Is judicial inefficacy increasing the weight of the house property market in Spain? Evidence at the local level

Mora-Sanguinetti, Juan S. Article Is judicial inefficacy increasing the weight of the house property market in Spain? Evidence at the local level SERIEs - Journal of the Spanish Economic Association Provided in Cooperation with: Spanish Economic Association Suggested Citation: Mora-Sanguinetti, Juan S. (2012) : Is judicial inefficacy increasing the weight of the house property market in Spain? Evidence at the local level, SERIEs - Journal of the Spanish Economic Association, ISSN 1869-4195, Springer, Heidelberg, Vol. 3, Iss. 3, pp. 339-365, https://doi.org/10.1007/s13209-011-0063-6 This Version is available at: https://hdl.handle.net/10419/77734 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/2.0/ SERIEs (2012) 3:339–365 DOI 10.1007/s13209-011-0063-6 ORIGINAL ARTICLE Is judicial inefficacy increasing the weight of the house property market in Spain? Evidence at the local level Juan S. Mora-Sanguinetti Received: 2 September 2010 / Accepted: 25 April 2011 / Published online: 31 May 2011 © The Author(s) 2011. This article is published with open access at SpringerLink.com Abstract The weight of the housing tenancy market in Spain is very low. It is frequently argued that an ineffective judicial system, implying a cumbersome procedure to evict a non-paying tenant or simply requiring a long period to execute a decision, may be an important determinant of the tenancy market’s weakness, as it constrains the effective supply by reducing the profitability of landlords. This research studies this effect econometrically using a panel data approach and exploring differences in judicial efficacy among the Spanish provinces. After controlling for several other factors, this study concludes that the degree of inefficacy of the judicial system has a positive impact on the property share among provinces in Spain. Keywords Judicial efficacy ·Property market ·Tenancy market · Contract enforcement JEL Classification K40 ·R21 I am grateful to participants in the seminar of the Banco de España-Eurosystem and to Stefano Battilossi, Ángel Estrada, Pablo Hernández de Cos, Llanos Matea, Carmen Martínez Carrascal, Paloma López García, Jorge Martínez Pagés, Claire McHugh and to the three anonymous referees who reviewed this paper and its working paper version for their useful comments and suggestions. I am also in debt with María Gil for her assistance and preparation of the graphs. The views expressed are those of the author and should not be attributed to the Banco de España, the Eurosystem or the OECD. J. S. Mora-Sanguinetti (B ) Economic Analysis and Forecasting Department, Banco de España-Eurosystem, Alcalá 48, 28014 Madrid, Spain e-mail: [email protected] J. S. Mora-Sanguinetti OECD, Paris, France 123 340 SERIEs (2012) 3:339–365 1 Introduction Since the Spanish Civil War (1936–1939) the weight of the housing property market has persistently increased in Spain (in contrast to developments in the rental market). Although the official statistical information available is very scarce, the census database reveals that the proportion rose from 63.4 to 82.2% between 1970 and 2001. Moreover, according to the estimations of the Spanish Ministry of Housing (2008), the average property rate rose by a further 2.1 percentage points in the period 2001–2007. Several factors may have affected the evolution of the property share in Spain over the last decades. These factors include the fall in interest rates (Blanco and Restoy 2007), especially after 1995; liberalization of the banking sector since 1980 (Kumbhakar and Lozano-Vivas 2004;Iacoviello and Minetti 2003), more stringent tenancy laws having been adopted following World War II (Mora-Sanguinetti 2011), and a fiscal regime which favors buying over renting (López García 1996;GarcíaVaquero and Martínez 2005). Several studies have pointed out that the factors mentioned above are not exclusive to Spain and that the increase in the property rate can be found in several other markets of the European Union as well as in the United States (Louvot-Runavot 2001). Nevertheless, the relative weakness of the tenancy market as compared to the property market in Spain is somewhat exceptional. This situation is generally regarded as undesirable for several economic reasons. The most important one is perhaps that a weak tenancy market is linked to lower mobility of individuals and workers (Maclennan et al. 1998;Barceló 2006) which tends to increase the unemployment rate (Layard et al. 1991) and to reduce the economy’s efficiency (Hardman and Ioannides 1999). More recently, Arce and López-Salido (2007) have emphasized that a well-developed housing rental sector can be a crucial device to avoid housing price bubbles and the excessive concentration of resources in the building sector. Despite these problems, and the imbalances entailed thereby, especially during the recent housing boom in Spain, the Spanish authorities have only paid attention to them recently. One such example is personal income tax deduction (IRPF) for the purchase of a primary residence. This measure has been maintained by all national Governments (irrespective of political orientation) for public choice reasons, despite its effect being to favor buying over renting (as mentioned above), its undesirable redistributive effects (Sanz 2000;Bilbao Terol et al. 2006) or its effects in prices (López García 2004). Only very recently it has been proposed its abrogation (under the 2011 Budget Law).1 Some time earlier, in 2009, the Government passed a Law2introducing new regulatory measures with the objective to protect the owners of rented dwellings. These reforms were intended to improve the functioning of the tenancy market, and thus, to reduce the weight of the property market. Those measures included, on the one hand, reform of the Spanish “Civil Procedural Law” (CPL)3in order to expedite evictions 1Law 39/2010 of 23 December 2010 (de presupuestos generales del Estado para el año 2011). 2Law 19/2009 of 23 November 2009 (de medidas de fomento y agilización procesal del alquiler y de la eficiencia energética de los edificios). 3Law 1/2000, of 7 January 2000 (de enjuiciamiento civil). 123 SERIEs (2012) 3:339–365 341 and to facilitate the collection of rents by property owners (see further discussions in Sect. 2) and, on the other hand, reform of the Spanish Tenancy Law4giving the owner more legal grounds to reduce the term of the tenancy contract. The latter reform was quite limited in scope and, as with all other substantive rules of the Spanish Tenancy Law, does not form part of this study as it is not subject to any regional variation. Underlying those latter measures is the idea that both a slow judicial system (implying a cumbersome procedure to evict a non-paying tenant or simply a lengthy period to execute a decision) and unduly onerous rules governing tenancy contracts (such as rules limiting the landlord’s ability to recover the flat for his own use) have been detrimental to the tenancy market as they have reduced the effective supply and may have contributed to lowering the share of rented dwellings. A theoretical explanation of how judicial inefficacy may affect the flow of demand and supply of housing is provided later in this paper. In any case it is important to emphasizeattheoutsetthatthedecisiontorentahouseonthemarket(orpartthereofas is assumed below) involves transaction costs that do not exist (or at least are different) in the market for home ownership. In other words, the renting of a home implies a sustained relationship in time with the figure of the tenant, which does not exist in the opposite case. That relationship is linked to the functioning of the judicial system in the sense that, in the presence of an ineffective judicial system (and therefore, under weakly controlled contractual arrangements), some tenants may default on their contractual obligations (such as paying the rent or taking care of the property). Those circumstances imply a reduction in the landlord’s profits. The flow of benefits provided by rental housing thus depends on the correct application in time of the rental contracts. This need not occur in the case of complete ownership of a dwelling. In addition to the judicial system, there are other control mechanisms in the market for home ownership (the discussion of which is beyond the scope of this paper) such as the property registry. An analysis of the role of property registration and other forms of property enforcement may be consulted in Arruñada (2003). In any case, as discussed below, it is possible to control by various inefficiencies that may exist in the mortgage lending market through alternative judicial system measures (discussed below). In summary, in this context, a proportion of potential landlords may disappear from the market in the event of weak enforcement being provided by the judicial system. This would imply an increase in the rate of property in the economy.5 That is the general result found on an international basis by some papers in the economic literature, for instance Casas-Arce and Saiz (2010). Those authors used the measure of judicial formalism (as a proxy of inefficacy) of Djankov et al. (2003)to explain the decision between owning and renting in a set of countries and found that more formalism may be expected to reduce the weight of the tenancy market. Djankov et al. (2003) proposed a measure of formalism of the judicial system when evicting a non-paying tenant. They concluded that higher formalism is related to more difficult 4Law 29/1994, of 24 November 1994 (de arrendamientos urbanos). 5A discussion of how the judicial system works when dealing with tenancy conflicts is included below in Sect. 2. 123 342 SERIEs (2012) 3:339–365 evictions and higher unpredictability of the procedures.6This paper specifically aims to analyze the impact of an ineffective judicial system in Spain on the housing tenure outcomes by exploiting the cross-province variation existing in the weight of the house property market in Spain and in the performance of the judicial system. Landlords are supposed to exit the tenancy market when they face an environment in which it is difficult to enforce their tenancy contracts. Thus, this paper aims to assess to what extent the efficacy functioning of the judicial system explains the varying weights of the property market in the Spanish provinces once other concurrent factors are taken into account. In order to do that, I have constructed an index of judicial efficacy for each province of Spain based on official judicial data.7Then, its impact in the property share is estimated after controlling for a set of other relevant economic and demographic factors. The structure of this paper is as follows. Section 2presents a descriptive analysis of the cross province variation of the property rate in Spain and constructs the judicial efficacy measure used in the main estimations. Section 3explores the empirical literature on the topic and provides a theoretical justification for the hypothesis tested in this paper. Section 4presents the estimations using panel data techniques. Finally, Sect. 5presents the findings of this study. Appendix A presents alternative estimations when other judicial efficacy measures are taken into account. 2 Measuring judicial efficacy and the property rate in the Spanish economy An owner who wants to collect an unpaid rent or evict a tenant for any reason (nonpayment of rent, vandalism of property) in Spain has to use the procedures established by the CPL (2000).8The CPL (2000) is the basic procedural regulation of the judicial system. It establishes the rules of access to the court system, the formal procedures that the parties must observe, the role of the judge, the rules governing the admission and use of evidence, the control exercised by superior courts and all related issues. Therefore that Law is a main determinant of the “aggregated” slow (or fast) performance of the judicial system in Spain (Mora-Sanguinetti 2010). Although it is a national Law, its application differs among the Spanish provinces. A rational explanation for that is that the workload of the judges may differ among the provinces. This may be partially implied by the difference in populations falling with the jurisdiction of the court or the various levels of litigation (generated by either the population or the number of 6The results and methodology by Djankov et al. (2003), although very relevant, cannot be used in the experiment proposed in this paper because, as mentioned, they concern international elements for a specific year. Therefore, they cannot capture the variations in efficiency within a specific country. The latter may be caused by differences in the application of the Civil Procedural Law and not by the Civil Procedural Law itself (see further Mora-Sanguinetti 2010). 7Other indices are also constructed and tested in Appendix A of this paper. 8It should be noted that some extrajudicial solutions may be found by the parties, such as sending the case to arbitration. However, only a judge can order an eviction in Spain. 123 SERIEs (2012) 3:339–365 343 lawyers per capita).9At the same time, the resources invested in the administration of justice differ from region to region.10 The key issue is thus to define a correct measure of judicial inefficacy. This section (and Appendix A of this paper) is directed to that task. It is in fact possible to observe that the efficacy of the judicial system diverges among the different regions of Spain over time (allowing the construction of a panel with information on the functioning of the judicial system). As argued before, we may expect that in the most ineffective regions, where it is more difficult to evict a nonpaying tenant or to have the payment of rent enforced through the judicial system, landlords will opt to quit the tenancy market, and thus the share of tenancy in the province will diminish (see Sect. 3.2 for a further theoretical discussion). In order to construct a set of judicial system indicators useful for the analysis of the housing market, a relevant question arises: what are the specific procedures required to recover an unpaid rent in Spain? The CPL (2000) establishes a specific procedure for recovering such debt: first, a “declaratory judgment” will “declare” the existence of the debt and the obligation of the debtor to pay. We can call that element the “first stage” or “first procedure” because the possibility remains that the tenant decides not to pay the debt. In that case, a final or definitive procedure (“executory process”) takes place. In the “execution” stage, the creditor asks the judge to “execute” the debt. As a result of this final procedure, the judge will seize the amount from the bank accounts of the debtor and will probably order his eviction from the dwelling. The General Council of the Judicial Power (Consejo General del Poder Judicial, CGPJ) neither collects nor publishes data on the duration of legal proceedings in the Spanish courts. However, information can be gleaned from its database reporting the number of cases filed, resolved and still pending in the Spanish judicial system by subject, region, court11 and year. Using that database, a relative measure of efficacy can be constructed for the enforcement of each procedure: the congestion rate (see Eq. 1below). The congestion rate is defined as the ratio between the sum of pending cases (measured at the beginning of the period) plus new cases in a specific year and the cases resolved in that same year (Padilla et al. 2007). A lower congestion rate is related to greater efficacy of the judicial system. Congestionratei,t=Pendingcasesi,t+Newcasesi,t Casesresolvedi,t(1) Three more relative measures of inefficacy are computed in Appendix A: the “resolution rate”, the “pending cases rate” and the “pending-plus-new cases rate” (ppnc in the tables). As discussed below in Sect. 4.1, alternative estimates including a measure of 9It should be noted, however, that both the effect of population and litigation in this relationship are taken into account in the different specifications of the model developed in this paper. 10 The “Comunidades Autónomas” (regions) have some powers related to the administration of justice in Spain. Even though the “judicial power” is not properly transferred to the regions, management of resources of the judicial system is influenced by the policies developed by the regions. For instance, they decide how much money is invested in new courts each year in their territories, even though the new courts are integrated in a centrally governed system. 11 The courts analyzed in this study are the “juzgados de primera instancia” and the “juzgados de primera instancia e instrucción”. Those are the courts available for the parties at first instance. 123 344 SERIEs (2012) 3:339–365 Table 1 Judicial system variables Type of procedure Variable Obs Mean Std. Dev. Min Max Declaratory Prtcongestion 350 1.53 0.36 1.03 4.17 Declaratory Prtresolution 350 0.92 0.14 0.39 1.18 Declaratory Prtpendency 350 0.41 0.16 0.13 1.59 Execution Excongestion 350 3.97 1.20 1.20 9.99 Execution Exresolution 350 0.87 0.20 0.42 2.02 Execution Expendency 350 2.77 0.98 0.46 7.59 Execution ppnc*1000 350 17.39 7.78 0.78 36.96 Mortgage Mortcongestion 350 2.72 0.98 0.36 8.50 Mortgage Mortresolution 350 1.03 0.41 0.22 3.67 Mortgage Mortpendency 350 1.62 0.62 0.09 5.29 Mortgage Mortppnc 350 0.72 0.45 0.01 2.81 Source: CGPJ (2010) and self elaboration “litigation” by province (calculated as the number of lawyers per capita) (following Carmignani and Giacomelli 2010) have been provided. The introduction of these “per capita” measures allow for a robustness exercise as the population differs quite significantly between different provinces. The CGPJ offers homogeneous data for the various procedures for the period 2001– 2007.12 For the purposes of the analysis herein, I have chosen the provincial level, although more disaggregated data on the judicial system is available (at the level of a particular court and judicial district (partido judicial)). This is due to the lack of more disaggregated data in other variables such as income per capita (or the PPP corrections). Another disadvantage of this should be noted. The rental market is mainly a local market, so when agents take their decisions, their probable point of reference is the situation in a sub-provincial market (their specific city and not the situation in other cities within the same province). However, an analysis at the provincial level remains valuable as we can assume that agents in a province are more likely to be aware of, and influenced by, problems in the judicial system in the surrounding markets than those of the rest of the country. In the tables the prefix “prt” precedes the efficacy measure related to procedures in the “declaratory stage” (or as we called it, “first” procedure): prtcongestion. The prefix “ex” precedes the efficacy measure related to the executions: excongestion. Table 1 shows the descriptive statistics for those computations (also for the alternative efficacy measures studied in Appendix A). Table 2shows the results for the congestion rate when studied for the executions (excongestion) in the period 2001–2007.13 An average congestion rate of 3.97 over the period 2001–2007 (see Table 1) indicates that around four cases (summing up the pending cases and the new cases arriving 12 Note that the new CPL (2000) entered into force on 7 January 2001. This new CPL radically changed several aspects of civil procedure in Spain (Mora-Sanguinetti 2010) and therefore it is not advisable to relate the data after 2001 with previous observations. 13 Excluding Ceuta and Melilla (no information is available for these regions). 123 SERIEs (2012) 3:339–365 345 Table 2 Judicial congestion rate by province (execution) Province 2001 2002 2003 2004 2005 2006 2007 Álava 1.20 1.88 1.25 2.62 4.28 1.23 2.28 Albacete 3.81 3.00 4.78 3.01 2.95 2.46 3.49 Alicante 3.77 4.68 6.01 4.46 5.72 6.23 5.64 Almería 3.08 4.38 3.92 4.14 3.54 3.60 4.11 Avila 2.52 1.85 3.48 2.95 4.19 5.64 3.74 Badajoz 3.30 3.76 3.25 3.28 3.81 3.95 4.52 Baleares 3.44 3.30 4.94 6.70 6.36 8.99 9.47 Barcelona 4.07 4.80 5.34 4.79 4.76 4.99 4.98 Burgos 2.31 3.28 3.14 2.79 3.36 3.16 2.95 Cáceres 3.92 5.93 3.41 4.62 3.31 3.28 4.32 Cádiz 3.55 3.29 3.71 3.99 3.08 4.89 3.91 Castellón 4.72 5.50 9.99 5.33 5.40 6.42 5.95 Ciudad Real 3.62 5.50 6.89 4.11 5.02 5.02 5.30 Córdoba 2.13 3.08 3.52 4.92 3.69 3.15 2.79 A Corurña 3.56 3.96 3.24 3.70 4.27 4.39 4.60 Cuenca 2.99 4.81 4.11 4.26 5.48 5.56 4.84 Girona 2.87 4.33 3.77 4.24 4.23 4.70 5.30 Granada 2.62 3.07 3.48 4.04 3.81 5.94 4.53 Guadalajara 6.14 3.99 4.58 5.20 2.80 4.43 5.78 Guipúzcoa 2.12 1.94 1.65 2.00 2.52 2.68 2.39 Huelva 2.89 3.51 2.76 3.52 3.92 4.82 3.79 Huesca 2.69 3.88 4.31 2.90 2.97 3.27 3.93 Jaén 2.54 2.47 3.63 3.45 3.37 3.32 3.16 León 3.46 3.98 4.88 3.49 4.36 3.18 5.54 Lleida 4.52 4.31 5.01 4.13 4.47 4.50 5.30 La Rioja 2.75 2.32 2.93 3.99 3.95 3.15 3.43 Lugo 2.83 2.75 2.57 2.51 2.89 3.67 4.30 Madrid 3.83 4.66 5.22 5.23 4.89 5.74 5.53 Málaga 3.04 3.45 3.30 3.89 4.05 3.98 4.07 Murcia 5.34 4.88 4.53 4.83 5.32 5.39 4.78 Navarra 2.87 4.67 3.84 3.99 4.56 5.16 4.06 Ourense 3.92 2.91 3.16 3.43 4.04 4.47 4.86 Asturias 4.05 3.90 4.26 3.91 4.31 4.01 4.14 Palencia 2.88 3.27 2.58 4.40 4.58 3.08 4.13 Las Palmas 3.07 4.56 6.16 4.61 5.13 5.16 4.89 Pontevedra 2.72 3.25 3.19 3.46 3.86 5.23 4.11 Salamanca 2.16 3.42 2.55 2.90 2.32 3.35 3.04 Santa Cruz de Tenerife 2.91 3.03 4.65 4.51 5.55 5.17 4.99 Cantabria 2.85 2.89 3.44 3.45 4.05 3.84 3.15 Segovia 2.51 2.68 3.20 3.08 2.54 3.85 3.96 123 346 SERIEs (2012) 3:339–365 Table 2 continued Province 2001 2002 2003 2004 2005 2006 2007 Sevilla 2.83 3.25 3.81 3.58 3.33 4.23 5.17 Soria 4.42 2.62 3.84 1.90 2.32 3.43 2.96 Tarragona 4.22 4.62 4.64 4.69 3.81 4.88 4.75 Teruel 3.25 6.07 5.56 5.41 5.17 6.11 4.75 Toledo 4.38 3.98 4.48 4.77 4.40 5.27 3.88 Valencia 5.23 5.71 6.12 5.29 5.64 6.39 6.13 Valladolid 1.30 4.28 2.10 4.03 3.86 4.07 3.72 Vizcaya 1.76 1.80 2.69 1.91 2.83 2.64 2.21 Zamora 3.62 3.58 3.22 2.77 2.76 3.75 3.93 Zaragoza 2.98 4.70 4.84 3.52 4.18 5.05 5.20 Source: CGPJ (2010) and self elaboration to the courts in a specific year) were awaiting resolution while the courts were able to resolve just one. In the worst case, this amount was almost 10. As we may observe, therewas,onaverage,adifferenceof5.98congestionpointsbetween the most efficient and the least efficient province throughout the period. Figure 1represents this quotient for the years 2001 and 2007. A decrease in the efficacy of the system can be observed throughout the period. Looking at the graph, it is also clear that no specific provincial pattern seems to emerge in the reduction of the efficacy of the judicial system. However, the Basque Country has a better performance over the entire period. Note that Law 19/2009 (see Sect. 1) introduced several minor changes to the procedures of the CPL. These reforms can be summarized as follows: First, it generalized the use of a specific type of procedure (juicio verbal) to resolve all eviction-related conflicts. Secondly, it removed some of the options that the tenant had to hinder the declaratory judgment. Finally, it accelerated the execution of the declaratory judgments. These changes could accelerate the functioning of the judicial system, improving the figures discussed above. However, there is still no data available in order to analyze those effects. What was the evolution of the property rate during this period (2001–2007)? The proportion of property among the total number of primary dwellings in Spain (called “Prprop” in the tables) is in fact chosen as a dependent variable in this research. That proportion is the aggregate counterpart of the individual housing tenure decision. The data are obtained from the Spanish Ministry of Housing (2008) and are available for the period 2001–2007 for 50 Spanish provinces (excluding Ceuta and Melilla).14 This classification divides the primary residences into three groups: dwellings in the property market, dwellings in the tenancy market, and “transferred dwellings” (cessions or non-profitable use of the houses). On average, in 2007, 88.2% of the dwellings were 14 Note that the data is provided in November of each year and not in January. That fact is taken into account in the estimations. 123 SERIEs (2012) 3:339–365 353 On the other hand, the tenant cost function could take the following simple form that does not depend on the judicial efficiency: τ(u)=u2 where τ(u)is also a convex function. τ(u)>0 and τ(u)>0. With those two cost functions, the equilibrium condition of the Henderson– Ioannides model will take the following form: rP 1+r=R−u2[αJ−1] 1+r That is, u=R(1+r)−rP αJ−1 where, ∂u ∂J<0,∂u ∂R>0,∂u ∂r<0 and ∂u ∂P<0,if 1 −αJ>0. Thus, following that derivation, in equilibrium the rate of utilization will depend negatively on the judicial inefficacy. As previously stated, judicial inefficacy can be understood as a cost for the landlord. As a result, if judicial costs increase, less “space” will be put into the tenancy market (and thus, theoretically, we would observe a less developed tenancy market). Another theoretical argument about how agents behave when they confront a risk of non-payment of rent can be found in Casas-Arce and Saiz (2010). u=g(J − ,r −,R + ,P − ) Even though we consider Jand ras exogenous variables affecting the equilibrium, Rand P(together with the quantity of housing services in the market) are defined within the model. In an econometric implementation they should therefore be treated as endogenous and thus they must be instrumented. For instance, an exogenous shock increasing judicial inefficacy will affect the equilibrium price and the quantity of housing services through a shift in the supply side (or investment) of housing services but not through the demand curve as defined before. Thus, in the case of an econometric estimation of the supply curve, we will have to instrument the price (or the user cost) using for instance strictly “demand” instruments (that is, demand shifters which are not affecting the supply). 4 Empirical strategy 4.1 Model The objective of this research is to offer estimations of the effect of the inefficacy of the judicial system on the proportion of property in the economy. As discussed above, 123 354 SERIEs (2012) 3:339–365 judicial inefficacy can be understood as an extra cost that landlords face when they rent their properties in the market. Therefore, following the reasoning offered in Sect. 3,I propose to estimate a supply curve. Then, the following model is proposed (Eq. 2): Prpropi,t=c+ctTt+β1Usercosti,t+β2Excongestioni,t+β3Densityi,t +(ηi+νi,t)(2) β2is the effect of judicial inefficacy (“excongestion” in this setup), on the share of property, “prprop”. As previously discussed, we expect that the population density, “density”, is negatively related to the property share. The price should enter the equation with a positive sign as we are estimating a supply curve. A detailed description of the variables used in this study is provided in Table 4. Themeasure of judicial inefficacycould beaffected by the litigation in the province. In order to provide results that take this possibility into account, alternative estimates including a measure of “litigation” by province are provided in this paper. “Litigation” is approximated by the number of lawyers per capita,“Lawyerspc” (see for instance, Carmignani and Giacomelli 2010). In that case, the model takes the following form (Eq. 3): Prpropi,t=c+ctTt+β1Usercosti,t+β2Excongestioni,t +β3Lawyerspci,t+β4Densityi,t+(ηi+νi,t)(3) The “price” (taking the form of a user cost, “usercost”) will be an endogenous variable as we face a simultaneity problem. That is, the price and quantity are jointly determined by the demand and supply curves of the market. Thus, I will instrument the price using several demand shifters. I choose as instruments a set of variables directly affecting the demand side of the market: the proportion of young people in the province, ppob2039 and its lagged value, the proxy to credit constraint, credit and its lagged value, the lagged value of income per capita,ln GDPpc, and the proportion of social housing in the province, Shousing. Also the lagged user cost will be included as an instrument. As explained in Table 4, the measure “credit” takes account of the effects on mortgage lending implied by the inefficacy of the judicial system when solving mortgage conflicts. To choose the set of instruments and provide evidence of their validity, the Hansen J statistic (as over-identification test) is computed with satisfactory results in all cases.16 Note that, in general, the strategy of including the lagged dependent variables of Eq. 1(or 2)as instruments has been avoided (thus providing a more robust experiment). Following Sect. 2, judicial efficacy has been studied at both stages of the procedure (declaratory and execution) in the form of a congestion rate. Prtcongestion and excongestion enter the equation lagged several periods, up to four, taking into account that the decision to put a dwelling into the tenancy market may take into account the “judicial environment” observed some periods before. This fact would 16 Note that I did not assume homoskedasticity. Otherwise, the Sargan’s statistic would be reported. 123 SERIEs (2012) 3:339–365 355 Table 4 Variable descriptions Subject Variable in Sources Discussion the estimations Housing prices Pviv,ΔPViv Secretaría de Estado de Vivienda y Actuaciones Urbanas Price per squared meter of the average house in the province. PViv stands for the inter-annual increase in the housing price Rental prices Prent Secretaría de Estado de Vivienda y Actuaciones Urbanas and INE Rent paid per squared meter in the average dwelling offered for rent in the province. The secretaría de Estado de Vivienda Actuactiones Urbanas of Spain only provides the average rent for 2006 so the series have been enlarged following the evolution of the component of the consumer price index that captures the evolution of the rents. The resulting variable is defined for the period 2001–2007. Interest rate iBanco de España Interest rate on lending for house purchase. This variable has no provincial variation. House de preciation rate δINE (Census) House depreciation rate (2%). According to the Spanish Census of 2001, 2% of the buildings were in poor condition. I have chosen to use that percentage. Other sources point to higher rates: Naredo et al. (2005) propose a rate of house demolition of 0.397%, the American Housing Survey provides a rate of 0.295% and in the case of France the rate would be 0.25%. User Cost Usercost See above User cost =PViv(i+δ−PViv) PRent Population density density INE (Padrón) The population distribution in Spain differs greatly among the provinces. On the one hand the population in Spain is concentrated in the coastal provinces (Barcelona, Valencia, Málaga, etc.) On the other hand, some inland provinces have low population density and have not attracted many new immigrants (Soria, Teruel, etc). Proportion of “young” population ppob2039 INE (Padrón) Ratio of the population that is 20 to 39 years old. Other demographic variables (rate of nuptiality and the share of foreign population) using the same source (INE, Padrón) are not included in the final estimations as they prove to be insignificant. Wealth Ln GDPpc INE (Regional accounts), Alcaide Inchausti et al. (2004) and Alcaide Inchausti and Alcaide Guindo (2008) “Ln GDPpc” represents the logarithm of the current GDP per capita once adjusted by provincial PPPs. The information on provincial PPPs is obtained from Alcaide Inchausti et al. (2004)andAlcaide Inchausti and Alcaide Guindo (2008). Another typical macro variable (rate of temporary employment) proved to be insignificant in this study. 123 356 SERIEs (2012) 3:339–365 Table 4 continued Subject Variable in Sources Discussion the estimations Litigation Lawyerspc Consejo General de la Abogacía (2009) “Lawyerspc” is calculated as the number of lawyers registered with the Bar associations multiplied by 1000 and divided by the population of the province. Coastal provinces Coast Dummy variable which takes value 1 for Mediterranean and Andalusian coastal provinces plus the Balearic islands and the Canary islands. Thus, “Coast” takes value 1 for the following provines: Girona, Barcelona, Tarragona, Castellón, Valencia, Alicante, Murcia, Almería, Granada, Málaga, Sevilla, Cádiz, Balearic Islands and Canary Islands (provinces of Santa Cruz de Tenerife and Las Palmas). Credit constraint Credit INE, Banco de España, CGPJ and self elaboration “Credit” captures unexpected easiness of credit after controlling for the most typical and expected factors in the granting of mortgages. Therefore it would be taken as a proxy for the inverse of credit constraint. It is calculated as the residual (μi,t)of the following estimation. Number of Mortgages i,t=c+λ1GDPpc i,t+λ2 Ppob2039i,t+λ3Mortcongestion i,t+λ4Coasti,t+μi,t.. The residual of the regression will assign a positive sign to the provinces and years in which the number of mortgages granted to families is still positive (on average) after controlling for wealth, population, “coast” (see above) and after taking into account the effects on mortage lending implied by the inefficacy of the judicial system when resolving mortgage conflicts. “Mortcongestion” is calculated according to the structure of Eq. 1but computing only cases related to the mortgage market. It should be noted that the measure “Mortcongestion” is replaced by “Mortresolution”, “Mortpendency” or “Mortppnc” when the model estimated includes a judicial efficacy measure of that form. “Coast” is included as those provinces are a typical destination for tourism and foreign real-estate investments. This may influence the number of mortgages observed in the statistics. This may influence the number of mortgages observed in the statistics. The dependent variable of the regression is chosen to be the number of mortgages and not the quantity of those mortgages. Taking the quantity of the mortgages would lead to a bias in the estimations in favour of provinces such as Madrid, Barcelona, Valencia, San Sebastián etc. in which the housing prices are much higher than in the rest of Spain. Another variable measuring a related factor (banking competition) was tested with no significant results. 123 SERIEs (2012) 3:339–365 357 Table 4 continued Subject Variable in Sources Discussion the estimations Regional tax deductions (Fixed effects) (Ppob 2039) See above Provinces have no power to adopt specific tax deductions for renting or buying a dwelling, although the regions (Comunidades Autonómas)dohavethatpower. Several regional deducations have been applied to home ownership, to tenancy or to both in the period after 2002. The Basque Country and Navarre have a special (foral) tax system. None of the deductions that can be found are “general deductions” because they apply to very specific population groups (to young citizens in the most part). This circumstance would be already covered by the variable “ppob2039”. Regional differences, especially those of the Basque Country and Navarre, can be considered covered by the inclusion of Fixed Effects (FE) in the estimations. Social housing Shousing Secretaría de Estado de Vivienda y Actuaciones Urbanas Proportion of social housing (houses sold or rented at prices below market price by the public authorities) over the total number of houses in the specific year and province. 123 358 SERIEs (2012) 3:339–365 also mitigate any problems of endogeneity of the judicial variables. In any case, there are no reasons to suspect the endogeneity of the judicial variables in this research. The courts considered in this study (“juzgados de primera instancia” and “juzgados de primera instancia e instruccion”) are not specialized courts and resolve a wide range of conflicts, from inheritance conflicts to bankruptcy proceedings. Thus the distribution of tenancy conflicts (generated in part by the amount of tenancy and property contracts in the province) is not necessarily influencing the distribution of “juzgados de primera instancia” and “juzgados de primera instancia e instrucción”. Judges in Spain are also obliged to process and resolve cases in chronological order of entry, and therefore cannot give preference to a specific type of conflict. The model is estimated following a two-step (instrumental variables) generalized method of moments (GMM estimation) (Woorldridge 2001;Arellano 2002;Baum et al. 2003).17 Fixed effects (FE) are included in all the estimations. Standard errors are clustered in order to make them robust to both heteroskedasticity and serial correlation. Time dummies are also included to take the cycle into account. Wald tests of significance for those time dummies are reported in the tables. Table 5reports the results when estimating both Eq. 2(models 1 and 2 in the Table) and Eq. 3(models 3 and 4 in the Table). 4.2 Results As an initial result, it is important to note that the efficacy of the declaratory stage has no significant impact on the share of property.18 Therefore, this paper focuses its analysis on the final or definitive step (execution). Nevertheless, this is an interesting result in and of itself, as will be discussed in the conclusions. First of all, it is worth noting that the user cost (Table 5) enters the equation with a positive sign. The sign confirms that we have estimated a supply curve once we take into account that the over-identification tests were passed satisfactorily. The variable density has the expected (negative) sign in all cases and is significant at 5 or 1% level. Finally, looking at the results for the judicial variables, we find the expected effects. First of all, it is found that a higher congestion rate, that is, a lower efficacy of the judicial system, attracts more houses to the property market. That is to say that a “problematic” tenancy market prevents the owners/landlords from placing their dwellings on the tenancy market. Table 5shows that an increase in one point in the congestion rate would increase the share of property by around 0.14–0.16 percentage points. Thus, taking the example of Madrid, the decrease in the congestion rate would attract around 3,400 dwellings to the rental market. In Barcelona, the increase would be around 3,100 dwellings and 17 Under the presence of heteroskedasticity, GMM estimators are more efficient than IV-robust ones. 18 The estimations are available on demand. 123 SERIEs (2012) 3:339–365 359 Table 5 Effects of the judicial congestion rate Model 1 2 3 4 Method of estimation 2-Step GMM 2-Step GMM 2-Step GMM 2-Step GMM Data transformation FE FE FE FE Excongestion (t-3) 0.14 0.13 0.05*** 0.043*** Excongestion (t-4) 0.16 0.16 0.07** 0.07** User cost 0.07 0.04 0.07 0.04 0.03** 0.03 0.03** 0.03 Density −0.07 −0.05 −0.07 −0.05 0.02*** 0.02** 0.02*** 0.02** Lawyerspc (t-3) 1.00 1.65 Lawyerspc (t-4) 0.93 1.83 Time effects Yes Yes Yes Yes Observations 250 200 250 200 Groups/Clusters 50 50 50 50 Hansen J statistic (P-value) 0.85 0.89 0.80 0.87 Wald Test for time dummies 0 0 0 0 Dependent variable: Share of property (Clustered) Standard errors robust to heteroskedasticity and serial correlation beneath coefficients lnstrumented: User cost Instruments: User cost (t-1), Ppop2039, Ppop2039 (t-1), Credit, Credit (t-1), In GDPpc (t-1). Shousing *** p < 1% **p<5% *p<10% in Valencia around 1,400 dwellings.19 Those results are significant at around 5 or 1% respectively. The variable “Lawyerspc” is insignificant and does not affect the results. Appendix A provides a discussion of the results of the model when alternative measures of judicial efficiency are computed (“pending cases rate”, “resolution rate” and the “pending-plus-new cases rate”). The results are consistent with those presented here. 5 Conclusions This research presents some estimations of the effect of the efficacy of the judicial system on the proportion of property in the Spanish provinces. The problem is analyzed 19 An interesting experiment is to calculate how many dwellings, according to the model, would be won by the tenancy market of Madrid if the rate of congestion in Madrid improves to the level of the average congestion rate of the Basque Country in 2007. In that case, the increase in the number of dwellings for rent in Madrid would be between 9,600 and 12,100. The same experiment in the case of Barcelona would lead to an increase in rental housing of between 7,300 and 9,100 dwellings. 123 360 SERIEs (2012) 3:339–365 econometrically through panel data techniques. Specifically, the generalized method of moments (2-step GMM) is used in the estimations as several instrumental variables are taken into account. This study is the first one in the economic literature to tackle the case of Spain at the local level. The judicial efficacy is measured through the construction of a “congestion” indicator at two stages of the procedure: the declaratory stage and its final executory stage. This research does not find any significant impact of judicial efficacy at the declaratory stage on the housing property share. However, this research concludes that an increase in judicial efficacy at the execution stage would have a positive, although minor, impact on the share of property in the Spanish provinces. The effect amounts to around 0.15 percentage points of the housing market (higher effects are found if other efficacy measures are taken into account) (see Appendix A). That effect would denote that homeowners avoid the tenancy market when they cannot enforce their contracts. The discussions presented in this research give some grounds to improve the efficacy of the judicial system, at least at the execution stage, in order to develop the Spanish tenancy market. In this sense, future research based on this paper could evaluate the effects of the recent reforms of the tenancy market in Spain, introduced by Law 19/2009, when the relevant data is available. Appendix A: Estimations with alternative judicial efficacy measures Judicial efficacy can be measured in different ways. This paper has opted to study the “congestion rate”, even though other efficacy measures could be computed. This Appendix presents the results of the study if three alternative efficacy measures are taken into account: the “pending cases rate”, the “resolution rate” and the “pending-plus-new cases rate”. The pending cases rate is defined as the ratio between pending cases in a specific year and the cases resolved in the same period. The resolution rate is defined as the ratio between the cases resolved and the cases that entered the system for a specific year. Finally, the “pending-plus-new cases rate” (ppnc)is calculated as the sum of the new cases and the pending cases (measured at the beginning of the year), divided by the population of the province in that year (see equations grouped as 4). Pendingcasesratei,t=Pendingcasesi,t Casesresolvedi,t Re solutionratei,t=Casesresolvedi,t Newcasesi,t(4) Pending −plus −newcases ratei,t=Pendingcasesi,t+Newcasesi,t Populationi,t As a measure, the “pending-plus-new cases rate” tries to take into account that the workload of the courts may vary due to the different populations of the provinces. 123 SERIEs (2012) 3:339–365 361 Table 6 Effects of the judicial pendency rate Model 1 2 3 4 Method of estimation 2-Step GMM 2-Step GMM 2-Step GMM 2-Step GMM Data transformation FE FE FE FE Expendency (t-3) 0.18 0.16 0.09* 0.08** Expendency (t-4) 0.27 0.27 0.13** 0.14** User cost 0.08 0.06 0.07 0.06 0.03** 0.03* 0.03** 0.03* Density −0.07 −0.05 −0.07 −0.05 0.02*** 0.02** 0.02*** 0.02** Lawyerspc (t-3) 0.87 1.73 Lawyerspc (t-4) 0.51 1.95 Time effects Yes Yes Yes Yes Observations 250 200 250 200 Groups/Clusters 50 50 50 50 Hansen J statistic (P-value) 0.73 0.81 0.70 0.82 Wald Test for time dummies 0 0 0 0 Dependent variable: Share of property (Clustered) Standard errors robust to heteroskedasticity and serial correlation beneath coefficients Instrumented: User cost Instruments: User cost (t-1), Ppop2039, Ppop2039 (t-1), Credit, Credit (t-1), In GDPpc (t-1), Shousing *** p < 1% **p<5% *p<10% A higher “resolution rate” and a lower “pending cases rate” are related to greater efficacy of the judicial system. A higher “pending-plus-new cases rate” is again a proxy of congestion of the judicial system (although related to the population of the province) and therefore it can be expected that a higher workload per capita is related to a lower efficacy. Some summary statistics are included in Table 1. With respect to the first measure of efficacy related to executions, expendency,we can observe the following: On average (see Table 1), almost three times more cases were pending (waiting to be resolved) in the execution stage with respect to the cases that the courts were able to solve. Although some provinces had, on average, very good results (pendency rate of 0.46), other provinces had more than seven times more cases waiting to be resolved than the average workload they were able to resolve in a year. With respect to the resolution rate, we can say that, on average, the judicial system was able to resolve nearly the same amount of cases as were entering the courts (resolution rate of 0.87) at the execution stage. This does not imply a constant workload because some conflicts may be waiting on the list at the beginning of the year (this aspect is better analyzed with more complete measures of efficacy as the pendency cases rate and the congestion rate). Even though some provinces underperformed quite radically (minimum of 0.42), others were able to resolve two times more cases than the 123 362 SERIEs (2012) 3:339–365 Table 7 Effects of the judicial resolution rate Model 1 2 3 4 Method of estimation 2-Step GMM 2-Step GMM 2-Step GMM 2-Step GMM Data transformation FE FE FE FE Exresolution (t-3) −0.03 −0.01 0.31 0.29 Exresolution (t-4) −0.82 −0.88 0.32** 0.34** User cost 0.07 0.07 0.06 0.07 0.03** 0.02*** 0.03** 0.02*** Density −0.07 −0.05 −0.07 −0.06 0.02*** 0.02** 0.02*** 0.02** Lawyerspc (t-3) 1.42 1.88 Lawyerspc (t-4) 1.35 1.86 Time effects Yes Yes Yes Yes Observations 250 200 250 200 Groups/Clusters 50 50 50 50 Hansen J statistic (P-value) 0.79 0.87 0.67 0.91 Wald Test for time dummies 0 0 0 0 Dependent variable: Share of property (Clustered) Standard errors robust to heteroskedasticity and serial correlation beneath coefficients Instrumented: User cost Instruments: User cost (t-1), Ppop2039, Ppop2039 (t-1), Credit, Credit (t-1), In GDPpc (t-1), Shousing *** p < 1% **p<5% *p<10% number of new cases entering the system, and thus were able to reduce the workload for future periods. Finally, with respect to the “pending-plus-new cases rate” we can say that, on average, 17.38 cases (summing the pending and the new cases) per 1,000 inhabitants were waiting to be solved. As in other cases, some strong differences by province can be found, from around 1 case per 1,000 inhabitants to 36. As occurred with the case of congestion, no significant results are found when the models are computed taking into account the efficacy at the “declaratory” stage. Table 6shows the results of the estimations when we consider the pendency cases rate instead of the congestion rate as a measure of efficacy. The results are consistent with the previous ones. An increase of the pendency rate of one point would increase the property share of the province by around 0.16–0.27 percentage points (around 3,700–6200 houses in Madrid, around 3,300–5,600 in Barcelona and around 1,500–2,500 in Valencia). The results are significant at 5% level. Then, an increase in one point of the resolution rate (see Table 7) implies a reduction in the property rate of around 0.82–0.88 percentage points (that would be approximately 18,800–20,100 houses passing from the property market to the tenancy market and related options in Madrid, 17,000–18,300 in the case of Barcelona, and 1,500–2,500 in the case of Valencia). 123