1# RECORD 2b0979784a826ff5181dbb43492b0239ca3062e82# TITLE: The Value of a Sea View: Hedonic Estimates Using 3D Simulation and Natural Language Processing3# YEAR: 2023.0 VENUE: Land Economics DOI: 10.3368/le.100.2.071522-0056r4# AUTHORS: Tom Gillespie; Ronan C. Lyons; Stephen Hynes5# PAGES: 42 CANDIDATE TABLE PAGES (1-based): [36, 37, 38]6# ABSTRACT: This article develops a new measure of sea view breadth and depth, based on a combination of GIS 3D simulation and natural language processing (NLP) techniques. NLP identifies properties that have a sea view, while GIS measures its extent. We estimate the value of a sea view using a hedonic housing price model and a sample of over 100,000 sales listings in Ireland, 2014–2019. We find that a sea view can add significant value to a property, all else being equal, with an average premium of 8.1% rising to 15% for a wide breadth of sea view.789===== PAGE 1 (score 4, numeric-paren 0, table-words 0, var-words 4) =====10111 12The Value of a Sea View: Hedonic Estimates Using 3D Simulation and Natural 13Language Processing 14 15Dr. Tom Gillespie 16Prof. Ronan C. Lyons† 17Prof. Stephen Hynes‡ 18 19 20Abstract 21This paper develops a novel measure of sea-view breadth and depth, based on a combination 22of GIS 3D simulation and Natural Language Processing (NLP) techniques. NLP identifies 23properties that have a sea view, while GIS measures its extent. It then estimates the value of a 24sea view using a hedonic housing price model and a sample of over 100,000 sales listings in 25Ireland, 2014-2019. We find that a sea view can add significant value to a property, all else 26being equal, with an average premium of 8.1% rising to 15% for a wide breadth of sea view. 27 28 29 30 31 32 33 34 35 Socio-Economic Marine Research Unit, Whitaker Institute, University of Galway. 36Thomas.gillespie@universityofgalway.ie 37† Department of Economics, Trinity College Dublin, and Spatial Economics Research Centre, LSE. 38Ronan.lyons@tcd.ie 39‡ Socio-Economic Marine Research Unit, Whitaker Institute, University of Galway. 40Stephen.hynes@universityofgalway.ie 41doi:10.3368/le.100.2.071522-0056R42by guest on October 20, 2026. Copyright 202343Downloaded from 444546===== PAGE 36 (score 22, numeric-paren 0, table-words 4, var-words 8) =====474836 49Table 3: Baseline results 50Treatment of SE 51Robust Standard errors & Clustered at level of SFE 52Level of spatial FE 53Local 54Market 55Micro 56market 57Electoral 58District 59Small 60Area 61ED*Ye62ar 63SA + 64MM*Year 65Dependent Variable: natural log of the listed sale price 66 67Coastal Variables 68 69Distance To Coastline 70(log) 71-0.038 72-0.046 73-0.040 74-0.061 75-0.038 76-0.061 77 780.00 790.00 800.00 810.00 820.00 830.00 84 85 86Distance To Transitional 87(log) 88-0.012 89-0.012 90-0.018 91-0.014 92-0.018 93-0.014 94 950.34 960.10 970.00 980.01 990.01 1000.01 101 102 103Sea View Variables 104 105Seaview breadth 1060.010 1070.011 1080.011 1090.011 1100.011 1110.011 112 1130.08 1140.01 1150.00 1160.00 1170.00 1180.00 119Seaview depth 1200.023 1210.020 1220.018 1230.013 1240.017 1250.013 126 1270.04 1280.00 1290.00 1300.00 1310.00 1320.00 133 134 135Controls 136YES 137YES 138YES 139YES 140YES 141YES 142Observations 143100,030 144100,030 145100,030 146100,030 147100,030 148100,030 149R-squared 1500.800 1510.841 1520.854 1530.882 1540.868 1550.890 156RMSE 1570.289 1580.258 1590.248 1600.228 1610.240 1620.222 163Number of absorbed SFE 16450 165261 166832 1675,571 1684,509 1695,571 170Notes: Robust p-values, clustered at the level of spatial fixed effect, are reported below coefficients. Columns show 171different levels of spatial fixed effects denoted at header, while controls include location, dwelling and time listed on the 172market, as discussed in the text. 173 174 175 176 177 178 179 180 181 182 183 184 185by guest on October 20, 2026. Copyright 2023186Downloaded from 187188189===== PAGE 37 (score 20, numeric-paren 0, table-words 4, var-words 7) =====19019137 192Table 4: Separating GIS and NLP measures of views 193Treatment of SE 194Robust Standard errors & Clustered at level of SFE 195Level of spatial FE 196Small Area & Micro-Market*Year 197Specification Number 198Baseline 1991 2002 2013 2024 2035 204Dependent Variable: natural log of the listed sale price 205Coastal Variables 206 207Distance To Coastline (log) 208-0.061 209-0.062 210-0.062 211-0.062 212-0.061 213-0.060 214 2150.00 2160.00 2170.00 2180.00 2190.00 2200.00 221 222 223Distance To Transitional (log) 224-0.014 225-0.015 226-0.014 227-0.015 228-0.012 229-0.013 230 2310.01 2320.01 2330.01 2340.01 2350.02 2360.01 237 238 239Sea View Variables 240 241Seaview breadth 2420.011 243 244 2450.00 246 247Seaview depth 2480.013 249 250 2510.00 252 253Sea View Variables (GIS only) 254 255Seaview breadth 2560.003 2570.008 258 259 2600.14 2610.00 262 263Seaview depth 2640.013 2650.015 266 267 2680.00 2690.00 270 271 272 273Sea Views NLP 2740.082 275 276 2770.00 278 279Sea Views NLP (GIS=0) 2800.079 281 2820.00 283 284 285 286 287Controls 288YES 289YES 290YES 291YES 292YES 293YES 294Observations 295100,030 296100,030 297100,030 298100,030 299100,030 300100,030 301R-squared 3020.890 3030.890 3040.890 3050.890 3060.890 3070.890 308RMSE 3090.222 3100.222 3110.222 3120.222 3130.222 3140.222 315Number of absorbed SFE 3165,571 3175,571 3185,571 3195,571 3205,571 3215,571 322Notes: Robust p-values, clustered at the level of spatial fixed effect, are reported below coefficients. Columns show 323different robustness specifications numbered in the headers, while controls include location, dwelling and time listed on 324the market, as discussed in the text. 325 326 327 328 329 330by guest on October 20, 2026. Copyright 2023331Downloaded from 332333334===== PAGE 38 (score 21, numeric-paren 0, table-words 4, var-words 9) =====33533638 337Table 1: Controlling for the total viewshed of a property 338Treatment of SE 339Robust Standard errors & Clustered at level of SFE 340Level of spatial FE 341Small Area & Micro-Market*Year 342Specification 343Baseline 344Adding Land 345Viewshed 346Adding Slope 347Adding Land 348Viewshed and 349Slope 350Dependent Variable: natural log of.. 351Listed sale price 352Coastal Variables 353 354Distance To Coastline (log) 355-0.060 356-0.061 357-0.060 358-0.061 359 3600.00 3610.00 3620.00 3630.00 364 365 366Distance To Transitional (log) 367-0.015 368-0.016 369-0.015 370-0.016 371 3720.01 3730.00 3740.01 3750.00 376 377 378Sea View Variables 379 380Seaview breadth 3810.011 3820.011 3830.011 3840.011 385 3860.00 3870.00 3880.00 3890.00 390 391 392Seaview depth 3930.013 3940.012 3950.013 3960.012 397 3980.00 3990.00 4000.00 4010.00 402 403 404Land view and Slope Controls 405 406Log of Land Area Viewshed 4070.005 4080.005 409 4100.00 4110.00 412 413 414Log of % Rise of Slope 4150.000 4160.001 417 4180.97 4190.83 420 421 422 423 424 425 426Controls 427YES 428YES 429YES 430YES 431Observations 43299,447 43399,447 43499,447 43599,447 436R-squared 4370.890 4380.890 4390.890 4400.890 441RMSE 4420.221 4430.221 4440.221 4450.221 446Number of absorbed SFE 4475,566 4485,566 4495,566 4505,566 451Notes: Robust p-values, clustered at the level of spatial fixed effect, are reported below coefficients. Columns show different 452levels of spatial fixed effects denoted at header, while controls include location, dwelling and time listed on the market, as 453discussed in the text. 454 455 456by guest on October 20, 2026. Copyright 2023457Downloaded from 458459460===== PAGE 39 (score 7, numeric-paren 0, table-words 2, var-words 3) =====46146239 463Figure Titles 464Figure 1. Distribution of visible points as a function of distance to coastline 465 466Note: The above figure shows summary statistics of the number of (inner) points visible for properties with at least one inner 467point visible in their viewshed, by distance to the coastline, with sample sizes for each distance bin reported in the horizontal 468axis. 469Figure 2. Graph of sea view coefficients by level of SFE – Illustrating the results from Table 3. 470Notes: The above figure displays point estimates and confidence intervals for the results in table 3. 471 472 473by guest on October 20, 2026. Copyright 2023474Downloaded from 475