Commuting speed as a proxy for the value of time

Closed vs. open city models: A new empirical approach to an old question

“…As a second measure of commute costs, we employ DiBartolomeo's (2020) driving speed ( Speed ) constructed from Google Maps driving data for all census tracts contained within our sample of UAs. Higher driving speeds indicate lower time costs of commuting in the UA.…”

“… Notes : This table reports descriptive statistics for the sample of urbanized areas (UAs). The variables are defined as follows: geographic land area ( Area ), total population ( Pop ), median household income ( Inc ), the per‐acre dollar value of agricultural land including buildings ( Ag Value ), the percentage of households with at least one vehicle available ( Vehicles ), driving speed ( Speed ) estimated following DiBartolomeo (2020), the percentage of UA employment in agglomeration industries ( Agglom ) of employment in manufacturing, information, and administration (NAICS codes 31–33, 51, and 56), and governmental fragmentation ( Fragment ) as proxied by the number of Census Designated Places (CDPs).…”

“…The conclusion is robust, holding for parsimonious basic specifications as well as extended frameworks incorporating limited land supply in coastal locations, amenities, intergovernmental competition, and agglomeration economies associated with industry concentration and proximity to ocean-going ports. Following the decision rule laid out above: the DWH test results argue for shifting from the single 1 See Brueckner and Fansler (1983), Su and DeSalvo (2008), Song and Zenou (2006), Spivey (2008), McGrath (2005), Geshkov and DeSalvo (2012), Paulsen (2012), Oueslati et al (2015), DeSalvo and Su (2017), DiBartolomeo (2020), and DiBartolomeo and Turnbull (2023) for applications using the close city model. The closest antecedent to this study, DiBartolomeo and Turnbull (2021), uses an ad hoc instrumental variables approach to deal with endogenous population and income in the single equation model for their sample of smaller single county urbanized areas.…”

On urban sprawl: Closed city, open city or does it even matter?

“…DiBartolomeo (2020) also compares alternative commuting cost proxies for UAs of all sizes in 2000 and 2010 and, like Song and Zenou (2006) for agricultural land values, obtains estimates consistent with the Mills–Muth closed city model. It is worth noting that, whereas Spivey (2008) sometimes finds negative income effects, DiBartolomeo (2020) finds significantly positive income effects on city size for all commuting cost models.…”

“…Agricultural land rents are also not consistently significant across all empirical specifications, the significance varying with different commuting cost proxies, and the income effect is surprisingly negative and significant in several specifications. DiBartolomeo (2020) also compares alternative commuting cost proxies for UAs of all sizes in 2000 and 2010 and, like Song and Zenou (2006) for agricultural land values, obtains estimates consistent with the Mills–Muth closed city model. It is worth noting that, whereas Spivey (2008) sometimes finds negative income effects, DiBartolomeo (2020) finds significantly positive income effects on city size for all commuting cost models.…”