THE IMPACT OF THE DEEPWATER HORIZON GULF OIL SPILL ON GULF COAST REAL ESTATE MARKETS

Size: px

Start display at page:

Download "THE IMPACT OF THE DEEPWATER HORIZON GULF OIL SPILL ON GULF COAST REAL ESTATE MARKETS"

Brian Lynch
5 years ago
Views:

Cole, PhD, CRE Kaye Family Endowed Professor Florida Atlantic University

1 THE IMPACT OF THE DEEPWATER HORIZON GULF OIL SPILL ON GULF COAST REAL ESTATE MARKETS Rebel A. Cole, PhD, CRE Kaye Family Endowed Professor Florida Atlantic University Department of Finance Richard J. Roddewig, JD, MAI, CRE. FRICS Charles T. Brigden, CRE FRICS Jones Lang LaSalle Valuation & Advisory Services

2 INTRODUCTION In this study, we examine the impact of the Deepwater Horizon oil spill on the value of gulf-front residential properties located in Baldwin County, Alabama. Using both hedonic regression analysis and traditional real estate appraisal techniques, we estimate the average diminution in value to more than 8,000 condominiums located along the Gulf of Mexico. We use condominiums located on the eastern U.S. coast in Volusia County, FL and Horry County, SC as control properties.

3 INTRODUCTION We find no evidence of a statistically significant diminution in value during the periods after the spill date. Instead, we find strong evidence of price appreciation, beginning six months after the spill date and strengthening during the subsequent four years. We attribute this, at least in part, to the clean up effort & the influx of clean-up workers to the Gulf area as well as to tourism promotion efforts focused on success of the cleanup effort.

April 20, 2010 Macondo Well blew out The explosion killed 11 and became the worst oil spill in history 210 million gallons of oil spilled into the ocean before the well was capped on

4 April 20, 2010 Macondo Well blew out The explosion killed 11 and became the worst oil spill in history 210 million gallons of oil spilled into the ocean before the well was capped on September 19, 2010 Oil eventually reached shore from Grand Isle, LA to Destin, FL. The resulting shoreline assessment (SCAT) process and shoreline cleanup process extended into March 2014.

5 LITERATURE REVIEW Epley, D. (2012). The Gulf Oil Spill and Its Impact on Coastal Property Value Using the Before-and-After Procedure. Journal of Real Estate Literature, Vol. 20, No. 1, p Siegel, C., S.B. Caudill, and F.G. Mixon Jr. (2013). Clear Skies, Dark Waters: The Gulf Oil Spill and the Price of Coastal Condominiums in Alabama. Economics and Business Letters,Vol. 2, No. 2, p Winkler, D.T., and B.L. Gordon (2013). The Effect of the BP Oil Spill of Volume and Selling Prices of Oceanfront Condominiums. Land Economics, Vol. 89, No. 4, p

6 DATA Our sample consists of relatively homogenous condominium properties located in high-rise towers. We collect sales data from Jan. 01, 2000 through Dec. 31, We limit the early data to sales after Dec. 31, 2005 to limit the influence of the sharp run-up in prices observed during Our treatment area is the Gulf coast county Baldwin, AL, consisting primarily of properties located in Gulf Shores and Orange Beach. AL. Our control areas are the Atlantic coast counties Volusia, FL and Horry, SC.

7 SALES DATA ANALYSIS Baldwin County, Alabama Horry County, South Carolina Volusia County, Florida

9 HORRY COUNTY, SOUTH CAROLINA

10 VOLUSIA COUNTY, FLORIDA

11 GULF OF MEXICO - MAXIMUM OILING

12 METHODOLOGY To estimate the price impact of the oil spill, we use both univariate and multivariate tests. First, we use deed-based sales transactions from real-estate-data vendor MMT and then perform extensive analysis to remove non-arms length transactions and partial interest transfers Sales are organized by resort or development so that the data may be verified for accuracy The top 75% of the towers are the focus to help ensure data integrity and homogeneity Sales trend analysis and repeat sales analysis are performed

13 METHODOLOGY To estimate the price impact of the oil spill in a multivariate context, we use hedonic regression models. Our dependent variable is the natural logarithm of the observed sales price of residential properties. Our explanatory variables are key property characteristics, such as square footage and building age, as well as locational characteristics, such as location in an ocean-front condominium, likely to be affected by the oil spill. Our identification strategy is the "difference-in-differences" of values in the contamination area and in the control areas.

14 METHODOLOGY To control for general price trends over time, we time-adjust the sales prices in each county using a condominium price index for a nearby county. On the Gulf Coast, we use the Destin area condo price index. On the Atlantic Coast, we use the St. Augustine Beach condo price index. We cannot use condo price indices for our counties because they are based for the most part on sample condominium sales. We estimated the correlation of Destin with Baldwin and found it to be greater than We estimated the correlation of St. Augustine with Volusia and with Horry, and found them to be greater than 0.90.

15 METHODOLOGY We also investigated using the FHFA housing price index for each county, We found that the county-wide indices do a very poor job of tracking oceanfront condominium prices, which saw much greater appreciation much earlier in the 2000s and saw earlier and more rapid declines in prices following onset of the financial crisis.

16 METHODOLOGY To test how long any price impact lasted, we iteratively limit our postspill period to six months, one year, two years, three years, and four years after the spill date. The coefficient on PostGulf then measures the average price impact of the spill over that period.

17 METHODOLOGY lnprice i = F (property characteristics i, Post i,gulf i, Post i Gulf i ) lnprice i is the natural logarithm of sales price. Property characteristics include: square footage, building age, number of bedrooms, number of bathrooms, etc. Gulf i is an indicator for an ocean-front condominium located in a Baldwin County, Alabama. Omitted category is location in one of our two control counties on the Atlantic Coast (Volusia County, FL and Horry County, SC).

23 EVIDENCE FROM REPEAT-SALES

24 DESCRIPTIVE STATISTICS Variable Mean Median Minimum Maximum Adj. Sales Price 191, ,381 16,566 1,448,121 Adj. Square Footage 1, ,504 Building Age Beds Baths REO Sale Post Post-Gulf

25 HEDONIC REGRESSION RESULTS Full Sample Full Sample Full Sample 48 Months Post 36 Months Post 24 Months Post 12 Months Post 6 Months Post (1) (2) (3) (4) (5) (6) (7) (8) Variables Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Constant *** *** *** *** *** *** *** *** ln(adjusted Square *** *** *** *** *** *** *** *** ln(building Age) *** *** *** *** *** *** *** Baths *** *** *** No Bedrooms *** *** *** *** *** *** *** One Bedroom *** *** *** *** Tw o Bedrooms *** ** *** *** *** Three Bedrooms *** *** *** *** *** *** REO Sale *** *** *** *** *** *** *** Gulf *** *** *** *** *** *** Post *** *** *** *** *** ** Post-Gulf *** *** *** *** Observations 10,333 10,333 10,333 9,471 8,004 6,479 5,135 4,469 R-squared

26 HEDONIC REGRESSION RESULTS Full Sample Full Sample Full Sample 48 Months Post 36 Months Post 24 Months Post 12 Months Post 6 Months Post (1) (2) (3) (4) (5) (6) (7) (8) Variables Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Constant *** *** *** *** *** *** *** *** ln(adjusted Square *** *** *** *** *** *** *** *** ln(building Age) *** *** *** *** *** *** *** Baths *** *** *** No Bedrooms *** *** *** *** *** *** *** One Bedroom *** *** *** *** Tw o Bedrooms *** ** *** *** *** Three Bedrooms *** *** *** *** *** *** REO Sale *** *** *** *** *** *** *** Gulf *** *** *** *** *** *** Post *** *** *** *** *** ** Post-Gulf *** *** *** *** Observations 10,333 10,333 10,333 9,471 8,004 6,479 5,135 4,469 R-squared

27 HEDONIC REGRESSION RESULTS Full Sample Full Sample Full Sample 48 Months Post 36 Months Post 24 Months Post 12 Months Post 6 Months Post (1) (2) (3) (4) (5) (6) (7) (8) Variables Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Constant *** *** *** *** *** *** *** *** ln(adjusted Square *** *** *** *** *** *** *** *** ln(building Age) *** *** *** *** *** *** *** Baths *** *** *** No Bedrooms *** *** *** *** *** *** *** One Bedroom *** *** *** *** Tw o Bedrooms *** ** *** *** *** Three Bedrooms *** *** *** *** *** *** REO Sale *** *** *** *** *** *** *** Gulf *** *** *** *** *** *** Post *** *** *** *** *** ** Post-Gulf *** *** *** *** Observations 10,333 10,333 10,333 9,471 8,004 6,479 5,135 4,469 R-squared

28 HEDONIC REGRESSION RESULTS Full Sample Full Sample Full Sample 48 Months Post 36 Months Post 24 Months Post 12 Months Post 6 Months Post (1) (2) (3) (4) (5) (6) (7) (8) Variables Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Constant *** *** *** *** *** *** *** *** ln(adjusted Square *** *** *** *** *** *** *** *** ln(building Age) *** *** *** *** *** *** *** Baths *** *** *** No Bedrooms *** *** *** *** *** *** *** One Bedroom *** *** *** *** Tw o Bedrooms *** ** *** *** *** Three Bedrooms *** *** *** *** *** *** REO Sale *** *** *** *** *** *** *** Gulf *** *** *** *** *** *** Post *** *** *** *** *** ** Post-Gulf *** *** *** *** Observations 10,333 10,333 10,333 9,471 8,004 6,479 5,135 4,469 R-squared

29 HEDONIC REGRESSION RESULTS Full Sample Full Sample Full Sample 48 Months Post 36 Months Post 24 Months Post 12 Months Post 6 Months Post (1) (2) (3) (4) (5) (6) (7) (8) Variables Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Constant *** *** *** *** *** *** *** *** ln(adjusted Square *** *** *** *** *** *** *** *** ln(building Age) *** *** *** *** *** *** *** Baths *** *** *** No Bedrooms *** *** *** *** *** *** *** One Bedroom *** *** *** *** Tw o Bedrooms *** ** *** *** *** Three Bedrooms *** *** *** *** *** *** REO Sale *** *** *** *** *** *** *** Gulf *** *** *** *** *** *** Post *** *** *** *** *** ** Post-Gulf *** *** *** *** Observations 10,333 10,333 10,333 9,471 8,004 6,479 5,135 4,469 R-squared

30 HEDONIC REGRESSION RESULTS Full Sample Full Sample Full Sample 48 Months Post 36 Months Post 24 Months Post 12 Months Post 6 Months Post (1) (2) (3) (4) (5) (6) (7) (8) Variables Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Constant *** *** *** *** *** *** *** *** ln(adjusted Square *** *** *** *** *** *** *** *** ln(building Age) *** *** *** *** *** *** *** Baths *** *** *** No Bedrooms *** *** *** *** *** *** *** One Bedroom *** *** *** *** Tw o Bedrooms *** ** *** *** *** Three Bedrooms *** *** *** *** *** *** REO Sale *** *** *** *** *** *** *** Gulf *** *** *** *** *** *** Post *** *** *** *** *** ** Post-Gulf *** *** *** *** Observations 10,333 10,333 10,333 9,471 8,004 6,479 5,135 4,469 R-squared

31 HEDONIC REGRESSION RESULTS Full Sample Full Sample Full Sample 48 Months Post 36 Months Post 24 Months Post 12 Months Post 6 Months Post (1) (2) (3) (4) (5) (6) (7) (8) Variables Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Constant *** *** *** *** *** *** *** *** ln(adjusted Square *** *** *** *** *** *** *** *** ln(building Age) *** *** *** *** *** *** *** Baths *** *** *** No Bedrooms *** *** *** *** *** *** *** One Bedroom *** *** *** *** Tw o Bedrooms *** ** *** *** *** Three Bedrooms *** *** *** *** *** *** REO Sale *** *** *** *** *** *** *** Gulf *** *** *** *** *** *** Post *** *** *** *** *** ** Post-Gulf *** *** *** *** Observations 10,333 10,333 10,333 9,471 8,004 6,479 5,135 4,469 R-squared

32 SUMMARY AND CONCLUSIONS In this study, we have analyzed treatment and control samples of waterfront condominiums for evidence on the price impact of the Deepwater Horizon oil spill using multi-variate regression modeling, scatter plot/time line comparisons & repeat sales analysis. Scatter plot/time line comparisons & repeat sales analysis show possible small temporary impacts on prices during the first few months to one year following the spill. Hedonic regression analysis provides strong evidence of price appreciation beginning six months after the spill and strengthening during the four years after the spill date. The quick end to the possible temporary adverse effects on prices and values and the strong rebound in prices is due, at least in part, to the clean up efforts, the influx of cleanup workers, and the tourism promotion efforts.

Internet Appendix. For. Birds of a feather: Value implications of political alignment between top management and directors

Internet Appendix. For. Birds of a feather: Value implications of political alignment between top management and directors Internet Appendix For Birds of a feather: Value implications of political alignment between top management and directors Jongsub Lee *, Kwang J. Lee, and Nandu J. Nagarajan This Internet Appendix reports