The effects of land price in the peri-urban fringe of Mexico City: Environmental amenities for informal land parcel purchasers

Abstract

Chinese

In many developing countries, urban growth is characterised by the emergence of informal housing at the periphery. Nevertheless, there is little evidence based on data from informal land markets and, in general, studies focusing on such markets often neglect environmental factors. Therefore, to contribute to these research gaps, this article aims to enhance our understanding of land markets in informal land parcels and their relationship to environmental amenities, by providing empirical evidence from Mexico City. The article estimates a hedonic pricing model using robust ordinary least squares with a SHAC (Spatial Heteroskedasticity and Autocorrelation Consistent) inference, including structural, environmental, neighbourhood and accessibility features. Results provide empirical insights regarding the way this land market behaves in the peri-urban area. Our findings reveal that informal land parcel purchasers are willing to pay for basic services such as access to piped water, proximity to schools and accessibility features such as being close to city centre, motorways and underground stations. Although a positive relationship between land price and distance to the nearest forest or Protected Natural Area is highlighted, it is low, meaning that individuals are largely ambivalent about environmental amenities. Therefore, the problem of irregular settlements could be approached from two different angles. Firstly, informal land buyers will not desist from invading and modifying natural areas without a comprehensive urban and environmental policy, oriented towards changing the perception of green areas as potential urbanisation opportunities. Secondly, public policy needs to solve the housing supply crisis, considering the characteristics presented here.

Keywords

agglomeration/urbanisation environment environment/sustainability hedonic model housing informality land use

Introduction

In many developing countries, urban growth is characterised by the emergence of informal settlements at the periphery (Hardoy and Satterthwaite, 2014; Wigle, 2010). In 2014, approximately 883 million people lived in such settlements in developing regions (United Nations, 2018), and for Latin America and the Caribbean the estimated population was about 105 million, almost 21% of the total population (United Nations-Habitat, 2016).

This rapid urbanisation has outstripped the capacity of local governments and private housing markets to plan and provide land for housing, especially for low-income people. This, in turn, encourages the creation of informal land markets. This phenomenon can be explained by continued rural–urban immigration, growing demand for living space in urban areas, elevated conditions of poverty and land use deregulations (Crane et al., 1997; Das et al., 2017).

The land market is not elastic, and it can be either formal or informal, or a mix of both (Birch et al., 2016). The formal market is based on national or local laws and government regulations on either land rights or specific zoning. Otherwise, people can obtain land through informal instruments such as informal purchase, illegal occupation, inheritance, donation, among others (Twarabamenye and Nyandwi, 2012). In the case of informal land sales, their cross-elasticity of demand is low since no substitutes exist. Yet, these kinds of settlements are managed by actors who have some weight in the real estate market. Therefore, any assessment of urban housing markets in developing countries should also consider the informal sector to avoid misleading estimates (Jimenez, 1982). It is important to note that the behaviour and effectiveness of the market depend on the context, the local history of informal settlements and the political regimes of each country. Hence, it is necessary to conduct empirical research worldwide (Zhang and Zhao, 2018).

In this regard, two main categories of informal settlements have been distinguished (Durand-Lasserve and Selod, 2009: 106). The first category includes unauthorised commercial land developments – frequently built on private land – where land is parcelled illegally, mostly by informal developers, and sold out as parcels. The illegal character of the subdivision must be highlighted, which arises either due to a violation of zoning and planning regulations, or because the obligatory permit for land subdivision has not been granted. When the land is illegally occupied, against the will of the landowner, it refers to the second type: squatter settlements on public or private land. In this article, the latter will not explicitly be discussed, as we focus mainly on unauthorised commercial land developments. An additional interest in this case study is that several unauthorised commercial transactions are carried out in a declared ecological area in the south of Mexico City.

Urban informality is common in peri-urban areas (Allen et al., 2006; Cohen, 2006) and frequently has environmental impacts (Benítez et al., 2012; Zeilhofer and Topanotti, 2008). Some authors (Ravetz et al., 2013: 16) point out that the dominant urban form might be precisely the peri-urban one, and that consequently it is the most important urban spatial planning challenge of the 21st century. However, most research presents and analyses the land access crisis and contemporary environmental problems separately.

Local values such as neighbourhood features, public goods, location-specific facilities and environmental amenities are capitalised in land values (Cheshire and Sheppard, 2004). Therefore, hedonic pricing models (HPMs) have been used for decades, mostly in developed countries, to provide useful information about the willingness to pay for different characteristics and amenities. Nevertheless, to the best of our knowledge, there are few studies on informal markets in informal land parcels in environmental areas (Everett, 2001; Moschella, 2018). Previous studies have focused mainly on slum and squat dwellers in developing countries, under the concern of land tenure and property rights. In India, Nakamura (2017) applied a spatial hedonic model to study how slum residents appraise formalised land. In Peru, Hawley et al. (2018) carried out an assessment of how squatting and property rights affect monthly rental values. Zhang and Zhao (2018) analysed how informal institutions affect the pricing mechanism in the Chinese informal housing market. In the 1980s, another concern was to estimate the demand for housing characteristics in countries such as Korea, the Philippines, Colombia and El Salvador (Follain and Jimenez, 1985; Follain et al., 1980; Lim et al., 1984; Quigley, 1982). However, environmental amenities have often been neglected. Some authors (Arimah, 1992; Crane et al., 1997) have estimated hedonic models in developing countries including environmental amenities, although they have focused mostly on infrastructure conditions (such as waste disposal, water supply, neighbourhood roads) rather than ecological assets. Das et al. (2017) showed slum occupants’ preferences in India, including variables related to a clean environment, but they did not consider green areas. An HPM in Chile measured the influence of a green area in slums, but the variable was not statistically significant (Espinoza and Balaguer, 2009).

In Mexico, as in many other Latin American countries, the bulk of urban housing production corresponds to irregular settlements (Connolly, 2009). The main reason is that access to housing for a high percentage of the population is only possible through informal settlement means, due to low income or to not meeting the requirements for mortgage credit or legal rental contracts (Connolly and Wigle, 2017). In this study, we only consider Mexico City proper (CDMX), not Greater Mexico City which also comprises municipalities surrounding Mexico City.

As the informal land market and environmental values trade-off is a challenge that needs attention, this article aims to enhance our understanding of land markets of informal land parcels and their relationship to environmental amenities, by providing empirical evidence from Mexico City as a case study. Therefore, this article addresses the following research questions: which characteristics determine the land value under a land market in informal parcels in a restricted environmental area in theperi-urban area of Mexico City? Does the value of environmental amenities play a role in such a market?

The outline of the article is as follows. The next section presents the basic features of the Mexican context. In the third section, the methodology is given, followed by the estimated results and a discussion of the findings. Finally, some policy implications are offered.

The institutional context in Mexico

From 1900 to 1970, Mexico City’s population increased rapidly, from 0.7 million inhabitants in 1900 to 6.9 million in 1970. By 2015, the population accounted for 8.9 million inhabitants (7.5% of the country’s total), of whom 99.5% were urban and 0.5% were rural (Instituto Nacional de Estadística y Geografía (INEGI), 2015).

The city is comprised of 16 districts. Four of these constitute the central city. The remaining 12 surround this core and have boundaries with other federal provinces. Districts with natural areas (i.e. Conservation Land, CL) are shown in Figure 1: Cuajimalpa de Morelos, Álvaro Obregón, La Magdalena Contreras, Tlalpan, Xochimilco, Milpa Alta and Tláhuac. From 1970 to 1980, the population of the central city decreased, while the surrounding districts presented an accelerated growth. Although the rate of growth has diminished in the last decades, it is still higher (12%) than in the central city (3%) (INEGI, 2015).

Figure 1.

Study area.

With population growth, urban expansion gradually spread into the countryside without following a development plan, and thus many irregular settlements appeared. For example, from 2003 to 2007, 5300 ha of CL were lost, corresponding to a rate of about 1300 ha per year (Aguilar and Santos, 2011). Besides this, about 867 irregular settlements have been registered in almost 2800 ha (Aguilar, 2008, 2016).

In fact, urban sprawl into conservation areas has been facilitated by the nature of the Mexican land property regime. Rural and peri-urban land can be private or social (communal or ejidal) (Jones and Pisa, 2000). The Mexican ejido is a land tenure system that resulted from the Mexican Revolution of 1910. Article 27 of the 1917 Mexican Constitution guaranteed land to the landless (ejidal land) and restored land to the indigenous communities (communal land). Thus, a given community is the owner of their corresponding land, while each individual has a given parcel of land with the right to work on it (Vázquez-Castillo, 2004). Ejidos and rural communities were legally blocked from selling their land. However, a Constitutional amendment in 1992 allowed them to sell their land, accelerating the urbanisation process throughout the country (Aguilar and Santos, 2011).

Changes in the legal framework have been implemented since the 1970s in an attempt to hinder urban sprawl. Hence, in 1978, the Urban Development Plan of Mexico City divided the city in two categories, urban and non-urban, this latter with two zones: a Buffer Zone or transitional area between urban and rural spaces; and a Preservation Zone, which in 1987 was classified as Conservation Land (Aguilar, 2008). CL covers about 59% (88,442 ha) of Mexico City’s territory and is defined as an area that should be preserved since it offers multiple ecosystem services (PAOT and SEDEMA, 2012). Nevertheless, despite being listed as a conservation area, it has many human settlements within its boundaries, in particular traditional towns (ejidos and communities) that used to be dedicated to agricultural activities (Aguilar, 2008).

Figure 1 shows in green the area of the original CL polygon. It is not a completely forested area – there are large areas with forest cover and several transition zones corresponding to different forest, agricultural and cultural uses, in addition to urban recognised areas – but urban developments are not authorised here (GDF, 2000).

It is important to highlight that only 9.3% of the CL is considered a Protected Natural Area (PNA), under the jurisdiction of the local Ministry of the Environment (SEDEMA).

Methods

Background to hedonic pricing modelling

One of the most widely used tools for evaluating the economic consequences of public policies is the hedonic pricing model (HPM), particularly where the supply of environmental amenities is concerned (Kuminoff et al., 2010). This revealed preference method of valuation relies on information provided by individuals when they make decisions in real markets to assess household preferences. The fundamental hypothesis is that complex goods (e.g. houses or parcels) are a combination of attributes. Consequently, the total price is composed of the elementary prices of each of the characteristics (Rosen, 1974).

In general, researchers take real market transactions and from these data estimate the marginal rate of substitution, thereby revealing the buyer’s marginal willingness to pay for a marginal variation of this attribute. Nevertheless, asking prices have also been used, for example in the study by Cheshire and Sheppard (1989) to analyse urban planning in Great Britain, and by Martínez and Viegas (2009) to evaluate the effect of transportation accessibility on housing prices. Furthermore, Van Eggermond et al. (2011) found in their study of Singapore that using asking prices and transaction prices yields similar results, despite the large differences between these types of prices. In our study, given that it is an informal market, no transaction records are available, so our analysis relies on asking prices (see data section below).

Regarding the urban fringe, the HPM has been applied to assess the value of farmland (Wasson et al., 2010) environmental, landscape (Cavailhès et al., 2009; Schläpfer et al., 2015) and forest amenities (Tuffery, 2017). Few analyses of this type have been conducted for Mexico City, despite it being a megalopolis. This might be due, according to Sobrino (2014), to a lack of reliable information on the housing market. Lara-Pulido et al. (2017) assessed the value of suburban housing construction.

We base our analysis on the preliminary results reported in a conference paper by Martínez-Jiménez et al. (2017). These authors find that several factors have an influence in determining the price of a land parcel, but that the presence of a sewage system is the main feature that determines a higher price. In the present article, we expand the analysis by allowing nonlinearity in the impact of distance variables and by using a spatial-heteroskedasticity and autocorrelation-consistent variance-covariance matrix for inference, which is a more robust form for dealing with both unknown forms of heteroskedasticity and spatial autocorrelation, as is the case here. We also include a thorough discussion of policy implications.

The hedonic pricing method consists of two stages (Anselin and Lozano-Gracia, 2009: 1217). First, the hedonic price function is specified in a set of characteristics with a model that relates an individual’s choices to the relevant prices. Assuming a market in equilibrium, the implicit price is estimated by regressing the land price against the parcel characteristics, calculating the partial derivative of price concerning the relevant attribute. The result is a price differential that reflects the marginal rate of substitution or marginal willingness to pay for the non-market attribute. The second stage refers to the construction of the inverse demand function for property characteristics. However, in most applied work this stage is not carried out. Accordingly, in this article, we only develop the first stage.

The model

We apply a spatial-heteroskedasticity and autocorrelation-consistent hedonic pricing model. Our first assumption is to consider a private market containing n separate land parcels, each one of these presenting k characteristics, which can be either structural or locational. Our second assumption is that the price of a parcel i is considered to be the combination of these characteristics. Our third assumption is that such prices can be estimated with a semi-logarithmic model, as is usual:

1 n y = X β + ε

(1)

Where 1n y is the (N, 1) vector of observations on the parcels’ prices expressed in logarithm and X is the (N, k) matrix of observations for the explanatory variables which were grouped in four categories (see data section below). Furthermore, β is the (k, 1) vector of the model parameters which can be interpreted as the semi-elasticities between hedonic prices and the explanatory variables. Finally, ε is the error term, an (N, 1) vector.

As in Ayouba et al. (2020), we use B-spline functions for taking into account the non-linearity of the impact of some variables on the parcels’ price in a parametric framework. This approach consists of estimating piecewise polynomial functions with smoothing constraints on the knots to ensure continuity of the function. B-spline functions require that the number of knots and the order of the splines are exogenously chosen. We choose third-order polynomials, as they yield enough flexibility to capture non-linearities in our specification without inducing too much multicollinearity.

From an econometric point of view, in the presence of heteroskedastic and/or spatially autocorrelated error terms, the OLS estimator remains consistent but is no longer efficient and statistical inference is biased. Consistency is lost when a spatial lag is wrongly omitted. To tackle such issues, we first estimate our models including district dummies to control for market heterogeneity. Second, we take into account spatial error autocorrelation and heteroskedasticity by following the method proposed by Kelejian and Prucha (2007): the SHAC (Spatial Heteroskedasticity and Autocorre-lation Consistent) estimator of the variance-covariance matrix of the error terms. This estimator of the variance-covariance matrix remains consistent in the presence of unknown forms of heteroskedasticity and spatial autocorrelation. However, we do not include a spatial lag term in our model, following the argument set forth by Anselin and Lozano-Gracia (2009). Indeed, Anselin and Lozano-Gracia argue that hedonic price equations define a market equilibrium after all interactions between supply and demand have taken place. Then, in a purely cross-sectional setting, as is the case of our application, it is not relevant to maintain that sellers and buyers simultaneously consider prices obtained in other transactions. Admittedly, many papers include a spatial lag term in their hedonic equation, but this is done in a somewhat ad hoc way without sound theoretical foundations. In this article, we therefore only control for spatial autocorrelation in the error term. Using the SHAC methodology allows doing so without making strong parametric assumptions about the form taken by spatial autocorrelation and by heteroskedasticity (see Ayouba et al. (2020) for a more detailed explanation).

Data collection

Dependent variable: Parcel prices

Given that the sampling time was short (six months), it was expected that market fluctuation would have little influence on land prices. Since land parcels are obtained through informal mechanisms, there are no records of the purchase transactions, so data were collected from March to August 2016, in two ways: online and on the spot. The former method consisted of searching asking prices (i.e. land parcel prices) on real estate websites (www.metroscubicos.com; www.inmuebles24.com). The latter method was to use the ‘mystery shopping technique’, a technique of participant observation that has researchers pose as customers or potential customers (Wilson, 1998). In our study, we identified banners along the roads and we also approached people as if we were buyers at small shops, asking whether they had information on people selling properties. We noticed that buyers had little bargaining power, and thus the final selling price did not change significantly, since people in general purchase small parcels (< 1000 m²). We recorded a total of 345 observations (parcel prices) from the seven districts (Figure 1). All selected parcels were located in the Conservation Land.

Explanatory variables

We classified the data into four categories: structural, environmental, neighbourhood and accessibility features, as hedonic studies usually do. Information was gathered using geographic information systems (GISs). Distances were measured from each land parcel to the external variable, while information regarding urban characteristics of parcels was obtained by telephone or direct inquiry from local landowners. Also, we used the census from the National Institute of Statistics and Geography (INEGI, 2010), to include the socioeconomic situation for each parcel as a dummy variable.

Our first group of variables is the structural one. The land parcel size and the provision of public services are considered (e.g. water, electricity, sewage), with the expectation of a positive effect on sale prices. Parcel size is included using splines to capture its potential non-linear impact on price.

The second set of variables includes environmental amenities. The distance to green areas is expected to have a negative sign indicating a positive effect on land prices, as previous studies reflect (Saphores and Li, 2012). Open spaces, which provide many ecosystem services, have important amenity values that include leisure opportunities and aesthetic enjoyment, as well as the positive effect on human health of controlling pollution and noise (Tyrväinen and Miettinen, 2000). However, most of these values lack a market price (Kong et al., 2007). Thus, for this analysis, two types of open space were considered: (1) Forest, which refers to unprotected forests surrounding the observed parcels; and (2) Protected Natural Area (PNA), which denotes an area that is preserved in native or natural vegetation habitat, with restricted access.

We also consider whether the parcel is sloped or not, within this group of variables, as it could offer a landscape view. Nevertheless, the expected sign is negative, due to the impact that slopes might have on house construction and parcel development costs, since we assume that buyers of these parcels are primarily focused on real estate value.

For the neighbourhood characteristics, straight-line distances to markets, schools, hospitals and grocery shops were included. We also considered the distance to traditional towns. All these variables are expected to have a negative sign. Socioeconomic conditions were included in the model as a dummy variable for social marginalisation. This variable refers to levels of urban marginalisation and income inequalities, as well as the access to education, health services, housing and goods (Consejo Nacional de Población (CONAPO), 2010) assigned to each land parcel. It is expected to have a negative sign because parcels situated in highly socially marginalised areas may have a lower sales price than parcels located in wealthier areas.

To capture the effect of accessibility, we computed four variables, referring to the distance to roads, motorways, the metro and the city centre. All of these are expected to have a negative sign, meaning that the longer the distance, the lower the land value. Note that we have systematically tried to include all distance variables using splines; however, they were never significant in our estimations, except for distance to the city centre. Therefore, we present our estimation results with a non-linear effect of the distance to the city centre on prices and all the other distances variables included linearly.

Variables included in the model are presented in Table 1.

Table 1.

Included variables and expected sign.

Variable code	Variable name	Expected sign	Definition
PriceTot	Total parcel price	Dependent variable	Total price of land parcel in Mexican pesos (2016)
Structural characteristics
ParcelSize	Parcel size	Positive	Total parcel size in square metres
Water	Water	Positive	Dummy variable indicating whether the parcel has a piped-water connection
Sewage	Sewage	Positive	Dummy variable indicating whether the parcel has a sewage service already established
Electri	Electricity	Positive	Dummy variable indicating whether the parcel has electricity already established
Environmental characteristics
Forest	Forest	Negative	Distance to nearest forest in metres
PNA	PNA	Negative	Distance to nearest PNA in metres
Slope	Slope	Positive	Dummy variable: 1 if the property is sloped and it is not appropriate for construction,0 otherwise
Neighbourhood characteristics
Town	Traditional town	Negative	Distance to nearest traditional town in metres
Market	Market	Negative	Distance to nearest market in metres
School	School	Negative	Distance to nearest school in metres
Hospital	Hospital	Negative	Distance to nearest hospital in metres
Commerce	Small shop	Negative	Distance to nearest small shop in metres
High	Socioeconomic	Negative	Dummy variable indicating whether the parcel is located on economic marginalised land
Accessibility
Road	Roads	Negative	Distance to nearest road in metres
Motorway	Motorways	Negative	Distance to nearest motorway in metres
Metro	Metro	Negative	Distance to the nearest metro station in metres
CdCen	City centre	Negative	Distance to the city centre in metres

Note: Table is based on estimation sample of 345 land parcel prices during 2016.

Source: Prepared by the authors.

Results

Table 2 presents descriptive statistics for the data in the sample. The average asking price per square metre was Mex$2058 (US$112 in 2016), with a maximum of Mex$8500 (US$462). The average total price was Mex$2,907,889 (US$158,209). The parcel size average was 1523.16 m². Regarding urban services, 44% of parcels had access to piped water, 32% to the sewage system and 63% to electricity.

Table 2.

Descriptive statistics.

	Mean	SD	Max.	Min.	Units
Parcel price	112	73.8	462	21.8	US$^a
Total price	158,209	346,404	2,557,127	4080	US$
Structural characteristics
Parcel size	1523.16	3355.48	25,250	80	Square metre
Water	0.44	0.50	1	0	Binary
Sewage	0.32	0.47	1	0	Binary
Electri	0.63	0.48	1	0	Binary
Environmental characteristics
Forest	988.27	1316.37	6221.3	1.0	Metres
PNA	3264.65	2154.81	9604	1.0	Metres
Slope	0.10	0.29	1	0	Binary
Neighbourhood characteristics
Town	584.95	732.71	3248.43	0.43	Metres
Market	3153.91	2557.45	11,293	197	Metres
Commerce	3340.23	2174.97	7967.7	105.7	Metres
School	923.73	744.27	4134.75	56.74	Metres
Hospital	1461.06	961.8	21.44	4564.05	Metres
High	0.68	0.47	1	0	Binary
Accessibility (locational characteristics)
Road	431.93	568.17	3441.4	0.3	Metres
Motorway	1515.02	2079.07	11,557	0.1	Metres
Metro	10,604.28	3273.63	18,039.1	880.8	Metres
City	15,674.59	4405.20	26,324	5258	Metres

Note: ^a Exchange rate in 2016: US$1 = 18.38 Mexican pesos.

Source: Prepared by the authors.

Table A1 in the appendix shows the correlation matrix between the continuous variables. No correlation is above 0.67 (between Town and School), and most correlations are below 0.4.

Model results are summarised in Table 3. Regression analysis and data processing were carried out using ordinary least squares (OLS), with a SHAC robust inference¹ on the model expressed in a semi-logarithmic form. We estimated four models, adding a different set of characteristics for each one² to analyse how the sign, size and significance of the coefficients associated with the various sets of characteristics evolve with the inclusion of the other attributes, given the correlations between some of our variables.

Table 3.

Estimation results.

Variable name	Model 1	Model 2	Model 3	Model 4
Variable name	S	S + E	S + E + N	S + E + N + A
Intercept	12.474*** (0.146)	12.954*** (0.171)	13.178*** (0.304)	14.422*** (0.568)
Structural characteristics
bs(ParcelSize)1	10.395*** (0.821)	9.738*** (0.734)	9.574*** (0.711)	9.634*** (6.812)
bs(ParcelSize)2	−2.207 (1.736)	−1.727 (1.599)	−1.249 (1.540)	−1.282 (1.371)
bs(ParcelSize)3	5.226*** (0.511)	5.271*** (0.535)	5.187*** (0.515)	5.109*** (0.453)
Sewage	0.159 (0.138)	0.172 (0.137)	0.187 (0.115)	0.102 (0.113)
Water	0.229 (0.140)	0.239** (0.114)	0.078 (0.087)	0.166** (0.074)
Electri	0.181 (0.135)	0.115 (0.166)	0.167 (0.133)	0.185 (0.124)
Environmental characteristics
Forest		−6.972e-05** (3.385e-05)	−6.920e-05 (4.277e-05)	9.125e-05* (5.544e-05)
PNA		−9.132e-05*** (1.428e-05)	−7.0018e-05** (3.387e-05)	1.537e-05 (2.283e-05)
Slope1		−0.392*** (0.149)	−0.344*** (0.132)	−0.414 *** (0.107)
Neighbourhood characteristics
Town			7.493e-06 (9.603e-05)	−9.221e-05 (6.679e-05)
Market			4.108e-05 (3.783e-05)	7.339e-05** (2.908e-05)
School			−0.0002*** (5.928e-05)	−2.107e-05 (7.746e-05)
Hospital			−4.780e-05 (9.281e-05)	−0.0001 (9.735e-05)
Commerce			−1.400e-05 (3.279e-05)	5.768e-05* (2.963e-05)
High			−0.146 (0.090)	−0.027 (0.072)
Accessibility
Road				4.268e-05 (0.0001)
Motorway				−9.953e-05*** (3.391e-05)
Metro				−6.559e-05** (3.083e-05)
bs(CdCen)1				−1.507* (0.775)
bs(CdCen)2				−1.590*** (0.310)
bs(CdCen)3				−1.760*** (0.413)
R²	0.7632	0.7888	0.8037	0.8314
R² Adjusted	0.759	0.7832	0.7948	0.8204

Notes: Robust standard errors are reported in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels, respectively. Bs refers to B-spline function.

Source: Prepared by the authors.

The first set only includes the structural characteristics (Model 1: S). With an adjusted R-squared of 75.9%, the quality of adjustment is very satisfactory. The variable ‘parcel size’ has a nonlinear effect in the dependent variable, as two of the coefficients of the spline function, indicated by bs(Parcel Size)1, bs(ParcelSize)2 and bs(ParcelSize)3, are significant at 1%. Its effect on price is shown in Figure 2, and seems to be positive since the significant spline coefficients are positive: if the lot size increases, the price will also rise. Regarding urban services, none of these variables have a significant impact in this benchmark model.

Figure 2.

Parcel size effect on land price.

In Model 2 (S+E), we added the environmental features to Model 1. A connection to piped water now becomes significant and positive. Concerning the environmental features, as a semi-log specification is used in all models, estimates can be interpreted as semi-elasticities, meaning that one unit change in the independent variables (in our case 1 m in the distance) will affect a percentage in the price. Distance to the forest and protected natural area (PNA) is significant with the expected sign, showing that as distance increases, land value declines. Nevertheless, the extent of the effect is negligible (for instance, a 1 m further away distance from the forest decreases the price by less than 0.01%). As expected, the slope variable is negatively related to the dependent variable.

Model 3 (S+E+N) further includes neighbourhood variables. The distance to school is the only variable with a significant impact at 1% with the expected sign. However, the inclusion of neighbourhood characteristics has the consequence of making the impact of distance to the forest and connection to piped water insignificant. The results for the other variables remain unchanged.

Finally, Model 4 (S+E+N+A) integrates the accessibility variables. The distances to the motorway and metro are negative and significant at 1% and 5% respectively, with the predicted sign, although the impact on the price is very low. As expected, distance to city centre was also statistically significant, with a non-linear yet globally decreasing impact (see Figure 3). With these new variables, some changes are worth noting. Access to piped-water connection becomes significant again, while the coefficient pertaining to PNA becomes insignificant and the forest’s estimate changes its sign and is only weakly significant. These changes can be explained by the correlation between distance to the forest/PNA and distance to the motorway, which is around 0.5 (see the appendix).

Figure 3.

City centre’s effect on land price.

Discussion

Our empirical analysis allowed us to identify several variables that determine the prices of land parcels in the CL of Mexico City. Such variables were classified into four main groups of factors that help to understand how this market behaves, providing some insight into urban policy issues. Hence, in this section, we analyse each group of variables in turn (i.e. structural features, environmental variables, neighbourhood characteristics and accessibility attributes). The final section is then devoted to policy implications under this context.

Structural features

As expected, all four hedonic pricing specifications showed that there is a positive and strong significant relation between land price and parcel size, which is a standard result. As stated by Turner (1967), low-income households bet on a ‘progressive development’ of their homes; they prefer larger unfinished houses to small finished ones, as they are concerned about the future of their families and so look for more living space. Therefore, the price will increase as the parcel becomes larger. Furthermore, the availability of infrastructure services has a positive relationship with land price. However, piped water was the only significant variable in two of the models, whereas services like electricity and sewer systems apparently are not highly valued in this irregular market. This is in tune with Jimenez (1982), whose findings showed that sanitation services did not have a significant impact on housing value in squats in Manila, the Philippines. In fact, during our field work, it was observed that people with no access to water at home obtain it either from water tanks or street vendors. Although accessible, it involves an additional expense. Hence, new buyers might consider water availability as a fundamental factor in their decision to acquire a land parcel, while electricity and sewage systems are secondary.

Other studies elsewhere have demonstrated similar relations between infrastructure services and low-income housing markets. For example, Crane et al. (1997) showed that services associated with infrastructure provision are seen to be as valuable as, or more valuable than, other features. Analysing slums in two Asian mega-cities, the authors suggested that in Jakarta, residents would pay more for a piped-water connection than for a toilet, a sink, access to a well or a larger house. However, residents of Bangkok would prefer to incur the largest expense in paying for a legal connection to the electrical power grid, followed by for a high-pressure water connection and, a distant third, for sanitation. Similarly, Choumert et al. (2014) found a positive effect of piped water and electricity on rents in Kigali, Rwanda. Furthermore, Nakamura (2017) pointed out that for informal settlements, basic services may be a factor in improving the tenure security of residents, hence the great importance of having them. In the context of Mexico City’s outskirts, informal land parcel buyers place value first of all on the availability of a water connection. Indeed, water is vital and has no perfect substitute in other variables, thus purchasers expect that other public services will eventually be delivered with increasing urbanisation.

Environmental variables

There is a large amount of literature on environmental amenities in formal markets; nonetheless, studies focusing on informal markets often neglect environmental factors. Hence, the evidence presented here is important, as it highlights a statistically significant relationship between the price of land and the distance to the nearest forest and PNA. However, although significant, the regression coefficients have values close to zero, which means that buyers are aware of but rather indifferent to environmental features in this informal market, in contrast with formal housing markets elsewhere (Morancho, 2003; Saphores and Li, 2012; Sylla et al., 2019). Even in formal markets in developing countries, environmental quality has a large impact on property prices, as reported by Humavindu and Stage (2003) for Namibia. According to this result, Mexico City’s authorities should promote environmental education along with other strategies that we discuss below. Otherwise, as households do not seem to value environmental characteristics, this laissez-faire scenario will lead to the inexorable degradation of environmental assets in the short term.

In addition, our study confirmed that a sloped parcel will command a lower price because potential buyers seeking land parcels with residential purposes worry about the inclination of the land, given that building houses on slopes is costlier than on flat parcels.

Neighbourhood characteristics

In general, neighbourhood amenities influence housing and land prices. Distance to school was significant in one model. This is in line with the results of Sylla et al. (2019), which showed that peri-urban inhabitants seek to live near a school.

In the sense of valuing the neighbourhood amenities, it was expected that the distance to the market would have a positive effect since markets are commercial areas of supply, which would imply a benefit for the inhabitants. However, this variable proves to have a counterintuitive sign in both models (significantly so in Model 4). In other words, the closer to the market, the lower the price of the land. One way to account for this finding is that markets are also frequently associated with negative externalities (garbage, traffic congestion), and buyers tend to avoid them (Brasington and Hite, 2005).

Accessibility attributes

Although the percentage change in land value per metre of distance is very low, the distance to motorways, the metro and the city centre had a significant positive impact (negative coefficients) on land values. Thus, accessibility and distance to the city centre are as important features in this informal market as they are in formal markets. These results are in line with Zhang and Zhao (2018), whose findings showed that residents pay a higher price for informal homes if these are situated in better locations, and also with Sylla et al. (2019) who found the main determinant of the price of parcels in a peri-urban area to be closeness to the city.

Policy implications and conclusion

The empirical results presented here are important as policy recommendations for two reasons. On the one hand, land parcel purchasers in the CL of Mexico City mainly value urban characteristics. This ‘urban premium’ is the driver of the expansion of urban areas in many cities in the developing world. Furthermore, as shown elsewhere (Moschella, 2018), this is an unsolved problem of spatial planning (both urban and environmental). In urban and real estate terms, rising house prices have pushed low-income people to look for settlements in conservation areas. Furthermore, poor or non-existent enforcement of preservation areas has allowed uncontrolled urban sprawl. Therefore, the expansion towards natural areas is led by people looking for urban amenities rather than environmental ones. Indeed, individuals are willing to pay for basic services such as access to piped water, proximity to schools and accessibility features such as proximity to the city centre, motorways and metro stations. Even when these are not present, buyers arguably expect that such services will eventually be provided by local authorities, resulting in growing urbanisation.

On the other hand, people are aware of the distance to the forest and/or PNA but are rather indifferent to this. Indeed, distances to both the forest and the PNA influence the price of informal land parcels, but the value is very low. This is an interesting result because it provides evidence of the difference between how informal markets and formal markets value environmental amenities. In fact, Das et al. (2017) discovered that slum dwellers value a clean environment.

Thus, our findings suggest that these individuals look to cover their basic housing needs and will continue to occupy areas despite negative environmental effects that might be generated. Hence, the problem of irregular settlements must be approached from two different angles.

Firstly, informal inhabitants cannot be persuaded to halt the invasion and modification of natural areas without a comprehensive and properly enforced urban and environmental policy. Policy planners need to develop a multisectoral and integrated spatial policy which considers urban and environmental areas as a unified system, to improve urban sustainability. Thus, changing the perception that green areas are potential urbanisation opportunities is paramount.

Moreover, urban dwellers must be sensitised to the whole value of the CL. For example, recognition of the importance of natural reserves to society through ecosystem services should be strongly promoted for all inhabitants, both land buyers and traders. In this vein, compensation schemes could be implemented and, consequently, prevent landowners (ejidos, communities and private) from selling their properties.

Furthermore, the idea of creating containment zones or green belts has been implemented in several countries for many decades as an effective land use planning policy (Pendall et al., 2002). However, in the case of Mexico, it has not been properly enforced, and people continue to settle in unauthorised territories. Indeed, purchased lands, despite their conservation status, eventually end up being regularised when the green spaces are already decimated, and a certain degree of urban consolidation is reached. In this sense, two ways of action can be undertaken.

The first path is to re-locate settlements from natural areas, to restore the affected areas and ensure their conservation. Nevertheless, the eviction of inhabitants is a delicate matter (Everett, 2001). The second path is containment, and focuses on preventing settlements from growing and spreading. In such cases, authorities can apply market measures (e.g. taxes and fines). As stated above, the expectation of buyers that the land-use change of conservation areas into urban use is a profitable opportunity should somehow be penalised. Land readjustment strategies such as re-location, on the one hand, and legalisation, on the other, can be considered but must be carefully planned.

In this sense, the second recommendation is that public policy needs to solve the housing demand and provide areas to cope with the increasing challenge of dwelling. This hedonic pricing analysis showed irregular land parcel buyers seeking urban features, such as drinking water connection, basic education and transport supply, while not paying a premium for green space proximity. Parcel size is also an important characteristic, as buyers look for parcels large enough to allow incremental growth of their home as their incomes improve and their families grow. Hence, satisfying the housing demand must include the attributes presented here. Effective housing projects need to be designed considering the preferences of target groups (Nassar and Elsayed, 2018).

Over the past 40 years, the urban expansion of the southern periphery of Mexico City has failed to be controlled despite the establishment of Conservation Land, multiple restrictions on land use and containment zones. The findings from this article contribute to the understanding of how land markets work when dealing with informal parcels, providing helpful guidelines for policy planning purposes. Consequently, an effective spatial and social policy should consider environmental protection, on the one hand, and access to housing, on the other. At present, little is known about the values that individuals in illegal settlements assign to green spaces, and therefore further research is essential.

Footnotes

Appendix

Table A1.

Correlation matrix.

	ParcelSize	Forest	PNA	Town	Market	School	Hospital	Commerce	Road	Motorway	Metro
ParcelSize
Forest	−0.050
PNA	−0.028	−0.066
Town	0.002	−0.142	0.293
Market	0.082	−0.282	−0.611	0.499
School	0.094	−0.212	−0.183	0.666	0.460
Hospital	0.084	−0.148	−0.404	0.465	0.375	0.542
Commerce	0.032	−0.141	0.138	0.368	0.115	0.491	0.155
Road	0.037	−0.159	0.052	0.548	0.078	0.564	0.325	0.485
Motorway	0.030	0.554	−0.480	0.083	0.138	0.093	0.059	0.166	−0.221
Metro	−0.007	−0.512	0.454	0.085	−0.072	0.212	−0.216	0.363	0.428	−0.571
CdCen	−0.031	0.494	0.514	−0.164	−0.391	−0.004	−0.227	0.160	0.144	−0.013	0.157

Acknowledgements

The authors thank colleagues from CESAER for helpful remarks and suggestions on a former draft, and the anonymous reviewers for their insightful comments.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The work was supported by the Mexican Council for Science and Technology (CONACYT) [grants 295890, 291250 and 179301] and by UNAM-PAPIIT IN301919. ETMJ was a visiting scholar at CESAER, France in August to December 2018.

ORCID iDs

Estebania Teyeliz Martinez-Jimenez

Julie Le Gallo

Notes

References

Aguilar

(2008) Peri-urbanization, illegal settlements and environmental impact in Mexico City. Cities 25(3): 133–145.

Aguilar

(2016) La Ciudad de México En El Siglo XXI: Realidades y Retos. Mexico: CDMX-SECITI, Miguel Ángel Porrúa.

Aguilar

Santos

(2011) Informal settlements’ needs and environmental conservation in Mexico City: An unsolved challenge for land-use policy. Land Use Policy 28(4): 649–662.

Allen

Dávila

Hofmann

(2006) The peri-urban water poor: Citizens or consumers? Environment and Urbanization 18(2): 333–351.

Anselin

Lozano-Gracia

(2009) Spatial econometrics. In: Mills

Patterson

(eds) Palgrave Handbook of Econometrics: Volume 2: Applied Econometrics. Houndmills: Palgrave Macmillan UK, pp. 1213–1292.

Arimah

(1992) Hedonic prices and the demand for housing attributes in a third world city: The case of Ibadan, Nigeria. Urban Studies 29(5): 639–651.

Ayouba

Breuillé

M-L

Grivault

, et al. (2020) Does Airbnb disrupt the private rental market? An empirical analysis for French cities. International Regional Science Review 43(1–2): 76–104.

Benítez

Pérez-Vázquez

Nava-Tablada

, et al. (2012) Urban expansion and the environmental effects of informal settlements on the outskirts of Xalapa city, Veracruz, Mexico. Environment and Urbanization 24(1): 49–166.

Birch

Chattaraj

Wachter

(2016) Slums: How Informal Real Estate Markets Work. Philadelphia, PA: University of Pennsylvania Press.

10.

Bivand

Piras

(2015) Comparing implementations of estimation methods for spatial econometrics. Journal of Statistical Software 63(18): 1–36.

11.

Brasington

Hite

(2005) Demand for environmental quality: A spatial hedonic analysis. Regional Science and Urban Economics 35(1): 57–82.

12.

Cavailhès

Brossard

Foltête

J-C

, et al. (2009) GIS-based hedonic pricing of landscape. Environmental and Resource Economics 44(4): 571–590.

13.

Cheshire

Sheppard

(1989) British planning policy and access to housing: Some empirical estimates. Urban Studies 26(5): 469–485.

14.

Cheshire

Sheppard

(2004) Capitalising the value of free schools: The impact of supply characteristics and uncertainty. The Economic Journal 114(499): F397–F424.

15.

Choumert

Stage

Uwera

(2014) Access to water as determinant of rental values: A housing hedonic analysis in Rwanda. Journal of Housing Economics 26: 48–54.

16.

Cohen

(2006) Urbanization in developing countries: Current trends, future projections, and key challenges for sustainability. Technology in Society 28(1–2): 63–80.

17.

Connolly

(2009) Observing the evolution of irregular settlements: Mexico City’s colonias populares, 1990 to 2005. International Development Planning Review 31(1): 1–35.

18.

Connolly

Wigle

(2017) (Re) constructing informality and ‘doing regularization’ in the conservation zone of Mexico City. Planning Theory & Practice 18(2): 183–201.

19.

Consejo Nacional de Población (CONAPO) (2010) Indices de marginacion. Available at: http://www.conapo.gob.mx/en/CONAPO/Indices_de_Marginacion (accessed 6 December 2016).

20.

Crane

Daniere

Harwood

(1997) The contribution of environmental amenities to low-income housing: A comparative study of Bangkok and Jakarta. Urban Studies 34(9): 1495–1512.

21.

Das

Mitra

Kumar

(2017) Do neighbourhood facilities matter for slum housing? Evidence from Indian slum clusters. Urban Studies 54(8): 1887–1904.

22.

Durand-Lasserve

Selod

(2009) The formalization of urban land tenure in developing countries. In: Lall

Freire

Yuen

, et al. (eds) Urban Land Markets. Dordrecht: Springer, pp. 101–132.

23.

Espinoza

Balaguer

(2009) Valuing environmental characteristics in urban slums: A spatial hedonic approach for Chilean cities. In: Valuing Environmental Characteristics of Slums Fifth Urban Research Symposium, pp. 28–30.

24.

Everett

(2001) Evictions and human rights: Land disputes in Bogotá, Colombia. Habitat International 25(4): 453–471.

25.

Follain

Jimenez

(1985) The demand for housing characteristics in developing countries. Urban Studies 22(5): 421–432.

26.

Follain

Lim

G-C

Renaud

(1980) The demand for housing in developing countries: The case of Korea. Journal of Urban Economics 7(3): 315–336.

27.

GDF (2000) Programa General de Ordenamiento Ecológico del Distrito Federal (2000–2003). Mexico: Gaceta Oficial del Distrito Federal.

28.

Hardoy

Satterthwaite

(2014) Squatter Citizen: Life in the Urban Third World. London: Routledge.

29.

Hawley

Miranda

Sawyer

(2018) Land values, property rights, and home ownership: Implications for property taxation in Peru. Regional Science and Urban Economics 69: 38–47.

30.

Humavindu

Stage

(2003) Hedonic pricing in Windhoek townships. Environment and Development Economics 8(2): 391–404.

31.

Instituto Nacional de Estadística y Geografía (INEGI) (2010) Censo general de Población y Vivienda 2010. Mexico: INEGI.

32.

Instituto Nacional de Estadística y Geografía (INEGI) (2015) Encuesta intercensal 2015. Available at: http://www.beta.inegi.org.mx/programas/intercensal/2015/ (accessed 9 April 2018).

33.

Jimenez

(1982) The value of squatter dwellings in developing countries. Economic Development and Cultural Change 30(4): 739–752.

34.

Jones

Pisa

(2000) Public–private partnerships for urban land development in Mexico: A victory for hope versus expectation? Habitat International 24(1): 1–18.

35.

Kelejian

Prucha

(2007) HAC estimation in a spatial framework. Journal of Econometrics 140(1): 131–154.

36.

Kong

Yin

Nakagoshi

(2007) Using GIS and landscape metrics in the hedonic price modeling of the amenity value of urban green space: A case study in Jinan City, China. Landscape and Urban Planning 79(3–4): 240–252.

37.

Kuminoff

Parmeter

Pope

(2010) Which hedonic models can we trust to recover the marginal willingness to pay for environmental amenities? Journal of Environmental Economics and Management 60(3): 145–160.

38.

Lara-Pulido

Estrada Díaz

Zentella Gómez

, et al. (2017) Los costos de la expansión urbana: Aproximación a partir de un modelo de precios hedónicos en la Zona Metropolitana del Valle de México. Estudios Demográficos y Urbanos 32(1): 37–63.

39.

Lim

G-C

Follain

Renaud

(1984) Economics of residential crowding in developing countries. Journal of Urban Economics 16(2): 173–186.

40.

Martínez

Viegas

(2009) Effects of transportation accessibility on residential property values: Hedonic Price Model in the Lisbon, Portugal, metropolitan area. Transportation Research Record: Journal of the Transportation Research Board 2115: 127–137.

41.

Martínez-Jiménez

Pérez-Campuzano

Aguilar Ibarra

(2017) Hedonic pricing model for the economic valuation of Conservation Land in Mexico City. WIT Transactions on Ecology and the Environment 223: 101–111.

42.

Morancho

(2003) A hedonic valuation of urban green areas. Landscape and Urban Planning 66(1): 35–41.

43.

Moschella

(2018) Peri-urbanization and Land Management Sustainability in Peruvian Cities. PhD Thesis, Université de Strasbourg, France.

44.

Nakamura

(2017) Tenure security premium in informal housing markets: A spatial hedonic analysis. World Development 89: 184–198.

45.

Nassar

Elsayed

(2018) From informal settlements to sustainable communities. Alexandria Engineering Journal 57(4): 2367–2376.

46.

PAOT and SEDEMA (2012) Atlas Cartográfico Del Suelo de Conservación Del Distrito Federal. Mexico: Procuraduría Ambiental y del Ordenamiento Territorial y Secretaría de Medio Ambiente (SEDEMA).

47.

Pendall

Martin

Fulton

(2002) Holding the line: Urban containment in the United States. Report, The Brookings Institution Center on Urban and Metropolitan Policy, USA, August.

48.

Piras

(2010) sphet: Spatial models with heteroskedastic innovations in R. Journal of Statistical Software 35(1): 1–21.

49.

Quigley

(1982) Nonlinear budget constraints and consumer demand: An application to public programs for residential housing. Journal of Urban Economics 12(2): 177–201.

50.

Ravetz

Fertner

Nielsen

(2013) The dynamics of peri-urbanization. In: Nilsson

Pauleit

Bell

, et al. (eds) Peri-Urban Futures: Scenarios and Models for Land Use Change in Europe. Berlin and Heidelberg: Springer, pp. 13–44.

51.

Rosen

(1974) Hedonic prices and implicit markets: Product differentiation in pure competition. Journal of Political Economy 82(1): 34–55.

52.

Saphores

(2012) Estimating the value of urban green areas: A hedonic pricing analysis of the single family housing market in Los Angeles, CA. Landscape and Urban Planning 104(3–4): 373–387.

53.

Schläpfer

Waltert

Segura

, et al. (2015) Valuation of landscape amenities: A hedonic pricing analysis of housing rents in urban, suburban and periurban Switzerland. Landscape and Urban Planning 141: 24–40.

54.

Sobrino

(2014) Housing prices and submarkets in Mexico City: A hedonic assessment. Estudios Económicos 29(1): 57–84.

55.

Sylla

Lasota

Szewrański

(2019) Valuing environmental amenities in peri-urban areas: Evidence from Poland. Sustainability 11(3): 570.

56.

Tuffery

(2017) The recreational services value of the nearby periurban forest versus the regional forest environment. Journal of Forest Economics 28: 33–41.

57.

Turner

(1967) Barriers and channels for housing development in modernizing countries. Journal of the American Institute of Planners 33(3): 167–181.

58.

Twarabamenye

Nyandwi

(2012) Understanding informal urban land market functioning in peri-urban areas of secondary towns of Rwanda: Case study of Tumba Sector, Butare Town. Rwanda Journal 25(1): 34–51.

59.

Tyrväinen

Miettinen

(2000) Property prices and urban forest amenities. Journal of Environmental Economics and Management 39(2): 205–223.

60.

United Nations (2018) The Sustainable Development Goals Report 2018. New York: United Nations.

61.

United Nations-Habitat (2016) Slums almanac 2015–16. Tracking improvement in the lives of slum dwellers. Available at: https://unhabitat.org/slum-almanac-2015-2016 (accessed 8 December 2018).

62.

Van Eggermond

MAB

Lehner

Erath

(2011) Modeling hedonic prices in Singapore. In: 16th International Conference of Hong Kong Society for Transportation Studies, Hong Kong, 17–20 December 2011. Hong Kong Society for Transportation Studies. Available at: https://doi.org/10.3929/ethz-a-006742047 (accessed 21 April 2018).

63.

Vázquez-Castillo

(2004) Land Privatization in Mexico: Urbanization, Formation of Regions and Globalization in Ejidos. New York: Routledge.

64.

Wasson

McLeod

Bastian

, et al. (2010) The effects of scenic and environmental amenities on agricultural land values. Land Economics 89(3): 466–478.

65.

Wigle

(2010) Social relations, property and ‘peripheral’ informal settlement: The case of Ampliación San Marcos, Mexico City. Urban Studies 47(2): 411–436.

66.

Wilson

(1998) The use of mystery shopping in the measurement of service delivery. Service Industries Journal 18(3): 148–163.

67.

Zeilhofer

Topanotti

(2008) GIS and ordination techniques for evaluation of environmental impacts in informal settlements: A case study from Cuiabá, central Brazil. Applied Geography 28(1): 1–15.

68.

Zhang

Zhao

(2018) The determinants of informal housing price in Beijing: Village power, informal institutions, and property security. Cities 77: 117–129.