Residential parking costs and car ownership: Implications for parking policy and automated vehicles

Highlights

• We estimate residential parking costs using local linear techniques.

• We examine the effect of these costs on household car ownership, taking endogeneity issues into account.

• The implied price elasticity of demand for car ownership is about −0.7.

• Residential parking costs explain around 30% of the difference in car ownership rates between the city centre and the periphery.

• Estimates are applied to infer the potential impact of automated vehicles.

Abstract

Residents are often offered on-street parking at a fraction of the market price which may cause excess car ownership. However, residential parking costs are difficult to observe, so we propose an approach to estimate implicit residential parking costs and then examine the effect of these costs on household car ownership. We apply our approach to the four largest metropolitan areas of the Netherlands. Our results indicate that for city centres, annual residential parking costs are around €1000, or roughly 17 percent of car ownership costs, and are more than double the costs in the periphery. Our empirical estimates indicate that the disparity in parking costs explains around 30% of the difference in average car ownership rates between these areas and corresponds to a price elasticity of car demand of about −0.7. We apply these estimates to gauge the potential implications of automated vehicles which suggests that, if residents no longer require parking nearby their homes, car demand in city centres may increase by 8–14 percent.

Introduction

Parking has far reaching consequences on urban life. In cities, where land is scarce, the opportunity cost of parking is high, as on-street spots compete with pedestrian, cycling, commercial, residential and recreational uses. Nevertheless, cities devote a substantial amount of space to implicitly subsidised parking which may induce excess vehicle demand (Shoup, 2005). This raises an important open question, to what extent do parking costs affect vehicle demand in cities?¹ We address this question by estimating residential parking costs and examining to what extent these costs affect household vehicle demand.

Theory indicates that cheap residential parking reduces the (fixed) costs of owning a car and thereby increases vehicle demand (Shoup, 2005; Arnott, 2006). The empirical literature that quantifies this effect is scarce, but supports the idea that higher residential parking supply and lower residential parking rents are associated with higher car ownership (Guo, 2013; Seya et al., 2016).² Furthermore, waiting time for an on-street parking permit is shown to negatively affect vehicle demand. Residents in Amsterdam that have to wait an additional year are 2 percentage points less likely to own more than one car, corresponding to a price elasticity of demand for car ownership of −0.8 (De Groote et al., 2016).³

In order to estimate the impact of on-street parking costs on car ownership, one would like to observe market prices for on-street parking or close substitutes (for example off-street parking). In some countries, we are able to observe market rates for residential parking, as there is a thick rental market of privately-owned parking (for example Japan). However, in most countries, such a market is absent, as privately-owned parking is bundled with housing. Therefore, private off-street parking prices are not directly observed as residents mainly pay for parking through the purchase (or rental) of residential property or via regulated parking permits. Furthermore, in areas with excess demand, parking costs also include the time cost associated with cruising for parking.

This paper contributes to the literature on residential parking and car ownership by developing and applying a two-step approach which enables us to estimate local private parking costs and test to what extent these costs affect household car ownership.⁴ In the first step, we identify the implicit price for parking through the effect of an outside private parking spot – arguably an almost perfect substitute for on-street parking – on house prices.⁵ We exploit variation in the supply of private parking within a parking district to identify district-specific residential parking prices using semi-parametric hedonic house price methods.

Households considering car ownership face the same parking cost, on average, if they live in the same parking district. Hence, in the second step, we estimate the effect of residential parking costs on car ownership using variation in residential parking costs between districts. Endogenous parking costs are instrumented using the median construction year of properties in a district. Arguably, this instrument affects the supply of parking, while having no direct affect on parking demand, as it is determined in the past, often before cars were present.⁶ We acknowledge that the construction year of properties is not random over space. Therefore, more precisely, we argue that, conditional on location controls, including, most importantly, distance to the nearest major train station, and household characteristics, historical supply decisions impact current building costs of a parking space, without directly affecting current demand for cars. We discuss this identifying assumption in more detail in the methodology section.

We focus on the Netherlands. In this context, residents who do not own private parking receive parking permits at very low fees and households with private parking are, in principle, not eligible for a parking permit. Hence, it is reasonable to assume that in equilibrium, the residential parking price for households that own private parking is equal to the opportunity cost of parking on-street, which equals the sum of the permit fee and cruising costs. The latter includes private search costs, walking time and uncertainty (Van Ommeren et al., 2011).⁷ In case there is no cruising and street parking is not priced, residential parking prices should approach some underlying value of private parking, such as the security value or convenience of always having the car on hand. This approximately equals the value of private parking in locations where on-street parking is free.

We apply our approach to the four largest metropolitan regions in the Netherlands and estimate residential parking costs at the parking district level for owner-occupier households. On average, annual parking costs are around €1000 in city centres but are less than €400 in the urban periphery. We identify the impact of these costs on car ownership and find that owner-occupier households facing a one standard deviation increase (€503) in annual parking costs own 0.085 fewer cars on average, corresponding to a price elasticity of car demand of about −0.7. Our findings indicate that the disparity in parking costs between the city centre and the periphery explains around 30% of the difference in average car ownership rates between these areas.

Our results have implications for related literature on the urban spatial structure and transportation. Dense urban form is associated with lower vehicle ownership and kilometers travelled (Bento et al., 2005; Bhat and Guo, 2007; Duranton and Turner, 2018). Furthermore, transport infrastructure has been shown to affect residential location and mode choice, however parking is usually ignored (Baum-Snow, 2007; Garcia-López et al., 2015; Baum-Snow et al., 2017; Levkovich et al., 2017; Heblich et al., 2018). Our findings shed light on one of the mechanisms which explains why car ownership levels are lower in dense urban areas and indicates that residential parking costs are a significant determinant of mode choice.

Our findings also have implications for residential parking policy and relate to the growing literature on estimating the potential effects of automated vehicles (AVs).⁸ We employ our estimates to consider the potential implications of raising fees of parking permits to the market value and eliminating parking costs from a widespread adoption of AVs. Increasing permit fees in the city centre of Amsterdam to the market value is expected to reduce average car ownership by 17–24 percent. Furthermore, the average annual gains per household from facing lower parking costs are estimated to be between €450 and €850 in city centres, depending on whether AVs are privately owned or shared. This is associated with an increase in average car demand between 8 and 14 percent. The effects are smaller in the periphery where parking costs are lower.

The paper proceeds as follows. In Section 2 we introduce the research context, data and provide some descriptives. In Section 3 we elaborate on the methodology. We report the main results in Section 4 and provide a counterfactual analysis in Section 5. Finally, Section 6 concludes.