Printer Friendly

Adjusting house prices for intra-neighborhood traffic differences.

When valuing property, an appraiser considers the impact of external factors on valuation. These external factors include any feature of a property that is not an actual property characteristic. For instance, street conditions, police protection, proximity to employment or shopping, and air quality are all factors that influence the value of property without being characteristics of the property itself. Typically, adjustments for these factors are grouped together as a location adjustment. This study focuses on one of those external factors: traffic.

Real estate appraisers have long recognized the effect of traffic on residential property value. In The Appraisal of Real Estate, eighth edition, appraisers are cautioned that "Excessive traffic, odors, smoke, dust, or noise ... can limit residential desirability."(1) Unfortunately, measuring these effects can be difficult. Many questions arise when an appraiser attempts to adjust value for traffic: Does the housing market actually price the different traffic levels within a neighborhood? Are the negative effects of traffic greater than the positive aspects such as better access? This article addresses these considerations and suggests a process an appraiser can use to adjust comparable properties for differing traffic levels.

A recent study by Hughes and Sirmans was the first to examine the effects of various traffic levels within a neighborhood.(2) This study shows that a neighborhood location adjustment would not account for this intra-neighborhood valuation effect. An absolute traffic adjustment is inappropriate as well, since all streets have some level of traffic. An adjustment must be made for the relative difference between the traffic associated with the comparable property and the traffic associated with the subject property.

This article provides results of a study that show selling prices for single-family detached homes are affected by traffic levels. Further, this adjustment occurs not only in the case of the external effect of high versus low traffic, but also the relative effect of differing levels of traffic among high-traffic streets. The following is an analysis of these effects and a procedure for determining the appropriate adjustment for individual markets.


Because all characteristics of real estate are bundled together for sale, one must rely on the implicit price of each feature. Therefore, the multiple regression technique is applied to residential transactions to examine the impact of high traffic.(3) Examination of the coefficient on the traffic variable yields a measure of the impact of traffic on residential property values. The model applied to the data is as follows:

|Price.sub.i~ = ||Beta~.sub.0~ + |Sigma~||Beta~.sub.j~(|X.sub.j~) + |Sigma~||Beta~.sub.k~(|Y.sub.k~) + ||Beta~.sub.T~(T)


|Price.sub.i~ = Selling price of the ith house

|Beta~ = Estimated coefficients

|X.sub.j~ = Physical characteristics of the house

|Y.sub.k~ = Market conditions

T = Level of traffic

The |Beta~'s from this model represent the marginal effect of each value-influencing factor. The marginal effect of traffic on the sale price is demonstrated through ||Beta~.sub.T~. If ||Beta~.sub.T~ is significantly different from zero, then varying levels of traffic are priced in the housing market. Both the sign and magnitude of this coefficient are important because an appraiser must know in which direction and by how much to adjust for the effect of traffic.


For this study, sale price, property characteristics, date of sale, and property location were collected from the local multiple listing service (MLS). These data were collected for two MLS areas within the Baton Rouge, Louisiana, metropolitan area. In addition, average daily traffic counts for various streets were collected for the same two areas from the Louisiana Department of Transportation and Development. Data from these two sources are used to estimate the previously described model to test the hypothesis that intra-neighborhood traffic affects property value.

Observations for which traffic counts are available are labeled high-traffic streets. A subset from the remaining streets within the defined areas is defined as low-traffic streets. Sales on streets that do not fit the low-traffic classification and for which traffic counts are not available are not included in the analysis. In this study, a low-traffic street is defined by the inability to advance using the street. For example, cul-de-sacs, looping streets, and dead-ends limit traffic to those seeking access to or from the street. The street is not used by the local traffic to move from one place to another unless the street itself is the destination.

The sample consists of 288 sales for which all variables are available.(4) The data cover the period from January 1985 to December 1989; summary statistics are provided in Table 1.

The average sale price for a single-family home in the sample is $111,887. Mean living area is 2,340 square feet while mean other area (e.g., porch, patio, garage, carport, storage) is 780 square feet. TABULAR DATA OMITTED The mean lot size is 16,658 square feet. Ninety-four percent of the properties have central air conditioning while the remaining 6% are equipped with window units. The average property age is approximately 15 years. Fifty-four percent of the homes are owner occupied; non-owner-occupied homes are occupied by a tenant or are vacant. Average marketing time is 101 days.

Dummy variables, representing the year in which a property sold, are used to capture housing market conditions in Baton Rouge over the sample period. This approach is used instead of a single time variable to allow for non-linear economic conditions. It is particularly important to capture the effects of external market conditions during this period because Baton Rouge experienced a peak followed by decline and recovery in just five years.(5) As seen in Table 1, approximately 20% of the sales took place in each of these five years. The location variable is equal to one if the transaction was located in the MLS area closest to the central business district (CBD) and zero for transactions in the other MLS area. The location variable shows that 44%, or 126, of the observations lie in the area closest to the CBD, leaving 162 observations in the other area.

Fifty-eight percent, or 168, of the observations are on high-traffic streets. This leaves 120 properties on low-traffic streets. The cars variable indicates the average daily traffic count is 6,234 cars for the 168 transactions for which traffic counts are available, with a minimum traffic count of 1,114 cars per day. This sample compares traffic effects for extreme cases: very high versus very low traffic. Standard deviation, minimum values, and maximum values for all of the variables are shown in Table 1.


Table 2 provides the results of estimating two price equations on the data described previously. Model A employs a traffic-dummy variable to distinguish the effect of location on a high-traffic street versus a low-traffic street. Model B is applied to only the sales on high-traffic streets using the cars variable (i.e., the actual traffic count) to detect the relative effect of traffic in high-traffic locations. The dependent variable in both equations is the natural logarithm of sale price. The semilogarithmic form yields a traffic/cars coefficient that is the percentage value adjustment for increased traffic count. Tests for heteroskedasticity and multicollinearity indicate well-specified models as both conditions are rejected.(6)

Regression Results of Impact of Traffic on Single-Family House

Variable Model A Model B

Intercept 10.5164 10.5184
 (159.988) (112.992)
Living area (|10.sup.-3~) 0.3370 0.2916
 (18.597) (12.755)
Other area (|10.sup.-3~) 0.2250 0.2798
 (6.033) (6.102)
Lot size (|10.sup.-3~) 0.0029 0.0035
 (4.245) (4.783)
Air conditioning 0.1013 0.0545
 (2.153) (0.991)
Age -0.0075 -0.0038
 (-3.721) (-1.490)
Age squared 0.0001 0.0001
 (2.605) (1.259)
Days on market -0.0002 -0.0003
 (-1.996) (-2.165)
Owner occupied 0.0797 0.0977
 (3.629) (3.383)
Sold in 1986 -0.0101 0.0106
 (-0.291) (0.229)
Sold in 1987 -0.0870 -0.0492
 (-2.497) (-1.038)
Sold in 1988 -0.1196 -0.1443
 (-3.412) (-3.194)
Sold in 1989 -0.1160 -0.0989
 (-3.341) (-2.139)
Location 0.2108 0.1640
 (9.344) (5.028)
High traffic -0.1090
Cars (|10.sup.-3~) -0.0085
|R.sup.2~ 0.82 0.75
F-Value 97.05 36.32
Sample size 288 168

* T-statistics are in parentheses. Dependent variable is

Basic results

Given recent findings in regression studies and a knowledge of the local housing market, the results are as expected. About 80% of the variation in selling price is explained by the variation in the independent variables. Living area, other area, and lot size are all positive and significant. This indicates that additional units of any of these variables, all other things being equal, have an increasing effect on sale price. Age of house is negative while age squared is positive, indicating that properties depreciate over time at a decreasing rate. For example, the depreciation effect from year 2 to year 3 is greater than the depreciation effect from year 10 to year 11. These results are consistent with past studies. The days-on-the-market variable has a significantly negative coefficient. Owner-occupied homes sell for a premium when compared to rented or vacant homes as indicated by the positive coefficient. This premium probably reflects a higher maintenance level of owner-occupied homes.

The annual dummy variables, sold from 1986 to 1989, describe the market conditions previously suggested. Using 1985 as the comparison year, the coefficients on the annual variables show the difference between 1985 and the specified year. Therefore, 1986 shows no change from 1985 since the coefficient for 1986 is insignificant. The significantly negative coefficients on the other variables, however, indicate lower housing values in each of the specific years as compared with 1985. A possible recovery begins in 1989 as indicated by the less negative coefficient in 1989 than in 1988. The significant location coefficient indicates a value effect between the areas. Properties sell for more if located in the area closest to the CBD.

Adjusting between high and low traffic

Focusing specifically on Model A, the coefficient of -0.1090 on the traffic variable indicates a significantly (t-stat = -4.821) negative effect on price for houses on a high-traffic street as compared with houses on a low-traffic street. Since the variable is a dummy variable, this coefficient yields a downward adjustment of 11.49% for high traffic.(7)

When appraising a subject property that is located on an extremely high-traffic street, and when the most appropriate comparable sale is located on a very low-traffic street, the local equivalent to this adjustment must be applied. For example, consider a subject property located on a high-traffic street within the study region. A comparable is similar to the subject in every way except that it is located on a low-traffic street such as a dead-end street. If the comparable property sold for $100,000, the traffic adjustment is negative $11,490, or $100,000 multiplied by negative 11.49%. The comparable sale price must be adjusted down to $88,510, since it has the advantageous feature of being located on a low-traffic street. The magnitude of this adjustment may seem large, but our sample compares the extreme cases of high versus low traffic. Adjustments for properties located on intermediate traffic streets will receive smaller adjustments.

Effect of increasing traffic

Does the housing market actually price the different traffic levels within a neighborhood? Model B shows the results of the test when the average daily traffic counts are used in place of the traffic dummy variable. We use the sample of "high-traffic" transactions only for this model. The cars coefficient is significantly negative, which provides evidence that the market does make price corrections for relative traffic differences. Given the semilogarithmic form of the model, the negative coefficient, -0.00000847, is the percentage discount a property receives for an additional car in the average daily traffic count. Translating this coefficient into usable terms, a property located in the sample area receives a 0.847% discount for each additional 1,000 cars traveling on the adjacent street in a day.

To illustrate, suppose a comparable property that sold for $112,000 is located on a street with 1,000 fewer cars than the street on which the subject property is located. (A key to this comparison is that both properties are located on high-traffic streets with the subject property having a higher traffic count.) The appraiser will make an adjustment to the comparable sale price that reflects the impact of higher traffic on the subject property. As shown previously, the market will apply a 0.847% discount for the additional 1,000 cars in the average daily traffic count. Therefore, the comparable property's sale price will be reduced by $949 to adjust for the discount that the market applies for higher traffic.


Intra-neighborhood traffic is a factor in the valuation of residential property and the market makes price adjustments for increased traffic flow. Therefore, an appraiser should consider the traffic effect when valuing residential property. In the sample described in this study, the traffic effect is a negative one, implying that the detracting effects of increased traffic (e.g., odor, noise, dust, danger) outweigh the positive effects of access.

A multiple regression model applied to local data can provide an appraiser with an adjustment for the differences in traffic between a comparable property and the subject property when using the sales comparison approach to valuation. One of the two models provided can be applied to the appraisal, depending on the nature of the comparison. If the subject property is located on a street with an extremely different traffic flow than a comparable property, the appraiser must determine the value effect at extremes. If both the subject and comparable are located on high-traffic streets, the appraiser must adjust for the relative difference in traffic.

Although the actual magnitude of the traffic effect may be market specific, this study outlines a procedure that can establish the local traffic effect. Matched pairs can be used to establish the market adjustment for traffic if similar properties exist across traffic differences in the recent sale data. If the data are more restricted, a simple regression model may be used to calculate the traffic effect. In either case, an appraiser will enhance the accuracy of appraisals by combining average daily traffic counts with sale data and computing the magnitude of the local traffic discount.

1. American Inst. of Real Estate Appraisers, The Appraisal of Real Estate, 8th ed. (Chicago: American Inst. of Real Estate Appraisers, 1987), 71.

2. William T. Hughes, Jr., and C. F. Sirmans, "Traffic Externalities and Single-Family House Prices," Journal of Regional Science (November 1992): 487-500. Previous studies have addressed the value impact of being located close to interstates and major highways. For a summary of this literature, see J. R. Nelson, "Highway Noise and Property Values," Journal of Transport Economics and Policy (May 1982): 117-138.

3. The frequent use of multiple regression in articles appearing in The Appraisal Journal suggests that this technique is widely used and accepted.

4. All observations are for single-family detached homes. To ensure comparability of transactions, property sales that were not cash-equivalent sales are not included in the sample.

5. These conditions are established by trends in employment, population, and constant quality house prices as collected by the Louisiana State University Real Estate Research Institute in An Analysis of Baton Rouge Real Estate: Spring 1992, (Baton Rouge: Louisiana State University, 1992).

6. The White test was used to detect heteroskedasticity and was unable to reject homoskedasticity. The Belsley, Kuh, and Welch procedure was used to test for multicollinearity, which was not detected. For more information on these techniques, see Halbert White, "A Heteroscedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroscedasticity," Econometrica (May 1980): 817-838, and D. Belsley, E. Kuh, and R. E. Welch, Regression Diagnostics (New York: Wiley, 1980).

7. Interpretation of a dummy variable coefficient when the dependent variable is in log form is described in Peter E. Kennedy, "Estimation with Correctly Interpreted Dummy Variables in Semilogarithmic Equations," American Economic Review (1981): 801.

William T. Hughes, Jr., PhD, is an assistant professor of real estate at Louisiana State University. He received a PhD from the University of Georgia in 1990 and has since written several articles. Mr. Hughes's primary areas of research include housing, real estate lending, and real estate securities.

C. F. Sirmans, SRPA, PhD, is the director of the Center for Real Estate and Urban Economic Studies at the University of Connecticut. He has directed research for both private and public organizations and has published extensively in academic and professional publications. Mr. Sirmans is the editor of The Journal of Real Estate Finance and Economics and The Journal of Real Estate Literature.
COPYRIGHT 1993 The Appraisal Institute
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Hughes, Willaim T.; Sirmans, C.F.
Publication:Appraisal Journal
Date:Oct 1, 1993
Previous Article:Computer-assisted real estate appraisal: a California savings and loan case study.
Next Article:Stein rule estimation in real estate appraisal.

Related Articles
Gross living area adjustment shortcut. (Features).
The effect of freight railroad tracks and train activity on residential property values.
City approves church zoning reform.
Builders geared up for another solid year.
Site size adjustments: a technique to estimate the adjustment magnitude.
Cost-based property tax inequity: evidence from Indiana.
Public school reform, expectations, and capitalization: what signals quality to homebuyers?

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters