findings in numerous hedonic price studies.

IV. Event Study Method and Results A. Methodology Used by Dale-Johnson 27. In economics/finance/accounting research, an event study is an analysis of whether there was a statistically significant market reaction due to a past occurrence of a given type of event that is hypothesized to affect market values. The event that affects market values usually involves an "announcement" date or window and/or an "event date." The event may involve a legislative act being passed (and its effective date), or a regulatory ruling being announced, or an announcement of other changes that affect future market prices in some way. The basic recipe for an event study is as follows:

(1) Define the event. In this case, it is the realignment of 156 F-18 C/D aircraft to NAS Oceana. (2) Define the "announcement" and the "event date." In this case, the announcement could occur as early as mid-September 1997, March 20, 1998, May 18, 1998, or July 1999 (DaleJohnson report, pp. 5-6). The event date is July 1999, i.e., the date of the alleged taking.

(3) Define a sample of sales that are unaffected by the event and hence experience "normal" performance returns. Dale-Johnson chooses to use properties with DNL < 45 as the "control group." However, other characteristics of these properties are unknown. There is no reason to suppose that housing returns for this group were totally unaffected by unobserved positive and negative factors (except for aircraft noise), and which had differential effects on the impact group. This is the aggregation bias problem. (4) Define a sample of sales that are affected by the event and which may experience "abnormal" returns. Dale-Johnson uses properties with DNL > 65, DNL > 75, and the DNL range for eleven complaint properties. (5) Run regressions that determine the size of the abnormal returns. If necessary, run additional regressions to explain the abnormal returns by property characteristics or time. I have reported additional time-adjusted regressions for the housing price indices, rather than their ratio.

Case 1:01-cv-00201-VJW

Document 219-3 12

Filed 10/05/2006

Page 14 of 28

No other study of aircraft noise and property values has applied the event study methodology used by Dr. Dale-Johnson. A lack of prior studies suggests that his econometric results are premature as a basis for litigation. A regression on the individual indices, rather than their ratio, suggests that there may be unaccounted events occurring in the "control sample" that are not revealed by his event study method. 28. I am not aware of any other study that has applied the event study method to aircraft noise, at least in the form used by Dr. Dale-Johnson.11 A lack of prior studies suggests that his results are premature as a basis for litigation. There are two major problems with this methodology as applied to aircraft noise exposure. First, it is unclear how market processes incorporate the new information about noise exposure and reflect it in "constant quality" price indices for Virginia Beach. Dr. Dale-Johnson used a "shift" or event dummy model that changes only the constant term in his regressions. 29. Second, the "control area" sample is subject to all of the criticisms levied against the samples used in repeat sale models. The larger and more diverse the control area sample, the more likely it is that the constant quality assumption is violated. The "control sample" is defined as the subset of properties with a BASE 2000 DNL < 45 dB. The sample size is 1,685 repeat sales. He argues that this subset of sales should not be affected adversely by the change in noise exposure (Dale-Johnson report, p. 17). The 45 dB cut-off is based on an EPA report that identifies noise levels less than 45 dB as within the normal indoor activity interference and annoyance levels. However, the BASE 2000 DNL for the "control area" is measured for aircraft noise only, and ignores other sources of noise (traffic, commercial and industrial activity, etc.). While the area in question might be removed from the Naval air bases, the noise levels and changes are measured inaccurately if they ignore other noise sources. This is the errors in measurement problem. In general, too little is known about the houses in the "control sample" to justify its use as a basis for normal returns in the Virginia Beach housing market. 30. Dr. Dale-Johnson's Exhibit 8 shows a graph of the housing price indices. This graph shows variation of the individual indices that is largely eliminated when ratios of the indices are constructed (Exhibit 9). However, the sample size is small (36 quarters), so it may be that the time span is simply too short for this type of study or the frequency of the data should be different (e.g., constructing the indices
11

I am aware of two recent studies that incorporate airport announcement effects within the hedonic price model. A study by Jud and Winkler examined the announcement of a new Greensboro/Winston Salem AirportHub on housing prices near the airport. They found housing prices declined in the post-announcement period by 5.7% to 9.2%; see G.D. Jud and D.T. Winkler, "The announcement effect of an airport expansion on housing prices," Unpub. paper, University of North Carolina at Greensboro, April 2005. A second study by Konda examined the announcement effect from construction of a new airport for Austin, TX on housing prices in the vicinity of the old airport, which is a complex setting with multiple announcements. Using hedonic methods and repeat sales, she found a positive effect of the closing, especially for properties located in the noisier areas (70 dB and above); see L.S. Konda, "A comparison of methodologies to measure the effects of airport siting decisions," Unpub. paper, University of Texas at Austin, October 2002. Neither of these studies uses the method employed by Dale-Johnson.

Case 1:01-cv-00201-VJW

Document 219-3 13

Filed 10/05/2006

Page 15 of 28

on a monthly basis would increase the maximum sample size to 108 observations). As a sensitivity exercise, I ran regressions on the individual indices in Exhibit 8, which are contained in a data file labeled "indices_9503.xls." A more complex model is estimated that incorporates a time trend, a shift in the time trend, and the shift of the constant term after 1999Q2.

B. Additional Regression Results for Price Indices 31. Table 10 shows the results of regressions on the individual indices for DNL < 45, DNL > 65, and DNL > 75. I have controlled for the constant term, the event dummy, general effects of time on housing prices, and an interaction term between time and the event dummy. All of the regressions have low Durbin-Watson statistics indicating serial correlation of the residuals, which were not reported by Dr. Dale-Johnson. The event dummies are significantly negative in all three cases, but there is a lack of an explanation for the negative coefficient for the DNL < 45 index (the "control sample" index). The event dummy coefficients are larger (more negative) for the areas that are impacted by aircraft noise. All of the interaction terms are positive, which is consistent with acceleration of housing prices in Virginia Beach after 1999. The time trends are positive, but the DNL < 45 trend is smaller. This result casts doubt on this sample as a basis for "normal" returns in this market. In general, it is very difficult to discern anything from the event study that either strongly supports or strongly refutes a negative effect of the realignment on property values. As applied by Dr. Dale-Johnson, this econometric method is too untested to serve as a basis for litigation.12

V. Effects (if any) of the Realignment on Housing Demand Dr. Dale-Johnson hypothesizes that the realignment had a positive effect on housing demand and values in Virginia Beach resulting from the transfer of 3700 personnel to NAS Oceana. In his opinion, this effect was large enough so that the net effect was a rise in property values, regardless of any adverse effects of increased noise. Using the best data available, I estimate that only 6-8% of yearly housing purchases were likely to be due to realigned personnel. It is not reasonable that this small percent of yearly housing purchases can produce a positive market-wide response in housing values that offsets the demonstrated negative effects of noise. My repeat sales results further refute his hypothesis. Dale-Johnson's hypothesis does not withstand additional scrutiny of the supposed positive impact of the transfer.

Because the price indices represent time-series data, there are statistical issues regarding the stationarity of these indices. Violation of the stationarity condition can result in spurious correlation, such as the supposed relationship between the control area index and the impact area indexes. On non-stationarity as a statistical concern, see F.C.N. Myer, M.K. Chaudhry, and J.R. Webb, "Stationarity and co-integration in systems with three national real estate indices," Journal of Real Estate Research, 13 (1997): 369-81.

12

Case 1:01-cv-00201-VJW

Document 219-3 14

Filed 10/05/2006

Page 16 of 28

32. Dr. Dale-Johnson hypothesizes that the realignment had a positive effect on housing demand and values in Virginia Beach. If a negative effect due to increased noise exposure is present in this market, this means that it was totally offset by the positive effect of increased demand for housing due to realignment of 3700 U.S. Navy personnel to the Virginia Beach area. He states that:

The realignment resulted in the movement of an estimated 8,300 people from Florida to the Virginia Beach area, including approximately 3,700 military and civilian personnel. In addition, the realignment was expected to add roughly $308 million annually to the regional economy and increase taxes by $7.8 million annually. Such increases in population and spending would be expected to result in an increase in the demand for housing which in turn would be expected to cause house values to increase ­ an increase as a result of the realignment . . . Since the positive impact of the realignment on house values (through increased housing demand) would likely accrue to both houses affected by and houses unaffected by the noise component on the realignment, if no impact is found for the noise component, then one can conclude the realignment did not negatively impact house values (Dale-Johnson report, pp. 4-5; emphasis added). 33. There are at least three economic effects that might follow from the realignment which appear to be incorporated in his statement. First, 3700 personnel were transferred to NAS Oceana and an unknown portion of them will purchase homes in Virginia Beach. This is a large urban area, so the percent of residential properties sold after 1999Q2 that were purchased by the new personnel might in fact be quite small. Second, some portion of the wages paid to the new personnel will "ripple" through the Virginia Beach economy, and might result in additional jobs or increased wages and other earnings for existing residents. An unknown portion of these increased incomes might be spent on new homes in Virginia Beach. Third, the "ripple" or multiplier effect might attract new workers to the Virginia Beach area, and some unknown portion of these new residents might purchase a new home in Virginia Beach. The size of the two indirect effects is a matter of speculation, and Dr. Dale-Johnson's report does not contain information that can be used to accurately pin-down the "ripple" effect on housing demand in Virginia Beach. Further, while $308 million is a large sum of money, total retail sales in 2002 in Virginia Beach were $4.7 billion (Economic Impacts of the NAS Oceana BRAC Decision, p. C-12). Hence, $308 million represents only a 6.6% increase in local spending, and some unknown portion of this spending will occur elsewhere in the greater Hampton Roads region.13

The exact dollar amount injected into the local economy has been variously reported as $280 million (Virginian-Pilot, 10/13/1998; 12/05/1998), $438 million (Virginian-Pilot, 10/03/1999), and $308 million (VirginianPilot, 5/19/1998). The $7.8 million of local tax receipts associated with the realignment are irrelevant for housing demand unless all substitution effects are accounted for, i.e., taxes are a transfer of income.

13

Case 1:01-cv-00201-VJW

Document 219-3 15

Filed 10/05/2006

Page 17 of 28

34. The only certain impact of the realignment is the transfer of 3700 Navy personnel to NAS Oceana. About 48% of NAS Oceana employees own their own home and 78% of homeowners live in Virginia Beach (Economic Impacts study, p. 28). About 65% of those transferring to NAS Oceana were married (Dale-Johnson report, p. 7). Applying these percentages to the 3700 personnel suggests that somewhere between 1400 and 1900 of the new personnel were most likely to purchase a residential home in Virginia Beach (1400 = 3700×0.48×0.78 & 1900 = 3700×0.65×0.78). However, all of these purchases will not occur in the same year. Absent data on the frequency of purchase, suppose that one-third of the purchases occurred each year during the three-year period 1999Q3 to 2002Q3. This amounts to 467 to 633 housing sales per year. 35. The Virginia Beach housing market is a substantial and active market. According to Dr. Dale-Johnson's Exhibit 1, there were 141,642 residential properties in the City of Virginia Beach and 70,520 of these properties sold at least once during the nine-year period 1995-2003. The Economic Impacts report contains additional information on Virginia Beach housing market activities. The number of owner-occupied housing units in 2000 was 101,265 (Economic Impacts report, p. C-26). By 2004, this number had increased to 108,997. During the year 2002, there were real estate closings on 1,567 new homes and 6,905 existing homes (Economic Impacts report, p. C-32). This is 8,472 sales in the year 2002, which is typical of the high turnover in this market (Economic Impacts report, p. 33). Further, the available supply of units modestly exceeded reported demand (i.e., closings) by approximately 930 units during 2001-2004 (Economic Impacts report, p. 33). 36. Dr. Dale-Johnson argues that the realignment of Navy personnel affected sale prices in both noisy areas and quiet areas (Dale-Johnson report, p. 5). In Virginia Beach, there are about 8,000 sales in a typical year or 24,000 closings during the three-year period 1999Q3 to 2002Q3. Suppose that one-third of the realigned personnel purchased a home in each year. Thus, somewhere between 6% and 8% of the sales closed in a given year might be due to realigned personnel (6% = 467/8000 & 8% = 633/8000). This estimate is rough and subject to revision if new information becomes available. However, the estimate suggests that home purchases by the realigned personnel were a small proportion of total real estate transactions in any given year in Virginia Beach. It hardly seems reasonable that only 6-8% of the homebuyers can produce the type of offsetting market reaction hypothesized by Dr. Dale-Johnson. My repeat sales results further refute his hypothesis, and indicate that the effect of aircraft noise was negative regardless of the measurable impact of the realignment on housing demand. To the best of my knowledge, no other study has advanced this hypothesis for aircraft noise or produced empirical results that support his hypothesis.

Case 1:01-cv-00201-VJW

Document 219-3 16

Filed 10/05/2006

Page 18 of 28

VI. Conclusions 37. I have examined the methods and arguments put forward in the expert report by Dr. DaleJohnson. The sampling methods used in his repeat sales analysis are subject to aggregation bias and sample selection bias. His sample ignores physical, environmental, and neighborhood changes that affect the value of the re-sold houses, except for the change in noise level. The selected sample(s) is not a random sample of the Virginia Beach housing stock, and this raises questions of the transfer of his results to the stock of housing in general. However, the larger problem is his use of the BASE 2000 noise data in lieu of ARS2 data. Some of his results also change the event date so that it is not the same as plaintiffs' date of taking. The ARS2 noise exposure data continue to be used by the U.S. Navy, BRAC, the City of Virginia Beach, and the Hampton Roads Planning District Commission. Replacing the BASE 2000 data with the ARS2 data leads to substantial changes in the results for the repeat sales model, given that the event date is July 1999. In my repeat sale results, the "Change in Noise" coefficients are significantly negative. The magnitudes of these negative coefficients agree with the meta-analysis average values in my Expert Report and with plaintiffs' damage formula (Nelson report, p. 29). 38. The event study methodology used by Dr. Dale-Johnson has not been applied in other studies of aircraft noise and residential property values. This method is untested and too speculative to serve as an econometric basis for litigation purposes. The statistical properties of the "control sample" index suggest that it is not representative of "normal" returns in this market. Finally, Dr. Dale-Johnson hypothesizes that the realignment of 3700 Navy personnel had a comprehensive positive effect on housing demand and values in Virginia Beach. Using available data, I found that only 6-8% of yearly housing purchases were likely to be due to realigned personnel. It is not reasonable that this small percent of yearly housing purchases can produce a positive market-wide response in housing values that offsets the demonstrated negative effects of aircraft noise.

Case 1:01-cv-00201-VJW

Document 219-3 17

Filed 10/05/2006

Page 19 of 28

Table 1 ­ Regression Results for ARS2 DNL > 65 (1996-2002) ­ Sample Area is ARS2 > 65
Dependent Variable: DLPRICE_DEP Method: Least Squares Time Period: 1996-2002 Sample: 1 3797 IF ARS2_DNL>65 Included observations: 1304 White Heteroskedasticity-Consistent Standard Errors & Covariance Variable DNOISE2 Q961 Q962 Q963 Q964 Q971 Q972 Q973 Q974 Q981 Q982 Q983 Q984 Q991 Q993 Q994 Q001 Q002 Q003 Q004 Q011 Q012 Q013 Q014 Q021 Q022 Q023 Q024 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient -0.006096 -0.090987 -0.113063 -0.091603 -0.082426 -0.086475 -0.090062 -0.057084 -0.041225 -0.032798 -0.032670 -0.030748 -0.049194 -0.007985 0.146769 0.131670 0.145103 0.140143 0.164189 0.167613 0.204044 0.202942 0.212724 0.218876 0.242201 0.286352 0.296163 0.360966 0.274073 0.258712 0.119026 18.07724 939.3109 Std. Error 0.001478 0.015123 0.013619 0.014790 0.012689 0.018342 0.013335 0.013873 0.014938 0.025942 0.012056 0.013129 0.019193 0.016985 0.030146 0.025661 0.025729 0.024455 0.022925 0.028638 0.029825 0.022082 0.022116 0.022671 0.022510 0.023168 0.027439 0.024464 t-Statistic -4.124060 -6.016416 -8.301724 -6.193766 -6.495853 -4.714653 -6.753567 -4.114684 -2.759804 -1.264274 -2.709907 -2.342045 -2.563087 -0.470104 4.868545 5.131123 5.639746 5.730657 7.161981 5.852873 6.841418 9.190447 9.618756 9.654647 10.75989 12.35959 10.79338 14.75477 Prob. 0 0 0 0 0 0 0 0 0.0059 0.2064 0.0068 0.0193 0.0105 0.6384 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.190289 0.138244 -1.39772 -1.28664 1.772017

Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat

Case 1:01-cv-00201-VJW

Document 219-3 18

Filed 10/05/2006

Page 20 of 28

Table 2 ­ Regression Results for ARS2 DNL > 65 (1996-2002) ­ Sample Area is Base 2000 > 65
Dependent Variable: DLPRICE_DEP Method: Least Squares Time Period: 1996-2002 Sample: 1 3797 IF BASE_00_DNL>65 Included observations: 760 White Heteroskedasticity-Consistent Standard Errors & Covariance Variable DNOISE2 Q961 Q962 Q963 Q964 Q971 Q972 Q973 Q974 Q981 Q982 Q983 Q984 Q991 Q993 Q994 Q001 Q002 Q003 Q004 Q011 Q012 Q013 Q014 Q021 Q022 Q023 Q024 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient -0.006516 -0.093578 -0.105878 -0.097202 -0.080509 -0.071451 -0.085796 -0.054003 -0.024955 -0.043314 -0.022164 -0.038493 -0.049127 -0.001643 0.173300 0.117308 0.143916 0.155049 0.173540 0.191188 0.223712 0.195821 0.217673 0.219110 0.238501 0.279107 0.311271 0.357586 0.297449 0.271535 0.110997 9.0185 606.5421 Std. Error 0.001704 0.018309 0.017365 0.016138 0.015483 0.021309 0.016436 0.016427 0.022486 0.018970 0.013460 0.017745 0.024826 0.016789 0.039529 0.030875 0.035123 0.029971 0.028172 0.031948 0.041911 0.026077 0.027064 0.027907 0.026270 0.025480 0.028424 0.029829 t-Statistic -3.824701 -5.111107 -6.097335 -6.023167 -5.199954 -3.353029 -5.220148 -3.287529 -1.109821 -2.283251 -1.646672 -2.169232 -1.978823 -0.097887 4.384102 3.799482 4.097430 5.173274 6.160086 5.984402 5.337746 7.509203 8.042938 7.851467 9.078739 10.95381 10.95091 11.98772 Prob. 0.0001 0 0 0 0 0.0008 0 0.0011 0.2674 0.0227 0.1001 0.0304 0.0482 0.922 0 0.0002 0 0 0 0 0 0 0 0 0 0 0 0

Mean dependent var 0.188031 S.D. dependent var 0.130049 Akaike info criterion -1.522479 Schwarz criterion -1.351778 Durbin-Watson stat 1.761016

Case 1:01-cv-00201-VJW

Document 219-3 19

Filed 10/05/2006

Page 21 of 28

Table 3 ­ Regression Results for ARS2 DNL > 75 (1996-2002) ­ Sample Area is ARS2 > 75
Dependent Variable: DLPRICE_DEP Method: Least Squares Time Period: 1996-2002 Sample: 1 3797 IF ARS2_DNL>75 Included observations: 386 White Heteroskedasticity-Consistent Standard Errors & Covariance Variable DNOISE2 Q961 Q962 Q963 Q964 Q971 Q972 Q973 Q974 Q981 Q982 Q983 Q984 Q991 Q993 Q994 Q001 Q002 Q003 Q004 Q011 Q012 Q013 Q014 Q021 Q022 Q023 Q024 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient -0.008335 -0.100405 -0.092789 -0.121079 -0.078867 -0.124163 -0.093401 -0.066450 -0.046524 -0.046605 -0.033148 -0.023317 -0.067712 -0.015348 0.227289 0.092951 0.145314 0.163283 0.190175 0.199715 0.275491 0.233133 0.259398 0.232694 0.249950 0.292296 0.313548 0.371499 0.286966 0.23319 0.12307 5.422317 275.4953 Std. Error t-Statistic 0.002633 0.031089 0.025458 0.027715 0.025840 0.044212 0.029707 0.026923 0.037426 0.028555 0.021857 0.026212 0.051091 0.024572 0.069323 0.056031 0.054043 0.053778 0.045513 0.046520 0.075405 0.044884 0.047748 0.047262 0.044225 0.041528 0.048255 0.043903 -3.165898 -3.229637 -3.644795 -4.368794 -3.052082 -2.808322 -3.144082 -2.468166 -1.243103 -1.632113 -1.516605 -0.889535 -1.325328 -0.624584 3.278689 1.658929 2.688862 3.036257 4.178457 4.293069 3.653510 5.194136 5.432621 4.923448 5.651730 7.038511 6.497673 8.461739 Prob. 0.0017 0.0014 0.0003 0 0.0024 0.0053 0.0018 0.014 0.2146 0.1035 0.1302 0.3743 0.1859 0.5326 0.0011 0.098 0.0075 0.0026 0 0 0.0003 0 0 0 0 0 0 0 0.183318 0.140542 -1.282359 -0.995407 1.811981

Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat

Case 1:01-cv-00201-VJW

Document 219-3 20

Filed 10/05/2006

Page 22 of 28

Table 4 ­ Regression Results for ARS2 DNL > 65 (1997-2001) ­ Sample Area is ARS2 > 65
Dependent Variable: DLPRICE_DEP Method: Least Squares Time Period: 1997-2001 Sample: 1 1782 IF ARS2_DNL>65 Included observations: 607 White Heteroskedasticity-Consistent Standard Errors & Covariance Variable DNOISE2 Q971 Q972 Q973 Q974 Q981 Q982 Q983 Q984 Q991 Q993 Q994 Q001 Q002 Q003 Q004 Q011 Q012 Q013 Q014 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient -0.008473 -0.066276 -0.070043 -0.038701 -0.039817 -0.043575 -0.001177 -0.005185 -0.039707 0.011094 0.207255 0.178249 0.199151 0.197249 0.222421 0.203317 0.247567 0.253737 0.249557 0.263558 0.116864 0.088278 0.113903 7.61567 467.5247 Std. Error 0.002320 0.021232 0.017830 0.018275 0.017497 0.018857 0.016780 0.017803 0.027777 0.024083 0.047342 0.039385 0.035770 0.033571 0.033829 0.041189 0.041399 0.033275 0.032333 0.033001 t-Statistic -3.651553 -3.121522 -3.928369 -2.117738 -2.275619 -2.310800 -0.070132 -0.291257 -1.429491 0.460659 4.377798 4.525837 5.567484 5.875599 6.574839 4.936145 5.980073 7.625393 7.718261 7.986285 Prob. 0.0003 0.0019 0.0001 0.0346 0.0232 0.0212 0.9441 0.771 0.1534 0.6452 0 0 0 0 0 0 0 0 0 0 0.138209 0.11929 -1.47455 -1.32929 1.935311

Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat

Case 1:01-cv-00201-VJW

Document 219-3 21

Filed 10/05/2006

Page 23 of 28

Table 5 ­ Regression Results for ARS2 DNL > 65 (1997-2001) ­ Sample Area is Base 2000 > 65
Dependent Variable: DLPRICE_DEP Method: Least Squares Time Period: 1997-201 Sample: 1 1782 IF BASE_00_DNL>65 Included observations: 343 White Heteroskedasticity-Consistent Standard Errors & Covariance Variable DNOISE2 Q971 Q972 Q973 Q974 Q981 Q982 Q983 Q984 Q991 Q993 Q994 Q001 Q002 Q003 Q004 Q011 Q012 Q013 Q014 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient Std. Error -0.010067 -0.075553 -0.072023 -0.051013 -0.036862 -0.036104 0.002713 -0.015513 -0.062948 0.013943 0.242847 0.148245 0.197954 0.210639 0.242483 0.229609 0.283966 0.260026 0.264470 0.272842 0.152665 0.102822 0.118501 4.535739 255.1689 0.002715 0.022962 0.020423 0.021701 0.022053 0.019232 0.018756 0.021808 0.042523 0.022351 0.064483 0.047011 0.047169 0.043488 0.043212 0.046567 0.058129 0.040696 0.040667 0.041251 t-Statistic -3.707677 -3.290293 -3.526616 -2.350716 -1.671490 -1.877266 0.144635 -0.711358 -1.480343 0.623848 3.766078 3.153434 4.196731 4.843625 5.611454 4.930729 4.885123 6.389524 6.503239 6.614133 Prob. 0.0002 0.0011 0.0005 0.0193 0.0956 0.0614 0.8851 0.4774 0.1398 0.5332 0.0002 0.0018 0 0 0 0 0 0 0 0

Mean dependent var 0.133921 S.D. dependent var 0.125107 Akaike info criterion -1.37125 Schwarz criterion -1.14747 Durbin-Watson stat 1.844767

Case 1:01-cv-00201-VJW

Document 219-3 22

Filed 10/05/2006

Page 24 of 28

Table 6 ­ Regression Results for ARS2 > 75 (1997-2001) ­ Sample Area is ARS2 > 75
Dependent Variable: DLPRICE_DEP Method: Least Squares Time Period: 1997-2001 Sample: 1 1782 IF ARS2_DNL>75 Included observations: 172 White Heteroskedasticity-Consistent Standard Errors & Covariance Variable DNOISE2 Q971 Q972 Q973 Q974 Q981 Q982 Q983 Q984 Q991 Q993 Q994 Q001 Q002 Q003 Q004 Q011 Q012 Q013 Q014 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient Std. Error t-Statistic -0.016160 -0.145970 -0.081360 -0.079550 -0.082590 -0.078060 -0.011650 -0.014620 -0.119030 -0.012800 0.335150 0.167491 0.252309 0.254123 0.314832 0.290859 0.386530 0.352513 0.364285 0.361028 0.244281 0.149817 0.138116 2.899576 107.0746 0.004824 0.044871 0.037274 0.036685 0.032935 0.038056 0.029883 0.035852 0.105523 0.030537 0.098267 0.080363 0.078366 0.076877 0.076735 0.073203 0.113426 0.073397 0.078642 0.077164 -3.350115 -3.253222 -2.182825 -2.168414 -2.507615 -2.051195 -0.389697 -0.407697 -1.127989 -0.418991 3.410612 2.084190 3.219641 3.305565 4.102870 3.973319 3.407773 4.802845 4.632222 4.678714 Prob. 0.001 0.0014 0.0306 0.0317 0.0132 0.042 0.6973 0.6841 0.2611 0.6758 0.0008 0.0388 0.0016 0.0012 0.0001 0.0001 0.0008 0 0 0 0.130238 0.149792 -1.0125 -0.64651 1.987784

Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat

Case 1:01-cv-00201-VJW

Document 219-3 23

Filed 10/05/2006

Page 25 of 28

Table 7 ­ Regression Results for ARS2 DNL > 65 (1995-2003) ­ Sample Area is ARS2 > 65
Dependent Variable: DLPRICE_DEP Method: Least Squares Time Period: 1995-2003 Sample: 1 6029 IF ARS2_DNL>65 Included observations: 2049 White Heteroskedasticity-Consistent Standard Errors & Covariance Variable DNOISE2 Q951 Q952 Q953 Q954 Q961 Q962 Q963 Q964 Q971 Q972 Q973 Q974 Q981 Q982 Q983 Q984 Q991 Q993 Q994 Q001 Q002 Q003 Q004 Q011 Q012 Q013 Q014 Q021 Q022 Q023 Q024 Q031 Q032 Q033 Q034 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient Std. Error t-Statistic -0.002880 -0.140350 -0.124330 -0.121920 -0.104370 -0.087470 -0.102030 -0.090230 -0.087160 -0.085370 -0.087210 -0.062520 -0.041100 -0.037010 -0.037360 -0.030890 -0.053120 -0.015660 0.088942 0.082349 0.094893 0.094650 0.120644 0.122840 0.160448 0.161717 0.167656 0.174099 0.195614 0.236264 0.256543 0.317661 0.353520 0.353835 0.386512 0.428647 0.471204 0.46201 0.119314 28.65681 1466.919 0.001231 0.020076 0.017993 0.012681 0.013020 0.016262 0.013530 0.012664 0.011864 0.016173 0.011904 0.012666 0.013355 0.021850 0.011152 0.011748 0.015907 0.014802 0.025261 0.022333 0.022808 0.021830 0.020337 0.024863 0.025526 0.019759 0.019380 0.019918 0.019733 0.020878 0.023852 0.021347 0.022121 0.022559 0.021874 0.020367 -2.336032 -6.990953 -6.909651 -9.613994 -8.015531 -5.378839 -7.540884 -7.125237 -7.346693 -5.278632 -7.326518 -4.935967 -3.077158 -1.693935 -3.350348 -2.629421 -3.339427 -1.057802 3.520899 3.687378 4.16043 4.33573 5.932302 4.940597 6.285657 8.18461 8.650991 8.740679 9.912814 11.31651 10.75549 14.88068 15.98091 15.68475 17.67012 21.04602 Prob. 0.0196 0 0 0 0 0 0 0 0 0 0 0 0.0021 0.0904 0.0008 0.0086 0.0009 0.2903 0.0004 0.0002 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.24781 0.162669 -1.3967 -1.29787 1.668993

Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat

Case 1:01-cv-00201-VJW

Document 219-3 24

Filed 10/05/2006

Page 26 of 28

Table 8 ­ Regression Results for ARS2 DNL > 65 (1995-2003) ­ Sample Area is Base 2000 > 65
Dependent Variable: DLPRICE_DEP Method: Least Squares Time Period: 1995-2003 Sample: 1 6029 IF BASE_00_DNL>65 Included observations: 1216 White Heteroskedasticity-Consistent Standard Errors & Covariance Variable DNOISE2 Q951 Q952 Q953 Q954 Q961 Q962 Q963 Q964 Q971 Q972 Q973 Q974 Q981 Q982 Q983 Q984 Q991 Q993 Q994 Q001 Q002 Q003 Q004 Q011 Q012 Q013 Q014 Q021 Q022 Q023 Q024 Q031 Q032 Q033 Q034 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient Std. Error t-Statistic -0.003840 -0.129460 -0.114750 -0.112660 -0.109320 -0.081170 -0.082760 -0.095980 -0.085660 -0.076950 -0.083130 -0.061800 -0.029370 -0.048040 -0.024320 -0.042620 -0.058310 -0.019550 0.110736 0.076362 0.102597 0.119424 0.132158 0.148788 0.179790 0.167812 0.175775 0.187959 0.197329 0.233077 0.276812 0.316580 0.353998 0.349898 0.371668 0.415883 0.45556 0.439411 0.114605 15.49851 927.0183 0.001432 0.024934 0.016844 0.018344 0.016637 0.019846 0.018042 0.014322 0.015429 0.018618 0.015044 0.015644 0.018788 0.017120 0.013052 0.015890 0.020689 0.016667 0.032470 0.027422 0.030688 0.026806 0.024838 0.027900 0.034482 0.023570 0.023692 0.024940 0.023069 0.023783 0.026148 0.026014 0.026983 0.028246 0.026772 0.023909 -2.679454 -5.191874 -6.812337 -6.141272 -6.571192 -4.089765 -4.58692 -6.701591 -5.552042 -4.133054 -5.526046 -3.950025 -1.563431 -2.80618 -1.863225 -2.682118 -2.818491 -1.172923 3.410458 2.784738 3.343189 4.455095 5.320723 5.332919 5.214034 7.119667 7.419021 7.536481 8.553816 9.800316 10.58652 12.16982 13.11935 12.38773 13.88279 17.39463 Prob. 0.0075 0 0 0 0 0 0 0 0 0 0 0.0001 0.1182 0.0051 0.0627 0.0074 0.0049 0.2411 0.0007 0.0054 0.0009 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.241215 0.153067 -1.46549 -1.31441 1.652004

Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat

Case 1:01-cv-00201-VJW

Document 219-3 25

Filed 10/05/2006

Page 27 of 28

Table 9 ­ Regression Results for ARS2 DNL > 75 (1995-2003) ­ Sample Area is ARS2 > 75
Dependent Variable: DLPRICE_DEP Method: Least Squares Time Period: 1995-2003 Sample: 1 6029 IF ARS2_DNL > 75 Included observations: 617 White Heteroskedasticity-Consistent Standard Errors & Covariance Variable DNOISE2 Q951 Q952 Q953 Q954 Q961 Q962 Q963 Q964 Q971 Q972 Q973 Q974 Q981 Q982 Q983 Q984 Q991 Q993 Q994 Q001 Q002 Q003 Q004 Q011 Q012 Q013 Q014 Q021 Q022 Q023 Q024 Q031 Q032 Q033 Q034 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient Std. Error t-Statistic -0.005020 -0.165710 -0.137000 -0.110900 -0.138910 -0.090720 -0.073180 -0.105130 -0.073470 -0.125960 -0.096040 -0.075430 -0.051310 -0.053450 -0.046190 -0.043170 -0.079180 -0.032360 0.134142 0.038578 0.095969 0.100908 0.136623 0.141395 0.214956 0.187703 0.194901 0.191435 0.199882 0.236149 0.266191 0.307414 0.363281 0.324495 0.379476 0.432288 0.426553 0.392008 0.125478 9.14766 423.7229 0.002340 0.039978 0.027369 0.031835 0.024679 0.035113 0.029918 0.025525 0.027276 0.038292 0.027987 0.024449 0.031710 0.027230 0.021556 0.025136 0.038851 0.024708 0.058608 0.049616 0.051269 0.051804 0.043482 0.044753 0.062885 0.043423 0.044341 0.045909 0.042789 0.040789 0.044846 0.042722 0.047139 0.049233 0.049554 0.046617 -2.143646 -4.145049 -5.005745 -3.483735 -5.628463 -2.583708 -2.446142 -4.118711 -2.693594 -3.289553 -3.431734 -3.085098 -1.617979 -1.962949 -2.142701 -1.717486 -2.038079 -1.309819 2.28881 0.77753 1.871851 1.947889 3.142064 3.159489 3.418222 4.322615 4.395504 4.169915 4.671312 5.789585 5.935639 7.195614 7.706644 6.591025 7.657826 9.273193 Prob. 0.0325 0 0 0.0005 0 0.01 0.0147 0 0.0073 0.0011 0.0006 0.0021 0.1062 0.0501 0.0326 0.0864 0.042 0.1908 0.0224 0.4372 0.0617 0.0519 0.0018 0.0017 0.0007 0 0 0 0 0 0 0 0 0 0 0

Mean dependent var 0.233649 S.D. dependent var 0.160923 Akaike