All Aboard Florida’s plan to return passenger train service to the Florida East Coast Railway tracks caused the value of smaller homes along the rail line to fall by as much as 29 percent, according to a study released Monday by the Martin County Property Appraiser’s Office.
The study, which examined the sale of more than 30,000 residential properties in Martin County over the last decade, found that All Aboard Florida had the greatest impact on homes that measured less than 1,532-square-feet and were located within 400 feet of the train tracks.
Measuring the Short-Term, Perceived Impacts of the All Aboard Florida Project Announcement on Residential Property Values in Martin County, FL
Below is the report measuring the impacts of the All Aboard Florida project announcement on residential property values in Martin County, Florida. Full report can be downloaded at the top of the page.
Prepared by: Dr. Jesse Saginor, Director of Housing, Economy, and Urban Development Center for Urban and Environmental Solutions (CUES) Associate Professor, School of Urban and Regional Planning Florida Atlantic University (FAU)
Prepared for: The Martin County Property Appraiser’s Office
Purpose of the Report: To assist the Martin County Property Appraiser’s Office in documenting value impacts of the All Aboard Florida project announcement.
All Aboard Florida (AAF) is projected to provide hourly passenger rail service to Orlando, West Palm Beach, Fort Lauderdale and Miami. To meet this proposed demand, AAF trains will make 16 round trips (or 32 one-way trips) from Orlando to Miami daily, passing through Martin County at speeds in excess of 100 MPH. While it is well-established in the peer-reviewed literature that property values increase in the presence of a rail station, it is unclear whether there is any benefit at all for areas, like Martin County, through which the passenger rail will pass without stopping. This paper examines the short-term perceived impacts of the AAF announcement in March, 2012 on single-family residential property values located within 400 feet and 1,000 feet of the rail line and waterfront homes west of the rail bridge. The results are based on 13,332 residential property sales throughout Martin County stratified by living area size from 2005 until February, 2015.
The findings of this study show that the AAF announcement had the greatest impact on smaller homes (homes less than 1,532 square feet) within 400 feet as well as from 400 feet to 1,000 feet of the FEC East Rail Line. The typical impact on sales prices for smaller homes that sold after the announcement was -$16,681 for homes within 400 feet of the rail and -$10,502 for homes from 400 feet to 1,000 feet.
For medium-sized homes (1,533 square feet to 2,184 square feet), the impact on sales prices for both distances was negative, but it was not statistically significant. For larger homes (over 2,185 square feet), the impact was negative within 400 feet but not statistically significant, while the impact of being from 400 feet to 1,000 feet of the railroad was actually positive and statistically significant. A major reason for the positive impact has little to do with the location of the FEC East line and more to do with the proximity to the Atlantic Ocean for larger homes east of the rail bridge. Despite the lack of an impact based on rail proximity, the premium for waterfront homes west of the rail bridge did not increase at a rate comparable to other waterfront properties east of the rail bridge. The reason may be the fact that the impact of the railroad is reflected in lower waterfront premiums just west of the rail bridge.
In summary, smaller homes closer to the rail exhibit the most negative sales price impacts. Medium and large homes located further away from the negative impact of the railroad, for a majority of homes, benefit from the positive impact of proximity to the water. These findings mirror the results found throughout the peer-reviewed real estate literature: premiums associated with proximity to water are often greater than declines associated with proximity to railroads.
The purpose of this study is to determine whether the proposed All Aboard Florida (AAF) project originally announced in March of 2012 is likely to have an impact on property values in Martin County. Given that AAF is proposed, and not completed, the results in this study are largely preliminary based on the actual effects on sales prices related to properties sold within 1,000 feet of the Florida East Coast east rail line and waterfront properties in Martin County. For waterfront properties, the model contains all waterfront properties, including but not limited to waterfront properties west of the railroad bridge that are most likely to be impacted by the future increase in rail traffic. This rail line runs through some of the most densely populated areas of the county. While the Draft Environmental Impact Statement (DEIS) attempts to address and resolve any environmental concerns regarding AAF, the purpose of this study is to focus on the economic impacts related to the real estate values in the county. These economic impacts are beyond the scope of the DEIS, yet measuring these impacts may result in determining whether the AAF will have a negative, positive, or possibly no impact on real estate values.
While the AAF announcement occurred in 2012, this study looks at property sales going back to 2005. Traditionally, Martin County took a measured approach to growth, implementing growth controls to foster balanced development. While Martin County was not completely unscathed by the real estate crash, this measured approach ensured that property development and values grew gradually. Unlike its county neighbor to the south, Palm Beach County, the changes in value in Martin County were not as drastic. The goal of this study is to analyze the properties proximate to the railroad in the larger context of the overall economy since 2005. The purpose for going back to 2005 is that real estate values are chained – looking at properties for a single year or two fails to account for larger economic and real estate cycles.
This study starts with a brief discussion of the related, peer-reviewed literature on the impacts of railroads on residential property values. Upon discussing the existing literature, there is an exploration of the dataset used for this study, including various residential characteristics and other variables integral to the study. The justification for the methodology follows the data section, leading to the results of the models. The last two sections outline caveats and limitations as well as overall study conclusions.
II. Studies on Impacts of Rail on Residential Property Values
While many studies have focused on the impacts of rail and residential property values, it is important to highlight that not all peer-reviewed studies are created equal. Several studies focus on the impacts of light rail, which often has the negative impacts of crossings and proximity to rail offset by the positive impacts of proximity to a rail stop. The speeds of these types of rail uses are most often lower than the speeds affiliated with All Aboard Florida. Other studies focus on commuter rail, but similar to light rail, even commuter rail goes at speeds far less than the proposed 110 MPH of All Aboard Florida. Additionally, many commuter rail lines and even freight lines run parallel to highways, causing some issues in delineating the negative impacts of highways from the negative impacts of the railroad. Finally, this research examines the impact of the announcement on property values as opposed to actual impacts, requiring some analysis of existing literature based on transportation improvements. Despite these issues, the studies that provide some insight into the likely effects of proximity issues and property values are consistently negative, with property losses ranging from two percent within a quarter of a mile to over 30 percent for properties with frontage on the rail line.
Several studies regarding rail highlight the positive impacts of proximity to a rail stop, often without mentioning any impacts of proximity to rail. These studies, to one extent or another, all focus on accessibility, which in turn implies having a rail stop. Whether it is high-speed rail (Loukaitou-Sideris, Higgins, Piven, and Wei, 2013) or light rail (Debrezion, Pels, and Rietveld, 2007; Baldwin-Hess and Almeida, 2007; Duncan, 2011; Dube, Theriault, Des Rosiers, 2013; Kim and Lahr, 2014; Wu, Dong, and Wang, 2015), the majority of studies focusing on accessibility denote positive impacts related to proximity to a rail stop, increased property values related to the rail stop, and even higher levels of public investment in these geographic areas for transit- oriented development. In terms of this last benefit, the AAF website already has renderings for the types of large-scale, transit-oriented development investment occurring or proposed for the stops in West Palm Beach, Fort Lauderdale, and Miami. But, without a train stop in Martin County, these investments are limited to railroad improvements.
It is worth noting, however, that a few studies actually found proximity to a light rail stop to have a negative impact on sales prices. Bowes and Ihlandfeldt (2001) actually found a negative impact of 19 percent associated with homes located within a quarter of a mile of a light rail stop in Atlanta, Georgia. They attribute this loss to the mix of retail in the area surrounding the stops. Additionally, while the light rail stop provides access to people looking for a transportation alternative to the car or have no car, this study also notes that access to a light rail stop may actually increase crime or result in other nuisances. Pan (2012) found similar negative impacts within a quarter mile of light rail and bus stops for the METROrail line in Houston.
More relevant to the situation with AAF and Martin County, there are studies that discuss announcements of rail improvements or expansions and the impact on property values, as well as studies that focus on impacts of rail lines in the absence of a stop. McDonald and Osuji (1995) studied the impacts of a proposed rail line on residential property values stretching from downtown Chicago to Midway Airport. While this type of rail line is not directly comparable to the AAF project, this article was included based on the attempt to measure the anticipated impact on residential land values proximate to the rail stations three years before the rail was built. Within half a mile of rail stations, residential land values increased 17 percent based on the announcement of the line and where the stations would be located. The down side of this study is that the impact of the rail in the absence of a rail station was not included. Ideally, this type of study would have focused on whether the positive impact of proximity to a rail station is offset by the negative impact of proximity to the rail only.
Strand and Vagnes (2001) conducted a study on residential values related to proximity to a railroad in Oslo, Norway. Their study used two methods to determine the impact on sales value for homes. In addition to hedonic regression, the authors also used a survey of real estate broker appraisals for the properties. For homes within 100 meters (330 feet), the loss in sales price was 7 to 10 percent, while homes within 20 meters (65.6 feet), the loss in value was 23 percent.
McMillen and McDonald (2004) published an article that was a follow up article to the previous article co-authored by McDonald. While the previous study had a relatively small sample size, this study contained 17,034 single-family sales within 1.5 miles of the rapid transit line between 1983 and 1999. While the actual construction was completed in 1993, home prices increased 4.2 percent before 1987, increasing to 19.4 percent from 1991-1996, and a decline down to 9.8 percent from 1997-1999. Despite the generally positive impact based on proximity to stations in the 1997-1999 time period, property values beyond 1.5 miles from the stations increases at a faster rate than those properties within 1.5 miles. Over the entire study period (1986-1999), the cumulative appreciation rates were 6.89 percent higher for homes within half a mile of a station.
Simons and Jaouhari (2004) examined the impacts of freight rail on residential homes in Cuyahoga County, Ohio in 1996 and 1999 using hedonic regression. For smaller homes (less than 1,250 square feet), the loss in value ranged from five to seven percent within 750 feet of the rail line. Medium-sized homes (1,251-1,700 square feet) had losses of about five percent, while large homes (over 1,701 square feet) had no statistically significant losses The purpose of this article, though, focused on additional freight trips based on the reorganization of CSX and Norfolk Southern coupled with the acquisition of Conrail. These additional trips resulted in a negative impact on smaller homes as well as larger homes. For smaller homes, a home within 250 feet had a drop in sale price of $194 per trip. This number diminished to $85 for homes 251-500 feet from the tracks and $94 for homes between 501-750 feet from the tracks. For medium-sized units, the loss per trip within 250 feet of the tracks was $262, which decreased to $107 for homes 251-500 feet from the tracks and $72 for homes 501-750 feet from the tracks.
Clark (2006) also examined the impact of an expanded, additional freight rail line on largely rural or low-density development property values in Ohio. While All Aboard Florida is not necessarily creating an additional line, the existing line will be getting added use from the new passenger trains. For every additional rail line, homes within a quarter of a mile demonstrated a loss in value of 2.1 percent to 2.8 percent. The findings also mentioned that for homes located near a crossing, the property losses were even greater due to train horns, ranging from 8.7 percent to 18.2 percent. For homes with frontage on the railroad, the loss was 32.5 percent in one county. For properties located 1,000 feet or beyond from the railroad, there was no impact.
Overall, the majority of research regarding rail and transit stops demonstrates a positive relationship between having a rail stop and property values. Where there is only the rail and no stop in sight, the impacts demonstrate a negative impact on residential property values, generally ranging from two percent to over 31 percent. This range acts as a frame of reference to determine whether Martin County residential properties fall within this range based on the existing real estate literature related to price impacts and rail proximity.
III. Data Sources and Methodology
The data used for this study are from the Martin County Property Appraiser’s Office. Annual residential sales data were originally obtained from the State of Florida Department of Revenue, but the final sales data were acquired from the Martin County Property Appraiser’s Office for sales from 2005 to February of 2015. Martin County also has datasets that include variables related to housing characteristics, lot characteristics, and other characteristics typically utilized in hedonic studies related to residential sales.
The full sales dataset originally contained 31,172 sales related to residential uses. The Florida Department of Revenue’s Property Tax Oversight Program requires county appraisers to use specific qualification codes to define the type of sales transaction that occurred. This fact is important when determining which sales to include in the analysis of residential property values. Using all 31,172 sales may result in a model that includes sales that would be disqualified in sales ratio analysis conducted by the Florida Department of Revenue (DOR). The Sales Qualification Codes required by the DOR contain 32 different codes in five groups. The five groups and a brief discussion of the sale qualification codes included are:
1. Real property transfers qualified and included in sales ratio analysis.
The codes in this grouping are the only sales qualification codes used in the DOR sales ratio analysis. There are only two codes in this group, with both codes serving as evidence that the transactions are arm’s length based on documentation. Properties in this group are the focus of this study.
2. Real property transfers qualified but excluded from sales ratio analysis.
The codes in this group related to properties whose characteristics changed after the transfer. One example could be new construction on the property, so a property originally sold as vacant may have a large change in value due to new construction. Conversely, a property with an existing residential structure on it might be demolished or destroyed in a natural disaster. This code incorporates these types of changes. Other codes included here are changes in legal characteristics, transfers involving multiple parcels with multiple parcel numbers, and single parcels that are located in more than one county. All of these properties were excluded from this study.
3. Real property transfers disqualified as a result of examination of the deed or other real property transfer instrument.
The codes in this group cover transfers to entities ranging from financial institutions to charitable organizations. Included in this group would be home sales transferred to banks in lieu of foreclosure as well as transfers related to trustees or governmental agencies. These properties were excluded from this study.
4. Real property transfers disqualified as a result of credible, verifiable, and documented evidence.
This category includes sales between families and related transactions where the transaction price might be zero or negligible. Other codes included here include sales under duress related to a home going through the foreclosure process, but the home sold before any final outcome from the judicial process. Other codes in this category include unfinished properties, mortgage fraud, atypical sales costs, and related factors.
5. Real property transfers qualification decision pending.
This category relates to errors that delayed or may eventually nullify the transfer. A property in this group may either be missing information, have incorrect information, or some other clerical error delaying the transfer. In the case of Martin County, the number of properties falling into this category since 2000 is fewer than five properties.
After accounting for the Sales Qualification Codes used in sales ratio analysis, there were a total of 21,647 residential sales. After accounting for non-comparable residential properties (apartment buildings with more than 10 units and condominiums) and properties with no or missing data (residential land sales without a structure), there were 16,254 properties.
To account for outliers based on sales prices, z-scores were calculated for sales prices. A z-score provides a statistical method for translating sales prices in to units based on standard deviations. In other words, an extremely expensive home can have an influential impact on the model because the values attributed to that home may show a much larger impact or benefit than the typical home. With this in mind, a z-score over two means that the transaction price is in approximately the top five percent highest sales in the county. In other words, homes that were more expensive than 95.4 percent of the homes in Martin County were excluded from this study as outliers, leaving a total of 15,834 property sales. The last step of the data process resulted in stratifying the housing sales into three groups for comparability. A 250-square foot mobile home within 400 feet of a railroad is not comparable to a 7,500-square foot home built on the Atlantic Ocean. Based loosely on the housing stratification methodology outlined in Simons and El Jaouhari (2004), homes less than 500 square feet or more than 4,500 square feet were removed from the dataset. A majority of units falling below 500 square feet consisted largely of mobile homes. Removing these properties from the dataset further minimized the likelihood of bias in the case of extremely small or large homes to provide a more accurate picture of typical sales transactions in Martin County. Based on square footage, there were a total of 4,473 units that were considered small units under 1,532 square feet; 4,429 medium-sized units ranged from 1,533 square feet to 2,184 square feet; and 4,430 large units that were 2,185 square feet or larger. This stratification by size resulted in a total of 13,332 property sales for inclusion in the models.
a. Housing Characteristics
Based on the dataset for this study, tables 1-4 provide descriptions of the size, sales, types, and characteristics of the housing stock in Martin County. The number of sales by year (Table 1) shows an active housing transaction market in the county for all unit sizes. The last two full years of data, 2013 and 2014, show the largest number of transactions in the county since 2005. Part of this trend is due to record low mortgage rates over the past two years. Another possible trend may be investors or people buying second homes, which is a trend occurring throughout Florida and largely fueling the real estate recovery from the Great Recession.
Excluding 2015 data that include only five weeks of sales transactions, the year with the fewest number of qualified sales transactions is 2012. Based on the transaction data, the All Aboard Florida announcement cannot be isolated as the sole reason for the decrease in the countywide transaction rate. Unlike the other years in the study, 2012 had an atypical number of transactions in two different categories. There were 574 sales that involved multiple parcels, which was the highest for any year in the study except for 2014. Unlike 2014, though, 2012 had far fewer overall qualified sales.
Perhaps the most important factor impacting the sales rate in 2012 is the fact that there were nearly as many qualified sales as there were sales falling under the disqualified category of quit claim/corrective tax/final judgment/court order deeds as a result of corrective deeds, possible transfers to or from banks involving foreclosure, and/or transfers to bankruptcy trustees, among other possible transfers. The total number of single-family residential sales in 2012 that were considered arm’s length transactions were 964, while there were 942 sales in the quit claim category. This transaction data means that, for every single-family home sold in 2012, there was a 50 percent chance that it was an arm’s length transaction and a 50 percent chance that there was some sort of atypical transaction that disqualified the property from being included in the Florida Department of Revenue’s sales ratio analysis. For comparison of 2012 to the other sale years included in this study, typically no more than 10 to 15 percent of all sales fall into the disqualified category.
Full report can be downloaded at the link above.